A tale of troubleshooting

[WhopperPlate]

Thought I would share with y’all a recent build troubleshooting experience to help any lost souls chasing gremlins inside the labyrinth Of your new prototype tube amp wonder , or whatever you might be struggling with

I built a two channel fender twin platform amplifier with bells and whistles blah blah and a steel string singer style cathode follower driver setup. Fun stuff.

I was pulling a lot of current of course , and wasn’t really getting the output and headroom desired so we upgraded the iron to higher voltage and higher current ratings and slapped a big fat output transformer to get a explosive low end . More fun

In the process I decided to compare a few different transformers I had collected , taking notes and confirming my choice . As I alligator clipped in each transformer I began to hear a slight buzz , almost too quiet to hear over background noise suddenly, but I only heard it with one transformer and swapped back to the others without any issues . I thought, “oh no it’s a bad transformer”, but then the other transformers began doing it as well .

I realized it was related to the presence control, turning it up made it go away , and turning up the master volume made it go away . Ironically it would only do this at below useable volumes . The scope showed very high frequency oscillations.

I chased that wild goose and nailed it down to basically one grid wire on the driver tube . This wire would dictate the volume, but even if I could get it to be inaudible it would still show up on a scope in the PI , and frustratingly sometimes the waveform would get bigger as I heard it less

After carefully re wiring the entire phase inverter and driver tubes with new unmolested wire and ran the leads in a more thoughtful manner than I previously laid out , I was almost rid of the noise. Now it was just coming in a shorter range . At this point I had been at it for longer than I will admit ,and I was getting a bit perplexed .

I checked the power supply , i reflowed many a connection , I removed the presence knob away from near the rectifier where it was mounted just to make sure it wasn’t interacting . I chased the goose all over .

Standing there with my oscilloscope probe poking around , I realized the tubes themselves were emitting this frequency and the probe was picking it up through the air . The wave form was huge . I thought to myself , this has to be a grounding issue .

I then proceeded to tweak down on every connection, realizing that yes ; they obviously should have already been tighter than they were in a few spots .

So now after all the hatches have been battened down and I fire it up and it’s still doing it …I am like ok…but I realize it’s much less severe and only comes on at a single point in the volume sweep. I begin to move the grid wire, lifting it up from the chassis . It drops out . Almost .the single point becomes smaller and smaller . I carefully adjust the phase inverter cathode lead and lift it a few mm . Gone . Stable amp . No more problems .

Whew that took awhile . A saga of grounding and lead dress combining to create a tricky nuisance that sounded like a hum and buzz , but was actually parasitic oscillations. Moral of story , when you swap components be careful of changing lead dress , and ensure all ground connections are cinched down .

Cheers

[bepone]

So classic p2p 3D modelling is better than cahssis runs! iI saw Bartel is doing "to the air building" by stages,

but i want to know one thing.. where is loudspeaker grounded and is it possible to see path from loudspeaker to the PI feedback wire and its dedicated ground wire?

[Stevem]

It's no just lead dress that needs to be considered, coupling caps have what can be termed a input and output side to them even though they are not label for such.

One oscillation effect of this was seen years ago in a reproduction of a Fender amp by a builder who built a number of them and sold them with the issue.

The sourse of the problem seen was due to a change in the position of a coupling cap that was very close to the steel tip of the input jack .
The biggest area of the cap( lenght wise was parallel to the tip, once the cap was spun around such that it was now perpendicular to the jack the oscillation was gone.

If you have a situation where a high level of signal is applied to the end of a cap who's foil connection is right under the skin of the cap, and then you happen to have a wire running next to that cap that is of a lower signal level, but in phase with it, then you stand I good chance of having a oscillation issue or maybe just just a issue of poor nasally tone with dead high end .

[Reeltarded]

Hold my beer! When the last 6 parts hit my mailbox I am moving a PI to the other side of the output tubes.. over by the rectifier.

[WhopperPlate]

So classic p2p 3D modelling is better than cahssis runs! iI saw Bartel is doing "to the air building" by stages,

but i want to know one thing.. where is loudspeaker grounded and is it possible to see path from loudspeaker to the PI feedback wire and its dedicated ground wire?

Forgot to mention that . Disconnecting the nfb eliminated the issue .spkr is grounded at the jack . All jacks are unisulated switch craft

The oscillation was visible at every point past the phase inverter input , and definitely visible in nfb loop

Moving the nfb made little difference . In fact separating the output transformer from the chassis and floating leads entirely away from circuit changes nothing.



