Marshall style build noise...

[Fiat_cc]

text deleted

[Fiat_cc]

The basic noise appears to be 100Hz, but there is all sorts of hash and distortion on it.

[EL34]

This post was corrupted
I had to delete some of the replies to fix the problem

[EL34]

Here's some of the text that was deleted




I am starting to seriously lose faith in my ability.  The last two builds I've done suffer from unacceptable noise.  Both are similar in that the output section seems quiet, and the phase inverter section seems quiet, but adding tubes prior to that starts adding an increasing level of unacceptable noise.  The first one was a Fender AB763 style build.  This one is a JCM 800 style build.  I have played with moving ground points around, disconnecting the relay power supply, re-flowing solder joints etc.  I have attached ample gut shots, a schematic, and a layout as it currently stands (I think they are all correct).  The schem doesn't include the relay supply, but it is just Hoffman's relay PSU board.  Currently I have pulled V1 and V2.  V3 generates noise.  Grounding the junction of R13, R14 and C7 significantly reduces it, but lifting R10 didn't improve things, and perhaps made things a little worse. 

This is so similar to the issue with my last build it is eery.  That one had a more hiss like quality to it (with some underlying buzz), and this one is more buzz/hum like in quality, but they both originate in the preamp, and seem to be virtually untraceable so far. Neither responded to moving grounds around (tried lifting the ground luck at the input and running it back to the main star at the power transformer, changing the order of things on the ground buss, moving the MV ground from the preamp buss to the phase inverter star, moving the output secondary winding ground reference from the phase inverter star directly to the main ground, rerouting grounds away from the PT etc).  Neither seemed to respond to lead dress. I've rolled so many tubes through each amp, it can't possibly be a tube. I've checked voltages and bias, and this amp in particular was so spot on voltages wise, and biased up so nicely, if you were going on that alone it would be text book. I've used shielded cable early in the preamp, which is indicated on my layout by the thicker yellow wires with silver stripes.  The shield is only ever grounded at one end. Both this and the AB763 have some noticeably microphonic components/turrets, and wires. Caps are all CDE silver mica, Mallory 150 630V foil, or Sprague, F&T, or Nichicon electrolytic.  Resistors are all Xicon carbon film 1W, besides the dropping resistors in the PSU, and screen resistors, which are either metal film, metal oxide, or wirewound, 3 or 5W depending on the location.

2 builds in a row.  Either I'm making a stupid fundamental mistake with something, or I have a batch of faulty components, or there is noise on my mains, but other amps I've built don't seem to exhibit the same noise (they would be classed as lower gain though).

Please.  HELP!

[sluckey]

Take the amp to a friend's house. Any better?

[Ritchie200]

I was just going to say - Is there a computer close by?  TV?  Iphone or tablet charging?  Florescent light?  Neon beer sign?

Jim

[sluckey]

There was a keyboard right beside the amp in one of the pics that became corrupted.

[Fiat_cc]

There was a keyboard right beside the amp in one of the pics that became corrupted.

That was my laptop. The noise doesn’t change with it on another room.
I have turned off heater, lights, soldering iron etc.
Completely reworked the ground scheme (it is much better grounding now), but the noise has not changed.

[Ritchie200]

Are your input jacks properly tensioned for grounding?

Jim

[Fiat_cc]

I’m not even worried about input jack. It is noisy further down the chain. I’m working my way back from phase inverter. I’m at V3 at the moment.

[Fiat_cc]

What does this scope image tell people?
The nasty jagged one is the speaker output with preamp volume and master volume at full 0.1V per division.
The smoother one is the ripple on the first filter/reservoir cap, at 5V per division.
The ripple is hardly there on the next filter, and is not measurable on my scope on the 3rd filter.

I have this same basic issue in two completely different amps with completely different circuits.  I don't have scope traces of the other amp, but the symptoms are essentially the same.

[shooter]

working my way back

Your scope shot is jibberish - no point of reference, assuming NO input signal?

I'd start at the jack with a clean verified signal, follow into the gain stages n find who/what is ucking with it

[Fiat_cc]

working my way back

Your scope shot is jibberish - no point of reference, assuming NO input signal?

