Troubleshooting buzzing

[cdemike]

Just finished wiring up a "space saver amp" for my wife and I to do shows together. The idea of the amp was to have one amp that has 2 truly separated channels so we could play small coffee shop gigs together and haul less. Channel 1 is basically a brown panel Princeton (6G2) and channel 2 is a Vox AC30 using a plan triode input stage rather than a pentode. Each feeds an independent output section based around one ECC99 each with 300V plate volts feeding a Hammond 125C set at 12.8K primary impedance. I used the output section from the "AC30-XS," which is a similar ECC99-based amp using the same output transformer at the same primary impedance. I've attached a schematic for reference. Please forgive the crappy scan. The filaments are on an artificial center tap formed by 2x 200R resistors. The part that's cut off on the bottom just shows the filter caps, the tail of the bridge rectifier, and the power transformer's "shield" lead connecting to ground.

The problem I'm having is that the "Fender" side oscillates at full volume (8W) no matter where the volume is set. The pitch changes if the volume control is above zero, suddenly switching from a low-pitched sound to a very high pitched sound. The tone control also impacts the buzzing when the volume control is above zero. The buzzing goes away when V1 is unplugged, and the pitch of the buzz changes depending on which 12AX7 I swap into the V1, but only slightly. The buzzing also stays even when V2 is unplugged, which tells me the issue is with the power section, is a lead dress issue, or a ground issue. Could be something else, but those are the things that come to mind. The 12AX7s I used work fine in other amps, so I think I can rule out microphonic or otherwise bad tubes. Weirdly, the pitch is responsive to probing at V2A's anode when I touch it with a multimeter probe, so it seems like there is amplification happening. I can also hear clicking from the amp chassis when it oscillates.

The "Vox" side, on the other hand, is completely dead -- nothing at all when all controls are full up, full down, half way, or any combination between (including independently setting the volume control at different levels with other controls at half, zero, or full up). I've tried plugging a guitar in and tapping the input stage triode, but it's totally dead.

I've tried the chopstick method to resolve the oscillation issue on the Princeton channel but didn't get any results (tried it with and without the guitar plugged in). I have also made sure my connections were good and found a resistor value error on one of the output grid stoppers (read: nothing major that would cause damage, just misread a 220K as a 22K), but the problem was there before and after without much change. With no tubes in it it runs draw very little current through the light bulb current limiter (40W bulb barely glowing), but with the output tubes in it glows around where it would if it were connected to a half-up dimmer switch. I figure that may have something to do with the full-output oscillation, but I could be off-track.

The amp was put together going against some "conventional wisdom" rules about lead dress (power transformer is mounted in the middle of the chassis -- space saving due to the weird mounting problems that come with using a toroidal transformer), but I figured the amp has such a low output that hum wouldn't really be too bad especially given the lower field coming from the toroidal transformer. I should mention this is my first time using a toroidal transformer, though, so I'm definitely feeling green trying to troubleshoot it.

If it helps, I made a video showing the problem: https://youtube.com/shorts/QCEwbCXoeZk?feature=share

[ChopSauce]

I'm sorry I can't help you beyond mentionning that whatever you meant by "buzz", it most likely is "of type signal/noise ratio", that is independent of the "low output" thing you mentionned - yet more audible as the output volume increases.

I even had noise issues when following the "conventional wisdom" rules ...

[Stevem]

For starters let's test one amp at a time.

First unpower all the B+ nodes feeding the Vox amp section, lets work on one amp at a time here!

Next swap the primary wires on the fenders output tube that feeds the output transformer.
When you then power the amp on does the Oscillation issue go away or get worse?

Also please post a complete schematic of your power supply with the nodes marked as to what feeds the Fender section and what feeds the Vox section.

A basic rule of Thumb for any power supply is that if any one node is feeding more then 3 gain stages then your asking for problems like your having!

It's far better to power no more then two gain stages from any one power supply node.

[R.G.]

Spot on again, Steve.

@cdemike:
Go read this: https://audioxpress.com/article/Glass-A ... Champ/1867

As stevem says, part of the issues you're having with the Fender side is lack of proper decoupling. You're probably having grounding issues as well. And also as Steve says, work on one amp at a time. Pull out all the Vox tubes. Get the Fender side working. Then work on the Vox side.

