Real World vs Theoretical Output Power

[Alexo]

Howdy folks, how many of you are in the habit of reading your output power with a scope, a dummy load, signal generator and all that?

Do you calculate it on your load line?

I like numbers, and I do both. The discrepancies are baffling. Am I alone in this? Here are some examples:

Push-pull 6550 amp, 568 Va, 312 Vg2 5K load:
GE datasheet says this should be close to 100 watts
Load line analysis using P = (HT-Vmin) * Ipeak / 2 (thank you, Merlin) says 95 watts
Measurement via scope, load, test signal yields almost 20 volts AC RMS, squared = 400, divide by 8 and you get 50 watts clean, at MOST, maybe 70 watts hard clipped. Double-checked with a true RMS meter, various test frequencies. SS rectified plate and screen supplies with minimal sag on each. Tried KT88's and got up to 60 watts, still pretty low! Pt rated at 465ma, OT at 100W.


6V6's @425V, 8K load, 400 screens,
Load line, common sense and datasheet extrapolation say we should get around 25 watts before clipping (think beefed up Deluxe).
Analysis of actual amp yields 15 watts clean, 30 when clipping into total square wave territory. 150ma PT, beefy OT.

5 watt 6AQ5 amp, built exactly to datasheet 5 watt specs
2.75 watts clean, close to 6 when slammed

PP Cathode biased EL34 amp built exactly to Mullard specs for a clean 40 watt cat bias amp: 425 plates & screens, 6K6 load, (load line analysis predicts more like 50W) yields 25 watts clean, 35-40 when slammed. 200 ma PT, JTM45 style OT.

..and on and on.

All of these amps are ss rectified, except for the EL34 one, the driving stages are powerful enough to push the output stage to cutoff before they saturate, screen and plate voltages hold, bias is in check, yet I am very reliably getting almost exactly half the expected output power!

I am using (RMS voltage squared)/load - are these just the inefficiencies of real life? Losing 50 watts across them seems rather drastic.

[Alexo]

...working backwards on my 6550 amp:

5K to 8 ohms gives an impedance ratio of 625:1.

sqrt (625) = 25

20 volts rms on the output x 25 = 500 volt RMS swing on the primary.

x 1/.3535 = 1,414 volt peak to peak swing - exactly what I see on the load line! The same load line that says we should get 95 watts.

Are the datasheet wattage ratings peak to peak, maybe? Don't most people talk about watts in RMS?

[Alexo]

edit: double post!

[Structo]

Yeah, that was always the problem with the audio industry.
They wanted big numbers to sell the amps.
Remember when the home stereos were advertised having 200 watts per channel!

Well what they didn't say was that was peak to peak or worse, it was in music power, what ever that is.
Was it before clipping or after a certain amount?
Or was it RMS which everybody really wants to see.

I think you only getting 5 watts between the measured and the load line is pretty good really.

[Jackie Treehorn]

Yes, I just started doing this measurement. On my Rocket/AC30 type build, 4 EL84's, appr 300volts on the plates and screens, 47 ohm bias resistor, 4.3k load, it measured 28 watts clean and 32 watts at full distortion.

I've got a fixed bias 2xKT88 build, 400 volts on the plate, 380, I think, on the screens, which did 35 watts clean and around 62 watts at full clip on a very tired pair of KT88's.

On your 6550 build, did you do a separate supply for the screens? Edit: Nevermind, reread your post and it looks like you did!

[VacuumVoodoo]

Check your dummy load. Is it 4 or 8 Ohm? Your math checks out.
There is no such thing as Watt RMS. It's just Watt.

[passfan]

Check your dummy load. Is it 4 or 8 Ohm? Your math checks out.
There is no such thing as Watt RMS. It's just Watt.

I think this problem stems from modern manufacturers using bs specs like "total music power" If you do the math it's simple what their doing , divide by 2 then multiply by .707 and you have a "closer to real life" spec for the junk that their selling. Of course "watts rms" is used by some manufacturers to let the consumer know the spec was calculated using the usable available voltage and current in the equation for wattage.

[jjman]

I'm still learning load lines but it seems to me that both axes are "instantaneous" figures. So the "P" power (watts) line would be instantaneous/peak watts and would not be RMS.

RMS is sort of an "average" figure over the time of the full wave. M=mean=average.