I'd start at the jack with a clean verified signal, follow into the gain stages n find who/what is ucking with it

No input signal, no.

[Fiat_cc]

More scope pics.
Hopefully in the right order.

All these are with no input to the amp. Most of them large portions of the amp are disconnected anyway. There are steps in between these that I've left out as it is just more of the same, but I did this experiment very incrementally.

So, first pic is the whole amp, no input, preamp and master volumes on full, other controls at zero. Top trace is noise at the speaker output, bottom trace is the HT AC.  With the standby switch open, this awful square wave goes back to a relatively clean (relatively, its far from perfect) sine wave, and the speaker output disappears.

Second pic, is of HT AC, under same basic conditions but with a couple of preamp tubes removed.  No huge difference there really.

Third pic is HT AC with only the first reservoir cap can in circuit.  There are no tubes or circuitry, just the reservoir cap connected to the bridge rectifier. This is also essentially what the wave looked like with all circuitry connected, but no tubes installed (therefore virtually no current draw at all).

Last pic is the unloaded HT AC (filter cap disconnected by removing the HT fuse).  This is the same wave form as having the standby switch open as mentioned in the first pic.

The heater AC looks pretty much the same as the HT AC. Any load makes it really square.

Now, maybe this is normal behaviour, but I have a mate who put a Super Reverb on the bench and scoped the HT AC with standby switch open and closed, and didn't observe any noticeable change in either heater or HT AC waves.

Please, does this mean anything to anybody? I have tried two different probes, and both channels on my scope, to try and rule out the scope being faulty.  Why does even the most miniscule capacitive load destroy my power supply? It can't be EMF from the choke or OT, because the wave form changes significantly without those attached to power.  Yes, the change is more pronounce with them in circuit, but the issue seems to exist independently of them.  I've tried various things with grounding, until I literally had no grounds connected besides a physical jumper from the reservoir cap to a ground lug.  Nothing seemed to change it a great deal.

It looks like both the heater and HT windings are saturating badly, but why?

[Fiat_cc]

Just thinking out loud...
Is it possible for faulty filter capacitors to cause something like this? The 2 amps that I am having this problem with all use 50/50uF 500V caps as reservoir caps (configured differently, but the same part).  All of the caps cans of that value in both the amps have the same batch number on them.  All my quiet amp builds recently don't use that part value. They are good quality caps from a reputable manufacturer, supplied by a reputable supplier, but perhaps they were the end of a batch that had been sitting on the shelf too long and dried out?  I have some caps of that value from a different manufacturer, and also the same manufacturer but a more recent batch. Maybe I should just try it, but they are somewhat tricky to access and remove if that turns out not to be the issue.

Just trying to find the thing that is the same here. 

[Fiat_cc]

Well, I clipped in another cap can of a different brand (still just the rectifier and a cap can, no other circuitry connected), and whilst the wave form does change a little when opening or closing the standby switch, it is far more symmetrical, and looks much less saw tooth/sharks fin.
I’m a bit loathe to build the whole amp back up only to find out this is a complete red herring, but I’ve run out of other ideas, besides diodes, and they are relatively easy to replace, should caps not solve the issue.

[shooter]

give your scope a 1 VAC p-p 1Khz signal, no amp needed.  does it measure n look correct?

what is your probe rated? IF it can handle 120Vac do that next.  Look correct

back to Amp;
does it power up with all tubes in, proper DC volts, Proper bias volts? (leave ALL pots at 0) ?
IF SO;
List ALL the Plate, and Cathode VDC for ALL tubes (no signal, ALL pots 0)
List EACH PS voltage TAP BOTH VDC AND Vac

[Fiat_cc]

Yes. Scope and probes are good for the voltages, and yes, mains (240v here) looks correct.

All voltages were completely spot on. Like, text book perfect. I’m not near the workshop where my note book is with the voltage chart for this amp, but it was perfect.
Amp biased up really nicely, and well balanced at about 65% dissipation.

The audio sounds great. It sounds like a Marshall should, but signal to noise is unacceptable.

[shooter]

when you get back to it - grab these then.

