Power Supply Design

[Rockn]

Is there any definitive simple tounderstand guide to building/designing the power section of a tube amp? I have found bits and pieces here and there and I understand some of it, just wondering about what others have used as a guide.

Thanks

MODS CAN DELETE. I DIDN"T DO A THOROUGH ENOUGH SEARCH

[Chuck H]

Go to the Duncan amps site. You can download a PSU simulator there that is configured around basic tube amp power supplies. Don't forget to install current taps in the right locations for your simulations.

Chuck

[Rockn]

I have tried using PSUD and have not fared very well. I understand how to add RC/LC filters and the different loads, but it still comes out wrong. Unless I am looking at the datasheets for the tubes incorrectly as far as current draw. I have even tried time tested designs like Fender Champs and it didn't come out right. I am sure there is more that goes into the design than just flipping components in and out of the software and would like the ability to take a PT I may have lying around and incorporate it into a new amp.

I have never seen a good tutorial on how to use this software either.

[J M Fahey]

*Very* short rule-of-thumb tutorial.
1) Filaments: make a tube list and add all filament currents ; consider (and wire) 12AX_ types as 2 x 6.3V filaments in parallel. Get the total value.
Specify a 6.3V winding with, say, 20% extra current capacity.
2) +B : you should know the maximum current consumption in mA at full power , or the idle current if single ended Class A; add, say, 20mA for the screens and another 15 or 20 mA for the preamp (say 3 mA per 12AX7); 20mA extra if driving a Reverb with a 12AT7, get a total value.
Specify a High Voltage winding of 0.75x(+B) at that current + 20%, if using a bridge rectifier; or 2x that if full wave 2 diode rectifier á la Twin or Plexi 50W.
Example: +B= 430VDC, 180mA.
If bridge rectifier: 430 x 0.75=322VAC
180mA x 1.2=220 mA.
To play it safe: if you do not get commercially your exact specs, err by going slightly low on voltage and somewhat high on current.
Forget about tube rectifiers, choke input and the like for now.
Read and mentally dissect every schematic that falls under your curious eyes (I do)
Good luck.

[Rockn]

Thanks for the info. I understand the majority of it as this applies for specing out a PT as well as an OT for an amp. The part about dropping the voltages to create the different bias points for each section (pre-amp, gain stage, inverter and power) is not getting there. I have an amp right now that I would like to convert and it seems to have all of the elements I am looking for with for a PP EL84 with 2 x 12AX7's for the preamp and phase inverter.
B+ is 325V
Pin 7 of the EL84 is 310V
Pin 9 of the EL84 is 275V
Pin 1 of the PI is 230V

Didn't measure the preamp voltages at this point.

The power filter section looks like the following attachment. I am assuming the 62K resistor is dropping the voltage for B2, but it seems the 275V should be closer to the 310V from every design I have seen. I would like to convert this to something like a Baby Will and it does not seem that far off.

ExpressSCH.pdf

Thanks for the replies

[Enzo]

If node B were not connectd to any load, then the voltage there would be closer to 310v. But the two power tube screens draw off that node, so their added current causes additional voltage drop across the 3.3k. Ohm's Law.

If no current flowed down the row of B+ resistors, no voltage would drop. Would be the same voltage all the way across.

Look at the posted circuit. Node B is 278v and node C is 263v separated by a 10.7k resistor. Um, who on earth has 10.7k 5 watt resistors? Let5's call it 10k just for ease of calculations here. 278 to 263 is a 15v drop, and that is across a 10k resistor. Ohm's Law tells us that the current through that resistor must be 15/10,000 = .0015 amp, or 1.5milliamp. SInce node C is used by both side of both 12AX7s, then the four triodes would be conducting a total of 1.5ma. SOUnds low to me, but that is how to figure out what voltages you will find at a given node. Add up all the currents coming from the node, then use Ohms law to find the voltage the resistor must then have across it. Or if you know the currents of the tubes and have a target voltage for the node then use Ohm's Law to calculate the resistor value.