[Alexo]

Thanks guys, Here is the root of the issue: I can't reconcile Rms squared over load with the valvewizard formula (P = (HT-Vmin) * Ipeak / 2)

Below is the combined AB load line for my 6550's.

You'll see that the total voltage swing for one tube is about 620 volts.

Multiply this by 2 and you have an overall voltage swing of 1240 volts. Converted to RMS = 438. Apply the old RMS squared/load ((438*438)/5,000) and you get about 40 watts.

Bench measurements confirm that this is more or less what is actually happening (I've got it swinging about 1400 volts for 50 watts, but close enough).

But then plugging into the valvewizard formula, P = (HT-Vmin) * Ipeak / 2:

P = (566-55) * .42 / 2 = 107 watts, which is also what the datasheet gives for running at these specs in its example circuit.

Watts up with that?

[VacuumVoodoo]

Below is the combined AB load line for my 6550's.

You'll see that the total voltage swing for one tube is about 620 volts.

Multiply this by 2 and you have an overall voltage swing of 1240 volts. Converted to RMS = 438. Apply the old RMS squared/load ((438*438)/5,000) and you get about 40 watts.

In push-pull each tube when fully conducting sees a load = Raa/4, here 1250 Ohm.
This will give you a theoretical 153W but looking at your real world peak swing 560-55V you will get (505/1.4)^2/1250=102W.

So data sheetzs and approximative simplified maths give same result while your measurement is off by more than 50%.

Assume you have a 4 Ohm dummy load on 8 Ohm tap of the OT.

OT impedance ratio is 625
4 Ohm reflects to Raa=625x4=2500 but each tube sees only 1/4 of that so Ra=625 Ohm
Assume same peak plate swing 505V

505 V over 625 Ohm load would require peak current swing 505/Ra=505/625=0.8A!! You can only get ca 0.4 Ampere because that's what the 6550 is capable of and I don't think your PSU can source it without sagging quite significantly..
It's what you are getting and it produces ca [(I^2xRa)/2]xRa= [(0.4^2)/2]x625=50W.

Your measurements tell me your dummy load is 4 Ohm not 8 or you're making some other systematic error in your measurements.
Or your OT is not what you think it is.

[jaysg]

Since I love to quibble, Vrms related to a load, results in 'Watts Average'..or averge watts, which mostly goes up as heat. The same logic tells you that Vpeak into a load yields peak watts.

I do use an HP function generator, a dummy load, and a scope to evaluate power. Different tubes yield different output power, even when they seem about the same audibly.

[Alexo]

Thanks again for the suggestions, guys.

But I'm reading 500 V RMS between the plates and 19 V RMS cross the secondary, very close to (but just under) clipping at 200hz..

500 / 19 = 26.3

26.3 x 26.3 = 692

692 impedance ratio x 8 ohms = 5K5.

..and it is an 8 ohm dummy, at least according to the same meters that tell me everything else (and agree with my scope on the voltages).

I understand that some power is lost as heat, but if it was 50W worth, I'd be pulling out the fire extinguisher!


...according to my 1 ohm bias resistors, we're also getting about 110mv RMS per tube, under load, aka 300ma peak when you subtract screen current. about 100ma under what the load line predicts. Yurgh.. Screens still holding strong at 310V and B+ at 562. Tubes are out of balance by about 15ma RMS at this output level.

[VacuumVoodoo]

Is your voltmeter a true RMS measuring one?- anything else will give false reading over 1 Ohm current monitor resistor as the the current is not sine but half wave rectified in each tube.

If you measure 500Vrms between plates and your OT really reflects 5k5 to primary side then there's 100mArms flowing into T's primary center tap at 50W output, you'll need 200mA rms into CT for 100W.

Big question: how much power can your PT deliver? everything else being OK your HV PSU must be capable of sourcing at least 200mA rms (ca 300mA peak) at 560V.

[Alexo]

Thanks again, Vacuum Voodoo, the PT is a great big Hammond spec'd for 465ma.

I've checked out the B+ under these conditions and there is virtually no sag.

Not sure honestly of the spec's on my voltmeter but the AC readings it gives always work out to .35 of the peak-to-peak readings on my scope (for sine wave, more for square), so I trust that it is good for RMS.

I'm starting to think that the tubes just don't quite behave in real life as they do on the datasheets. Maybe the 0V cutoff point is happening earlier than the load line predicts? Maybe the tubes are just tired, although KT88's gave similar results...