List EACH PS voltage TAP BOTH Vdc AND Vac

like a Marshall should

which one so I can hunt down a schematic, that way we can start at V1a with your known good input sinewave and walk it right to the problem.

[Fiat_cc]

It’s a 2204 JCM800.
V1 is a boost stage (work in progress), but the noise is there with that removed from circuit by bypassing relay switching.

[Fiat_cc]

I can’t seem to post my schematic. Tried several times.

[Fiat_cc]

Trying a different way.

https://www.facebook.com/photo.php?fbid=10158647249152646&set=pcb.2267970873514835&type=3&theater&ifg=1

[Fiat_cc]

These are the voltages I have.
Note that, if the link I posted above works, I'm pretty sure I flipped the triodes in the tone stack driver CF arrangement between schematic and layout as it worked better for the layout (just incase you're wondering why the voltages look weird).
I can't measure all of the B+ nodes currently because the amp is currently partially disassembled.  I have literally only got a bridge rectifier and the first filter cap connected, with a 110K load resistor across the cap.  That's it. Anyway...

VAC on the HT winding is 351VAC.
Current unloaded B+ is 490VDC.
Loaded B+ with tubes biased up correctly was 455VDC.
Heaters were 6.5VAC

V1 - 1 - 147.5
2 - 0
3 - 1.26
6 - 147.5
7 - 0
8 - 1.26

V2 - 1 - 269.7
2 - 0
3 - 2.79
6 - 225
7 - 0
8 - 1.98

V3 - 1 - 308.8
2 - 167.7
3 - 169.3
6 - 167.7
7 - 0.003
8 - 1.194

V4 - 1 - 234.9
2 - 28.2
3 - 43
6 - 240
7 - 26.5
8 - 42.9

V5 - 3 - 453
4 - 444
5 - -36.5
8 - 0

V6 - 3 - 453
4 - 443
5 - -36.6
8 - 0

[shooter]

since it's all torn apart anyway, I'd try to get you PS OT wiring to look closer to these 2 examples, these layouts go 1000% toward noise free amps

[Fiat_cc]

I installed a humdinger pot, and replaced all the 1N4007’s with UF4007’s. The humdinger is interesting. Pretty much dead centre of the pot is the lowest amount of ‘hum’, but there are noticeable transient spikes in the trace. I can reduce the level of the transient spikes considerably, by then the heater voltage is way off balance and the hum becomes prominent.
So...
Why do I have such prominent spikes? I’ve done some more mucking around with trying to sort out lead dress, but nothing so far effects it.

[Fiat_cc]

For what it’s worth, you can see that the transient peaks in the signal trace line up with notable distortion in both the heater trace, and the HT trace. I’m assuming that the heater and HT windings are coupling through the power transformer, but I guess lead dress is a possibility.

Question about using my scope.
The scope specs say Max input: 600V p to p or 300V DC plus AC peak.
What is actually safe to probe? I’ll stay away from output plates, because I believe their voltage swing could exceed 600v at times, but for example, if I wanted to probe directly onto a preamp plate, is that safe? My probes are 1 to 1 or 10 to 1 switchable and rates to 600V. I’ve probes early preamp plates in 10 to 1 mode, but there is not enough signal to trigger (and I’m not 100% sure I should be probing there). I would like to probe 1 to 1 if I could, as I might be able to see where the noise is.
Likewise B+ nodes.

[Fiat_cc]

I have a couple of lead dress
And grounding things I’m going to try, but would it be worth bypassing each diode in the Bridge with a 0.047uF 630V Mallory, as Marshall do in some newer amps? That might get rid of the switching transients?

[shooter]

I’ll stay away from output plates

good idea, I have a HV probe (5Kv) and I never measure PA plates anymore too much interaction/odd things.  Grids are another place to stay away from, and "inside" a TS.