[cdemike]

Thanks everyone for the help so far! I'm currently reading through the Champ article -- thanks for passing it on, R.G.! [Edit: just finished the article: I think the lack of decoupling may be the issue here -- would decoupling the output transformers be a good starting point? Since the issue appears to be in the power amp, my hunch is that working from power amp to preamp would be a good strategy, but tell me if that's off. My next thought would be to increase the resistance between the power section nodes and each preamp node 20K, i.e., 2x 20K resistors branching from the shared power section node (assuming sharing that node is okay). From there, I'm not really sure how to proceed since the 6G2 schematic doesn't show decoupling past there, so I'm not sure if there's a guiding principle in determining the values. Here's the schematic I used as my reference: https://schematicheaven.net/fenderamps/ ... _schem.pdf]

Sorry for the confusion with the schematic. I really should have done it in pen to get it to scan better, but I did transfer it to an electronic version to better show what I did (attached). Each side is divided into 2 nodes with one feeding an output transformer center tap and the other feeding the preamp. On the Fender side it appears that reaches the rule-of-thumb maximum of 3 gain stages, and on the Vox side that's 5 triodes on a single preamp node. I might have made a mistake in copying the "AC30-XS" if that's bad practice (attached layout for reference). Just so I'm learning what I need to, would the cathode follower be considered a separate gain stage from the anode-coupled gain stage that typically precedes it? In other words, would V3 in this schematic be considered 2 gain stages or a single stage? Same question for the LTPI -- that'd also be two gain stages, right?

I'll get to work on taking apart the B+ feeds to the Vox side and swapping the output transformer feeds and report back. If it helps at all in the meantime, each test was done with one 1 output tube installed at a time, with V1 unplugged when testing the Vox side, and with V3 and V4 unplugged when testing the Fender side.

[Stevem]

Also it needs to be mentioned that since you have two separate amps in one chassis and both will be played thru at the same time you need two separate power supply sections other then the Power transformer if it’s big enough to supply the total current needs.

Also there may be a grounding / out of phase issue with each output section taking place since one amp may be out of phase with the other due to a different number of gain stages.

I will look over the schematic again when I have more time later, or maybe someone else can chime in before that in regards to the phase of each speaker output.

[maxkracht]

Picture(s) of the inside of the amp would probably be helpful.

As others have mentioned, you probably need more decoupling and more isolation between the two amp sections. In practice, no more than 3 gain stages usually means one filter cap and dropping resistor per preamp tube. You can sometimes get away with less, but the cost saving isn't worth the trouble. It is probably also wise to have separate plate and screen supply caps for each power amp section, and to isolate each amp's grounds from each other as much as possible.

Aside from that stuff, assuming it isn't a typo, 120pf coupling cap on the vox side v1 is awfully low.

[maxkracht]

Just looked at your schematic again. I think you are missing grid leak resistors on the vox output and you should have 5000pf instead of 50pf on the fender tone control. Also, I didn't notice you were using triodes for the output tubes, so ignore my mention of a screen supply. Instead, maybe use a large shared cap after the rectifier then split into two separate power supplies with separate grounds. Crude drawing, you can ignore values, but this might give you a better idea. Others might have different ideas, but that's probably what i would do.

IMG_9557.jpg

[cdemike]

Y'all are the best -- thanks so much for helping me figure this out. So I tried gradually reducing the power supply as suggested and found the following didn't resolve the oscillation:
-Disconnecting the Vox side completely
-Disconnecting the parallel 30uF capacitor at the reservoir (didn't fix the problem but there was some improvement, actually)

When it finally came to completely disconnecting the preamp section (including the NFB lead), there was no more problem. Weirdly, the output section got biased hotter the further I run the amp, so in the process I increased the cathode resistor to 500R, which also reduced some oscillation. It's still bypassed with the 100uF cap. The cathode resistor and bypass cap were previously mounted on the board, but during the process of changing the resistor, I migrated them to the socket and to a ground tab by the socket.

Also it needs to be mentioned that since you have two separate amps in one chassis and both will be played thru at the same time you need two separate power supply sections other then the Power transformer if it’s big enough to supply the total current needs.

Also there may be a grounding / out of phase issue with each output section taking place since one amp may be out of phase with the other due to a different number of gain stages.

I will look over the schematic again when I have more time later, or maybe someone else can chime in before that in regards to the phase of each speaker output.

Thankfully I think the PT I bought should be adequate, since it's rated for 130mA. I'll likely need to rebias the Vox side as well, so would mounting the cathodes to the socket and to a ground tab by the socket solve some of the phase issues? It sounds like you and maxkracht are in agreement over that first node potentially being a source of problems. I'll give his proposed power section filtering arrangement a shot.

Picture(s) of the inside of the amp would probably be helpful.

As others have mentioned, you probably need more decoupling and more isolation between the two amp sections. In practice, no more than 3 gain stages usually means one filter cap and dropping resistor per preamp tube. You can sometimes get away with less, but the cost saving isn't worth the trouble. It is probably also wise to have separate plate and screen supply caps for each power amp section, and to isolate each amp's grounds from each other as much as possible.

Aside from that stuff, assuming it isn't a typo, 120pf coupling cap on the vox side v1 is awfully low.

