Version 1.9, October 2011 Copyright 1999-2011 R.G. Keen. All rights reserved. This article served from GEO - http://www.geofex.com

The Thomas Vox solid state amplifiers are notorious for their poor reliability. I love the way these amplifiers look (and sometimes the way they sound) and in fixing a couple that I have, I came up with several things that should make them vastly more reliable, and possibly sound better. There are also a lot of tips here on how to get a dead one running again.

I own a solid state Berkeley II, a Berkeley III, a Buckingham, a Beatle, a Royal Guardsman, a Viscount, a Cambridge, a Pathfinder, a Pacemaker and a Nova. In doing this, I found that all Thomas Vox amps all share some similarities in their designs, particularly in power supply and output power amp sections that can be improved a great deal. In this article, I'll mention some things that are trivial updates, and some that might be seen as defiling the temple. Use your own judgement, and do as you best see fit with your amp.

For those of you who are all ready to go, there is a table of reliability upgrades at the end of the article.

Understanding the Thomas Vox Amps

The Thomas Vox (which I usually abbreviate "TV") amps are all clearly of a single design philosophy and share certain elements inside. Their kinship is much, much more than similar styling on the external covers. 

Power Amp Designs

All of the TV solid state amps from smallest to largest use a driver transformer style power amp, which places the date of the designs very closely. This style of power amplifier design was only current from the time power transistors worthy of the term appeared in the mid to late 50's and died out rapidly as suitable complementary devices became available in the early 60's. In fact, I can find almost no reference to this circuit in most of the design literature I can find from 1963 on. 

The driver transformer is a legacy of the vacuum tube. "Stacked" power tube circuits were devised to avoid the output transformer, which was widely viewed as a serious impediment to getting good sound. These stacked tube designs used driver transformers because a practical equivalent to a "PNP" tube is forbidden by the laws of physics - there's only one polarity of charge carriers that are easily mobile in a vacuum. To have one tube pull up and the other down, a driver transformer is a clever solution that simplifies thing a lot. The first power transistors were only one polarity as a practical matter, so the step from designing with single polarity tube to a single polarity transistor was a natural one. In any case, the driver transformer is there to do two things; isolate the output transistors from the DC conditions in the driver transistor circuit, and invert the signal so that one output transistor is driven on while the other is driven off. It's a phase inverter and complementary driver.

It's important to note that the use of a driver transformer makes a complete break between the output transistors and driver transistors in the DC sense. Transformers simply cannot transform DC between windings, so all biasing concerns for the output transistors are carried out solely through the resistances attached to the transistors on the output side of the driver transformer. All the output transistor biasing that is done in the TV amps is done purely by the bias resistor strings on the secondaries of the driver transformers. You are in fact free to get the bias set up properly even if the driver circuit is not attached to the primary of the driver transformer at all. This will be a great help when we start servicing them.

The output sections of all the TV amps is the stacked "totem pole" style output. Both transistors are the same polarity, PNP for early ones, and NPN for later ones. Each transistor has an emitter resistor to provide some stabilizing local feedback, and each (top half or bottom half) has a biasing string which stretches from the transistor's collector to the outer end of the emitter resistor. The junction of the two resistors sets the bias voltage to which the base of the transistor is held.  For germanium devices this is about 80 millivolts, and for silicon this is about 450 to 550 millivolts, which corresponds to the barest beginning of conduction for the respective materials. 

The current in the resistance string is set to be some 30 to 60 times the expected idling base current so that variations in the base current needed for idling will not change the bias string current much. Since most bipolar transistors in a Class AB setup are just about perfectly biased at a collector current of 25 to 35 ma, and typical output transistor gains for the time were 20 to 75, the base current could be half to a bit more than one ma, so the bias string conducted typically 30 to 60 ma per output transistor( there were two paralleled transistors in the case of the 120W Beatle). With the base-emitter voltage so low compared to the collector base voltage, effectively the high resistance from the collector to the middle of the bias string set the bias string current, and the resistor from the base to the outboard side of the emitter resistance set the base bias voltage. There is very little interaction here, because the voltages across the two resistors are so different. The base-collector resistor varies from 220 ohms in the Beatle to 1K or so in the smaller amps. The base-emitter resistor bias resistor is only a few ohms, from maybe 2.2 up to as much as 13-15 ohms depending on the output transistor material. This is an important point for upgrading the power transistors, as it allows us to substitute in more commonly available silicon power transistors  and rebias from germanium's 80mv to silicon's 500mv by changing only one resistor per output transistor. 

The bias current is conducted from the junction of the two biasing resistors to the base of the transistor through the DC resistance of the driver transformer secondary. The winding resistance is a few ohms at most, so it does not change the bias conditions much at all. Since the two secondaries are out of phase, the net DC in the transformer from output transistor bias is zero. This avoids having a net DC offset on the transformer core caused by the DC conditions on the secondaries.

Driver Circuit

The driver circuit for all the TV amps is itself a small Class A transformer coupled amplifier. In the case of the bigger output amps like the Royal Guardsman, Westminister, and Beatle, the driver "power amp" provides maybe 4-6W to drive the output section. I finally found enough references to the stacked output configuration to make an educated guess about the impedance the driver section sees in driving the bases of the output devices. For the "big head" amps above, it turns out that the equivalent loading on the driver transformer secondary is around 8 ohms. You could actually disconnect the output power transistors from the driver transformer outputs and hook up an 8 ohm speaker for a few watts of output. 

The driver transistor's collector is tied to the driver transformer primary, pulling an idling current through the primary, with the usual problems of DC offsets in the driver transformer. The driver transformer is a gapped design to support the DC offset, as in all single ended Class A designs. The driver circuit being a single ended class A amplifier has some direct results on the output stage on the other side of the driver transformer.

The idling driver current sets the limit of drive for the output devices - in the direction of turning off the primary current, the output device that does that polarity simply can't be driven any harder than the idling current in the primary times the turns ratio. This and DC saturation issues is the reason for setting the idle current of the driver transistor in the larger amplifiers. Too little current and you can't drive one of the output transistors hard enough. Too much and the driver transformer saturates on large signal swings.

The driver circuit itself is of fairly conventional design, and is one transistor in the smaller-output amps, two transistors in the bigger amps.

 I've done some reverse engineering on a Beatle driver transformer. The Beatle driver transformer is wound on a square stack of EI-75 scrapless laminations. The laminations are paralleled rather than interleaved and have a spacer at the joint where the E's meet the I's because of the DC current in the primary side. Only the two outside laminations are reversed to the others to cover the gap. It might be  is possible to replicate a functional version of this from a small power transformer. 

The primary of the Beatle driver transformer is about 325 turns, the secondaries about 130 turns each. I wound ten turns of fine magnet wire into the crevices around the coil of the driver transformer of my Beatle and pulled out the driver transistor and the output transistors. That way the primary and secondaries were opened and I could drive the transformer and measure both the existing turns ratio and the voltage across my added ten turns. This let me calculate the turns ratio and the volts per turn, so I now know that the primary is about 325 turns and the secondaries are about 2/5 of that, about 130 turns each. The currents in the driver collector and into the bases of the outputs let me make a fairly good guess about wire size, so I'll eventually be able to write down a recipe for one of these things.

