Best way to cure output osscilation

[trobbins]

Can you show the power supply cap values, and also the bias circuit details, just for completeness?

[catalin gramada]

Power supply sketch with actual voltages,iddling and full pwr.,please

[J M Fahey]

1) forget testing at 40 Hz, period.

2) original 8x10 SVT, the ruler by which all bass cabinets are measured, dropped like a brick below **100 HZ**

Have you ever heard anybody complaining about how an SVT *sounds*???
Not how it measures or scopes or calculates or Thiele Smalls.

Here´s what a dedicated cabinet expert found, to his dismay.

And this, loading a regular Ampeg cabinet with Eminence PA woofers, bassier than original CTS:



notice that although they produce loud output at 40Hz , some 95dB/W , it´s all but irrelevant compared response 100Hz and higher yielding 105dB/W
Which are achieved by close coupling 8 of them, what helmholtz hints above.

Don´t be surprised, that relatively flat response down to a certain frequency which is not "deep" by any means and dropping below like a sandbag thrown from a roof is typical of *all* closed cabinets, simplest of all cabinet types (ok, infinite baffle is even simpler, just a sheet of plywood with a hole) and NOT providing any acoustic help at all to the raw driver.

You want louder deeper? ... ok, tune your box, enlarge it, put driver at a large horn throat, mount it at the end of a long pipe or transmission line ... NONE of that is used in an Ampeg 8 x 10"

And yes, introducing the cabinet in an amp discussion is very relevant if it clearly shows amplifier has NO business reproducing 40Hz flat anyway.
Or clean, which is the flip side of the same coin.

Suggest you repeat all tests at, say, 250Hz ... then if you need peace of mind, repeat at 100 Hz.

Any lower is just satisfying a personal curiosity but has no bearing on actual stage sound.

[catalin gramada]

Thank for very instructive explanation JMF. I still want to know why my output response at 40cps look like that. Have no possibility to found the specs of my OT. Can I conclude it have a limited bandwith and is not able to provide full output for certain frequencies below, please?

[Helmholtz]

Thank for very instructive explanation JMF. I still want to know why my output response at 40cps look like that. Have no possibility to found the specs of my OT. Can I conclude it have a limited bandwith and is not able to provide full output for certain frequencies below, please?

As mentioned above, it's transformer core distortion, sometimes called (partial) saturation.

It works like this:
Any real world transformer has a finite primary inductance and its impedance acts in parallel to the load, stealing current from the power tubes/load. As long as this additional current is low compared to the load current, you won't notice it. But as frequency decreases, the inductive (so-called magnetizing) current increases and reduces output. Furthermore, higher magnetizing current drives the core into partial saturation, which decreases the primary inductance and thus increases the unwanted additional current even further depending on its momentary value. This non-linear current load produces signal distortion at the output.

[catalin gramada]

Ok, Thank You for support. This is how the output looks like before clipping in respect with preamp input signal at 100 cps including tone stack with flat controls. Pretty flat for 13mV peak input sensitivity.



I will post further investigation what's going on in respect with certain amount of input signal at 40 cycle. Thanks.

[J M Fahey]

If you are showing 100Hz sinewave, one trace being power amp Input and the othjer power amp Output, then both look very good and your amp is handling both easily.

Now if exact same amp clearly distorts, shows wonky signal waveform at same level (justy before clipping), well, then , there you have answered yourself, transformer cannot handle 40 Hz full power, period.

If you have any available, just test any 200/300W (or more) SS amplifier (rack power amp, PA box mixer, etc. ... even an SS SVT ) at 100Hz and then again at 40Hz ... you´ll see that they have NO trouble at all.

[catalin gramada]

If you are showing 100Hz sinewave, one trace being power amp Input and the othjer power amp Output, then both look very good and your amp is handling both easily.

Now if exact same amp clearly distorts, shows wonky signal waveform at same level (justy before clipping), well, then , there you have answered yourself, transformer cannot handle 40 Hz full power, period.