And if node C is 263v and the plats of V2 are 230, that means 33v is dropped across the 67k load resistors. Ohms law says 33/67,000 = .0005 amp - 1/2 milliamp


Just because a 10k resistor sits in some Fender and such and such voltages are on either end of it doeswn;t mean ALL 10k B+ resistors will act that way. It all depends upon the current through the node.

[Rockn]

Gotcha! I have been doing some modeling with PSUD and it is all starting to make sense now. Aside from all of the different tube manufacturers laying out there datasheets out differently I should have a handle on this now.

In the schematic I sent, why is there a 62K resistor on the second voltage tap to ground? Was this intended to lower the voltage even further?

Thanks for the helping the newbie.

[tubeswell]

In the schematic I sent, why is there a 62K resistor on the second voltage tap to ground? Was this intended to lower the voltage even further?

Yep it forms a voltage divider with the 3k3 to lower the B+. With that divider, the voltage at the 'knee' of the divider (where the 2nd filter cap node is) would be: 62k / (3k3 + 62k) = ~95% of the voltage at the first (reservoir) filter cap node. The 62k also acts as a bleeder resistor to discharge the filter caps quickly when the amp is switched off. Having said all that 62K seems (to me?) to be an uncharacteristically low value for a bleeder resistor. Most of the ones I've seen are around 220k (which if it were in this amp would make for a 98.5% reduction in voltage - not a big difference).

[Enzo]

Well, there is something darned odd about the whole circuit. Where does someone come up with 10.7k power resistors and 67k plate resistors? And what is that strange RC sticking out from the plate of V1b?

[Chuck H]

And how does that PI work???

Chuck

[J M Fahey]

Well, we can'y analyze this schematic very much, because it is still incomplete.
Some things, as shown, won't work or are useless.
For example, that phase inverter does not invert , there is no volume control between the first and second triodes (don't think it's a Thrash amp), why both a .01 and a 500pF in series with 2M2?, the 46dB loss because of loading it with a 10K pot, and so on.
I think we'd better wait for a corrected/complete schematic.
Meanwhile, , good effort to start with, keep going ahead.
Simply do no try to reason *too much* from the data you have, try to complete (and check) it a little bit more.

[defaced]

Looks like alot of the resistances are weird. Maybe measured in circuit?

[J M Fahey]

Hi Mike, I think that's exactly the case.
I guess some were read from the color codes, other were measured, either in circuit which gives *very* weird results, or even out of circuit, but because of real world tolerances values are not exactly as printed on them.
Somebody was wondering here a couple of days ago about his ".01" caps showing "106.9" and "107.8" on his meter screen; obviously they were ".01069" and ".01078"uF , or only 7% above nominal (excellent) and an even more amazing 1% difference between them (talk about consistency).
Physics can worry about .01% differences, good for them, that's their job, but we in the practical production/Engineering side of the road, can be very happy with 10% tolerance and usually accept 20% on most parts, unless we want to price ourselves out of the market.
Boutique builders, of course, live in a special niche of their own, good for them too.
To each his own.

[Enzo]

And having an idea what you are looking at helps. How often have we heard from someone who grew up on analog meters, and they look at a digital meter trying to measure 10v, and the display says: 10.072, 10.066, 10.049, 10.003, 10.101, 10.051, etc. They freak out because the numbers are always moving, without ever really looking to see that indeed they have pretty darn close to 10v.

[Rockn]

It is a one off add-on for a Hammond amp. I cannot find any info on it or what it was used for and am assuming that it is a reverb amp with the R designation in the model number. The 10.7 resistor is a sand resistor with the value clearly printed on it. It also has some weird sort of riser type tube socket for the preamp with a metal shielding around it. It was hard to find where all of the resistors and caps went inside of that thing since they were all crammed in there so tight so the drawing is more than likely not correct. I was also wondering about the PI because of the way I drew it out and had never seen anything like it. I may go through it again and see if there are any corrections to be made. Most if not all of the resistors are readable so the values are what is on the resistor. I also know better than to read resistors in a circuit with a capacitor in the same connection. I am thinking it would be best just to blow out everything and start from scratch with a known and tested design. I would probably have to end up changing most of the components anyway. I will take some pics and show you how different this thing is, maybe someone here has seen something similar. Thanks for the input guys.