Isolation Transformer Upgrade for Old Guitar Amps

I'm copying this from my own instructable about tube amp rebuilding :

DISCHARGE THOSE POWER FILTER CAPACITORS!!!!!

Seriously. Do this EVERY TIME you work on the amp. If you don't, DO NOT complain if you loose the use of your hand. DO NOT come back and haunt me if you die....

The power 'filter' caps can store fatal amounts of electrical current, and are sometimes termed "reservoir" caps. The caps are connected near the rectifier and are part of the power supply, and aid in converting AC to DC. In fact, they are a standard component in any power supply.

If you're completely lost, and don't understand this , DON'T MODIFY YOUR AMP . You haven't enough knowledge to work on high voltage/current circuits safely...

There are several ways to discharge caps, but here's the easiest:

FIRST, UNPLUG THE AMP! (But that doesn't make it safe....)

THEN,

-- Jumper the positive (+) lead of each large cap to GND for several seconds. A jumper with a built-in resistor (10K or so) will help prevent sparks here... If your jumper has a resistor, leave it connected for at least 30 seconds before you touch anything.

-- OR short the caps with a screwdriver. Lay the shaft on the chassis, then bridge to the positive (+) lead of the cap. Be sure the screwdriver handle is insulated (if it's painted, it might not be.)

This may result in a spark... Obviously, your flesh can act as a jumper also (that is NOT a challenge.)

First, mains-rectified amps were generally small output, 1-5 watts. Manufacturers usually didn't skimp on the larger amps.

If your amp has only one transformer (the output transformer) the answer is YES, you need one. If your amp has two transformers, odds are you don't need an isolation transformer.

Power transformers, the type that's missing from these unfortunate amps, are the largest transformers. They also tend to get warm, so 19 out of 20 times they'll be mounted on the outside of the chassis. The lack of one will be obvious.

Output transformers (and no vintage tube amp can be without one) however are smaller, and might be mounted in various ways, some of which are hard to see. They could be on the outside of the chassis, yes--but also under the chassis, or on the speaker itself. But rest assured--there will be an output transformer somewhere.

But wait--it's not that simple. Some amps isolated the signal path from the mains, but not the filament voltage. If equipped with a three-prong cord, these amps are somewhat safer, as they do offer isolation in most cases.

One sure-fire way to know if your amp lacks isolation is to examine the tubes. American tubes are prefixed with the filament voltage (12ax7 has a 12V filament, 6V6 has a 6V filament, etc.) The AC/DC circuits were designed to run all the filaments in series on a 110V supply. They therefore have high prefixes:

One common set: 50C5, 35W4, 12AU6

...which together equals 97V, so a small resistor was also added in series to drop the 110V voltage an additional 12 to 15V. It should be immediately evident that this was a cheaper way to build an amp. And many were built.

So, from a safely perspective--does your amp need isolation? YES.

I picked up this funky little Gregory Mark I amp from Craigslist for ~$25. Gregory put date stamps on their cabinets, and this one dates to March 25, 1955. So this little guy is over 50 years old!

Paul Marossy has a great website dedicated to Gregory amps (in fact, the photos of the Mark I example on his site are mine.)

It's a typical low-wattage practice amp of the time. No tone control, only volume. Probably 1-2 watts of output power. It's great "living room" or recording amp.

Among the mods I've already done was adding a 1/4" jack for the speaker output. I just unplug the small speaker, and run the amp into one of my 4 ohm cabinets. The amp is easily twice as loud through a 2 X 12 cab... (with loads of bass, too.)

But it's also a typical non-isolated amp, and that safety issue need to be addressed...

Tools

Soldering iron and solder
Drill and bits
Stepped drill bit (for large holes--fuse holder)
Screw drivers, etc.

Parts

-- Isolation transformer
-- Fuse holder and fuse
-- Scrap wood
-- Heat-shrink tubing
-- Three-prong cord (scavenged from an old computer)
-- Line wire, misc wire, wood screws, etc.
-- Metal plate for mounting fuse holder
-- Strain-relief for the cord

Here's a schematic for the amp (complements of Paul Marossy's website.)

It's very typical of this amp type. Note the following:

-- the lack of a power transformer.
-- no fuse in the circuit.
-- the 35w4 diode is directly connected to the mains.
-- the GNDs are directly connected to the mains (this one doesn't even have the protection of a "death cap!")
-- the tube filaments are all connected in series, directly to the mains.

How do we fix it?

-- add an isolation transformer
-- add a fuse
-- reroute the ON/OFF switch
-- add a three-prong cord, and a proper earth ground

One issue will be dealt with later: using an iso transformer with a half-wave rectification circuit.