Will definitely get a layout and pictures posted. I have a weird attraction to the Hiwatt style approach to layouts, but I find it makes it hard to see what's happening with so much hidden under the board when I'm troubleshooting. Hopefully a drawing would help if others encounter similar frustrations. Thanks for passing that on about the filtering -- I think that answers my question about the phase inverter and cathode follower. I currently have a shared ground buss for everything on the board, which seems like it may have been bad practice based on what I'm reading here and in the Champ article. Regarding the first stage's coupling cap on the Vox section, that was indeed a mistake -- that's the original bright cap value across the volume pot, which I omitted. I used the preamp circuit from Mojotone's interpretation of the top boost circuit (https://www.mojotone.com/Amp_Kits/Briti ... LjYwLjAuMA..). Good catch, and sorry for the confusion! Another mistake on there was the way I drew the grid leaks (late night drawing) on the Vox section. The weird 220K resistors in series actually is set up to be the grid leaks for that side's output section. Because this amp is really a mashup of the AC30, AC30-XS, and 6G2 circuits I skipped to drawing layouts rather than using a schematic, so that schematic was drawn when I decided it was time to ask for help with the amp.

I have a 30uF @500V capacitor lying around. I know you said you weren't too specific about the values in your power supply drawing, so I'll try wiring that as the reservoir and split the sections as you drew. Thanks, by the way, for that! I'll give that a shot and see what happens when I re-supply the PI and preamp.

[maxkracht]

When it finally came to completely disconnecting the preamp section (including the NFB lead), there was no more problem.

Not sure if I am following. Have you tried just disconnecting the nfb? If no, you should try reversing the primary or secondary of the OT.

Weirdly, the output section got biased hotter the further I run the amp, so in the process I increased the cathode resistor to 500R, which also reduced some oscillation.

If you mean further as in, turn the amp on and the bias ramps up more and more until it gets way too hot, you could be missing or have improperly installed grid leak resistors, or you could have an oscillation that is building more and more. Biasing colder won't fix either problem.

I currently have a shared ground buss for everything on the board, which seems like it may have been bad practice based on what I'm reading here

Buss grounds are fine, as long as they are in the correct order. When combining two amps in one, you could be easy to put a noisy ground from one amp next to a sensitive ground of the other or have weird interactions between the two. Probably best to keep all grounds from one amp on a bus, and all grounds from the other on another bus. No idea what the actual best practice is in this situation, but having more control is probably better. I'm guessing you have other problems, but you might need to amend how the grounding is done to keep the noisefloor down.

[Stevem]

I have seen amp conditions/ problems like I will present here.

The amp was oscillating, osscillating up at a frequency that I could not hear and the speaker could not reproduce.

In fact the amp was oscillating so hard that it could barely pass a guitar signal.

I noted this because the amp was outputting nearly all the output voltage that it could because even with out my scope hooked up to see the Oscillation, I saw all of this output voltage on my meter.

Also D.C. Voltage wise I saw that all of that output voltage was pulling down the amps power supply.

This of course was also seen in how much cathode current the output tubes where pulling

Long story short, a amp can be working just fine other then a strong ossciation issue that is blocking any normal signal from passing thru the amp.

[cdemike]

When it finally came to completely disconnecting the preamp section (including the NFB lead), there was no more problem.

Not sure if I am following. Have you tried just disconnecting the nfb? If no, you should try reversing the primary or secondary of the OT.

Weirdly, the output section got biased hotter the further I run the amp, so in the process I increased the cathode resistor to 500R, which also reduced some oscillation.

If you mean further as in, turn the amp on and the bias ramps up more and more until it gets way too hot, you could be missing or have improperly installed grid leak resistors, or you could have an oscillation that is building more and more. Biasing colder won't fix either problem.

I currently have a shared ground buss for everything on the board, which seems like it may have been bad practice based on what I'm reading here

Buss grounds are fine, as long as they are in the correct order. When combining two amps in one, you could be easy to put a noisy ground from one amp next to a sensitive ground of the other or have weird interactions between the two. Probably best to keep all grounds from one amp on a bus, and all grounds from the other on another bus. No idea what the actual best practice is in this situation, but having more control is probably better. I'm guessing you have other problems, but you might need to amend how the grounding is done to keep the noisefloor down.

My apologies for the unclear response. I didn't try that -- I disconnected all connections between the preamp and power amp all at once. I'll see what happens when I reconnect it with the OT secondaries reversed. With respect to the bias issue, I just noticed it was hotter than when I first tested it. The mains in my building (could be a Birmingham issue) is all over the place, especially with the heat wave, so I imagine that might have something to do with it. It does get wildly hot when the oscillation gets bad, like when the volume is all the way up, so I'm glad to learn that is related to the oscillation and not another issue with the amp. I'll work on getting the grounds separated when I pull the board tonight.

