Causes of inadequate output power

[Matt C.]

I've been working on repairing a couple amps lately that both have outputs that are too quiet, and I'm trying to sort out the possible causes of an issue like this.  I know aging power tubes can develop weak output and that might be all it is, but I'm curious to know if there's a way to confirm this without just dropping in a whole new set and measuring again.

The amps are a Sovtek MIG 100-H (quad of 5881), and a Music Man HD130 (quad of EL34).  Both amps are putting out only around 50-60 watts.

My method for measuring output power has been to plug the amp into a dummy load, input a 1kHz sine wave, turn it up as much as possible without significant clipping, read output voltage from the scope or DMM, and calculate power from there.

One thing I'm curious about is whether the biasing of the output stage affects the total output power.  It seems like acceptable bias ranges from about 50%-70% of max dissipation, and I'm wondering if changes within that range are reflected in the total power output of the amp.

I'm also guessing that I could measure the maximum size of the signal on the grids of the power tubes, and use the tube charts to figure out if that size signal should be able to push the power tubes into "full" power output, but I am not quite that fluent at reading tube charts.  I guess this would just tell me if the problem is in fact in the output section, or if it's somewhere upstream.

Another wrinkle with the Sovtek amp is that the output transformer was replaced at some point.  I'm suspicious that it might not be a correct match for the circuit, but can't confirm because I don't have any specs for the original.  My understanding of Transformers-101 says that it shouldn't be the culprit since input power roughly equals output power, but I suppose if the transformer was presenting an incorrect impedance to the output tubes it could throw things out of whack.

Thanks for reading, this is my first post and I'm glad to have found a good forum for this stuff.  Any help or ideas would be appreciated!

[kagliostro]

50% seems way too conservative to me

I would like to set the bias 70-75%

K

[shooter]

Tubes would be my best guess also, but if  you don't have any on hand;

So when you measure power, where is it compared to "rated"?  like 1/2?  btw, I use the same method.

DC volts on all tubes compared to schematic is a good place to look for odd things

your PA drive is basically your bias, if bias is -30 and you're only driving with 20vac then the power tubes are probably just yawning.  The datasheets usually spec G1 Vac max.  I like to have about twice that available, just in case.

Does the amp work well besides lack of power, good sound etc.?

[Matt C.]

your PA drive is basically your bias, if bias is -30 and you're only driving with 20vac then the power tubes are probably just yawning. 

meaning if I have 4 output tubes all biased to (for example) 20 watts, then that 20 watts also represents the maximum signal output of each tube?  so I'd get a max output of 80 watts?

Yeah right now the actual outputs are roughly half (or a little more) of rated output.

The amps sound good and work well otherwise.  The schematic for the Sovtek does not have DC reference voltages on it but nothing seemed way out of whack.

[shooter]

Your bias, neg, is how much signal you need to get the grid to "near zero" (max plate current) , but until it gets "close" you're not overdriving the PA tubes, so there's "more" left.  If that makes sense.  so If you bias cold, further from zero, then the tubes aren't pushed, assuming you didn't change the drive, but if you bias hot, closer to zero, then the same drive gets it closer to max.  Don't quote me though, I have dyslexia 

[PRR]

> whether the biasing of the output stage affects the total output power.

Insignificant. (For push-pull fixed-bias.)

Try it yourself. Bias-down as far as you can, say 5mA. Around 1 Watt out you may get horrid wave-shapes as each tube "wakes up" from its "too cold" bias. But up near full power the peaks will be fine. And you will get very nearly the same maximum unclipped power whether biased cold or hot.

OTOH----

> biased to (for example) 20 watts, then that 20 watts also represents the maximum signal output of each tube?

This is true for *class A* amps. Input power is constant. Signal diverts power from tube to load. Efficiency (on sine wave) is 50% at best. One 6V6 at 12 Watts DC can make about 5.7 Watts of clean sine output. Two EL84 P-P with cathode resistor bias makes about 18 Watts (violates the 50% efficiency because we often let them run hotter than 12W/plate, push-pull keeps distortion lower, bias-shift happens, and the EL84 does not hard-clip so "18 Watts" may be significant rounding).

However quad 5881/EL34 amps are most-often "fixed" bias, idle cool, and the exact idle does not affect maximum output.

> output transformer was replaced

Loading is critical to getting rated power. I have a 12 Volt tractor and I put a 12 Ohm lamp on it. I get 12 Watts of power drain. Now I replace the bulb with a 24 Ohm job, I only get 6 Watts. The tractor will hold 12V despite a few lamps; tubes are not so solid. Too-low loads also reduce power.

As a rough guide, quad big tubes run about 2K plate-to-plate; however unusual plate or screen voltages will shift this optimum. Post your voltages and expected power output.

There is not a ton of difference between 100W and 50W. If the tubes are old and abused, this may be your entire problem.

> measure the maximum size of the signal on the grids

If you have a DC-coupled 'scope, put the trace on the top line, poke the grid. Adjust gain so the grid's negative DC bias voltage puts the trace mid-screen. Apply signal until lightly clipped at the output. If the peak grid voltage almost touches the top line, and is otherwise clean and symmetric, the driver is doing fine. If it has to touch the top line to get "maximum" at the output, the power tube has no margin and is weak.

If the amplifier has NFB, as you exceed clipping you will get very wonky wave shapes. The NFB is trying to "force" the power stage to give more than it can give. Interesting, but not relevant if diagnosing low power output.