How to lower even harmonic, keep odd one

[ervinl]

Hi,

I just wonder is there a way to lower even only harmonics, keep odd (especially 2nd harmonic), like in SE.

I like SE sound, but the wattage is small vs Push-Pull. But in PP, odd there is much even harmonic than odd. How to reduce this even only, either local or global feedback?

Thanks,

Ervin L

[TheGimp]

It is not clear from your message (English not primary language is not an insult, just difficult too understand). I think you have even and odd harmonics switched.

I think you are asking how to enhance 2nd, 4th, etc harmonics while suppressing 3rd, 5th, etc.

If you are running class AB1, you can increase the bias of one output tube slightly to increase even harmonic generation.

Likewise, you can either increase or decrease the bias on an input buffer to push it into a non-linear region where more even harmonic distortion will be prevalent.

[ervinl]

Originally Posted by TheGimp

It is not clear from your message (English not primary language is not an insult, just difficult too understand). I think you have even and odd harmonics switched.

I think you are asking how to enhance 2nd, 4th, etc harmonics while suppressing 3rd, 5th, etc.

If you are running class AB1, you can increase the bias of one output tube slightly to increase even harmonic generation.

Likewise, you can either increase or decrease the bias on an input buffer to push it into a non-linear region where more even harmonic distortion will be prevalent.

Yes, how to enhance 2nd, 4th, etc harmonics while suppressing 3rd, 5th, etc. Like what SE amplifier has. But I would like more power like in PP (anyway, I think this PP is already in Class A, depends on Rk and current calculated from the load line).

I heard that PP will loose all odd harmonic (due to cancelling effect in PP) but leave even harmonics, which is un-expected. Then how to make even harmonics lower but enhance odd one, I don't think a feedback (local/global) can.

Do you mean making "unmatched" bias (or using unmatched tube) will do this job?

Thx,

Ervin L

[Miniwatt]

Originally Posted by ervinl

Yes, how to enhance 2nd, 4th, etc harmonics while suppressing 3rd, 5th, etc. Like what SE amplifier has. But I would like more power like in PP (anyway, I think this PP is already in Class A, depends on Rk and current calculated from the load line).

I heard that PP will loose all odd harmonic (due to cancelling effect in PP) but leave even harmonics, which is un-expected. Then how to make even harmonics lower but enhance odd one, I don't think a feedback (local/global) can.

Do you mean making "unmatched" bias (or using unmatched tube) will do this job?

Thx,

Ervin L

First part is correct, you got mixed up again in the end. Even harmonics are cancelled out because of the PP arrangement. You could get some more even harmonics by biasing differently or using unmatchd tubes

[Evenharmonics]

Originally Posted by ervinl

But I would like more power like in PP (anyway, I think this PP is already in Class A, depends on Rk and current calculated from the load line).

Try these pages.
Noble Amplifier Company - Custom Vacuum Tube Amplification (Push Pull amp)
Stereophile: Boulder 500AE power amplifier

The best solution is to get speakers with higher efficiency (>95db) and use SET amp. If not, you may want to tweak SET amp to get higher power (online search) instead of trying to increase even harmonics from P-P amp.

[ervinl]

Originally Posted by Miniwatt

First part is correct, you got mixed up again in the end. Even harmonics are cancelled out because of the PP arrangement. You could get some more even harmonics by biasing differently or using unmatchd tubes

Thanks for the info. Sorry that I rather mixed-up (swapped) between odd and even harmonics of PP vs SET :-)

Ervin L

[Miniwatt]

Originally Posted by ervinl

Thanks for the info. Sorry that I rather mixed-up (swapped) between odd and even harmonics of PP vs SET :-)

Ervin L

No problem, a year ago I wouldn't have had a clue, we all learn here

[Wavebourn]

It is not the problem to add distortions;
the problem is how to lower them.
It is easy to compensate even order harmonics making a curve symmetrically bent. But to minimize odd order harmonics the only way is to make the curve straighter. Either by extra gain compensated by negative feedback, or by extra power consumption compensated by room cooler and money spent on electricity. The first method is preferred for Hi-fi, the second one is used for High-End.
However, what I personally call High-End, is a combination of both methods, but it is a different story...

[DF96]

Note that a P-P output stage only cancels even harmonics generated within itself. It will faithfully pass on any even harmonics in the input signal. There are two ways to deliberately add evens without adding too much odds as well:
1. Include an amplifying stage which has a square-law response - ECC82/12AU7 can do this if biassed correctly (for this application). Or use a variable-mu valve such as EF92 or 6BA6, as these will generate lots of evens. Then don't have any overall feedback, as this will mix the even with the original signal to create more odds.
2. Design a good linear amp, with negative feedback, but include in the negative feedback path a square-law stage. Achieving loop stability might be interesting, though!

Alternatively, just buy a badly-designed preamplifier or buffer. There are plenty of these around on the internet. Some are discussed on this forum, after the hapless new owner wonders why he has spent good money to degrade his sound.

I won't ask why you want to do this.

[mashaffer]

Let me just throw out something for the experts to bat around. How about using an output tube designed for lower 3rd harmonic like the 6L6 and then bias carefully and instead of using global feedback use ultra linear or triode mode or maybe a touch of plate to plate feedback.

The track I am shooting for is to get a favorable spectrum not so much by adding 2nd harmonic but my minimizing higher order distortion in the first place. Then if some masking is needed a little shift in the bias could be used I suppose.