I don't believe Transients are known for hum.  possibly hiss, more likely could cause a tube to become parasitic(?)  The period times are so short their not audible.  I would just leave them till you get hum fixed.  (adding things to an existing problem only causes more rabbit holes)

Probing pre plates should work, even V1a.  start with ~~80mVrms in you should have close to Volts at plate.  1st play with your scope n signal source to see were it's at for small signals.  I can discern a coherent signal in white noise but I have 35yr practice 

[sluckey]

but there is not enough signal to trigger

I almost never use internal scope trigger/sync. Everytime you move the probe you gotta readjust the trigger level. I always trigger the scope externally when chasing a signal through an amp. The external trigger not only allows you to quit messing with trigger level, it allows you to see phase changes as you move through the signal chain. When looking at power supply ripple I use "line" as the trigger source.

[Fiat_cc]

I’ll stay away from output plates

good idea, I have a HV probe (5Kv) and I never measure PA plates anymore too much interaction/odd things.  Grids are another place to stay away from, and "inside" a TS.

I don't believe Transients are known for hum.  possibly hiss, more likely could cause a tube to become parasitic(?)  The period times are so short their not audible.  I would just leave them till you get hum fixed.  (adding things to an existing problem only causes more rabbit holes)

Probing pre plates should work, even V1a.  start with ~~80mVrms in you should have close to Volts at plate.  1st play with your scope n signal source to see were it's at for small signals.  I can discern a coherent signal in white noise but I have 35yr practice 

The humdinger pot has made it very obvious there are two different noises (a third if you include the hiss, which I’m not right now). One is hum, which I can all but get rid of with the humdinger. The other is very much a buzz. You can watch the scope and listen at the same time as you adjust the humdinger, and watch/listen to the hum decrease in volume until it’s all but disappeared, but the buzz increases in volume with a corresponding increase in the size of the switching transients that are on the corner of every cycle (100Hz where the hum is 50Hz).

[shooter]

100Hz where the hum is 50Hz).

so leave humdinger set for no hum, and move to PS.

I like schematics so I can think, so I'm thinking with this one til you have your version.
IF you have a handful of .1uf 1kv's laying around strap the diodes and test, IF you have new diodes, start there.
IF you have neither take your volt-meter and measure VAC at the 1st cap after the standby switch.  (typical from my observations ~~~ 4vac at 1st PS tap ~~ <.2vac at last tap)

[Fiat_cc]

100Hz where the hum is 50Hz).

so leave humdinger set for no hum, and move to PS.

I like schematics so I can think, so I'm thinking with this one til you have your version.
IF you have a handful of .1uf 1kv's laying around strap the diodes and test, IF you have new diodes, start there.
IF you have neither take your volt-meter and measure VAC at the 1st cap after the standby switch.  (typical from my observations ~~~ 4vac at 1st PS tap ~~ <.2vac at last tap)

I’ve tried multiple times to
Upload my schematic, both from my phone and laptop. This forum just doesn’t like the file annoyingly. I might try and export to a different file format.

Yeah, I have installed all new UF4007 diodes. The transient peaks appear narrower on the scope now, but they’re still there. I will set for minimal hum, and take measurements as you suggested.

[sluckey]

I’ve tried multiple times to
Upload my schematic, both from my phone and laptop. This forum just doesn’t like the file annoyingly. I might try and export to a different file format.

It's not complicated. Your file is either too big or it has a file extension that's not allowed. Ever see this...

Allowed file types: doc, gif, jpg, jpeg, pdf, png, txt, zip, sch, jsch, xls, xlsx, csv, dxf, vsd, vss, diy
Restrictions: 4 per post, maximum total size 2048KB, maximum individual size 1024KB

[Fiat_cc]

I’ve tried multiple times to
Upload my schematic, both from my phone and laptop. This forum just doesn’t like the file annoyingly. I might try and export to a different file format.

It's not complicated. Your file is either too big or it has a file extension that's not allowed. Ever see this...

Allowed file types: doc, gif, jpg, jpeg, pdf, png, txt, zip, sch, jsch, xls, xlsx, csv, dxf, vsd, vss, diy
Restrictions: 4 per post, maximum total size 2048KB, maximum individual size 1024KB

I’ve uploaded plenty of other files and never had an issue. It has been resized to the correct size, and is an approved file extension. If the file is too large, or the wrong file extension, the forum will take you to a page that says that. When I try to upload this one particular file, it takes me to a blank white screen. When I hit refresh, it says I’ve already posted that reply, but the post never appears.
Like I said, I’ll try converting it to another supported file format and hope it works.