Unlike many power transformers, isolation transformers have a 1:1 voltage ratio. The output voltage is (for practical purposes) identical to the input voltage. They serve only to "isolate" the device from the high-current potential of the mains. DON'T use an auto-transformer--they don't isolate.

Transformers also have a Volt-Ampere or  VA rating. VA is roughly analogous to wattage (remember, wattage = voltage * amperage, or wattage = V * A.) for resistive circuits, but not for inductive loads. For inductive load, you can "guesstimate" wattage capacity = VA * 0.7, or the wattage of an inductive load is ~70% of the VA.

Wiki page on the Volt-Ampere.

So the first question is:
What is the total power consumption of the amplifier?

I.E., NOT the output wattage, it's only a fraction of the total wattage it takes to run small amps.

Most amplifiers have a power consumption rating on the back. My Gregory doesn't, but it's safe to compare it to other three-tube amps. My little Kay amp consumes 28 watts. My Danelectro DM-10 (4 tubes) is closer to 40 watts.

It's a safe guess that most three-tube amps don't consume anywhere near 40 watts of power, and probably not 30 watts. Since more than half the load of a small amp is resistive (the tube filaments), and 70% of 50VA is 35 watts, then a 50 VA rated transformer should be fine.

So we're going with a Triad N68-X isolation transformer, with a 50 VA rating. Good stuff.

The N-68X is inexpensive, and can be purchased at various online electronics stores. One example:
Allied Electronics (for $11.41 USD.)

Mouser has it, and Digikey probably does, too.

If your amp requires more than 50 VA, Triad also makes larger transformers. Of course, isolation transformers from other manufacturers will work just as well...

Here's where we decide how to implement the changes.

Wiring the N-68X iso transformer

Primary--

The N-68X can be used with either 120V or 240V AC systems. 

US 120V

For 120V, place the two primary coils in parallel.

Tie these colors together, and connect to the mains (through the switch, etc.):

-- Black and Red/Black
-- Yellow/Black and Green/Black

Euro 240V

For 220-240V, wire the N-68X primary coils in series:

220V / 240V mains-- Black and Black/Green .
Connect Yellow/Black and Red/Black together.

Secondary--

Use only the two Red secondary wires. The white wire is the shield. Connect it to the chassis (or earth ground) if it's mounted there, or if you experience any noise.


Re-routing the switch

The original ON/OFF switch is mounted on the chassis panel. To keep the switching truly functional, we'll have to route it differently.

We could leave the switch as-is, but then the primary of the isolation transformer would be in a permanently ON condition. Only unplugging the cord would cut the power to the trannie. The switch would still operate the amp, but there would still be some current draw. That's wasteful and "bad form."

To use the original switch, a simple two-conductor wire can be attached, and run down to make/break the incoming AC connection to the isolation transformer.


Connect the earth ground

With the three-prong cord addition, a true earth ground is available.

Attach a wire from the center prong (should be Green, but verify) of the plug and connect it to the chassis.

Optionally, the transformer casing can also be grounded.


Power -- connecting the isolated AC

OK, here's where things get a little "iffy."

The Simple Way:

The transformer's secondary can be connected directly where the old power connections attach. In this case

Wire 1 ) to the rectifier plate, and the series filaments
Wire 2 ) to the chassis ground

The order of the secondary wires doesn't matter--the AC from the transformer is isolated, so there's no Hot or Neutral side. They are both Red for a reason...

The Correct way:

Read the next Step--it deals in depth with half-wave rectification...

6 More Images

But wait--the 35W4 tube is only a single diode , so the rectification is half-wave , rather than full-wave.

Is that bad?

Well, yes. As the name implies, half-wave rectification only uses one half of the AC waveform, and blocks the other half. Power transformers are really designed to be symmetrically loaded. The flux field collapses as one peak falls, and the transformer expects an equal load--and an equal amount of magnetic force from the complementary peak.  Without a load on half the cycle, the collapse of the field causes the transformer core to become saturated much more quickly than normal. That puts a "standing" DC voltage on the transformer. The N-68X, being a small transformer, isn't designed to handle this.

Half-wave rectification isn't quite as much a big deal on your household "mains." The current draw is small compared to the available current. The resulting asymmetry only changes the total waveform fractionally. But even that could be enough to create noise in other devices...

When I first installed it, I tried to use the N-68X with the circuit, as-is. But it immediately became obvious that the transformer became too hot, considering a current draw less than 30 watts.