There is an Achilles heel of this style of power amp design. It's relatively intolerant of shorted output connections and of massive overdrive signals. Output short will usually cause one or both of the output transistors to die, and massive overdrives will eventually do the same thing. This critical weakness, coupled with the fact that the transistors are just barely capable of handling the power they handle in the big amps is what usually gives the TV amps their bad rep.

Preamp Design

The preamp circuitry varied a lot more than the power amps did. However, there are some common facets. The overall style for the preamp and other signal processing circuitry is the prevailing "two transistors and a coupling cap" style from the early 60's. This was a time when DC characteristics of transistors were not nearly as reliable as they are today, and integrated circuits were unknown. It's a matter of some interest that it was during the time when the TV amps were current, Texas Instruments was awarded a patent on the very idea that more than one transistor *could* be put into a single package to make an "integrated circuit". 

This is an advantage to us in servicing TV amps, because this style of circuitry chops the whole complicated mess up into bite-sized one and two transistor circuits that can be successfully serviced in relative isolation from the others. In almost all cases, the DC blocking caps are on the PCB, so that the off-board controls are not necessary at all for functioning of the on-PCB circuit - the controls can be unhooked or ignored and you can get the DC conditions correct before worrying about that complicated wiring. 

Design and Reliability Problems

The power amps are all a "stacked output" type power stage, where a pair (or two pairs, in the Beatle) of same-polarity (NPN or PNP) output devices are driven by an interstage driver transformer. The transformer provides both phase splitting and some current step-up to allow a single driver transistor to drive the output transistors. The driver transistor is run from an R-C decoupled +25 to +27Vdc supply, and is run at a high enough current level to allow it to drive the output devices as far as they can be driven. On the "big head" amps, power level is high enough that the driver transistor is connected to some kind of a secondary heat sink itself. In the "small head" amps like the Berkeley 2 and 3, the Cambridge, etc., the driver transistor is a TO-5 or TO-39 transistor and has a round "top hat" press-on heat sink. The "big head" amps like the Buckingham (and its combo twin, the Viscount), Royal Guardsman, and Beatle use TO-66 power transistors. These look like half-scale output transistors. 

From an engineer's point of view, the Thomas Vox power amps are of questionable reliability, given the state of today's knowledge. They may have been - and probably were - state of the art for transistor amp design when they were made, but the state of the art has improved a lot since then. Remember that power transistors of any kind were rare and costly in the early 60's when these things were on the drawing board. Essentially every one of the things a modern amplifier would have to prevent catastrophic failure is missing from these amps. Here are some of the problems:

• The heat sinks on the transistors of the high power amps are just barely sufficient to the thermal loading -- which is good engineering-to-cost practice, but not very good for reliability over the long haul. One might think that just the higher power Beatles are affected, but the other "big head" amps seem to have thermal problems too, especially the Guardsman and Westminister (both versions). The Buckingham and Viscount seem to have escaped this problem, though, as they're lower power amps. The thermal stress per device is smaller in these. 
• Airflow over the heat sinks is limited by the outer cabinet's small air vents. These things really could profitably use a fan; thanks to computer CPU's needing fans these days, you can put one in!
• There is no provision for thermal compensation of the output transistor bias. The devices are fixed biased by resistors, and so thermal runaway is always a possibility. From my reading, biasing with a low-impedance resistive source like these power amps may be more stable than one would expect, but it's an iffy replacement for modern thermal compensation techniques.
• There is no thermal protection for overheated sinks, power transistors, or anything else for that matter.
• There is no provision for limiting transistor current on peaks, with the exception of the "peak limiter" circuit in the preamp feeding the power amp.
• No speaker fuse or other protection was used.
• Any "current limiting" on the output transistors is inherent - that is, whatever condition exists; for example, the driver circuit is current limited by the bias current in that stage and the inter-stage driver transformer. The driver simply can't provide more than its bias current through the driver transformer. As such, the output transistors are "limited" to whatever current they can provide with their current gains times the fixed driver current from the transformer. While low gain devices will probably survive output faults like a short circuit, a high gain device will probably self-destruct. It is possible that the designers thought that the preamp peak limiter would give adequate protection.
• No safe-operating-area is provided to keep the power transistors out of the "second breakdown" region where they will fail catastrophically at higher voltages and currents that are within their power rating.
• The only explicit overcurrent protection provided in the whole amp is a single primary AC fuse in the power wiring. This is almost certainly too slow to keep the output transistors from killing themselves.
• A subtle flaw that I've found twice now: the screws that hold the output transistors into the transistor sockets on the heat sink are prone to coming loose. This can happen just from the repeated heating and cooling cycles that they will certainly undergo. If that happens, the transistors will see a high-thermal-resistance path from their case to the heatsink and they will overheat and die rapidly. There should be star washers under the heads of all these screws to keep them tight over lots of thermal cycles. Even better, the sockets and screws should be replaced with more reliable screw fittings.

If these amps were produced today, there would be a number of things done differently:

• power devices would be placed on a more generous heat sink with fins closer to the transistor
• heat sink(s) would be located where ambient air could flow over it more easily or some way to move air across them would be provided
• transistors would be fed a thermally-compensated bias voltage for their Class AB setting (still working this one)
• active circuitry limiting the peak current (and possibly volt-amp operating area) for the output devices to prevent short circuits and odd loads from letting the output transistor go over their current rating or into second breakdown.
• there would likely be a speaker fuse
• a well-protected amp might also have thermal cut-off devices mounted on the output transistor heat sink and shut down the whole amplifier if the sinks got too hot.

First, either the heat sinks need to be improved or the output stages need more output transistors to share the load. Gotta keep the output transistors cool! It would be best if the heat sinks could be better-finned and out in free air. A possible fix-up might be to put an internal fan where it would cause airflow over the sinks.  Thanks to the tiny fans used on computer CPU heat sinks, we can do this now. The thin (0.375") fans used for CPUs are usually 12Vdc. While they make a horrible whine when you run them at 12V, at 6V they're very quiet. They're small and light enough to fit onto the heat sinks. I'll include a picture of how I did this.

[6/12/00] I pulled out a junker Vox Beatle chassis and did some thermal measurements on one of its heat sinks. I clamped a power resistor to the sink, put 10W of DC into the resistor and measured the temperature rise. The narrow (3" wide) sinks in the Guardsman, Beatle, and high power Westminister measure 2.6 C/W - that is, they get 2.6C hotter for every watt you put into them. 

[6/18/00] I then glommed on a 1 9/16" square by 3/8" thick CPU fan, held directly over the power resistor by standoffs. This size fan is thin enough to do this and still be inside the outer envelope of the heat sink. I ran the fan at 6V instead of its rated 12V to keep it quiet, which was very successful. I could hardly hear it even with the chassis outside the enclosure. This blew enough air to get the equivalent thermal resistance of the heat sink down by just about 1C/W to 1.6 C/W. 

The worst thing you can do to an amp with an input signal is to force it to put out a square wave with a peak voltage equal to half the power supply. For all the "big head" amps, the power supplies are +/- 31V, so if you put out a 15Vpk square wave, the transistors really cook. While actual use on stage will not generally see a signal like this for long, they might see just a bit less for a whole song if you have a distortion like the MXR Distortion+ going. This is truly the input signal that will make the outputs hottest.

