Oscillation with high NFB on 5F6a

[Antoon]

I have installed a two position NFB switch on my 5F6a. On the high NFB setting (JTM45 level) the amp oscillates. The oscillation goes away when I turn the presence control higher. So the oscillation seems to be caused by too much high frequencies being fed back. I read at Rob Robinette's site that too much HF NFB can indeed induce oscillation.

Is there a way to reduce the high frequency content of the NFB signal somewhat? For instance by a small cap between the NFB signal and ground? Has anyone ever had to do this?

 

[Lowerleftcoast]

Traditional ways to reduce high freq. The 5F6a used a 47pF cap from plate to plate in the PI. I would suggest 100pF. Higher values can be used but audible content is reduced as the value of the cap is raised.

Reduction of highs can also be used early in the circuit. Marshall (and others) used both the methods shown on the first triode. Again, small caps will reduce ultrasonic highs and larger caps will reduce in the guitar audio range.

 

[Lowerleftcoast]

I will add... a cap may be placed across the NFB resistor. Usually this is done to reduce *ice pick highs* but it could be used for oscillation as well.

Let us know what solves the issue.

 

[Antoon]

View attachment 940365

I will add... a cap may be placed across the NFB resistor. Usually this is done to reduce *ice pick highs* but it could be used for oscillation as well.

Let us know what solves the issue.

Thanks for your suggestions, but this would worsen the oscillation I expect. The problem seems to be caused by too much HF being fed back.

 

[Lowerleftcoast]

but this would worsen the oscillation I expect.

Well then, you could try adding a small cap across the presence pot terminals 1 to 3, like you suggested in the op. 47pF to 100pF might be enough???

 

[Antoon]

Well then, you could try adding a small cap across the presence pot terminals 1 to 3, like you suggested in the op. 47pF to 100pF might be enough???

Yes something like that. Then I would rather add the cap directly to the JTM45-NFB resistor that is mounted on the NFB switch. The other NFB switch position (5F6a NFB) sounds fine so I can leave that one alone. I will give a 100pF cap a try.
Thanks.

 

[Lowerleftcoast]

I will give a 100pF cap a try.

Looking at some cut off frequencies, 500pF to 1000pF (.001uF) might be considered if the 100pF doesn't do it.

RC filter cut off frequency with 10k resistor:
100pF = 159,236Hz
500pF = 31,847Hz
.001uF = 15,924Hz

 

[robrob]

Have you tried reversing the power tube plate wires? You may actually have positive feedback. The 5F6A doesn't scream with PFB like most amps. Mine just had distorted over tones until I swapped to NFB.

 

[Antoon]

Can't seem to get rid of it. Tried a 100pF and a 500pF between the 13k FB resistor (I have only a 4 ohm OT, so I need 13k for a JTM45 type NFB) and ground. Perhaps the 500pF was little better, but not enough. Also a 500pF over the FB resistor just to make sure. Made it worse, as expected.

Looking at some cut off frequencies, 500pF to 1000pF (.001uF) might be considered if the 100pF doesn't do it.

RC filter cut off frequency with 10k resistor:
100pF = 159,236Hz
500pF = 31,847Hz
.001uF = 15,924Hz

Can I regard the FB resistor (13k) and the cap to ground as one LP filter? 500pF would then mean 22748Hz, as in 23kHz? Hm, okay. In that case I might even have to go with a higher cap. Will try 1000pF and up.

Have you tried reversing the power tube plate wires? You may actually have positive feedback. The 5F6A doesn't scream with PFB like most amps. Mine just had distorted over tones until I swapped to NFB.

Thanks, but I am pretty sure the NFB and presence work as they should. No indication of reversed OT connections. I also recall that I checked this some time ago.

 

[Antoon]

A 1.5nF cap between the FB resistor and ground finally got rid of it. I suppose the frequency causing the oscillation is between 8kHz (1.5nF) and 23kHz (500pF) then. Too much NFB of that frequency apparently caused the amp to oscillate. Weird..
I will test with a pot how much series resistance I can put between the cap and ground without the amp starting to oscillate again.

 

[Antoon]

I finally settled with a 1.5nF mica cap and a 10k series resistor in between (still using the 13k feedback resistor). That was what was needed to cancel the oscillation. I did try to find the real cause (leads, layout) but could not identify anything that made a difference.

 

[dan40]

I know from building several of the Marshall circuits that the routing of the nfb wire from the board to the presence pot can cause instability if routed to close to the tonestack wiring. You may be able to adjust that wire a bit and get rid of the cap.