Solving the problem

A larger isolation transformer might nullify the problem, but when using the N68X the best solution is to rectify twice -- once with a solid-state bridge rectifier to shift the negative voltage over to positive; then rectify again with the 35W4 tube. That will eliminate our asymmetry, since there will no longer be any negative voltages for the tube rectifier to block.

See the fifth illustration for this "combination" technique... Note that the output of the combination is full-wave , despite passing through a single diode rectifier after the bridge. So there's more current potential for the amp circuitry than before. Plus it's probably quieter, too.

And note that the peak voltages of the tube rectifier (diode) are lower than the solid-state bridge. Note also that half-wave rectification need not be done with a tube diode--a solid-state diode functions just as "well" for this application.

Where to insert the SS bridge

There are two good options:

Option A ) between the isolation transformer and the entire amp circuit. Since rectified AC (pulse DC) holds the same potential as regular RMS AC, the total voltage doesn't change.

If the filaments were fed solid-state rectified and filtered DC the voltage would be too high, because the total voltage would approach the peak voltage, rather than being an average. And the filaments would fail. However, the filtering caps come after the tube rectifier, so that's not a problem.

In addition, the SS rectifier could be mounted back on the iso module. Since I didn't do that initially, I placed it on the chassis.

Option B ) after the filaments, and feed the tube rectifier only (only the DC parts of the amp cause asymmetry.) This would work fine. But it also requires a bit more rewiring.

I chose the first option...

Why include the tube rectifier at all?

The bridge produces all the rectified current the amp needs...why keep the 35W4?

-- Leaving the 35W4 will keep the peak DC voltages at a lower level than the more efficient SS bridge by itself. The 50C5 power tube wasn't designed for plate voltages much higher 120V. Since AC peak voltage is higher than it's RMS value, rectification circuits tend to output a higher DC voltage (theoretically 1.414 times higher than the RMS.) But as stated previously, tube diodes are less efficient.

-- All the tube filaments are still connected in series, so removing the 35W4 would have created a new problem--how to drop the voltage on the series string of filaments (the remaining two tubes) by an additional 35V.

Leaving the 35W4 tube in place solves both these issues.

Necessity

Is all this absolutely necessary? Well, with a large enough Isolation transformer, maybe not.

A 100 or 150VA rated transformer could safely deal with halfwave issues for a <50 watt amp, I'd say.

OK, it's a year later, and then some...

These changes do seem to introduce hum to some AC/DC tube circuits. For a few reasons: SS rectifiers are more efficient, the filtration is a little lacking, and fullwave rectification shifts the PS wave peaks from 60Hz to 120Hz.

So in the quest for a hum-free amp, I've modified the circuit somewhat. This has made the little Gregory amp almost totally free of nasty hum. Your mileage may vary--each amp is a little different.

NOTE about this section:
There's cost for converting to higher voltage DC filaments--increased power consumption. The power draw for the 120V AC filaments is 18 watts; 25.2 watts for 168V DC filaments. Keep that in mind. Note also that this mod may raise the plate voltage for the 50C5 output pentode somewhat higher than the recommended voltage...this has worked fine for me, but YMMV.


Option C

This option Inserts another filter cap after the SS rectifier. It's a little odd, as the additional filter cap is placed between the two rectifiers. Nothing technically wrong here, just unusual...(as are two rectifiers, but we know that works.) We're just feeding the second rectifier a current source that's less...wavy .

However, Option C introduces a complication: With even a moderate filter cap, the filament voltage is much closer to DC than the original AC.

That's good, right? DC is quieter. Yeah, but the DC voltage resulting from rectifying and filtering AC is closer to the peak AC voltage, and can't be treated as an "average"... So the new DC voltage is higher--TOO high, in fact. The old AC-to-DC formula is in play...the DC voltage is approx 1.4 times the AC RMS, approx 168V. This would surely burn out the filaments.


Taming the Higher Filament Voltage

But there's already a series resistor inserted with the three filaments to drop the voltage--for line AC (115-120V). We just need to increase that resistance so it can handle the higher voltage.

So how do we figure the new resistance value for Rv? A few facts...

-- the three tubes (12AU6, 35W4, 50C5) drop a total of 97 volts (12 + 35 + 50 = 97).
-- each tube draws 150 mA (0.150 Amps ). That's important.
-- the stock Rv value is 160 ohms (for 120V).
-- the new filament supply voltage is 168V .

Hmmm, each tube draws 150 mA. AaaHa! Current is equal for all components in a series circuit. So the current draw of Rv must match.

Time for good ol' Ohm's Law (R = E / I, or resistance = voltage / current). Let's check the original value:

120 - 97 = 23 extra volts to drop.

To achieve the same current draw for Rv: 23 / .150 = 153 ohms. Good! That's almost spot-on to the 160 ohm spec'ed value.


