On Stability

[strelok]

Hey everyone, haven't been around much these past few years, been busy learning and designing power electronics so I haven't had much time for guitar amp stuff. Work has slowed down (for the time being) so I've been spending some time screwing around with that 6v6 express I built a while back. Its one of those spirals that's been going on the past few days. Initially I wanted to just open the thing up and look at the ripple in the power supply, only to discover the F&T 50/50uF main filter cap had bulged something fierce for whatever stupid reason. I tacked some sprague atoms in temporarily until a replacement arrives. Glad I caught it before it 'sploded its guts all over the chassis.

As these things go I started probing around some more. Found what I thought to be an oscillation. Turned out to just be induced noise from the filament circuit. Managed to fix that fairly easily, will make a separate thread on that in technical discussion as the method is somewhat novel in this application. In the midst of all that screwing around I found that if I replaced the shielded wire with unshielded going from the volume pot to the grid of the second stage things would get quite out of hand. It raised a question that's been in the back of my mind ever since I started playing with the express circuit.

What precisely is it that makes the express so prone to instability? I mean, the preamp is three gain stages. No crazy high plate resistor values or anything like that. Compare it to something like a Marshall 2203. 3 gain stages plus a cathode follower. Should be *roughly* the same gain right? You might be thinking the 2203 has a lot more loss between stages due to the dividers but its not as much as you might think. The 470pF across the top of the divider makes the response almost flat, and only when the miller capacitance of the subsequent stage starts loading the output impedance of the previous one does it start to roll off and that is largely out of the audio band. I've built several different 2203 style preamps all with different layouts, I've never had to resort to shielded wire or additional grid stoppers to suppress oscillations, save for on the input. So what gives? Why is the Express so finicky and either shielded wire and grid stoppers have to be used or the exact right layout meticulously tweaked until its quiet? I think its topology.

In a Marshall almost all of the gain proceeds the tone stack. In the express all but one stage is after. So whatever gets induced in all the stray reactance present in the tone stack, which is a good bit of area and wire, has 2 more gain stages, the PI, and the outputs to go through and get amplified many, many times. This certainly correlates with how important lead dress is and how sensitive the wiring on the tone stack is. Whereas the Marshall its just the PI and outputs.

To help support this argument this afternoon I whipped up a quick small signal model of both circuits up through the PI. For simplicity and presenting what should be worst case all controls are set at 10 in the model. Attached below is an image of frequency response taken from the inverting output of the PI and the model schematic. The green trace and top circuit is the Express and the blue trace and bottom circuit is the 2203. Attached also is the spice netlist if anyone wants to play around with it. This is a very simple model but I think the results are illuminating. From 1kHz to 20kHz the two circuits have very similar levels of gain. This is the region that matters most as its where all of the squealing and oscillations tend to take place. Yes supersonic oscillations do occur and can result in audible effects, and yes the 2203 rolls of a bit lower and that definitely helps. But I don't think that difference is enough to explain it on its own.

Hopefully I'm not beating a dead horse here and/or didn't make any boneheaded mistakes or oversights like I always seem to do. I just really felt the need to satisfy that nagging feeling in my head that something was missing or maybe I wasn't doing something right. Now that I think I've found it I feel less concerned about trying to get everything to run without shielding or grid stoppers for the most part. Anyway curious to hear what y'all think.

[bepone]

Lead dress is causing oscillations in higher gain... oscillations are positive feedback, and cable connected to sensitive grids is acting like an antenna..
Whatever this cable picks up, it is getting amplified. grid is multiple turn of small wire so it is real antenna..

In guitar amp there are many receivers and transmitters, spread them to reduce effect or join them together to increase, specially in in-phase stages

[pdf64]

There's a tendency with master volume amps not to turn the master up full, so perhaps instability with the 2203 doesn't get noticed?
Any gain boosting mods to the 2203/4, eg at the 2nd stage cathode, often necessitate stability mitigation, eg anode to cathoe or anode to grid cap at the input stage.
Marshall even resorted to it on the stock model once https://el34world.com/charts/Schematics ... w_2203.pdf

[strelok]

That's certainly possible. However in the 3 2203 style amps I've done I can dime everything and not have a problem. I've never included any of the additional stability caps either. At least that's what I remember, should go back and verify that. I think it'd be interesting to see how the express behaves with the tone stack moved to after various gain stages or just not have it in circuit at all and see how that affects its tendency to oscillate. Maybe I will play around with that at some point.

It could also be that the model is deceiving in some way. Perhaps the drive capability simulated is better than it is in reality by some non-negligible factor. Then it might be the case that the effects of the miller capacitance loading down each stage *might* be more significant in the Marshall causing lower than simulated frequency roll off.

Edit: Meant to also mention that yesterday after getting everything in the express to a place I was quite happy with, I decided to add a PPIMV. This brought some of the instability right back. Its located next to the bass control as I had an extra spot there so I guess that's a point in favor of my theory?

[romberg]

What precisely is it that makes the express so prone to instability?

Two big ones are:

• No grid stoppers in the preamp. So, there are no RC filters (using miller capacitance) low passing the signal.
• No voltage attenuation between stages. Many/most other amps attenuate the signal a bit between stages. The voltage dividers also act as grid stoppers. The Trainwrecks do none of this. So, the three stages of a trainwreck produce more gain than those of a marshall because the trainwreck does not throw away any gain.

Mike

[pdf64]

There's the V3 LTP a-a snubber cap on the 2203/4 rolling off a bit of high end.

[Reeltarded]

No stoppers is big, no attenuation is bigger BIGGER


BIGGER