Problem in Mark III style presence and feedback wiring

[Darkbluemurder]

Just for the heck of it I implemented the Mesa Mark III style phase inverter and presence control (250kA) in one of my amps. I really like the tone but there's an odd problem: when I turn the presence control down (0 ohms resistance) there's a ghost note/farting in the clean tone (something like "chrrsstpl" after the initial pick attack that goes away when the note fades out). Once I turn the presence control up (more resistance) it goes away. Therefore I think it is a parasitic oscillation coming from the feedback loop. It's not as much noticeable in the overdrive mode.

I tried without success:

1) rerouting the ground end of the feedback loop which I had first at the master volume ground to the speaker ground
2) shielding the leads to and from the master volume control.

If the feedback loop is the culprit where should its ground go to? Can/should I use shielded wire for the feedback loop or is the current too high?
Finally I saw that Mesa used a 0.1uF cap between the speaker jack and the presence control in the Mark IV. I am not sure how that could help since there should not be any DC at the speaker outputs (unless the output is clipped to square waves), right?

Of course I could just not turn the presence control down full - what I don't do anyway - but I consider this to be a defect and want to get it sorted.

As always thanks for your input.
Stephan

[drz400]

If the feedback loop is the culprit where should its ground go to? Can/should I use shielded wire for the feedback loop or is the current too high?

Ground from the presence control feedback section should go to the same place the speaker jack ground goes. For instance the OT ground goes to the speaker jack ground and then goes to the ground point on the buss which should be where the presence (feedback voltage divider) ground goes. This completes that loop. I like to run everything to a buss it works everytime, no need for shield but it should still work fine I have done it both ways.

What does this look like on a scope if you have one and do you have a schematic of the section in question?

[Darkbluemurder]

drz,

Thanks a lot.

Well, then at least I returned the ground to the correct position (speaker ground). Speaker ground in my amp is through the output jack at the chassis.

Unfortunately I do not have a scope. The link to the schematic is here: http://www.tubefreak.com/mk3-2.gif
And here the Mark IV for comparison: http://www.tubefreak.com/mk4-4.gif

I checked in KOC's TUT 3 and found that he recommends to twist the feedback wire with the associated ground wire, and the ground wire terminates at the negative filter cap end for the phase inverter. In my amp this ground is at the chassis, too, through a multi-section filter cap. Twisting the wires is what I will try next.

I rerouted some of the grounds in the amp (took them out of the chassis). While this did not solve the presence ghosting problem it made the amp more stable overall. I could turn up almost everything to 10 in the lead channel and got microphonic feedback only after I turned up the bass control as well.

[Darkbluemurder]

Twisted the feedback wire with the return wire - no luck. Still ghosting with the presence pot turned down. I am at a loss now - the amp sounds so good now at all volume levels but the presence control is not fully useable - turn it up to 3-4 and everything is fine. Below that it starts ghosting depending on the setting of the master volume.

I will try whether routing the ground end of the feedback wire from the speaker ground to the ground of the phase inverter filter cap makes a difference but my hopes are not high at the moment. I need to take a look at the Mark III how MB did it.

[PRR]

First off: swap the OT primary leads. Better? Worse?

This amp has "no NFB" with Presence full up, but rather a lot of NFB with Pres turned down. Without NFB, it does not matter which OT primary lead is which. With "NFB", you have a 50:50 chance of making Positive feedback, which will be unstable. On simpler amps with lots of feedback, this can be very obvious. On low-NFB amps, a little PFB may not be obvious. But since you do get a change as you increase feedback (turn-down Pres), see how it likes the other way.

I can't even read that mk3-2.gif schematic.

Confirm my squint. Reading down from "to 8 ohm spkr" the stack is:

Presence 250K
3300
56K||0.01uFd
3300||0.047
1300

?

That's a lot of R-C action, and this PC has decided to stop doing simulations. Where's my slide-rule... ah, a reactance chart.....

Set Presence where it does not quite oscillate. Disconnect one end and measure the pot. Get a fixed resistor of next larger value. Put it in series with the 0.01uFd cap. See how you like it.

[PRR]

> I think it is a parasitic oscillation coming from the feedback loop.

Seems likely.

> "chrrsstpl" after the initial pick attack that goes away when the note fades out

It's a cool AB amp. The output tubes idle near 40mA but peak around 400mA. The gain of the tubes varies with current, and in this case enough that it may be marginally stable at idle or decay, but unstable on the loud high-current part.

The "Presence" network is mostly overall NFB control. With the network disconnected, the amp would have gain near 25. With the "Presence" pot full-up, gain is reduced slightly in treble and a bit more in bass: classic guitar-amp, you need a little but not a lot of damping of the speaker bass resonance, and you may want a near-raw sound in the highs. Turning down increases NFB and reduces gain at both extremes. At some point a mild Fender Mid-Dip appears. At the extreme, as I ready the fuzzy values, it goes to a very dark tone with high-end gain only about 3.

