I don't get the expected power output.

[jazbo8]

Agreed about the increased current and shorter tube life, but I have never seen a tube amp that has increased output power when running into a mismatched load. Whichever setting gives the highest power output is the correct match.

You are correct, the perfect match should provide the highest output power, but in this case, perhaps the load was not correctly matched to begin with? But as mentioned, the tests were done under non-ideal conditions, so the readings could very well be off anyhow...

[Alan0354]

Keep it coming. I have not responded because I need to find a resistor load, also I have some other observation that is puzzling. I tested two pairs of 6V6. One is cheap Chinese 6V6GT by Weber and the other is JJ 6V6S. You would expect JJ has higher power under my condition of +B=420V and 25mA per tube. With just the THD 8 ohm, I got 37Vpp clipping from the cheap Chinese Weber, I only got 30Vpp from JJ!!!! I made sure the condition were identical.....+B=420, I=25mA, using 4ohm tap with feedback from 4 ohm tap......Literally swap tubes and adjust bias and test.

JJ is a well recommended tube, it can take higher +B and higher wattage, why the power is so much lower? From calculation, the cheap Weber gives 21.4W, the JJ only gives 14W!!! I have to dig up another pair of 6V6 to try.

Also, I am going to experiment using 12X7 for PI tube instead of 12AT7. 12AX7 has higher gain to make sure the power amp is not running out of forward gain. If the gain of JJ is lower, I might still not have enough drive to really clip the power amp.

[jazbo8]

That is odd, what is the screen voltage used for the tests? And the primary impedance?

[Alan0354]

That is odd, what is the screen voltage used for the tests? And the primary impedance?

The OT is Vibrolux/Bandmaster 4K:16/8/4 by Classic Tonehttp://www.classictone.net/40-18006.html. I put 8 ohm THD on 4 ohm tap to cheat. This is to make the primary reflecting 8K to the 6V6.

The screen voltage is about 3V below the voltage at the center tap of the primary of the OT. The center tap measured +420V, the screen is about 417V. I have a 1K 5W screen resistor. All conditions are identical.

I literally swap the tubes, only adjust the bias and test. Nothing else is changed.

[jazbo8]

Did some quick calculations, and JJ's 6V6S on paper should give you slightly more output, but not by much. However, the estimated output power with the current operating conditions should be well over the 20W that you are getting. Assuming no PS sag and the grids are actually driven to 0V, Po should be more than 30W, so something is not quite right... For starters, you should get a big dummy load and use a signal generator on the input in order to get more accurate readings.

[Alan0354]

Did some quick calculations, and JJ's 6V6S on paper should give you slightly more output, but not by much. However, the estimated output power with the current operating conditions should be well over the 20W that you are getting. Assuming no PS sag and the grids are actually driven to 0V, Po should be more than 30W, so something is not quite right... For starters, you should get a big dummy load and use a signal generator on the input in order to get more accurate readings.

Yes, I will try the generator next.

OK here are some results:

1) I tried a pair of NOS Ruby 6V6GT, result is the same as the cheap Weber, I got 37Vpp output to load which equated to 21 to 22W. So it's the JJ only.

2) I change to a new 12AX7 for the PI instead of 12AT7 when using the JJ, no difference, this means the internal gain of the power amp is not a problem( with NFB of cause).

3) I measured the swing of the plate of JJ, it is asymmetrical, it clipped between +570V and +220V with idling at 420V.

4) I measured the Weber tube, the plate clippe3d between +620 and +170 with idling at 420. So you can see the JJ swing a lot less.

Please keep giving me opinions, I am going to do resistor load and generator later today or tonight. But I don't think the THD load make a difference in this comparison test as this become relative, not measuring absolute power. The JJ is 6W lower RELATIVELY in every case.

BTW, I notice that idle current has no effect on the clipping amplitude at all. So I quit adjusting the bias. I can go from 30mA down to 20mA with no noticeable change the clipping amplitude.

[nickb]

Did some quick calculations, and JJ's 6V6S on paper should give you slightly more output, but not by much. However, the estimated output power with the current operating conditions should be well over the 20W that you are getting. Assuming no PS sag and the grids are actually driven to 0V, Po should be more than 30W, so something is not quite right... For starters, you should get a big dummy load and use a signal generator on the input in order to get more accurate readings.

Yes, exactly. Lack of standardization is a big issue here. In that vein, the onset of clipping is just being eyeballed. Once you have the signal generator and resistive load in place then if you have an FFT on your scope you can choose a specified level of 3rd harmonic to define onset on clipping. Or a use THD meter if you have one.

[g1]

As the JJ's specs are somewhat different than a "real" 6V6, is it possible the impedance is also different?

[Alan0354]

OK, now with a 10ohm 100W resistor ( I don't have an 8 ohm resistor) and signal generator at 1KHz sine wave.

1) I got 42Vpp clip from the cheap Weber, so Vrms= (42/2) X 0.707= 14.85V. W=V^2/R= (14.85)^2/10=22W.

2) I got 36Vpp clip from the JJ. so Vrms=12.73V. W= 16.2W.

I am working at home, I don't have distortion meter or spectrum analyzer.


Edit:

I parallel a 50 ohm with the 10 ohm to get 8.33ohm load.

1) I got 36Vpp from the Weber so Vrms=12.73, W = 19.44W.

