Best way to cure output osscilation

[catalin gramada]

If itīs not present at lower volumes then it is NOT crossover distortion by definition, although waveshape looks *somewhat* similar.

It is bias shifting which appears at high power, a very diffeent thing, in my way is "natural", it was "always there since day one" , is an important part of tube sound (so much so that Transtube and others create it on purpose, go figure) and people startted worryoing about it *only* after they started scoping amplifiers , misinterpret it and, of course, "listen with their eyes" so itīs now "unbearable"
Snowflakes.
[/rant mode]

So, JMF this is a consequence of rail supply sag? If bias voltage sag proportionally in respect with voltage rail at momentary current draw ensuring new bias point in respect with voltage plate at high power this artefact still appear, please ? Or in case of stiff power supply and regulated bias, for instance?

[pdf64]

No, the suggested issue is unrelated to power supply sag, but rather it's due to the signal at the power tube grids getting rectified and increasing the effective bias voltage, see https://www.ampbooks.com/mobile/ampl...ias-excursion/
It's most noticeable during overdrive, but, especially with tubes, such effects as grid conduction may not be a 'brick wall', eg zero grid conduction below Vg1-k<0 and then full on grid conduction for Vg1-k=>0; rather more of a gradual curve.
Of course, sag at the HT nodes (especially g2) will exacerbate this.
The point is not to worry about it, and definitely not to decrease bias voltage in an attempt to mitigate for it.
As Chuck notes, lowering the value of the coupling caps to the power tube grids may help.

[catalin gramada]

Hi.
I changed coupling caps to power stage to a smaller value as You suggested, 47nF.
In same time found for low freq. 50cps around since wave is very distorted with and without nfb . Why,please? The pics are with 1,5k / 100k nib divider for 7db around



And OT output



The PI is direct coupled from previous stage, why the shape of wave is frequency dependant at the outputs, please?

[catalin gramada]

Can be a distorted current waveform, please? How to cure that in a practical way in my circuit, please?

[Helmholtz]

Can be a distorted current waveform, please? How to cure that in a practical way in my circuit, please?

If the distorted waveshape only shows at low frequencies and not at higher ones and not at the PI plates, I assume it is OT core distortion. As the magnetizing OT current increases with lower frequencies, the non-linear inductance of the core shows as distortion.
It would need higher levels of NFB to reduce it.

Is this a guitar amp and does a little bass distortion (mostly below the frequency range of a guitar) really hurt?

[catalin gramada]

Thanks for answering. The distorsion is present at PI plate just in low freq range above some signal level at the inputs. Weird thing , connecting nfb it seems to do worst than better. I will do some investigation and post some pics. Thank you.

[Helmholtz]

The distorsion is present at PI plate just in low freq range above some signal level at the inputs.

Do you see the low frequency distortion at the PI plates when you disconnect the NFB?

How do the low frequency signals look at the PI plates if you disconnect the power tube coupling caps?

[pdf64]

Waveforms inside a feedback loop generally look bad, as components to correct any error will be significant.

[Helmholtz]

Another question: What was the load with the scope pictures in post #33, a resistor or a speaker?

[catalin gramada]

That's looks like the PI output without nfb connected before clipping (bias set 41V). The waveform is skewed little bit to the left.



And this is how OT output looks like with 4 ohm resistor



This is how PI output looks like with nfb attached for the same output voltage like in first pic



And the OT output



The test done at 40 cps

[Chuck H]

I don't see the skew in the first shot (is it just me?). If it is skewed it's not nearly so compared to the other shots. So, along the lines of what Helmholtz it saying, notice that the waveform without feedback from the OT is pretty horrible without NFB. Once you apply NFB the distortion on the PI is indeed increased, but the distortion on the OT is decreased. This is pretty much indicative of what Helmholtz was saying. The distortion is in the PT and NFB is attempting to correct it.

EDIT: On another note... If there is a little skew in the first shot it may be a coincidence. I don't know what frequency you're testing, but the lower you go the harder it is for both vintage and modern electronic gadgets to accurately produce. Since bench gear is not often as advanced as multi thousand dollar amplification systems it may be that your signal generator is slightly distorted at that frequency. It might be worth looking at it directly with the scope to see.

It may also be that even with the NFB disconnected there is still a small amount of interaction between the OT's distorted response and the power supply or the power tube grids and the PI measurements. That is, everything is connected a little bit. So there's some shared interaction (and therefor visible distortion) even when, in a perfect world, those circuits upstream are considered exclusively from those downstream.

[catalin gramada]

It is Chuck, take a look, the same like in first pic, no nfb aplied



So, to consider to inject more nfb to get rid of it? I just lowered cause have some oscilation problems arround the other bandwith end as I asked at the beginning of the topic
The test frequency was 40 cps. Build this amp for a friend to be used for bass and guitar aso

[pdf64]

As long as the amp puts out good power before clipping down to low E, what does it matter if the waveform looks 'less than perfect'? The additional low order harmonic distortion will probably just add richness to the tone. The important thing with a musical instrument amp may be that the transition into clipping happens smoothly and the amp remains stable, whatever the load and signal level.
What are the specs of the OT, eg power and bandwidth? Typical MI amps tend to be full power +-1dB 70Hz - 15kHz http://www.hammondmfg.com/pdf/EDB1750M.pdf so things can be expected to be getting 'out of shape' at 40Hz.

[catalin gramada]

I was surprised to found I never get this problems with Hammonds, at least from 1650 series, used in previous buildings. But I started to collect old British trannies from trashed amps from the same price. This one is happen to be a Partridge (1,9k measured) coming from 100w amp. I don't know it's specs and from what amp was commin from but for sure have a very limited bandwith, maybe was built for guitar application as you say

[Helmholtz]

Ok, here is my analysis of the scope pictures:

According to scope readings we have a signal of 40Hz. The PI plate signals without NFB look OK, no signs of power tube grid current. Also the PI output amplitude is just below bias voltage, where no grid current is expected. Applying NFB improves output signal somewhat.

The output signal distortion looks similar to a cold bias effect, but if - as you reported - the distortion vanishes at higher frequencies, we are back at OT LF distortion. To cleanly reproduce such low frequencies, the OT needs to have a primary inductance of several hundred Henries.
What are the specs of your OT?

As mentioned before, only increasing feedback may help. But this may cause other problems (like instabilities).

After all it's no HiFi amp (and OT).

(Sorry if some points have already been covered in other posts above.)