Nevermind the response curve. Just set up adjustable NFB around an amp. As you turn from no-NFB hi-gain to medium NFB and medium gain, the amp gets cleaner. In theory you can go to high-NFB and very clean. But in real life, at some point the amplifier oscillates. There are always phase shifts. Depends on tube and resistor values, layout, and exact OT details. Very minor changes in OT winding can give large changes in frequency response and phase shift at and above the top of the audio band.

How much NFB can you put on a tube amp before it gets unstable? 10dB is easy, 30dB is hard, 20dB or 10:1 is borderline for most circuits and most OTs. 20dB NFB on an amp with gain of 25 is gain of 2.5. This Presence control takes the amp down to gain of 3. It's borderline-borderline.

It worked for them. It isn't working for you.

M-B is fanatical about layout and parts. They have to be, because then they push as hard as they can and still have a good amp. Something about your amp is different than the one the plan purports to document (and remember that M-B hand-drawn plans often have small errors).

Man, how did they hack this without computers????

We need to raise the max NFB (minimum gain) at the top of the audio band, without greatly upsetting the shapes of the curves. In fact it was a bad idea to "go dark" with NFB around the power amp, we'll have to let the gain rise at the top and you will turn your main Treb knob down a notch. Comes to about the same result.

Try the changes in the right plan, marked "!", roughly in order. Try just doubling the top 3300 resistor to 6K8. That mostly just raises gain overall, and only at minimum Presence. That may be enough to get stable. No? Then try change 2, double R10 and halve C5. That gives a top-lift, and you may want to turn Treb down to get the dark tone. Still chrrsstpl? Change 3, an R-C network off to the side, adds gain in the 5KHz-50KHz range where OTs like to go crazy.

Part vaues +/-20%. We don't really know what the amp needs, so use whatever is close, and whatever units the maker likes to mark with. "0.005uFd" will be 4.7nFd, etc.

Also try 500pFd+22K from plate to plate of V5 driver tube. That reduces forward gain above the guitar band just a bit, lessening the urge to chrrsstpl.

[Darkbluemurder]

I can't even read that mk3-2.gif schematic.

Confirm my squint. Reading down from "to 8 ohm spkr" the stack is:

Presence 250K
3300
56K||0.01uFd
3300||0.047
1300

PRR,

Many thanks for your input.

Your understanding of the values is mostly correct. The only two differences how I saw the schematic and did it in my amp are a 0.005uF (=5000pf) cap in parallel to the 56k, and a 1500 ground resistor instead of a 1300 (which would not be a standard value anyway).
I guess the smaller cap around the 56k makes the amp less dark so it stays. The difference between the 1300 and 1500 is probably negligible.

I am not sure whether the OT primaries have an impact since the amp had a Fender style feedback circuit with Dumble values (6k8 series FB resistor and 470 ohms ground resistor) without any problems. 6k8 to 470 is 14.5:1 but that does not appear to be much different from 59k3 to 4k8 which is 12.4:1, unless either the bypass caps throw it off or this is just enough to get it unstable if the OT primaries are reversed. If my calculations are correct then I would need to replace the 3k3 with a 15k to get the same amount of NFB. I'll try it!

[Darkbluemurder]

First off: swap the OT primary leads. Better? Worse?

WORSE - squealed like a banshee and got only quiet with the presence turned full up. At least we are sure now that this wasn't the culprit.

Replaced 3k3 before 56k/5000pf with 6k8. Still chrrsstpl with the presence down.
Replaced 56k/5000pf with 100k/2200pf. Oscillation still there but the tone was even better!
Put in a 250pf PI snubber cap a la Mark IV. Bingo! No more chrrsstpl but the tone got a little dull with it.
Put in a 120pf PI snubber cap. Better tone but a hint of chrrsstpl with the presence pot down.
Put in a 180pf snubber cap (did not have a 150pf). Everything OK so far even though I liked the tone better without the cap but for now stability is more important and the tone is good enough.

So right now the amp is a clean Fender preamp, an ODS overdrive preamp, a Matchless chieftain reverb coupled with the dry signal just like the one tube reverb into the PRR influenced Mark III phase inverter into 2 6V6 tubes. Is this now a "M-umble"? Anyway I'll take it to a gig tomorrow and see how it works with the band.

Once again, PRR, thanks for your help. Your contributions to this forum are invaluable.

Cheers Stephan

[PRR]

> that does not appear to be much different from 59k3 to 4k8 which is 12.4:1, unless either the bypass caps throw it off

At DC it is like 64.1K/4.8K. Above the audio band the caps become near-short and it is like 4.8K/1.5K. In between it wobbles. That gives the various shape curves, which are variations on response, damping, and distortion shapes. And plus the 0-250K pot, it all gets a bit much for a sliderule and a cocktail napkin.

> Replaced 3k3 ...Still chrrsstpl...
> Replaced 56k/5000pf ...Oscillation still ...
> Put in a 250pf ...a little dull
> Put in a 120pf ...a hint of chrrsstpl
> Put in a 180pf ........