For the Beatle, this forces the transistors to dissipate about 28W each; the calculated junction temperature for a transistor under these circumstances in a 35C/95F ambient is 128C. This is perilously close to the maximum the devices can stand. The Guardsman and Westminister will have almost the same heating per transistor. 

With the added fan, the transistors will run a full 28C cooler (28W times 1C/W) This is enough to get the junction temperature down to 100 C even on a hot stage. Looked at another way, it's well known that reducing the temperature of a semiconductor by 10C approximately doubles its expected life. This change with the fans is enough to increase the expected reliability by almost 8 times if your amp is regularly cranked to warp 9 or fed a distortion signal.

If a power transistor's chip temperature gets too high, it conducts more current, which makes it hotter, and they go into thermal runaway. You can reduce chip temperature two ways - better heat sinks or spreading the heat out over more power devices so that each one dissipates less heat. An example of this last technique is the Beatle output stage versus the smaller power heads - the Beatle doubled up the number of transistors so that each one had to dissipate only the same amount of heat that the transistors in the lower powered heads do. That enabled them to get the Beatle to put out twice the power to the speakers and run without (always) dying with no bigger heat sinks than the lower-power heads. Put another way, the power transistors in the Beatle are dissipating the same amount of heat as the (fewer) transistors in the lower power Guardsman and 60W Westminister heads do. 

From today's vantage point, what was once scarce and costly is now cheap. A very reasonable thing to do is to simply add more parallel output transistors to the ones that are there. This will mean some new holes drilled into the heat sink, but it should make the thermal reliability much, much better. It also adds more safe operating area / second breakdown capability. If, for instance, you doubled the transistors in the Beatle, each transistor would now be dissipating only 14W under the worst-worst case input conditions. This cuts its temperature rise from about 93C to about 46C, and its peak junction temperature from 127C to 79C - this is much safer, and the transistors can be expected to last much, much longer.

Notice that doubling transistors cuts the expected maximum temperature even more than a fan on the heat sink, and  makes no noise, although the slow fan speed is very quiet.

You can add heat dissipation capability to the existing heat sinks fairly easily. This will make the best use of the air that does manage to get into the cabinet. The existing sinks are thick aluminum sheet, and do make pretty good heat spreaders, but are limited in the amount of fin area close to the power devices. As a result, the heat has to travel through a long stretch of aluminum plate before it can get into the air, and the transistor gets hot. If you can either add air flow near the output device, or  put heat-dissipating fins closer to the power transistor, the device will stay cooler.

Fortunately, this is pretty fixable. There is lots of flat surface area there. With a lot of surface area, the temperature drop across a joint can be very low. We can do one of four things - (1) Bolt more heat sink pieces with fins immediately above and below the existing transistor mounting (2) bolt a flat-backed finned extrusion directly under the power transistor for some highly local relief or (3) replace the bent-aluminum heat sinks with a largish finned extrusion that fits the available space (4) add in a small fan to blow air across the existing sink.

You can find lots of surplus flat-backed heat sinks that can be attached to the flat areas of the Vox heat sinks. Clean the Vox heat sink, drill and tap holes to screw the finned sink to the existing plate, and use a *very thin* layer of thermal joint compound between the added finned sink and the Vox flat plate sink. With a hacksaw, you can cut sections of finned sink that will fit the flat parts of the Vox sink without interfering with the transistor mounting. The basic heat sink can now conduct heat over a large area into the finned section of the added heat sink, which is more efficient at getting heat into the air. It is important to get the best contact possible between the bolt-on heat sink and the existing aluminum plate. Both surfaces should be clean, possibly sanded smooth, and covered with the absolute minimum of heat sink compound that will ensure no air pockets between the two. 

The Royal Guardsman and Buckingham amps have a single transistor pair and the heat sinks are about six inches wide. You can use a flat-backed heat sink that is already drilled for a TO-3 transistor and mount the sink so its holes line up with the holes in the existing heat sink. The transistor leads are long enough to make this work. This way the transistor gets tied directly to a more efficient heat sink, and still gets some benefit from the existing aluminum sheet heat sink. Note that there is probably room to do the same thing from the back side as well - you can keep them really cool that way. The Buckingham amps probably don't need any special treatment, because of the low output power.

Unfortunately, the four smaller heat sinks in the Beatle head are too narrow to use the 4.5" wide normal heat sink extrusions.

I found a batch of microprocessor heat sinks that were about 2.25" square by 1" tall. These things bolted right up to the heat sinks on my Beatle, one above and one below each output transistor; they cut the temperature rise on the transistors just about in half. There are thermally conductive epoxies that could be used to attach such sinks as well if you were certain you would never want to take them off. Otherwise, they can be drilled and tapped to tie down to the aluminum plate sink. You could also bolt some on the back side, overhanging the circuit board. Since the heat sinks on these amps were simply bent aluminum sheet metal, they are not very flat. On my amp, the sinks were slightly bowed to the power transistor side, which made the sinks on the chassis/driver side only touch the 'flat' plate at the ends of the sink. This is a problem. If you go this route, count on messing with the bent plates to get them flat enough for good contact to the added sinks. If the added heat sinks are not firmly and flatly in contact with the existing sinks, you're wasting your time adding more sinks. The air space between the two parts will act as an efficient thermal insulator.

You can also take larger finned heat sinks and use a hacksaw to cut them to fit between the arms of the U's on the Beatle heat sinks and bolt them in.

An alternate approach is to find one or two large finned heat sink extrusions and replace the existing sinks with the extrusion. This will be a much more efficient heat dissipator. The only real limitation is that the fin height is limited by the positioning of the heat sink in the cabinet to about 1.25" to 1.5". If you do this, the actual transistors can be mounted to the back/flat side of the heat sink by bolting them to a thick aluminum L bracket, interfaced to the extrusion with insulators and heat sink grease, naturally. If you go for one of the transistor replacements I mention later, the replacements are TO-3P packages, and bolt to a flat heat sink without through-holes for the leads, so the L-bracket will be unneeded.

Next, add some thermal protection. There are often thermal cutouts available for $1-$2 each on the surplus market. These are switches that open when they get too hot. You can mount one of these on each of the heat sinks that are there for very little money. These cutoff switches can be used to disable the power to the amplifier if the heat sinks get too hot.

The germanium devices in the early solid state gear should not run hotter than about 85C on the junctions so they're really scorching under maximum warp drive. Allowing for some temperature drop from the transistor junctions to the sink, about 50C - 55C on the sinks should be close to a maximum safe temperature for Ge devices. Scan the ads in surplus suppliers for thermal switches - they're usually under $3.00 each. Mount the cutouts as close as you reasonably can to the power transistors on the back (non-finned side) of each heat sink, one per transistor. Daisy-chain the cutouts so that if any one opens, it breaks a chain. Connect the two ends of the daisy chain to your favorite protection circuit - more about that later.

If you have silicon devices (all silicon types are rated at 150C or more junction temperature) you can get thermal cutouts that are rated for about 65C-70C and get the same protection. Why so much lower than the rated junction temperature? It turns out that although you can run power device junctions at the specified max temperature, you can't do it repeatedly without lowering the life of the part. The part will keep operating, but the heat will accelerate the secondary failure mechanisms that will shorten the device's life. Transistor life probably doubles for every 10C cooler they are kept.

