Power Supply Design - Calculating and Simulating Circuit Load

[Zarrir]

When you guys are designing a power supply for a specific amplifier, how do you calculate the aproximate load of the entire circuit based on the schematic? And how do you simulate and connect a dummy load to the power supply to record the voltage drop? Sometimes the schematics provide very little information, specially the old ones.

[pompeiisneaks]

Basically look up the tube complement, and look at the expected current requirements for each and do the math, then make sure the main power winding can handle that, and do the same for heater current.  One of the best explanations I've seen of this was the valve wizard's : http://www.valvewizard.co.uk/smoothing.html he talks about power supply smoothing and filtering here, but he's got a graph at the bottom taht shows the power current requirement in a nice visual. 

As for load, I'm not sure, that's somewhat dynamic based upon the tubes themselves.  Again, I'd guess you could look at the expected plate loads listed in the charts for the voltage range, but that's variable with a tube...

Maybe someone else has tips on that part.

~Phil

[HotBluePlates]

When you guys are designing a power supply for a specific amplifier, how do you calculate the aproximate load of the entire circuit based on the schematic? ...

Hopefully you noticed from the Valve Wizard link pompeiisneaks provided that the output stage represents the lion's share of the load on the power supply.  So usually, you'd figure out how much current the output tubes draw at full output, add a milliamp for each preamp tube stage (which might be overstating their draw a bit), and call it done.

You could get an answer with more precision with more math, but it probably wouldn't change your end result enough to dictate a change of output transformer or rectifier.

[Zarrir]

Thank you guys, excellent link pompeiisneaks.  So after making that math i could roughly pick a dummy load resistor of that wattage, stick to the end of the filtering stage and start messing with the power transformer HV values till i get the B+ that i need?

[shooter]

after making that math i could roughly pick a dummy load resistor of that wattage,

Yup, I've done it a couple times, IIRC I used something in the 5K range.  If all is well with rail V-drops and stability, I would do the "tweaks" with tubes vs the dummy load, the tubes are gonna fluctuate till they find their happy place.

[HotBluePlates]

... i could roughly pick a dummy load resistor of that wattage ...

Well, a dummy load resistor of 2-3 times the expected wattage, unless you want to get the resistor so hot it'll burn/melt any surface you put it on.

But yes, you could do that.

[Zarrir]

Awesome!

[PRR]

Power stage is most of the load.

Power stage "acts like" a resistor to the DC supply.

A happy SE stage acts-like roughly the OT primary impedance. Use 7K resistor for Champ with 7K OT.

A push-pull stage at full output acts-like *roughly* half of the OT primary impedance(*). So a Bassman with 4KCT OT at full output is like a 2K load on the DC supply. Fixed-bias stages may idle cold. Taking 30mA/tube at 400V, 400V/0.060A is a 6+K load on the DC supply. You may want to know both conditions. WRONG, see below.

Each Fendery 12AX7+100K stage can be approximated by a 300K resistor. Compared to 7K-2K in the power stage, all of these together add up to "negligible". 5F6a is 50K smalls, 2K large, 1.92K total, bah, this isn't a Mars-rocket. An absurd 12-stage distortion chain in a Champ is 25K for little tubes, 7K big tube, 5.4K total, just enough to be worth adjusting the test.

Power dissipated will be at least double the amplifier expected audio output. 3X is safer.

Power rating: why are you testing? As a first-check, I would run only long enough to read the meter. I would not be afraid to use exact-Watts parts. Doing output testing, I have run 23W for many minutes in a "20W" resistor. It stinks, the ink fades, although I could not quickly ignite a Kleenex on it. OTOH on another job I have run load-tests for many hours (I hate fragile amps). Run AT rated wattage, a resistor "should" be good for hundreds of hours. But if you have to buy resistors for long or multiple tests it makes more sense to over-buy for many years of fault-free testing.

[sluckey]

I built this dummy load from junk box parts. All three resistors are 1500Ω/30W. I can get several different loads by using gator clip jumpers. With no jumpers you can load a 450V supply to 100mA. I've only used it a few times but it comes in handy when you need to test a typical guitar amp supply.

[jjasilli]

This topic is interesting & informative.  But as a practical matter, I'm not sure I see the point or need to actually dummy load the power supply.  Per Jack Darr's "Electric Guitar Amplifier Handbook", expect a 10 - 15% difference in PS voltage loaded vs. unloaded.


Can someone explain further?  E.g., sluckey, how has your dummy load helped you?

[pompeiisneaks]

My first thought would be to test if it's working well without risking a whole amp... If it's giving off bad voltages or gets overly hot itself, then it can't handle the expected load... ?  but am interested myself, great question :)

~Phil

[sluckey]

E.g., sluckey, how has your dummy load helped you?

Remember all those Hammond organ conversions I've done? I acquired those amps from eBay over a 2 or 3 year period. When I got those amps I would test the unloaded PT voltage as part of my evaluation. If the secondary voltages were as expected then I'd plug in a SS rectifier and slowly bring up the mains voltage on a variac to verify filter caps. Then I would use the dummy load to see if the PT could supply the current I needed for a project.

Most of my conversion projects went through this eval. And the power supplies were left running for at least 24 hours. This was just my way of verifying that the 50 year old iron was still up to the task. I've never used that load on new transformers.

[PRR]

> I'm not sure I see the point or need to actually dummy load the power supply.

Another thread another place, I was helping a guy reporting WAY low voltages when connected to circuit. I could not see if the supply was weak or the circuit sucked. Dummy loading the supply with a "known" load instead of a mystery circuit might have answered that. (But another reader noted that his PT was about 1/4 the size it needed to be....)

Yes, you can throw stuff together and if it "plays", ship it.

OTOH engines for important applications (aircraft, racing, fire pumps) are routinely connected to brakes or dynamometers (dummy loading) and run a while, light and full-power, before installation and flight/race/fire starts. Many failures happen early in life. Weeding-out while still near the bench avoids embarrassment in real life.

In an audio (guitar) amp, a later step will be to dummy-load the speaker output and beat audio through the amp until you are sure most early-failures won't happen.

[jjasilli]

In summary: You can calculate a load, then match it to the specs of a PT, assuming you know the specs, per pompeiisneaks' reference to the Merlin (Valve Wizard) site. If you now the amp a used PT is pulled from, you can calculate the load it was supporting.  With a mystery PT, I guess a dummy load test would be useful.  Or as sluckey says, to verify that an old PT is still up to spec.

[Zarrir]

Does PRR stand for "Prrofessor"?

[Willabe]

Does PRR stand for "Prrofessor"?

Might as well.   

[PRR]

> acts-like *roughly* half

Gentlebeings--

I was wrong.

I knew it right away, because I "know" the factor for a transistor amp is about 6. Actually 2*Pi, 6.28.

Transposing from totem-pole to CT transformer gives a factor of 2^2*2^2 or 16. So the right answer should be 2Pi/16, 0.393.

Not half.

However it is also wrong for all Class A cases.

As there are several cases, I made a picture. Attached.

This is "ideal". Device pulls to zero voltage drop and has no side-current. Transistors can come close enough that my 6X is seen in practice. Tubes will have significant voltage drop, and often non-negligible side-current in screens and drivers, so the numbers will differ, but not tonight.

The figures for Class B are for FULL output. Ideal Class B goes to infinite RDC (zero current) at no-signal.

[pompeiisneaks]

I was wrong.

All, we've seen history made... PRR was wrong! ;)

that's a pretty rare thing indeed!

~Phil