Marshall 2204 inverter strange signal shape

[oliver_mpt]

Hi Everybody,

I build a Marshall 2204 50W head from scratch, and it works although I'm not satisfied with the overdriven sound which is too crunchy to my taste.
I checked the design by injecting a sinusoidal signal 500Hz, Vpp = 0.65V and looked with the scope at each stage. At full preamp volume for max saturation, the output of the first valve is clipped on the positive side with the negative side still sinusoidal - looks OK to me, then after the second valve, the signal gets clipped both sides, and is rather "square" (for max preamp gain).
The strange thing is the output of the inverter (V3) : the signal at the output on both plates of Valve 3 is very triangular ! that is strange for me.
Any explanation for this strange shape, should'nt it reproduce the signal as it goes out from the preamp ?

Any hint welcome !

[nickb]

The tone controls will have a dramatic effect on the waveform. What does it look like at the top of the master volume control?

[Chuck H]

That^^^^^^^^^^^ And a 2204 with well over half a volt into a cranked preamp will definitely show some extreme clipping at the PI. Extreme as in abnormal except for what you might see at the initial pick attack of a normal guitar. For scope viewing I would reduce the input signal to about 100mVac. That will still clip a cranked 2204, but not abnormally If you want to see an actual sine wave reproduced with that 100mVac input you'll need to turn down the amp controls down to where it would normally be clean with a guitar. So, somewhere around 4 on the volume control, about five on the tone controls and keep the presence at zero for now. Otherwise, if you leave the controls all full up, you'll need to reduce the input signal still further below 100mVac until it shows clean.

[pdf64]

Bear in mind that the phase splitter is inside a feedback loop.
Strange things will happen there.
Open the loop, set the tone controls for as flat a frequency response as possible (mid max, treb & pres min, bass very low), and re-check.

As it's a new build, consider the possibility that the feedback loop is positive rather than negative; does opening the loop increase or decrease the signal level at the amp's output?
For positive feedback to result in oscillation, a particular set of conditions must be met; the (apparent) absence of oscillation does not preclude the possibility that the feedback loop is positive.

[Enzo]

Not only all the above, but if you look at a square wave, the leading and trailing edges are a very high frequency component. If you clip a more or less square wave, then run it through something lacking a lot of high end response, it will lop off the corners.

[J M Fahey]

Please tell us what frequency you are using.
Also recheck the value of the plate to plate capacitor; is it still 47pF or you used a larger one?
Remove it and retest to see if anything changes.

EDIT: a triangular waveform is what you get when you try to drive a too large capacitor with too little current available, hence my suspicions that you have a too lrge capacitor there
And please post a picture of your triangular waveform, we might be imagining something different from what you actually have there.

[oliver_mpt]

Hello to All !

I'm impressed with the reactivity of the forum. Thanks a lot to you for all the nice indications. I was away for the weekend, so I had no chance to test but I will try during the week.
A few more indications on how I did the tests :
All tone control at mid position (5), including the presence. Preamp was either mid position (5) or full at 10 a,d master volume at 3. The frequency used for testing was 500Hz sinusoidal, in the middle of the normal (non-harmonic) guitar frequency range.
So to sum up :
I'll test with 100 mV peak (200 mV pp right ?), I'll check without the presence and with full presence to check whether it could be reverse polarity, and I'll check the 47pf capacitor between the plates. I also suspected a charging capacitor due to the shape. By the way it's triangular with two different slopes, both at rising and falling parts. Before the tone stack, the signal at full preamp is square, with a small peak at the top of the rising edge and bottom of the falling.
I'll post my results as soon as I have done the new tests.
Thanks for the nice help

[nickb]

Equally important to the input signal level is where you set the gain control.

I say again, as it seems to have got missed, You really do need to look at the signal going into the PI at the top of the MV (i.e out of the tone stack) otherwise you cannot possibly know what effect the PI is having on the signal. Having the tone controls at five does not guarantee a flat frequency response. Assuming you have a square wave going in, the only way you'll get close is by monitoring the output of the tone stack and adjusting the controls to get as close as possible to square wave out. Further, by observing how the waveform changes with the controls you will learn more in a few minutes than a month on a forum.

[cjenrick]

it's a Marshall, what did you expect? use 50 mv rms input V-ac, 100 is a bit steep.

might have excessive grid current flowing in the cathode follower (V2)

try 220K plate resistor for 330 K.

or put tone controls after V1 for less distortion, change V2 from cathode follower to dual gain channel with resistance voltage dividers to set gain, turn it up and let it rip.