As to what to do with that protection... one thing you could do is to use it to disable the AC power to the amp. If you connect a triac's gate to its MT2 through about a 1K resistor, it keeps the triac fired all the time until the gate line opens, and then the triac does not fire on the next zero crossing. Using the thermal switches to do the connection of the triac gate to its MT2 through a resistor would provide this protection nicely. Another one that I like and am designing circuits for is a pair of power MOSFETs that will switch the DC power supply to the power stage off as needed. An example is shown below.

The availability of IRF's photovoltaic MOSFET drivers allows one LED to turn on and off two high voltage/high current MOSFETs. Although the switching times are glacial by switching power supply standards, we only expect these switches to work once or twice a night, so slow switching is not fatal.

The nice thing about this approach is that protection circuits don't have to be elaborate, as all they have to do is turn on or off one LED. This is well within the capabilities of even ordinary bistable thermal switches. Further, we can design more elaborate protection circuits to drive the LED on and off as we need it. This kind of circuit has the ability to act quickly enough to protect the output power devices and the power transformer from serious damage if the fault sensing circuits act quickly enough. Even though it's slow for a MOSFET switch, it's *fast* compared to a fuse.

With one of the smaller single supply amps, like the Berkeley, one triac in the secondary of the power transformer is all you need. The bigger amps all use a bipolar (+ and -) power supply so this single triac won't work. You could put the triac in the AC power line going to the power transformer, but then the wires to the thermal protectors would be hot with respect to the AC line, a possible safety hazard. It is probably better is to use the cutoffs to interrupt the power from the secondary of the power transformer to the coil of a small AC relay. The relay then opens the secondary of the power transformer leading to the rectifiers or that opens the DC lines from the filter capacitors to the power transistors. That way the thermal sense lines are not AC power line safety hazards.

A short circuit on the output of a Vox power amp will usually kill the output transistors without blowing the main AC power line fuse. I think that this may be one reason Thomas used those microphone-style speaker connectors - they can't short the outputs as they're plugged in. A better way to do this is with a speaker fuse and current limiting on the power devices themselves. I would add fuses to the power supply after the rectifier/filter section in the +V line if the amp has a single polarity power supply, and at the amplifier output before the output goes to the speaker jack. With appropriate sized fuses, this will add some immunity to power disasters. However, in many cases, the fuses will blow slightly *slower* than the output transistors die, so while the fuses will protect the speakers and power transformers, you might need something else for the transistors.

You could do this with a solid state current limiter to shut down the power to the amplifier or to the output transistors either on an instant by instant basis or a latching basis. Current limiting the output transistors will sound God-awful when it happens, but remember that you're not paying for a trip to the repair shop if you ever hear it. Chances are, you won't hear it because the preamp limiter will keep you from ever hitting the limiting in normal circumstances. Updates to this article will contain some suggested circuits.

It's not particularly a good idea to add fuses in the +/- power supply lines from the filter caps to the power output stage. Although at first glance this looks like a good idea, the problem is that both fuses do not blow simultaneously. If only one blows, or if you inadvertently remove one and leave the other one in, the full power supply voltage appears across the speakers - leading to a speaker life only a few seconds long. Maybe the other fuse will blow with current through the speaker before the speaker dies, maybe not. Power line fuses are OK if coupled with DC voltage protection on the speaker output.

The power output transistors really should be kept out of their forward biased second breakdown region by some kind of volt-amp (V-I) limiting circuit as is present in all modern amplifiers. At the time these amplifiers were designed, perhaps in 1963-65, neither forward- or reversed-biased second breakdown was well understood. All that was really known was that power transistors sometimes failed under conditions that were within their current, voltage and power ratings - that is, for no apparent reason. What is actually happening is that at higher voltages,  the edges of the base-emitter areas develop slight variations in localized gain, so some of these places have slightly higher gain than other areas. These regions would go into thermal current-hogging and run away locally, pulling currents from the nearby regions. The transistor would appear to fail under currents well within the current rating and close to but still below both voltage and power rating limits. To make matters worse, the designer would then usually specify a bigger power device. This often made matters worse, as forward bias second breakdown gets worse as the silicon chip area gets bigger.  Paralleling smaller devices would have helped a lot - except that this was a very expensive solution, usually forbidden by accounting and management.

As far as I can tell, no one has ever designed a modern VI limiter for a stacked transformer driven power stage like the ones found in the "big head" TV amps. I'm not certain what would happen on the driver side if you did - negative feedback will keep the driver trying to drive the output devices harder and harder. I'm going to take a swing at it, and see if I can come up with something to keep the outputs from tipping into destruction.

Output Transistor Replacements

All of the Thomas Vox NPN output transistors are silicon, replaceable with 2N3055. I think that they actually *are* 2N3055, just stamped with a special in-house number for Thomas. If you have an amp with a penchant for eating power transistors, you could replace the outputs with 2SC5200 (NPN) or 2SA1943 (PNP). These devices have probably the largest second breakdown area of any device available for audio style uses, as well as having high current gains to relieve the drive requirements on the driver transistor. The other alternative is to use the "big brothers" of the 2N3055. Motorola makes the MJ15015 transistor, which is a 120V, 15A, 200W NPN power device which otherwise has similar specifications to the 2N3055. It has roughly three times the second breakdown rating at high voltage that the 2N3055 does. You could also use the Motorola MJ15024, which is an even-higher voltage rating than the MJ15015. Also suitable are the MJ15003 and MJ15022.

The advantage of the 2SC and 2SA parts is that they are very modern designs, and not only do not go into second breakdown at all at the full DC power supply of the bigger heads, so these parts are only power limited. The other advantage is that this change will probably improve performance, and possibly output power, as their current gain does not drop dramatically at higher currents like the "stock" devices do. A 2N3055 has a gain that drops to about 20% of its low current value at currents of about 6A, where they will be working in a Beatle or Guardsman head. The 2SC5200 retains about 80% of its low current gain at 6A currents.

If you were really going to go for max reliability, you could add a third or even a fourth transistor per side. This should keep any conceivable power condition other than an output short from killing the output transistors. It works by adding a lot of brute force current capability as well as splitting the power dissipation over more transistors as noted in the "heatsink" discussion above. In fact, given the low prices for output devices in today's market, this may be the most sensible alternative. However, too much in the way of improving the output stage power capability puts more stress on the power transformer and filter caps. Even if the output stage can stand the strain, eventually the power transformer will blow. Maybe the AC fuse would save it. Maybe.

Stability

The existing Vox amplifier designs are sensitive to reactive loading. Modern amplifiers do two things to help with reactive loading. First, they use a series inductor to decouple the speakers and speaker cable at higher frequencies. Usually this is one or two layers of #18 wire wound on the body of a 10 ohm, 2W resistor. Second, there is usually a "Zobel" network there. A Zobel network is usually a 0.1 uF cap in series with a 4.7 to 10 ohm resistor to ground. This helps stabilize the amp loading. Both help with stability - modern amps usually have both. These things are so cheap, it's hard to understand why they're not already in there. I guess that the advantages of these simple networks were just not known to the designers of the amps at the time - but we know, and we can add them easily.