I need a computer …believe it or not I navigate this brave new world with only a phone… someday I will bust open the piggy bank

It's no just lead dress that needs to be considered, coupling caps have what can be termed a input and output side to them even though they are not label for such.

One oscillation effect of this was seen years ago in a reproduction of a Fender amp by a builder who built a number of them and sold them with the issue.

The sourse of the problem seen was due to a change in the position of a coupling cap that was very close to the steel tip of the input jack .
The biggest area of the cap( lenght wise was parallel to the tip, once the cap was spun around such that it was now perpendicular to the jack the oscillation was gone.

If you have a situation where a high level of signal is applied to the end of a cap who's foil connection is right under the skin of the cap, and then you happen to have a wire running next to that cap that is of a lower signal level, but in phase with it, then you stand I good chance of having a oscillation issue or maybe just just a issue of poor nasally tone with dead high end .

All caps are oriented with outer foil for proper noise rejection .

Hold my beer! When the last 6 parts hit my mailbox I am moving a PI to the other side of the output tubes.. over by the rectifier.

I almost did this ! After thinking I was trying everything I was about to punch a hole and shorten the driver grid leads lol. Fortunately I gained a hold of my senses and connected the dots .

[bepone]

So classic p2p 3D modelling is better than cahssis runs! iI saw Bartel is doing "to the air building" by stages,

but i want to know one thing.. where is loudspeaker grounded and is it possible to see path from loudspeaker to the PI feedback wire and its dedicated ground wire?

Forgot to mention that . Disconnecting the nfb eliminated the issue .spkr is grounded at the jack . The oscillation was visible at every point past the phase inverter input , and definitely visible in nfb loop

Moving the nfb made little difference . In fact separating the output transformer from the chassis and floating leads entirely away from circuit changes nothing.

ok.. maybe this is a problem.. many people doing this fault by default.. speaker cannot be grounded on chassis.. speaker gnd belongs to PI ground wherever it is because on this place is applied NFB loop, they are sharing the same reference gorund point. NFB signal is voltage signal and this voltage has + and - wire.. - cannot be on the chassis. makes wrong path to the PI capacitor.

speaker jack terminal must be isolated from the chassis, and NFB live wire and GND wire travelling together, to the PI, live wire need to go to the gnd reistor, and gnd wire must go to the PI last tail gnd resistro point, and there must be also connected PI supply electrolytic capacitor.

too much words but one pic will make it clear

[WhopperPlate]

So classic p2p 3D modelling is better than cahssis runs! iI saw Bartel is doing "to the air building" by stages,

but i want to know one thing.. where is loudspeaker grounded and is it possible to see path from loudspeaker to the PI feedback wire and its dedicated ground wire?

Forgot to mention that . Disconnecting the nfb eliminated the issue .spkr is grounded at the jack . The oscillation was visible at every point past the phase inverter input , and definitely visible in nfb loop

Moving the nfb made little difference . In fact separating the output transformer from the chassis and floating leads entirely away from circuit changes nothing.

ok.. maybe this is a problem.. many people doing this fault by default.. speaker cannot be grounded on chassis.. speaker gnd belongs to PI ground wherever it is because on this place is applied NFB loop, they are sharing the same reference gorund point. NFB signal is voltage signal and this voltage has + and - wire.. - cannot be on the chassis. makes wrong path to the PI capacitor.

speaker jack terminal must be isolated from the chassis, and NFB live wire and GND wire travelling together, to the PI, live wire need to go to the gnd reistor, and gnd wire must go to the PI last tail gnd resistro point, and there must be also connected PI supply electrolytic capacitor.

too much words but one pic will make it clear

Whereas I completely recognize the grounding methods described as inarguably ideal and I am familiar …having said that , this amp is stable, not a buzz or a hum with the master dimed , and so are many amps that deviate from the ideal method . I have had pretty good luck.

In fact I have gone out of my way the past to completely rewire an entire amp in a similar tail chasing debugging venture when I was building my very first amplifier . My novice skills left me with a super lead that was riddled with bad solder joints causing all kinds of parasitic oscillations .

In an effort to quelch the squeal I ended up changing the entire grounding scheme that separated everything as described .I suddenly couldn’t stand the character of the amp , and it still squeeled lol . For better or worse it all adds up .

Believe it or not , besides completely rebuilding that amp , know what fixed the majority of the issues with that amp ? Replacing the tungsol preamp and power tubes. Those things are so bright and open that they pickup like an antennae .

