'Scoping the phase inverter

[Matt T.]

I finally got everything I need to use my newly aquired oscilloscope...thanks for all the help with that. Now I'm just poking around my most recent homebrew amp (Trainwreck Express with power scaling). This amp doesn't really sound right to me...I'm able to get acceptable tones out of it but this amp is supposed sound much better I'm sure of it. The simplest way to put it is the attack is all thuddy/fuzzy.

So I adjusted the volume on the amp (while playing it) where the thuddy/fuzzy attack was present but it was otherwise as clean as I could get it...then hooked up the sig gen and 'scoped around. What I found was that the signal hitting the PI input was pretty clean (sinusoidal with just a very slight 'slant' on half of the waveform) but the two sides of the PI (plates) seemed to be distorting differently but here's the strange thing (I think?)...when I scoped the tied cathodes of the PI I only got the top half of the wave and complete cutoff on the bottom half. When I scope that junction point of those 4 resistors I get the same thing and also the same thing (except lower amplitude and looks slightly 'shifted') on the 'bottom' grid of the PI. But the signal on the input grid is big and clean.

I can get a good clean sinusoidal waveform on the tied cathodes (and those other points) if I turn the volume way down...but as I increase the volume it quickly cuts-off the bottom half of the waveform. And it's cut-off longer than it's making signal.

Does this mean the PI is clipping way assymetrically? That kind of explains the tone. Is this normal? What should I see on the tied cathodes when the PI is distorting 'correctly'? I tried different tubes...no difference.

BTW - the 'scope and sig gen work great together...the waveform is very focused and stable. Sig gen is made by Leader I think.

[stokes]

The output of the two sides of the PI should be close to equal,or ideally they would be identical,I dont have a schem for this amp,but the two sides of the PI should be close to or identical to each other.A slight mismatch in the 2 sides shouldnt cause a problem but if there is a severe mismatch the tone would suffer as you describe.Could be an imbalance in the plate resistors.If you have two different plate resistors,i.e an 82k on one and a 100k on the other,maybe you have them switched?

[Matt T.]

They were actually 100K/91K but I switched the 91K to 82K and then raised the 10K tail resistor to 22K (I put a pot in there and listened). Those two things helped a lot...it's still not blowing me away but it sounds OK.

Now the wave looks pretty square on both PI plates when cranked.

[stokes]

"Now the wave looks pretty square on both PI plates when cranked."
Are you comparing the PI outputs when they are distorted?If so I would suggest raising the volume just till you get the largest waveform before the tops square off.If they are close to the same size clean,the PI is okay.

[Matt T.]

Yea, I looked at it throughout the volume range and it starts clean, then morphs to what looks like 'gears' then morphs to square...happens about the same on both PI plates and after some more preamp mods it's sounding pretty good. Thanks.

[Matt T.]

While I'm here maybe someone can explain a few things I'm seeing with my new toy. This amp has London Power Power Scaling installed.

When I fully crank the amp and look at the output at the speaker jack I get very close to a square wave with close to equal duty cycles (I hope I used that term correctly). The 'walls' and 'tops' of the waveform are smooth with no zigs/zags or anything. If I turn down the power scaling control but leave the post-PI MV cranked, the wave starts to look like a top hat or a small step-ladder (two steps up then two steps down). That is, the 'walls' of the waveform develop this extra zig/zag. What is this? Crossover?

Another one; when looking at the waveform, say at the plate of a preamp triode (make it late in the preamp), I'll see a fairly sinusoidal waveform but then when I look at the same signal on the other side of the coupling cap, the waveform has been distorted a bit. To put it crudely, there appears to be a change in the 'slope' of the slightly flattened top of the waveform. This appears to be happening in two places when the cap is small-ish (.0022uF and .01uF) but I haven't verified that. What is going on here?

Thanks for any help...please bear with me...the curve is very steep for me right now.

[Steve Conner]

The top-hat effect is massive crossover distortion. It happens because turning down the B+ voltage will bias the tubes too cold. I'm surprised though, I thought Ko'C's power scaling circuit also adjusted the bias to compensate.

Even if it did, this could still happen, because if the power tubes are overdriven enough to draw grid current, they will start to charge up the coupling caps and bias themselves colder. This could happen more as the B+ is turned down, because the tubes will be driven harder as you turn it down, due to the negative feedback.

I don't know about the other thing.

[Matt T.]

Thanks for the response. The bias scales along with the B+ so it's just from the power tubes being driven harder and harder as the voltage is scaled down. That's why you have to have a post-PI MV when scaling the power tubes but not the PI (you need a pre-PI MV if you scale the PI along with the power tubes).

If I scale the voltage way back and crank the post-PI MV then I can even get the amp to pump and breathe...i.e. play and you get sound but stop and the amp goes dead silent for a second or so...then all the hiss/noise fades back in...and it sounds pretty bad like this. I assume it's so much grid current that it shifts the bias into cut-off? In any case, I assumed it was a normal 'quirk' with power scaling (i.e "doc, it hurts when do this."...doc: "well then don't do that.").

I guess it's good that I don't have any cross-over when fully cranked and when the post-PI MV is set appropriately regarding the power scaling setting.