2) I got 29Vpp from the JJ so Vrms=10.25V, W= 12.61W.


Ha ha, looks like both 6V6 like 10 ohm load better.

[Alan0354]

Now this is even funnier. For the fun of it, I plug the 8.33 ohm to the 8 ohm tap. The feedback still on the 4 ohm tap.

1) I got 40Vpp from the Weber so Vrms=14.14, W = 24W.

2) I got 36Vpp from the JJ so Vrms=12.27V, W= 19.44W.

[J M Fahey]

So any way you measure it, the "cheap Weber" is actually better than the ¿expensive? JJ ?
Why doesn´t that surprise me ?

[Leo_Gnardo]

Yes, exactly. Lack of standardization is a big issue here. In that vein, the onset of clipping is just being eyeballed. Once you have the signal generator and resistive load in place then if you have an FFT on your scope you can choose a specified level of 3rd harmonic to define onset on clipping. Or a use THD meter if you have one.

FFT, Fast Fourier Transform built into a scope, now that's way too modern. What are you trying to do, bring us up to the 1980's?

Bejeebers I'd LOVE to have a scope that had FFT! Point me to one please! Even if I can't afford, can always dream about one. I'm using a 1960 Hewlett Packard that was scrapped out from the US Air Force. Ol' reliable but no fancy features beyond dual trace. Good enough for audio I always thought. I can eyeball the clip just fine. Still in use every day. Back from the day when HP actually made excellent gear and before Carly Fiorina turned it into a printer ink sales company. Feh, don't make me think of it.... and she wanted to be elected what, senator, governor? Oh please no . . . . . .

More to the point, yes a resistor load is a big jump to standardization. With a speaker impedance jumping all over the place it's easy to "assume" the speaker is 8 ohms or whatever, but measure at some point where it's way over its nominal impedance and you calculate a power figure that's way inflated. That's what some of the other techs in my area do. Their customers are always impressed at all the "extra" power they get out of amps = baloney. Also consider the driving tubes have some effect via ACTUAL plate vs primary impedances so the experimenter doesn't know what he has going on when swapping from type of tube to the other.

With more power at 10 rather than 8.33 ohms, looks like the amp 'might' be happier driving a 16 ohm load. With a set of R loads, you could determine the best power transfer, and there's your ideal impedance match.

[Alan0354]

FFT, Fast Fourier Transform built into a scope, now that's way too modern. What are you trying to do, bring us up to the 1980's?

Bejeebers I'd LOVE to have a scope that had FFT! Point me to one please! Even if I can't afford, can always dream about one. I'm using a 1960 Hewlett Packard that was scrapped out from the US Air Force. Ol' reliable but no fancy features beyond dual trace. Good enough for audio I always thought. I can eyeball the clip just fine. Still in use every day. Back from the day when HP actually made excellent gear and before Carly Fiorina turned it into a printer ink sales company. Feh, don't make me think of it.... and she wanted to be elected what, senator, governor? Oh please no . . . . . .

More to the point, yes a resistor load is a big jump to standardization. With a speaker impedance jumping all over the place it's easy to "assume" the speaker is 8 ohms or whatever, but measure at some point where it's way over its nominal impedance and you calculate a power figure that's way inflated. That's what some of the other techs in my area do. Their customers are always impressed at all the "extra" power they get out of amps = baloney. Also consider the driving tubes have some effect via ACTUAL plate vs primary impedances so the experimenter doesn't know what he has going on when swapping from type of tube to the other.

With more power at 10 rather than 8.33 ohms, looks like the amp 'might' be happier driving a 16 ohm load. With a set of R loads, you could determine the best power transfer, and there's your ideal impedance match.

Ha ha, I am one up over you!!! I have the Tektronix 465......from the late 70s, 10 years newer!!!! And I have a function generator I got from the surplus store for $100. But I think both are plenty for guitar amp. To be honest, I don't even use these often. I had to pull them out from the closet yesterday when you guys here insist on using them.

Now a days, scopes are really cheap, I saw in Fry's they have some digital scope smaller, fancier and faster than mine for less than $500.

Yes, I think so, I put the 8.33 ohm on the 8 ohm tap, this is like 16 ohm on the correct transformer.

I had the fanciest scopes, spectrum analyzers, network analyzers at work back in the days. But in this guitar amp, or even audiophile amps, I really don't think you need any of those. Low distortion has nothing to do with good sounding.

[Alan0354]

So any way you measure it, the "cheap Weber" is actually better than the ¿expensive? JJ ?
Why doesn´t that surprise me ?

Yes, however which way I tested, the Cheap Weber tubes kick the butt of the JJ!!! Even the pair of junk Ruby tubes get higher power than the JJ. But notice when I put the load on the 8 ohm tap ( equal to 16 ohm for normal OT), the JJ is getting closer......19.44W vs 24W of the cheap tubes.

[jazbo8]

Now that you have all your old test gear out, you can do some real testing. One thing wasn't clear in the results posted was the actual grid drive voltage at the power tubes. They should be pretty close to the bias voltage you use, i.e., if the bias voltage is -30V, then Eg1 = 30Vpeak, but if they are far from it, and the output is already visibly clipped on the scope, then something is still amiss. BTW, is the amp cathode biased or fixed biased?