Next amp I build I will do exactly as Bepone described and maybe angels will descend and bestow glory tone with blackest background

[WhopperPlate]

It would be a cool experiment to build the exact same amp in as many grounding configurations as possible and document the results . People are always touting their specific method as the best , and it would be nice to hear them all side by side in the same application

[bepone]

Whereas I completely recognize the grounding methods described as inarguably ideal and I am familiar …having said that , this amp is stable, not a buzz or a hum with the master dimed , and so are many amps that deviate from the ideal method . I have had pretty good luck.

Next amp I build I will do exactly as Bepone described and maybe angels will descend and bestow glory tone with blackest background

Here point is not about the grounding or humm. NFB doesnt affect humm / too much.. Point is about feeding back the most critical part of the amp / feedback wire/ and voltage which is on the speaker terminals. To avod the problems good idea is to shield this wire, i did many times.
This voltage brought back to the PI is going to control a lot of things in the amp..total gain, bass control, treble, damping factor, output resistance, phase loop etc...

"Grounding" is wrong term here.. because speaker can be also in the air (in theory, but in practice this will make sqeal for sure), correct long expression is "zero voltage refference point for negative feedback" and need to go to the stage where is applied feedback. If is that PI then spk must be "grounded" to the PI.
If is that point V2, then speaker must be grounded to the V2 minus pole / where is electrolitic capacitor supplying this stage. Best on gnd side of some tail, or cathode resistor..

[WhopperPlate]

Whereas I completely recognize the grounding methods described as inarguably ideal and I am familiar …having said that , this amp is stable, not a buzz or a hum with the master dimed , and so are many amps that deviate from the ideal method . I have had pretty good luck.

Next amp I build I will do exactly as Bepone described and maybe angels will descend and bestow glory tone with blackest background

Here point is not about the grounding or humm. NFB doesnt affect humm / too much.. Point is about feeding back the most critical part of the amp / feedback wire/ and voltage which is on the speaker terminals. To avod the problems good idea is to shield this wire, i did many times.
This voltage brought back to the PI is going to control a lot of things in the amp..total gain, bass control, treble, damping factor, output resistance, phase loop etc...

"Grounding" is wrong term here.. because speaker can be also in the air (in theory, but in practice this will make sqeal for sure), correct long expression is "zero voltage refference point for negative feedback" and need to go to the stage where is applied feedback. If is that PI then spk must be "grounded" to the PI.
If is that point V2, then speaker must be grounded to the V2 minus pole / where is electrolitic capacitor supplying this stage. Best on gnd side of some tail, or cathode resistor..

Agreed, it obviously isn’t just grounding at play, but a weak ground connection (even if the meter reads full continuity and zero resistance) obviously revealed the potential issue from the layout . This wasn’t a hum ,it just sounded like one , the scope was reading around between 33-100khz depending on the settings.

For clarity the amp has pi and nfb grounded at fox loop jack and spkr gnd to spkr jack. Electrolytic tied to preamp at input gnd.

[sluckey]

Disconnecting the nfb eliminated the issue .

That indicates the NFB was wrong phase. You discovered this issue while using gator clip leads to connect several different OTs for testing. It's a 50/50 chance to get the primary plate leads connected to the correct tubes. Swapping the plate leads should correct this issue.

Wrong phase of NFB does not always result in ear splitting howls or squeals. Sometimes the issue will be subtle. Other times the squeal may be high enough frequency that you can't hear it, but your test equipment can. I suspect this was the case since your scope probe could even see it in the air without being connected.

[WhopperPlate]

Disconnecting the nfb eliminated the issue .

That indicates the NFB was wrong phase. You discovered this issue while using gator clip leads to connect several different OTs for testing. It's a 50/50 chance to get the primary plate leads connected to the correct tubes. Swapping the plate leads should correct this.

Wrong phase of NFB does not always result in ear splitting howls or squeals. Sometimes the issue will be subtle. Other times the squeal may be high enough frequency that you can't hear it, but your test equipment can. I suspect this was the case since your scope probe could even see it in the air without being connected.

Actually , this where my mistakes come in Handy . Because of some assumptions concerning transformer color coding, I had already swapped the primaries several times mistakenly …on each transformer .

“wrong phase doesn’t always result in ear splitting howls” , but it definitely did in this case on every instance

[R.G.]

[...] I begin to move the grid wire, lifting it up from the chassis . It drops out . Almost .the single point becomes smaller and smaller . I carefully adjust the phase inverter cathode lead and lift it a few mm . Gone . Stable amp . No more problems .