 

[robrob]

The 5F6A is a pretty standard circuit. If you have to add bandaid caps to get it to run there is something really wrong.

You can test a 5F6A for positive feedback by paralleling another 27k resistor across the NFB resistor to double the NFB. If you get no obvious oscillation then you're probably good.

You can also simply disconnect NFB and see if that fixes the issue. If it does then swap the power tube plate wires.

 

[Antoon]

I know from building several of the Marshall circuits that the routing of the nfb wire from the board to the presence pot can cause instability if routed to close to the tonestack wiring. You may be able to adjust that wire a bit and get rid of the cap.

The NFB wire in my amp does run a bit awkward. I will see if I can re-route it. Would a shielded NFB wire help?

The 5F6A is a pretty standard circuit. If you have to add bandaid caps to get it to run there is something really wrong.

You can test a 5F6A for positive feedback by paralleling another 27k resistor across the NFB resistor to double the NFB. If you get no obvious oscillation then you're probably good.

You can also simply disconnect NFB and see if that fixes the issue. If it does then swap the power tube plate wires.

Thanks but I really checked for positive instead of negative feedback. On your site (which is awesome) you write that installing an ice-pick cap over the feedback resistor sometimes results in oscillation. Wouldn't this be such case in which high frequency NFB is increased and the amps starts oscillating?

Something may be wrong with this amp but I am not going to rebuild it. I will look into the NFB wire routing but it sounds pretty good as-is.

 

[Lowerleftcoast]

Did you try paralleling another cap with the 47pF snubber cap?
That would be my first choice, before trying the other *fixes*. It is easy to just clip in more capacitance to have a listen before soldering.

 

[Antoon]

Did you try paralleling another cap with the 47pF snubber cap?
That would be my first choice, before trying the other *fixes*. It is easy to just clip in more capacitance to have a listen before soldering.

Yes I tried this as well after you suggested it. I even went up to 500pF. It just took some brightness away but it oscillated in the high NFB position none the less.

Meanwhile I looked closer at the NFB wire routing. After a recent mod it crossed the leads to the treble cap. I moved the tone stack wires away from the NFB wire. It is too late know to test the amp again on high volume but tomorrow I will report back.

 

[dan40]

The NFB wire in my amp does run a bit awkward. I will see if I can re-route it. Would a shielded NFB wire help?

I would try it without shielded wire first. Just make sure the wire runs straight to the pot as much as possible and don't let it run parallel to the tonestack wiring.

 

[Antoon]

After rerouting the NFB wire I have the impression that the oscillations are indeed reduced a little bit further. But I have decided to let it rest for now. As said it sounds great and is very usable. There is something else that is rather weird with this amp. I might start a new topic for this if I cannot figure it out myself..

Thanks again everyone.

 

[Antoon]

Not really related to oscillations, but I finally tried an ECC823 in V2 (= reverse wired 12DW7). It increased the clean headroom a lot! Actually quite strange that Leo did not use this tube in his 5F6a in the first place. I am sure he would have liked it a lot. But instead he chose to entirely delete the CF in his next models.

This experiment reminded me how much of the 5F6a overdrive tone is determined by the (barely coping) 12ax7 in the cathode follower. Without that it barely overdrives until the amp is dimed. This is probably true for the related marshall amps too. If you substitute an ECC823 in the CF, the 5F6 Bassman practically turns into a BF amp, except for the more pronounced midrange of the 5F6a. If you use a worn and weak 12ax7 in the V2 position, the distortion comes in even sooner (almost from the start). Actually quite nice

 

[Antoon]

Now that I think of it, the capacity of the 12ax7 in V2 (cathode follower) tube to provide the current needed to drive the tone stack and PI, appears to be so critical that I believe that in fact a lot of the exemplary differences between two used 5F6a's may very well be attributed to the particular condition of their V2 tube. A 12AX7 in which the second triode is worn (the one in the CF) gives a good amount of extra (actually good sounding) distortion at the cost of headroom. And in my experience that second triode in V2 wears out exceptionally fast in a 5f6a. A worn preamp tube suffers in the high frequencies first, so to avoid the tone from becoming overly bassy or mid-rangy due to the tube being worn, I had to wire the mid control as a variable resistor instead of a potentiometer. This enabled me to entirely remove bass and low mid range frequencies to balance the sound of the worn V2 tube. And I got a great marshally distortion, almost from the start of the volume dial. But very little headroom.