[Fiat_cc]

OK, I'm trying again with a PDF version of the schematic instead of the .jpg version. Fingers crossed.
Original layout too.  Grounding has changed but I've not updated the layout as I haven't yet solved the issue, and next time I build this circuit will be in a different chassis design.

I measured ripple.  My DMM wouldn't read it.  Bounced all over the place from 130V to zero and many places in between. The scope would get a trace on the reservoir cap, which was very close to 5VAC, and typically saw tooth looking. The node after the choke is weird.  The ripple itself is very small, but it seems to be riding on top of a much slower, much bigger, and slightly random wave.  I've not seen my scope do this before. It is hard to get a photo of unfortunately.

[Fiat_cc]

Excellent! FInally got a schematic uploaded!  I've been trying since I uploaded the very first batch of chassis shots.  not sure what the issue was, but converting to PDF solved it.  Here is the best I could do with a photo of the scope on the screen node (B+ B node in the schematic).  It's weird! Or at least to me.

I've found that grounding the junction of R13, R14, and C7 reduces the hum by 80 or 90%.  I also found that, even with no tube in V1, if the boost is active (which it is with the relay un-powered due to the way the LED in the foot switch works) that winding up the Boost pot dramatically increases noise, even though it is not being amplified any more than signal going straight to the grid of V2A.

There is also the odd pop or thud noise through the speaker, even with the master down.  Not horrendously loud, but definitely audible.  It also pops sometimes when taking it out of play mode, back into standby mode.  It doesn't do it every time, and sometimes it is loud, other times very slight.  I know some noise when using the standby switch is normal, but this seems pretty hit and miss and random.  I have the standby switch in the AC line, not DC, to try and give the switch an easier life, if that would have any effect.

[Fiat_cc]

I think I have narrowed down my issue.
Using a scope probe I sniffed around the chassis for EMF/noise, and what I found was that the switching transients from the diodes are in everything.  Heater power, masses of it in the HT AC, and it is quite clearly in the air around the rectifier board. As I've seen by scoping actual signal, it only needs a little in order to make a big noise at the output. Around 10mV on the plate of V2A turns into almost a volt by the speaker output, and evidently a volt is very audible!
The question becomes how to shut it up.  I have had Schottky diodes suggested, but I don't have any on hand. I've seen a 0.047uF 1kV tied from the HT fuse holder to ground in some Ceriatone Marshall style layouts, and I've seen 0.047uF 500V caps bypassing each diode in the bridge in some Marshall schematics.  I have 0.047uF 630V Mallory's on hand.  I could try the bypassing trick, or should I try the tying DC output of the rectifier to ground style? Will 630V be sufficient (B+ is 460 ish, so it seems safe to use a 630V cap...).  DO people have any suggestions, or feel I'm barking up the wrong tree?

[shooter]

from the diodes are in everything

typically suggests poor ground, lead dress, not enough filtering.
I suggested using .1uF 1Kv around diodes.

fwiw because I build xSE amps I filter the crap outta the PS.  Look at marshall PS's where they use "stacked" filter caps at 1st tap, I use that style for most builds.

You can "hang" the .1ufs(experiment from .001 to .1 MUST be 1kv rated) about anywhere in the B+ chain, i'll sometimes put one at the V1 tap close to where the B+ feeds the plate R

[Fiat_cc]

I’ll go to the parts shop tomorrow.

I haven’t com across this issue before, in all my builds, but now being aware of it I could’ve improved the layout I think. There is a lot of this noise present around the PT though.

[Fiat_cc]

The best I could do at the local electronics shop were 0.01uF 3KV ceramics. I bypassed each diode and it changed the shape of the transients but not really the amplitude. The peaks have a slightly slower ramp up and a longer hold time, but are still quite audible.
I tried powering heaters from a spare PT I had lying around. This removes the switching transients from the output signal, but I can now hear an noticeable buzz coming from within the chassis.
I feel like this maybe a step in the right direction to isolating the issue, but I’m beginning to wonder if the switching noise is an audible symptom of another bigger problem.

