Question about power supply filtering in a D-20 tone cabinet power amp

[normende]

Hi,

I am trying to see what is the best solution for this situation:

I have a D-20 tone cabinet. A tech replaced the 5U4 with silicon diodes, and changed the power transformer to lower the AC voltage to compensate the 5U4 internal voltage drop.

So, in terms of output voltage of the power supply we are ok. BUT, what about ripple?

The output waveform of the residual AC voltage is different in a rectifier tube than in a solid state diode. The diode generates has higher resulting ripple.
This means that the original LC filter is not enough to deliver a smooth DC.

I made some tests, and even with no output tubes on the amp a hum can be heard through the speakers (more noticeable in one of them) only because the field coils in the speakers also get the ripple as they are feeded with the filtered DC.

I can't get another transformer and put back a 5U4... So do any of you experimented with safe modifications to the original power supply when replacing the 5U4 with a solid state diode?

The voltage from the transformer is just what is needed to deliver the right voltage on the 2A3s, I don't have extra voltage that would allow me to add another choke and capacitor to lower the ripple (these chokes have resistance and drop the voltage).

Any ideas?
Thanks in advance for your advice!

[tpappano]

I would add additional capacitance to C12/C12, should improve filtering without affecting voltage. I have to ask, why did the tech make these changes in the first place?

[David Anderson]

I would guess that the original power transformer probably failed. I was given one of these amps years ago, and the B+ secondary winding was shorted. Since this is a choke-input power supply, the B+ secondary winding voltage has to be higher than it would be in a capacitor-input power supply, and that would limit choices in a replacement. Note the 1,050 VAC spec. When you combine that with 1930s magnet wire insulation . . .

[Alek Z]

I got one recently with my CV, when I took the amp out of the D-20 I noticed a large glob of wax that had leaked out of the power transformer. Given the age of those amps I can imagine it would be common for them to fail. Thing is also massive, one of the heavier tube amps I've dealt with, has several large transformers.

[David Anderson]

While a silicon diode can sustain higher peak currents than a vacuum tube rectifier, that doesn't necessarily mean that you will see higher peak currents. Much of that is determined by the other components in the circuit such as series losses in the power transformer and the input choke. With this being a choke-input power supply, ripple will be greatly reduced by that choke. In fact, putting more capacitance after the first choke will actually increase peak charging currents because you are lowering the impedance of the circuit.

I've written extensively about these amps on this forum before. In the 1930s when they were built, electrolytic capacitors were fairly unreliable, so Laurens Hammond and his engineers decided to use oil-block capacitors, which were much more reliable, but also much larger. Thus, they were limited in terms of how much capacitance they could put in the power supply. There is a good deal of AC ripple on the power supply in these amps. There always was. It was never even close to "smooth DC" even with a tube rectifier brand new at the Hammond factory. They rely on the fact that it's a fully balanced amp to cancel the ripple. The closer each "side" of the amp is balanced to the other side (tubes, resistors), the quieter it will be. At the time, the results were considered acceptable.

With modern capacitors, we can add more capacitance to the power supply, but there is one issue to consider. When the speaker field-coils are powered, there is a good deal of energy stored in them as inductors. When you remove the B+ voltage, that energy has to go somewhere, and the speaker fields can actually briefly flip the polarity of the power supply. Thus, if you used electrolytic capacitors, you might want to add a "flyback diode" to prevent power supply polarity reversal on turn-off and protect capacitors. (Electrolytic capacitors are damaged by reversed voltage on the leads.)

Alternatively, since we do now have large-value metallized film capacitors in smaller packages, you could add some of those under the chassis.

You really should not run this amp without output tubes because that puts all the B+ current across the 5k field-coil, which is normally in parallel with the four 2A3s. The 5k field-coils are the most likely to burn out due to the small diameter of the wire. Treat them gently. Trying to find anyone to rewind them is almost impossible.

There is also the possibility of power transformer / output transformer interaction. Amps are usually laid out to try to minimize field interaction between the PT and OPT, but since your PT is not original, we can't know whether or not it has the same field orientation.

These amps were never completely free of hum. How many millivolts AC hum are you measuring at the speaker voice coils AND with the inputs shorted such that you aren't hearing hum from an organ preamplifier?