Another potential problem with stacked-output transistor power amps is that the static DC level, low distortion, and RF stability all depend on having a matched pair of output transistors. Since the first time a TVSS amp gets opened up is usually when the output transistors are dead, you're usually faced with the replacement problem right off the bat. A bit of history that has been lost is that transistor makers and suppliers used to supply matched pairs of output devices. The stacked output amplifiers are one of the reasons they did. You might be able to get matched sets of replacements through NTE, ECG, or SK, but they will be expensive. It might be just as well to order a number of unmatched replacements and sub them in until you get the lowest DC offset. This is about all you can do without a curve tracer to match devices. Hand matching is not as expensive as that might seem. For all the NPN silicon amps, you can replace the transistors with the industry workhorse, the 2N3055. Even in steel TO-3 packages, I've seen these for as little as US$0.79 each, probably cheaper in quantity. That makes it reasonable to buy a batch and swap for lowest offset, although it takes some time.

At this point you also might want to ask yourself whether you really want to just put the same old devices in there. The 2N3055 is a workhorse of the industry, all right, but it too has been superceded. At least some of the failures in TVSS amps is from second breakdown of the output transistors as mentioned above. The much more modern 2SC5200 is available for about $3.00 to $6.00 each, and have about four times the safe operating area as a 2N3055. They are incredibly rugged, and should add a lot more safety factor for difficult loads. As I mentioned above, you could also use the 2N3055's big brothers, the MJ15015 NPN, which have higher ratings and lower thermal resistance. The biggest problems here are that modern devices are hard to find NOT counterfeited and that TO-3 metal packages are hard to find. If you can use plastic flat packs the number of suitable devices increases a lot. Toshiba, Sanken, and On Semi have several choices. Get devices with over 140V BVceo, over 15A, and over 100W dissipation ability per device. 

I recommend considering doubling up on power devices by installing double the number of  power devices on the heat sinks. If you match devices and use emitter resistors for each device like the paralleling on the Beatle, this halves the current and power dissipated per output device and dramatically improves reliability.

Even if you go for the high tech new output power devices, it isn't all that expensive to match them. The 2SC/2SA power devices I mention are available for about $3.50 each from several distributors. Much cheaper than speakers...

The earliest Buckinghams have germanium PNP transistors. While you might find matched pairs of silicon NPN's, you're going to be essentially out of luck with germanium. You **may** find NTE or ECG replacements, but they will be expensive and of at least questionable quality. It might be that you could hit it lucky in the surplus market. I have found 2N627 and 2N1073 germanium TO-3 devices that work well in the germanium heads at about $7 each. This is far better than the replacement ECG part price.

A better alternative is to change to silicon PNP's for these amps. Suitable devices are available at reasonable prices of a few dollars each, but will give you severe crossover distortion if you simply pop them in. To fix this, all you need to do is to replace the bias resistors with new ones that have the same respective values as the silicon NPN biasing resistors, but in the places where the PNP variety demands them. When you simply replace germanium PNP's with silicon without the bias change it will give a "fizzy" distortion sound at the quiet end of every note. Replacing four wire-wound resistors to re-bias will remove the fizz. An even better solution might be to replace the base-emitter end of the resistor bias string with a silicon diode wired to the heat sink. This will add some thermal protection as well.

Plus remember - with silicon devices you can let the heat sinks run hotter by about another 10C.

The 2N2955 is a complement to the 2N3055, and might work OK. It will still have marginal BVceo and safe area. Better to use more modern devices. See the discussion above about different output transistors. All of the mentioned NPNs have custom-designed complementary PNP types intended for the sam applications. TO-3 packages bolt right in, but modern flat pack transistors are a relative breeze to mount in this application.

It is slightly more complicated but possible to change what was PNP outputs to NPN output transistors. You plug the NPN transistors into the transistor sockets, then reverse the +V and -V connections to the output stage transistor string (NOT to the driver circuit, of course). Putting in NPN's inverts the phase of the output stage, but reversing the + and - supplies inverts it again, so there is a net no change in the phasing of the output stage. Since this change is only reasonable if you're changing from PNP germanium to NPN silicon, you'll also have to change the value of the biasing string of power resistors to get the NPN biased properly to eliminate crossover distortion. In most cases, you can just increase the value of the resistor in each output transistor's string with the lowest value, increasing it to give more bias voltage. 

The preamp circuits in Thomas Vox amps typically use pairs of transistors to do the miscellaneous amplifying tasks. Current-day designs would use an opamp instead of the two-transistor pairs in most cases. These two-transistor compounds are excruciatingly sensitive to any problem with power supply or emitter bypassing. If the emitter and bypass capacitors are not in very good shape, these things will usually oscillate, and the oscillation will be fed through the power supply to the entire rest of the amp. It is almost impossible to figure out where it comes from - I know this from personal experience, as a full Saturday of chasing oscillation did not find it on my first Beatle head. It was only when I just decided to do a wholesale replacement of the electrolytic capacitors that the oscillation quit. It turns out that the emitter bypass cap on the reverb driver section of the amp had developed a high internal resistance, and that was enough to make it sing all by itself. 

Oscillation problems in the reverb section are the most prominent place to find bugs caused by aging capacitors. The others will bite you sooner or later. On any solid state Vox amp you open up, replace the electrolytics with fresh ones. Just do it! - the amp *will* start oscillating and/or sounding bad when those components start decaying and you or the then-current owner will be back in there again.

Replacing caps is a pain because of the mechanical setup, but it's a really good idea. If you're a weenie about pulling the board loose for access to the back for removing and replacing caps, you can do a reasonably reliable job from the front/top side. Carefully note the orientation and value of the cap you're about to replace and clip the leads off right next to the body. using needle nose pliers, bend the remaining lead stub so it sticks straight up in the air. Use a pencil eraser or something similar to clean the lead so it solders very easily. Pre-tin the leads on the replacement cap. Put the replacement in place on the board and bend its leads around the old lead stub. Trim the leads and then solder the new lead onto the old one. You *must* do a quick, complete job on this joint to keep from melting the solder on the bottom of the board that the lead stub is still connected to.  If you're fairly deft with the iron, the pre-tinned and cleaned leads will solder quickly enough to not melt the solder on the stub on the bottom of the board. Get in with enough heat on the iron, and out quickly. It's a good idea to stick a small blob of hot glue to hold the new cap in place for max reliability. This keeps from flexing the new joints.

Note that on these amps, they used a fair quantity of 1uF and 2.2uF Non-Polarized electrolytics in the signal path for signal coupling and even for tone control frequency response setting. Using electrolytic types for frequency/tone setting capacitors is a bad idea in general, as the values change as the capacitors age, and eventually the cap will die. It's a really good idea to replace the small electrolytic NP caps with plastic film capacitors wherever possible. NP electrolytic capacitors don't fare any better than regular electrolytics with aging.

The distortion circuit inside the Buckingham, Guardsman, and Beatle is very, very close to being a Fuzz Face. If you were so minded, you could mod it into a real Fuzz Face circuit with little effort. More on this in later updates.