Steve -- that's really interesting. I made the mistake of not checking to see if there was any inaudible output from the Vox side before I disconnected everything. I wonder if that was another oscillation issue but above audible range like in your case. Thanks for passing that on!

The plan tonight is to redo the power supplies with better decoupling and to get pictures to post. My layout got water damaged, but I'll also try to redraw it.

[cdemike]

Turns out my hasty decision to remove the preamp and negative feedback at once hid an important clue. The oscillation is much, much better without NFB. I was able to redo the power supply to decouple the OT from the preamp in a way similar to maxkracht's drawing with some value changes. The stages are 30uF reservoir feeding parallel 10K decoupling caps, then 30uF/1.8K. I also did a small test with the preamp hooked up and with the NFB on alligator clips. It does have output, but very little (~0.5 watts) when NFB is connected, but I'm getting a low-ish output at around 1.25 watts when it's disconnected. The NFB-associated oscillation was present when connected to either secondary lead. I suspect the low output is due to over-correction when I changed the cathode resistor, which now elevates the cathode only 6V above ground (cathode resistor was still 500R, so about 4.5W total plate dissipation). I'll readjust, but I did some thinking and realized I did most of my bias testing while it was on the lightbulb limiter and also during a heat wave. Nevertheless, the limiter probably had a greater effect on the B+ than the lower wall voltage... When the oscillation problem is resolved, I'll bump it back down to 330R.

Currently the amp has reasonable output with the volume set below 75% and with the tone control below 20%. Above there it starts to sound like a radio tuner moving away from a station and a classic "motorboat" oscillation starts at around 3hz.

Looking back over the 6G2 schematic, there's no decoupling between V1 and V2, which I know is bad practice after the learning experiences that came through this build. My original hope was to maintain the preamps as closely as possible to the originals, so I'm considering redoing grounds per maxkracht's suggestion, i.e., 2 shared busses, as my next move. If that doesn't solve it, I'm thinking after that I'll need to decouple the PI from V1, which will likely mean major revisions to the board.

Also, as promised, here's a picture of the internals. There are a lot of flying wires from things being disconnected, and the power section is currently in shambles after messing around with this the past few days.

[maxkracht]

I have the same shoes. Damn things are getting hard to find...

The NFB-associated oscillation was present when connected to either secondary lead.

Your NFB is probably squealing because it is currently PFB. You need to change the polarity of the transformer leads. Either reverse the two primary wires going to the plates, or reverse the wires on the secondary going to the output jack.

The stages are 30uF reservoir feeding parallel 10K decoupling caps, then 30uF/1.8K.

classic "motorboat" oscillation starts at around 3hz.

Is there a cap after that 1.8k? Bare minimum, you need a decoupling cap for the power tube plate and one for the PI/preamp, but a separate one for the preamp is recommended. 1.8k is pretty low, 18k is more like it... I wouldn't worry too much about changing the power supply, It will be subtle, especially in comparison to different output tubes and other changes. Motorboat usually means not enough resistance or capacitance in the power supply or too many stages sharing a decoupling cap.

Hard to see exactly what is going on in the picture with big caps and underboard wiring, but it looks like the 1M resistor on the input is connected to v1's cathode instead of ground.

[cdemike]

So when I did some testing yesterday I detached the wire running from the speaker jack to the board on the jack side. That let me alternate between tip and sleeve, and in both cases the oscillation remained but also the total voltage across the jack also decreased, so it would seem that there is some negative feedback happening at least on the tip? The sleeve is grounded, so I really just was tapping the wire while watching the multimeter so that I didn't burn up V2 by bypassing its cathode resistor for long periods of time. The oscillation is still there regardless of NFB, though, so I'm not really sure how to proceed. I did also mess around with raising the resistor's value from the 1.8K that was there to 4.7K, then 15K, then 33K. Raising the value did reduce some oscillation, but going higher brought B+ at V1 down to around 90V, so I'm hesitant to go higher. I also added capacitors to the V1 node: 32uF, 30uF, 470nF, and 22nF. I know the last two values are way too small so they were really just to see how the oscillation was impacted. The small values made it much worse with oscillation starting earlier on the volume knob, and I saw a maximum of 12V across the 8ohm dummy load when it started to oscillate.

So what I'm left with now is that the amp can't have the tone control turned up above about 1/3, and the volume control can't go above about 2/3. The higher the tone setting, the lower the volume setting threshold to start oscillation. My multimeter has a frequency measurement function, and the pitch goes up above 20hz at higher settings on either control. B+ isn't reaching the Vox section at all, and the only tubes installed are those associated with the Fender side. I'm not really sure what to do past this other than raise the filtering overall, which I'm open to, but it's hard for me to believe that this preamp isn't working when it's exactly the same as the 6G2 circuit except for deletion of the vibrato section.

Is there anything I'm overlooking? I included a drawing of the version with the heaviest filtering if it helps.