Whew that took awhile . A saga of grounding and lead dress combining to create a tricky nuisance that sounded like a hum and buzz , but was actually parasitic oscillations. Moral of story , when you swap components be careful of changing lead dress , and ensure all ground connections are cinched down .

Great work chasing that down. Parasitic oscillation is the work of the devil.

With a victory beer in hand, it might be a good thing to think about how that pesky grid lead came to be so sensitive to its specific position in 3-space.

Oscillation needs only two things to happen, those being enough gain and a feedback path that lets enough signal through from an "output" somewhere in the signal chain to an "input" somewhere in the signal chain. Stopping oscillation amounts to lowering the gain between the "output" and "input" or increasing the signal loss in the feedback path. Moving that cathode->grid lead did something, obviously. But what?

If it was moving it away from the chassis, then either the chassis was carrying the feedback signal, or being near the chassis was tuning the feedback path to have a higher "gain" at one frequency. The chassis carrying output signal currents and therefore being the source of the feedback signal is what bepone was getting at, I think. The cathode of a tube is an input too, if the driver is low enough impedance, and speaker output currents are about as low an impedance as tube amp signals get. Stray capacitance has a hard time driving something as low impedance as a cathode.

If it was tuning in the feedback path, there is still a problem hiding in there. Obviously, if that wire isn't held in place so that over time and vibration it can't wander back toward the chassis, the oscillation can return. That brings up the question - was it nearness to the chassis, or near/far away from another wire or terminal?

One other issue raised its head in my mind. Do your output tubes have grid stoppers? Grid stoppers work by reducing the RF gain of the tube. The parasitic oscillation might be from an un-stoppered output tube. It's also possible that putting a stopper on the PI tube(s) could cure a sensitivity that the PI input is getting from somewhere.

[WhopperPlate]

If it was moving it away from the chassis, then either the chassis was carrying the feedback signal, or being near the chassis was tuning the feedback path to have a higher "gain" at one frequency. The chassis carrying output signal currents and therefore being the source of the feedback signal is what bepone was getting at, I think. The cathode of a tube is an input too, if the driver is low enough impedance, and speaker output currents are about as low an impedance as tube amp signals get. Stray capacitance has a hard time driving something as low impedance as a cathode.

So back to what Bepone said :

“speaker jack terminal must be isolated from the chassis, and NFB live wire and GND wire travelling together, to the PI, live wire need to go to the gnd reistor, and gnd wire must go to the PI last tail gnd resistro point, and there must be also connected PI supply electrolytic capacitor.”

Easy enough to do with my layout , simple as unsoldering two leads and moving them to the resistor and replacing speaker jacks .

If it was tuning in the feedback path, there is still a problem hiding in there. Obviously, if that wire isn't held in place so that over time and vibration it can't wander back toward the chassis, the oscillation can return. That brings up the question - was it nearness to the chassis, or near/far away from another wire or terminal?

I definitely I am not afraid to use an appropriate adhesive lol

The cathode follower grid wire closest to the LTPI tube was the biggest culprit, and the LTPI cathode lead a half inch away had a the effect . Lifting these leads slightly remedied this .

Moving the opposite grid ( the higher plate load LTPI fed half ) was much less effected, almost seemingly negligible . Moving the presence leads did nothing . Moving the nfb lead did nothing

Before finding the problem , attempting to remove transformer from chassis and running any secondaries as far away From any potential antenna as possible did nothing .

If there is still a problem lurking that I could better resolve through either what Bepone mentioned or through some yet undiscussed means I would be happy to implement . I had thought of shielding the grids to the cathode follower, but i that was going to be my Hail Mary after rearranging the grounding scheme.

One other issue raised its head in my mind. Do your output tubes have grid stoppers? Grid stoppers work by reducing the RF gain of the tube. The parasitic oscillation might be from an un-stoppered output tube. It's also possible that putting a stopper on the PI tube(s) could cure a sensitivity that the PI input is getting from somewhere.

Yes ,grid stoppers check 3.3k on each grid

[WhopperPlate]

One last obvious honorable mention that has quenched this further:

in the hunt for solutions I installed shielded cable from the master to the fx loop , but did so temporarily without the jacket soldered in place to the rear of the pot , I just alligator clipped it for a quick test. I soldered it in place permanently at the end .

I tend to leave this run unshielded , as I like minimal high end loss .

However the cathode follower grid driver is far more sensitive to interactions obviously , and while it didn’t do anything when it was really problematic ,the extra shielded run was enough in the very end to really stabilize everything further and make the leads that much less interactive to movement.

I guess it’s hard to keep track of everything after this long lol