The power supply filter capacitors on the TV amps are minimal capacitance values. It's a good idea to replace them for both aging and capacitance reasons. There is enough room on the power amp/power supply sub-chassis to add filter caps under the existing ones. There are also clamps that will adapt modern filter capacitors to fit the holes for the twist-lock capacitors that are already there.

RSR Electronics had at the time of the first draft of this article twist lock capacitors in 3200uF/50V ratings in their surplus bargains for $1 each. These are all you need in the Buckingham and Royal Guardsman, but are too small for the Beatle and the 120W (four-heat-sink) version of the Westminister. Note that these capacitors will have twist lock connections, but the twist locks will be the common minus, where the original capacitors had a (-) terminal isolated from the can. You'll have to make an insulating wafer from phenolic or glass-epoxy PCB stock with no copper on it so you can mount the capacitor and have its can be isolated from chassis ground for the - 31V power supply. Notice also that the outside/can of the capacitor will be riding at -31V, so you could get a serious short to ground from it. It's a good idea to cover it with a sleeve of large heat shrink tubing to insulate it.

[6/19/00] I tried out the generic capacitors, and they work great - except it's a true pain to manually make an insulating wafer. It might be better to hollow out the caps you have now and stick modern radial lead capacitors inside and putty them in with RTV, as described below. If you want to make your own capacitor insulating wafers, I've drawn up a template for this. Print this template on any postscript printer and then use rubber cement or spray adhesive to stick it to a 1.5" by 1.5" square of bare glass-epoxy circuit card stock. Cut the small rectangles for the twist legs first, then cut the central square out and drill the mounting holes. The holes should probably be for #6 machine screws. Once the wafer is made, you can use the wafer as a template for drilling the chassis by clamping the wafer on the chassis so that the four twist leg holes are lined up on both, and then drilling the corner holes in the chassis. Only then nibble away the mounting ear area on the chassis. Do it this way and you'll end up with perfectly aligned wafer and chassis mounting and no shorts. Once the wafer is installed on the chassis, you can install the capacitor. Double check to be sure that the twist lock legs don't short to the chassis.

Hollowed out originals are probably less work. Modern replacements with cap clamps don't look original, but are very effective. Installing capacitor clamps lets you put in 8000 to 10000 uF capacitors and get better filtering.

If you don't like the idea of having non-original looking capacitors, you can find a twist lock capacitor, good or not, that is the same size as the original twist locks. Remove the guts of the cap by carefully un-knurling the terminal end of the capacitor or just drilling out the terminal seals and removing all the insides, leaving the twist lock tabs intact. You can probably fill it up with a modern filter capacitor of higher capacitance and equal voltage rating. Just get the new capacitor in there, and RTV the new cap in place. Another thing that is possible is to put a new pair of filter caps on the underside of the chassis, and just leave the old twist locks in place but disconnected for their vintage look.

While you're under there, solder a V150LA20A MOV varistor across the AC power line lugs on the little terminal strip under the power transformer. This will capture and subdue the AC line voltage spikes that can kill parts further in. They're about $1.00. Do not use the lower-voltage V130Lxx devices, as they will become a fire hazard over time.

If you have a Buckingham or Royal Guardsman and want to get a bit more power out of it, you can replace the power transformer, filter caps, and power transistor drive circuit with the equivalent of the Beatle circuit. It's important to upgrade the power transformer and filter caps, not to simply rebuild the driver circuit, as the amplifier can only put out as much power as the power supply gives it. The Buckingham, Royal Guardsman, and Beatle all use the same raw power supply, +/- 32Vdc, taken from a 42Vct power transformer. The power transformers are different sizes, reflecting the different output powers the amps will do, but the power amp/power supply subchassis is punched to accept any of them. If you find a suitably sized transformer, you can probably just bolt it in. In a coming update of this article, I'll try to provide the measured sizes of the power transformers as a guide for updating.

It never hurts to upgrade the rectifiers. You can use the same 3A rectifiers as were in the original, or as I prefer, replace them with 6A, 200V devices. You can also get rectifier bridges with extended wire leads that can be bent so the whole rectifier bridge can be substituted for the discrete diodes. Likewise, if you're a sledgehammer kind of guy, you can put one of the 15-25A rectifier bridges onto the chassis and never worry about popped rectifiers again. It's possible to simply bolt this rectifier bridge to the bottom skirt of the chassis, so it causes no cosmetic problems. While you're doing this, solder 0.01uF/1kV ceramic disk capacitors between the terminals of the bridge. This will reduce radiated EMI buzz from the diodes when they snap off at 120 times a second. Some people have reported good results with fast, soft recovery FREDs (Fast Recovery Epitaxial Diodes) instead of the snubbing components.

Another problem with upgrading the power of the amp is that if you keep the same transformer-driven circuit as the Beatle, you'll have to either find or make a driver transformer equivalent to the one in the Beatle, or convert the circuit to one that does not use a driver transformer. There is more info on the driver transformer below. In a later update, I'll try to include a recipe for building a Beatle-style driver transformer.

While it is possible to update the Buckingham and Royal Guardsman amps to as much power out as the Beatle because they all use the same size cabinet and same chassis sizes, you really should pause here and think about whether you want to make a Beatle out of it. For about the same effort, you could make a more modern power amp inside the shell by just disconnecting the old power amp for re-conversion back if you ever sell the amp. This should properly be considered a major rebuild, but it is possible.

With modern power transistors, it's entirely possible to make a very robust power amplifier that simply bolts over the existing circuit board and uses the same heat sinks (preferably augmented and thermally protected as noted above). The existing circuit board bolts to the chassis with 6-32 screws. If you get 1/2" male/female standoffs with 6-32 threads, you can replace the circuit board hold-down screws with the standoffs and mount another circuit board above the existing one with whatever fancy new power amp you like on it. This could even be a power MOSFET design with greatly improved performance.

If you are not concerned with the "originality" of the power amp section (subject to being able to re-convert it, of course), you could even replace the two or four bent-aluminum heat sinks with one piece of aluminum finned extrusion. Save the original hardware and transistors for re-conversion back to stock, naturally. You could also just bolt in a replacement and more modern power amplifier circuit. A power MOSFET output amp would work well.

The sound of such an amp should be very close to the sound of the existing amp if not identical. The Vox design is intended to be limited by the signal limiter in the preamp, never by the actual saturation of the amp itself.

[6/18/00] More about that limiter. I now think that the limiter is why the Thomas Vox amps sound much better than other solid state amplifiers. It's a peculiar limiter, not nearly as harsh as clipping diodes, and it seems to have much more like a tube amp squash on the signal than clipping it off.

A corollary to Thomas Vox amps' good looks is the mantra of the beauty salon - there is no such thing as a natural beauty. There were some compromises made to get them to look good and still fit inside that neat-o cabinet, and the compromises only become evident when you start unwrapping the thing. I've already mentioned one - heat and the poor ventilation - but the one most people run up against is how to get those silly screws out of the ears of the preamp chassis to remove it from the cabinet.