If you have any available, just test any 200/300W (or more) SS amplifier (rack power amp, PA box mixer, etc. ... even an SS SVT ) at 100Hz and then again at 40Hz ... you´ll see that they have NO trouble at all.

I buit some powerstages with 1.9k Hammonds before. Never get this kinds of troubles as down as 40cps. This one was built from a respectable vintage company claimed to be a sort of Rolls Royce of output transformers. I bought a couple smashed perhaps from trashed vintage P.A. amps. It looks very well made with interleaved windings, fully impregnated...it is weird their bandwith is so limited.

[Helmholtz]

I buit some powerstages with 1.9k Hammonds before. Never get this kinds of troubles as down as 40cps. This one was built from a respectable vintage company claimed to be a sort of Rolls Royce of output transformers. I bought a couple smashed perhaps from trashed vintage P.A. amps. It looks very well made with interleaved windings, fully impregnated...it is weird their bandwith is so limited.

So the Hammonds will have higher primary inductance. This can be achieved by better quality steel laminations or by more primary turns.
The Partridge may have other advantages like lower losses or better HF response, though.

[trobbins]

You identify a minor but noticeable change in bias voltage. Perhaps that totally comes from a transformer winding level/waveform change, rather than anything related to EL34 grid conduction. You probably get a similar heater waveform peak/rms voltage drop.

If you had a storage scope and were keen, you could apply a tone burst, and see if the waveform deviation/distortion changed as the supply rails sagged - that would likely separate transformer and other forms of distortion from power supply voltage sag related changes.

[catalin gramada]

Now a question goes through my mind. It really need for a bass (clean) amp to have extended bandwidth OT at full power as time it is able to get power at low freq.? Thinking the average signal will no go up by a half power even compressed due the decay. So even a limited bandwith OT should be satisfactory if is capable to run undistorted at 100 cps. meant undistorted at 50cps. for half of its power...suppose. Most part of this extended bandwith full power capability is usefull just for peak attack for a bass instrument. It is that important if it is bit distorted or not, as time can reach the full headroom? That.s correct ?

[catalin gramada]

Hi, this is all I could do. The pics shows the response with signal applied to preamp input, bass input for 50mV input sensitivity, at 40 Hz. 9 db negative feedback was applied. The pics show the response slightly before clipping.

Full 100w clean



At 3/4 power dissipated on load resistor (ca.75 w it looks really neat)



Can be make it better, please.? Precise balance into power stage will improve somehow ?Thanks

[catalin gramada]

This is a pic at 50 hz full power, slight before clipping

[trobbins]

You have feedback around the output stage, within which is a CR filter which drops the forward gain at a nominal 16Hz corner (0.1uF, 100k + a titch more from bias supply). I'd be expecting that corner frequency will increase significantly as grid current starts to increase (as Vg starts to approach about -2 to -1V), and as such your feedback level drops, and so you lose the benefit of feedback to attenuate a growing level of distortion from the output stage valves and output transformer.

However, attempting to change circuitry and confirm an improvement is presently based on your scope - not a confident means of knowing what is changing and by how much.

[catalin gramada]

You have feedback around the output stage, within which is a CR filter which drops the forward gain at a nominal 16Hz corner (0.1uF, 100k + a titch more from bias supply). I'd be expecting that corner frequency will increase significantly as grid current starts to increase (as Vg starts to approach about -2 to -1V), and as such your feedback level drops, and so you lose the benefit of feedback to attenuate a growing level of distortion from the output stage valves and output transformer.

However, attempting to change circuitry and confirm an improvement is presently based on your scope - not a confident means of knowing what is changing and by how much.

Thanks for answer. Mainly the nfb network was changed from 100/1500 ohm to 2.2k/22k. The PI tail resistor was adjusted in consequence from 22k to 18k and the bias voltage was refited in respect with that. I choosed to supply the PI a litle high, close to the screens voltages. That.s it. No nfb compensation was added yet. Coupling caps was changed before to 47nF by Chuck H suggestion