[muckelroy]

You really should not run this amp without output tubes because that puts all the B+ current across the 5k field-coil, which is normally in parallel with the four 2A3s. The 5k field-coils are the most likely to burn out due to the small diameter of the wire. Treat them gently. Trying to find anyone to rewind them is almost impossible.

Tell me about it. I found a guy in the antique radio community who re-wound one of these 5k field coils for me. It sits in a box as a spare unit.
The guy who did that admitted to me after finishing it that he will charge a good bit more the next time some one wants one re-wound. It took a lot more raw material and labor than he originally estimated. It's something like 2 miles (or maybe more) of very fine wire. Not a job for the shade tree pickup re-winder.

[normende]

Thanks a lot for all your very useful and interesting answers, bringing light into my gaps of knowledge... :)

Thanks for the warning about the 5000ohm field coil. No more tests without the 2A3s.
Honestly I don't know the story of this amp and why they replaced the 5U4...

So, a couple of questions:

About capacitance (because I am not sure if I understood it correctly), would it be useful and safe to use for example 100uF caps instead of the 6uF original caps? Will it get things to go better or worse?

About the "flyback diode" (and sorry for the silly question), where am I supposed to put it? In series with each of the field coils? (flyback diode means a fast diode used in H-out TVs?)

I haven't measured the ripple voltage yet, for the moment I had to move the amp and speakers out of the working room but I will do it as soon as possible.

I have replaced all resistors and caps and the 2A3s are NOS and paired. In that sense cancellation should be favoured, but anyway there was a noticeable "base" ripple generated by the field coils so I know I should try to do something with it.

Thanks again.

[David Anderson]

If you look in the Hammond Service Manual -- which it seems not that many people do anymore -- you will find that the first version of this amp (which doesn't even have a model number) used electrolytic capacitors with values of 30uF. That's a typical value for a tube power amp, especially a very low-gain amp like this one. Using huge capacitors will greatly increase inrush current at turn-on, and there's no need for massive filtering here. It's not a mic preamp. There's no hard and fast rule for capacitor selection. You can use 22uF in the first stage and 47uF in the second & third -- or use 33uF for all of them. The second version of the amp (which you have) adds an additional choke for better filtering with the 6uF filters.

If you google "flyback diode," you will find tons of explanations of how they work and what their purpose it. They should be in parallel with the power supply capacitors wired normally off, band towards positive voltage. Thus, if the inductor attempts to flip the power supply polarity, the diode will turn on, limiting the reverse voltage to ~0.6V. I won't pretend to be an expert in flyback diodes. Maybe someone else in here knows more than I do.

The original oil-block capacitors are non-polar, so they are not affected by any polarity reversal. Electrolytic capacitors are damaged by it. This may have been one reason that Hammond decided to use the oil-block capacitors. In the 1930s, there were no silicon diodes available to use for this purpose. 85 years later, we have more options.

I learned this when I built a stand-alone power supply for a field-coil speaker and observed that it could briefly flip the power supply polarity when turned off.

And back to your original point about tube rectifiers, it is true that Si rectifiers switch off in a way that vacuum tubes don't. However, you can put snubbers on silicon diodes to slow them down and damp any switching noise -- if you're actually having a problem with it.

Finally, remember that this is a choke-input power supply. You do not attach a filter capacitor directly to the rectifier tube or the silicon rectifiers. That would send the B+ voltage much, much higher, destroying your 2A3s and 5k speaker field coil.

[normende]

David, thanks a lot.
You are a real expert and I will carefully follow your advices.

[Jyvoipabo]

Hello,

I guess the technician replaced the 5U4 with 2 x 1N4007.
But they are not the same components.
The 1N4007 has an abrupt U(I) diagram that results in shorter switching times when crossing On/Off states and therefore much higher di/dt. That is a possible source of noisy voltage.

I don't want to bore you with formulas, but manufacturers quite often add ceramic capacitors (~ 47nF 1000V) paralell of each of the diodes to filter this.

If you want, you can try ..

JP

[David Anderson]

I don't want to bore you with formulas, but manufacturers quite often add ceramic capacitors (~ 47nF 1000V) paralell of each of the diodes to filter this.

The term may not translate, but that's why I mentioned "snubbers" above. You find them in some modern Fender amps, little green ceramic disc capacitors.