There really is only one way. You have to get a extension or tool long enough to reach the wood screws that go through the ears into the particle board of the cabinet. This is at least 10" in the case of the big-cabinet boxes, the Buckingham, Royal Guardsman, and Beatle. The tool goes through the hole in the bottom of the cabinet and reaches the screw at an angle. If the screw or tool is not immaculate, no rounded-over corners on the edges, you will not have enough purchase on the screw to remove it if it's tight. I recommend putting in a new screw each time you remove it so the next removal is possible. It also helps to have a so-called "wobble extension" on the tool. These are square-drive extensions that are machined with a bit of wobble built into them, so they can effectively flex as they turn.

On amps where the chassis has been removed a lot (i.e. more than once) the particle board that the wood screw had a grip on will powder and fall out, leaving a chancy purchase for reinstallation. After a few removals, the holes in the particle board usually strip out, and the screws no longer support the chassis. This leads to the down-at-the-back way most of the chassis ride, as the front tabs will bend from the chassis weight.

About all you can do is to try to shore things up a bit. There is a boating product called "GetzRot" that is a water-thin epoxy designed to saturate into decomposed and rotting wood and affix what remains in place. This is very workable for the Vox cabinets.

Turn the empty cabinet upside down, and excavate the stripped out hole a bit underneath the particle board surface, leaving the entry hole intact. Then you pack the hole full of sawdust or powdered particle board. Mix the GetzRot according to the instructions (and following all safety instructions!) and pour the hole full. When the Getz Rot is cured, it will have epoxied the new sawdust and older particle board together much more strongly than they were originally. Drill a small pilot hole for the screws to be reinserted, being careful not to drill through the tolex on top, and screw them in experimentally before replacing the chassis in the cabinet.

The rotary power switch on the bigger Thomas Vox SS products has been a problem for many owners of the larger amps. The thing was a custom part for Thomas, and has a habit of wearing out. The thing that makes it irreplaceable is that it is not only a fairly complicated switch, it also carries a high current. There are 3P3T rotary switches available that could replace the switching function, but they have current ratings down in the sub-one-ampere region. The line current rating on the Beatle is about 3A; worse yet, the speaker signal runs through the switch contacts, so the switches have to handle the speaker currents as well, which may be as much as 15A, although you probably will not have to switch that much current. The switch contacts must just conduct it.

I have now repaired a couple of these. Usually the phenolic switch wafer breaks near where the support posts hold the wafer in line with the rest of the assembly. If this has happened to yours, you can often repair it by carefully crafting bits of 1/16" thick glass-epoxy PCB stock to fit over a non-used part of the wafer and epoxying these reinforcing bits in place. It's worked two out of two times for me.

If you must replace the switch, one clean way to do this with a lower current switch that controls relays to switch the AC line power and speaker loads. This allows you to use the low current rated and easy-to-find 3P3T rotary switches and still have the amp work properly. You'll need at least two relays, one for the main AC power and one for the speaker output. At least the AC power relay will have to be powered from 110VAC through the new 3P3T switch. For the speaker relay, you have the choice of driving the coil from either the AC line or from DC from the power supply. The third section of the 3P3T controls the indicator lamps, which are run from DC on the secondary side of the power transformer.

[5Oct2011] I'm testing out a replacement for the power switch. This uses a 100ma 4P3T Lorlin rotary switch from Mouser, and two PCB mount miniature SPDT relays with 120Vac coils. This lets me mount a small PCB to the rotary switch with the relays on the board, and have the switch control the power to the two indicator lamps and the relays. The relays are burly 16A/250V devices; one of them turns power on/off to the power amp, and the other connects the speakers to the output jack in the "on" position. I'll let you know how it works. 

You must be very, very careful when wiring this up that you don't set up a condition where the live AC could contact the DC power on the secondary side. This could not only kill you, it could damage the amplifier. It's a good idea to plan to use heat shrink and/or other supplementary insulation on any and all AC power rewiring to ensure that it remains safe.

I've had several questions about why a standby is used on a solid state amp at all. I think I finally figured it out. The fact that there is a stop position on the switch that momentarily powers the amp without connecting the speakers gives the power amp time to come up to a balanced and ready condition before the speakers are connected. This effectively keeps any power on thumps from reaching the speakers. It also keeps any power-off funnies from reaching the speakers. Some amps squeal as the power supplies go down. The "standby" position is really, really effective at keeping both of these out of the speakers. It's not for "standby" at all, it's for power on/off sequencing, but "standby" is a familiar term that tube amp users would accept. It's a good bit of terminology use by Thomas to make a better-behaved amp, but it's not quite what the term leads you to believe.

 AAAGHHHH!! Mistakes in the Official Thomas Service Schematics!!

Yep. The service schematics have bugs in them. So far I have not found examples of the real circuits being different from the schematics shown, but the "typical" voltages show are in error in a couple of places. I just ran into one, which is what prompted me to write this up.

I was looking at the Berkeley III, V1083*6 power amp schematics when I noticed this one. The schematics show not once but three times that the output of the power amp sits at 10Vdc. This cannot be accurate, because 10Vdc on the speakers would cook the speakers in short order. The voltage shown is just plain wrong. The output undoubtedly sits at near 0V, and maybe this was just a typo.

I've found similar bugs in the drawings for the reverb section of the Beatle. The DC voltages on the reverb driver just could not be as listed. I wasted most of a Saturday morning on trying to make the reverb section be what was listed in the schematics, and never could. A short bout with a calculator and some thought showed that it never could be that way.

The moral to this story is - use the schematics and service voltages as a guide, but be prepared to use your brain.

I will document other mistakes on the schematics as I find them. If you have questions about the schematics and would like me to do some circuit analysis to find out if the schematic is correct or not, contact me by email.

The Beatle, Royal Guardsman, Buckingham, Westminister, and Viscount all have the same raw power supply voltages out of the power supply, +/- 31Vdc. The big differences are only in the size of the transformer and amount of capacitance used in the first filter caps. The Beatle and the later high power version of the Wesminister bass amp are capable of 120W rms into 2 ohms - which is only 22V peak, 15+ Vrms signal out. 

That's why they use such low speaker impedances - to get high power from relatively low-voltage power supplies. Although the raw power supplies are +/- 31Vdc, the power amp will not push the output signal higher than 22V into a speaker load, althrough it will go to nearly 31Vdc if it's unloaded. This is due to the heavy loading of the driver stage and the falloff of current gain of the germanium and early silicon output transistors. I think that the design is this way because at the time the amps were designed, these transistors were state of the art for affordable power devices. They designed within the device limitations, including the limited Vce and second breakdown area. 

What's important about this is that the Beatle can be cut back in power by simply running it into a higher speaker impedance. The TV Beatle is one of the loudest guitar amps ever made. The Beatle will deliver 120Wrms into 2 ohms, 60W rms into 4 ohms, and only 30Wrms into 8 ohms and (gasp) only 15W into 32 ohms. If your Beatle is too loud, switch the speakers to change the loading. A four-speaker Beatle cabinet could be reworked with a power switching setup to make any of these come true, as the four speakers in there can be series/paralleled to get either one of 2, 4, 8, 16 and 32 ohms nominal impedance.

Similarly, the Royal Guardsman will do 60Wrms into 4 ohms, and 30Wrms into 8 ohms. The Buckingham will only do about 30Wrms into 8 ohms, and uses two 16 ohm speakers in parallel to do that. The ultimate limit on the amount of power that these amps will put out seems to be the limited amount of voltage in their power supply.