The New Rv Value

Estimated DC voltage for the filaments: 120 * 1.4 = 168V
168 - 97 = 71 volts to drop.

71 / .150 = 473 ohms. That's SO close to a standard value...

470 ohms is the new Rv resistor value. Rv is dissipating 10.5 watts, 15 watter is required.

This has been tested, and worked perfectly--the very first time (yeah!)

Yes, this ups the current draw (total wattage) of the amp, without increasing the output power. OK, not quite true--the output pentode now has a higher plate voltage, so the output is slightly increased. The higher filament voltage is drawing about 7 additional watts. The iso transformer does get a little hotter.

The New Filter Cap

Pick a reasonable value here. I used 22uF / 250V, but upped that to 100uF / 250V. It works just fine, and obviously the 100 uF cap is a bit quieter.


Other Anti-Hum Mods

I've moved the initial SS rectifier ground directly to the bolt that holds the rectifier to the chassis. Probably helps a bit. The first (filament) filter cap is also grounded here.

Also moved the isolation transformer a little farther away from the speaker voice coil. It's easy to experiment with this...just clamp the transformer "module" in different spots and test. Didn't have much effect, but it can't hurt.

Don't forget to clean and reseat input jacks, especially if they are grounded directly to the chassis. That's a common source of hum.

3 More Images

I built it as a small self-contained module, mounted on a block of wood.

There are other ways, of course. All the components can be mounted directly on the cabinet itself. The cab plywood is rather thin for this amp, so best to use the wooden block for a base.

Make the module base

A scrap piece of poplar 1x2 was used, cut to a length that easily fit all the components.

Add a fuse holder

The fuse holder is a pretty standard type. It's mounted in a small piece of galvanized metal plate (originally a truss plate.) Metal plate is definitely the best choice for securing this kind of fuse holder device. Thin plywood wouldn't be secure.

A stepped drill bit was used to drill the hole for the fuse holder. Wood screws were used to attach the plate to the base.

Mount the transformer

This is straight-forward. The N68-X transformer is attached with a pair of wood screws.

Make internal connections

Wire the module using the schematic / wiring diagram on Step 7. You can find it below.

Some pointers:

-- The switch and fuse should be on the Hot "mains" wire.
-- When routing the switch wire, avoid the signal path where ever possible.
-- Connect the transformer primary wires as noted. This is US, 120V wiring. Euro wiring will be different (and is explained on Step 7.)
-- I used "wire nuts" to connect the wires, but soldering is more secure. Once I'm satisfied with the setup, I'll replace the nuts with solder, and cover with heat-shrink tubing.

Add some strain-relief for the cord

I used plastic wire channels to fix the cord in place. Electrical cords must have some strain-relief, or flexing will quickly lead to disconnections or shorts.

5 More Images

Ok, now to hook everything up...

Fix the module in place

Yep. That means attaching the module somewhere inside the cabinet.  I used wood screws; whatever is adequate will work. Mounting it some distance from the chassis is fine, and might be advantageous in some circumstances.


Attaching the earth ground (from the three-prong plug & cord)

An important safety feature in any amp is a valid external earth ground. This helps protect the amp (and the player) in a very simple way:

Should any parts fail, or any connections loosen and cause a short circuit, the ground wire provides a "safe" current path, while ensuring that the current flow from a short will also blow the fuse. If the fuse blows, you know there's a problem to fix. And you won't be using potentially dangerous equipment.

The center prong wire from the three-prong cord is the earth ground. In the US, this should be the green wire. Test it anyway, to be sure.

Connect it directly to the chassis. It does not go through the isolation transformer.

Connect the Power switch

Route a two-conductor wire from the switch on the front panel, down to the incoming AC line. Line cord, like the type used in lamps or extension cords works fine. Buy it by the foot at hardware and home improvement stores (Home Depot, Lowes, etc.)

Drill a hole through the chassis if necessary (I did.) Install a rubber grommet in hole, to prevent the wire from rubbing across the chassis, a creating a short circuit.

Route the wire away from the signal path, if possible.

Connect the transformer secondary to the amp

As discussed in the "half-wave" step, there are several way to do so.

But in any case, a double-conductor wire should be connected to the RED secondary wires on the isolation transformer. The wire can then be fed through the chassis using the original power cord entry hole.

Add the solid-state bridge rectifier

This is discussed in depth in Step 8, and schematics are included. Check the photo below for a wiring example.

A bolt-on type of rectifier was used. A new hole was drilled in chassis to accept the mounting bolt.

Once soldered in place, heat-shrink tubing was added.