I'm sure this is how this network was "designed". Mr Mesa on guitar and Mr Boogie on soldering iron (whatever their names are), trying this and that in real time on a real chassis.

> WORSE - At least we are sure now that this wasn't the culprit.

Good. Sometimes a failure is the best success.

[Darkbluemurder]

I took the amp to the gig on Saturday. It was at a company party, and the volume was set quite low (6 on the preamp, 3.5 on the master) and my trusted Stephenson Stage Hog as my overdrive generator. Treble and Bass controls set to 6 in the beginning, later Treble to 7 and Bass to 4. Reverb on 10, no other effects.

Clean tone was warm, full but still clear, not harsh or spikey in any way. OD with the Hog was smooth and singing and gave better feedback (no pun intended) than ever before. All in all, a worthy improvement. What I like in particular is that when I use the bright switch with the OD pedal engaged it does not sound as fizzy anymore.

As to the feedback ratio I now understand that one cannot take the series/ground rule once bypass caps are used. That may also be the reason why the problem could not be solved by different wire routing and ground placement.

In fact I did a lot of experimenting by trying different things to see (and hear) how they work. I did this a lot with coupling caps (too muddy, too thin) or with bypass caps (not bright enough, too bright) until I got what I wanted to hear. I believe this is what Kevin O'Connor refers to as "empirical approach". The only downside is that it takes a lot of time to do things this way. The major learning experience from this exercise was that I better appreciate the contribution of the phase inverter to the overall tone of the amp.

Once again thanks to all who contributed to this.
Stephan

[skyboltone]

I think this answers my current question with regard my little bastard AO-35. I've got a 100pf in there. I'll try something a litttle larger. Thanks Stephen. I don't hear well enough to call it a "chrrsstpl", more like "fizzerstl". Great discussion!

Dan

[Darkbluemurder]

I think this answers my current question with regard my little bastard AO-35. I've got a 100pf in there. I'll try something a litttle larger. Thanks Stephen. I don't hear well enough to call it a "chrrsstpl", more like "fizzerstl". Great discussion!

Dan

"chrrsstpl" and "fizzerstl" are the same thing - an audible parasitic oscillation! I also had the "bzzz" and the "wheeechhh" variant of the same thing but they were preamp wiring related.

[skyboltone]

I clipped in a variety of sizes of snubber at the PI output, and first and second stage plate to cathode locations. Noticable tone changes but still chrrsstpl. The only piece of wire longer than a CM in the build is the PI output to .02 to final grids. I may have to shield those.

[drz400]

drz,

Thanks a lot.

Well, then at least I returned the ground to the correct position (speaker ground). Speaker ground in my amp is through the output jack at the chassis.

Unfortunately I do not have a scope. The link to the schematic is here: http://www.tubefreak.com/mk3-2.gif
And here the Mark IV for comparison: http://www.tubefreak.com/mk4-4.gif

I checked in KOC's TUT 3 and found that he recommends to twist the feedback wire with the associated ground wire, and the ground wire terminates at the negative filter cap end for the phase inverter. In my amp this ground is at the chassis, too, through a multi-section filter cap. Twisting the wires is what I will try next.

I rerouted some of the grounds in the amp (took them out of the chassis). While this did not solve the presence ghosting problem it made the amp more stable overall. I could turn up almost everything to 10 in the lead channel and got microphonic feedback only after I turned up the bass control as well.

Personally I would NEVER use the chassis as a ground path. Speaker jacks should be isolated, (OT ground goes to them and a wire then goes to a deliberate ground point on the ground buss).
Sure it has been done but you have no way of steering (assuring) actual ground path, ground is not ground, it has signal and I dont want signal on my chassis.

check the wiring location of the wires going to the power tubes in relation to their proximity to the speaker jacks or OT secondaries, best to keep them away from each other. Sometimes just running a wire to the front of the tube socket instead of the back (keeping it away from the speaker jack and secondary) will help big time in these oscillations.

[Darkbluemurder]

Personally I would NEVER use the chassis as a ground path. Speaker jacks should be isolated, (OT ground goes to them and a wire then goes to a deliberate ground point on the ground buss).
Sure it has been done but you have no way of steering (assuring) actual ground path, ground is not ground, it has signal and I dont want signal on my chassis.

check the wiring location of the wires going to the power tubes in relation to their proximity to the speaker jacks or OT secondaries, best to keep them away from each other. Sometimes just running a wire to the front of the tube socket instead of the back (keeping it away from the speaker jack and secondary) will help big time in these oscillations.

Yes, isolating the speaker jacks and running the grounds together with the power tube cathode grounds to the PT/first filter ground is the state of the art method. However I must say that the two amps I converted to this scheme did not hum less or more.

The wires going to the power tubes are twisted and going directly to the front of the power tubes.
Actually I considered swapping the phase inverter position as indicated by G. Weber in his third book to make the plate wires longer and the grid wires shorter. I haven't got around to do it yet. However, since the amp works well now with the 180pf PI snubber cap and I like the tone I don't think I'll take the hassle. I'd better rewire my modded Concert.