The Beatle power transformer is rated at roughly 45Vct to 48Vct at 6A. Later production Beatles seem to be on the low side of this, leading me to believe that Thomas was subtly correcting some of the reliability problems with the output devices by lowering the voltage stress on the transistors. 

A 48Vct/6A transformer should make a great replacement. Similarly, the Guardsman power transformer can be replaced by a 48Vct/3A power transformer. Do NOT replace the transformer with higher than 44Vct if you are not also replacing the output transistors with higher voltage devices as noted above. It's likely the originals will die. 

The chassis-mount cans used in all of these amps are fairly easy to replace, as they're rated at about 2000uF/50V in the smaller amps up to 5000uF/50V in the Beatle/Westminister. It's easy to find 15000uF/50V capacitors that are the same diameter and shorter than the original caps. It may even be feasible to gut the existing capacitor shell and stuff it with modern caps so that the original look is maintained. In any case, capacitor clamps will put modern caps in very easily, although you will have to drill holes to mount the clamps.

The driver transformers were fairly mysterious, and they do die sometimes in the chain of destruction that follows power transistor failure. I have successfully replaced a driver with a modern TO-220 audio driver device, but it took some tinkering with the feedback compensation caps to prevent oscillation. 

I've spent about two months researching the transformer-driver style of power amp and learned that

•  the transformer driven series output stage was old even for the early 60's; Thomas must have used "proven" amplifier designs from the mid to late 50's for the basis of these amplifiers. The then-current or progressive amp designs were based on the quasi-complementary design first espoused by Linn at RCA. The Linn design has formed the basis of most of the amplifiers in the world after the early 60's.
• the above notwithstanding, authentic JMI amps in the Supreme and Conqueror line use transformer driven stacked output stages as well.
•  the design of the driver transformer is critical. It must have vanishingly low leakage inductance between secondaries, as well as being able to deliver substantial power. The Beatle driver stage is a power amplifier in its own right, about 4-6W.
• The Beatle driver transformer is wound on a square stack of EI-75 scrapless laminations. The laminations are paralleled rather than interleaved and have a spacer at the joint where the E's meet the I's because of the DC current in the primary side. It may be possible to replicate this from a small power transformer. We'll see.
• the primary of the driver transformer is about 325 turns, the secondaries about 130 turns each. I wound ten turns of fine magnet wire into the crevices around the coil of the driver transformer of my Beatle and pulled out the driver transistor and the output transistors. That way the primary and secondaries were opened and I could drive the transformer and measure both the existing turns ratio and the voltage across my added ten turns. This let me calculate the turns ratio and the volts per turn, so I now know that the primary is about 325 turns and the secondaries are about 2/5 of that, about 130 turns each. The currents in the driver collector and into the bases of the outputs let me make a fairly good guess about wire size, so I'm close to being able to write down a recipe for one of these things.
•  The "headroom" on the output stage, putting out only 22V peak from a 31V power supply make it tempting to put a MOSFET power amp in there. The only real problem with power MOSFETs in amplifiers is that their gates have to be pulled substantially higher than their sources to get enough current flow. The Thomas Vox power amp stage makes this super easy, as there's enough extra voltage here that a MOSFET power stage with its own drivers will probably drop right in and work fine. I may design up a replacement output stage based on this.

What to do with a dead one

First - don't bury it!! It can be brought back to life. If you don't believe this, and just want to be rid of it, call me and I'll take it off your hands 8-)

I've been through a number of these things now, and a lot of the time there are a only few things that are the cause of death. I know they look intimidating inside, but hang in there. Here's how I reincarnate a corpse.

1. Get the power supply working.
• Inspect the power cord, power wiring, AC switch, fuse, and other AC power wiring items. Bring these up to a safe level before powering it up. It's common to find that the rotary AC power switch that gives off/standby/on in the larger solid state amps has cracked or broken wafers. You can often fix these with carefully shaped bits of blank glass-epoxy circuit board stock (no copper layer) epoxied in place to reinforce the breaks, as the breaks are usually at the supporting posts of the wafer. If you can't fix it, you can replace it with a rotary switch from Digi-Key or a low-current rotary and a relay board, as noted above.
• Power it up, using a variac or the series light bulb adapter to limit AC current into the amp. If the current looks excessive - like if it approaches half the normal power rating of the amp something is wrong - stop and go figure out why the amp is pulling so much power. I recommend unsoldering the transformer leads from the rectifiers and filter caps and seeing if the transformer alone pulls too much power. The transformer could be bad. If  the transformer puts out the right amount of AC voltage and does not itself pull too much current, check the rectifiers and filter caps for shorts, and then suspect shorted output devices. If that's not the case, you may have shorted electrolytic caps as bypasses.
• You may get normal for the transformer, good AC voltage and good first filter cap voltage, but still no operation. It may be an open resistor in the power supply decoupling chain. With the exception of the +/- power to the output stages, the power supply on TV amps is like that for tube amps; a series of RC sections that successively drop voltage and filter both ripple and stage-to-stage interaction. I recently had an open 3.9 ohm/5W wirewound resistor that was the only thing wrong with the amp.
2. Get the power stage working
• Once the power supply is working right, get the power amp running correctly. It doesn't hurt to just pull the power transistors out - they're always socketed - and test them with ohmmeter and/or curve tracer. Output device failure is common. Use the info above about the power transistors and get them working. If you find shorted or open output devices, measure every resistor in the power stage for the correct value and also test the driver transistor. The outputs may have killed the resistors and/or driver in their death throes. Verify full voltage swing into a no load, and then into a resistive dummy load with a function generator driving the power stage.
3. Clean the connectors
• Caig DeOxit or R-5 is great for this. Clean all the electrical connections you can, including RCA jacks and plugs, etc. These amps are old enough that the tin coating on the hardware and contacts is usually growing white oxide fuzz. Get them clean to the metal.
4. Replace the electrolytics
• Just do it. Make up a list of the electrolytics on the board, including non-polar ones, and replace them all. Even if they're not bad, they will be soon. Do this carefully and don't break wires. Note the polarity closely, of course, as you do this. If you run a repair shop, and you don't do this, the amp will either be back to you or will go to someone else's shop because you didn't fix it right in the owner's mind.
5. Clean the pots and switches
• Again, Caig to the rescue. Schpritz them, wipe off excess with paper towel, then operate several times while still damp.
6. Get the preamp running
• Boring and tedious, but it's gotta be done. Pick up your schematic, stuff 100mv of sine wave into the input, and follow it through the preamp sections, fixing as you go. It really helps if you've replaced all the electrolytic caps first. You may already be running right if you've done that. The second most common problem is a broken wire. Parts failures other than electrolytic caps are usually rare, with the exception of physical damage (broken parts or cracked solder joints, etc.) and relay failure in amps with relays in them.
7. Fix the damage to the cabinet
• See above for screw holes. Do any woodworking and/or covering repair, etc. as needed before putting the amp back together. The use of particle board for the cabinets means that it's fairly easy to fix stuff. Use a good glue, either epoxy or carpenter's glue and clamp the bits together for cracks and well-fitting pieces. You can actually mix up an epoxy-sawdust putty and sculpt in missing sections if you're good with epoxy.