[shooter]

or there is noise on my mains,

did you measure any?
did any PS fuses blow during the build/testing?

[Fiat_cc]

No fuses blown.
Mains looks clean enough.

It is switching noise. Cap bypassing the diodes did significantly change the switching noise in terms of shape of the wave, but didn’t reduce the amplitude. A snubber network on the PT has been suggested to me on another forum. I’m currently away from my workshop for a few weeks. Took the amp with me. If I get time I’ll try to install the snubber.

[Fiat_cc]

It's been a while.  This build is still giving me grief, but I have made some progress.

Back story, for those who don't want to re-read the whole thread.
This one is a Marshall 2204 style build. It has a foot switchable boost stage ahead of the standard 2204 first gain stage, as requested by the owner (an old friend of mine). He asked me to turn his TSL601 combo into something similar to John Sykes' Jose Arredondo modded Marshalls. Being as I am, I like to learn things as I go, and not just blindly copy, so I created my own version of the concept, being a 50W 2204 JCM800 Master Volume, with a parallel triode gain stage, which is foot switchable before the rest of the 2204 preamp.
Now, where I am at is, it all works, and sounds massive. The owner is super happy with the sound, (has told me it absolutely nails the sound he is going for) so I've gotten that much right.
I wrote a post recently on some Facebook groups about transformer replacement and hum. I installed a Hammond 290CEX in this build to replace the Dagnall transformer that came with the TSL601, and I now have a 2204 that is virtually hum free (obviously went from a bridge rectifier, to a regular two phase full wave style, and a Fender style bias supply).
The trouble starts when engaging the boost. Hum city. It hums like a m*****f*****! The interesting part is, it hums with the boost tube removed. I can stop the hum by grounding the output of the boost pot (which is basically grounding the input grid of the first actual 2204 gain stage). Hum wont stop if I ground the input to that same pot though. I have added an power extra filtering stage to feed only the boost stage since I drew this schematic, and I have also gone from the full bridge rectifier to a Fender style two phase rectifier, and Fender style bias feed as mentioned. This all helped the hum a lot in the non-boosted state, but not so much when boost is engaged (even with that tube removed).
I noticed that the hum changes considerable by plugging the foot switch in , or unplugging, and by waving my hand, or the foot switch plug around near the foot switch socket. I have had difficulties with this particular relay power supply before, so I am wondering if it is simply a relay power issue finding its way back into the power transformer (relay supply runs off the 6.3V supply). I am using Hoffmans relay supply, and I know plenty of people use it with no issues, but I have had hum issues with it before, and opted to run off the 5V supply with a ground reference, and it worked much better.

Anyway, I'm so close to having this perfect, and looking for suggestions, or even validation of my own ideas.
Cheers.

I always have attachment trouble with this forum, so here is a link to the Facebook group post with pics and schematic attached (hoping it works).
https://www.facebook.com/photo.php?fbid=10159037910312646&set=pcb.3190125341062484&type=3&theater&ifg=1

[sluckey]

I have no facebook account so I can't see a thing.

[Fiat_cc]

Here is schematic. It is a little out of date because my laptop got stolen, and I didn't have backups of the JSchem files, only the PDF.
Currently changes are R5 has been removed, I have swapped power transformer to a Hammond 290CEX and I'm using a 2 phase full wave rectifier, and Fender bias supply, rather than the bridge rectifier for HT and bridge rectified bias supply. Literally an hour ago I tried adding an extra power filter stage to feed V1, which uses a 4.7K dropper and a 22uF cap.

[Fiat_cc]

Trying for pics...

[sluckey]

I can't match what I see in the pics with the schematic.

[nandrewjackson]

I've got a few suggestions.  I've built several preamps/amps with the soldano/mesa rectifier/bogner/ 2204 preamp with added stage in front.


That extra stage in front of the 2204 changes the game plan.


The lead dress needs cleaned up. A centimeter here and there adds up. There's A LOT of spaghetti in those 2 pics that can be cut out of there. A LOT.


You may want to look at using terminal strips with a more point to point approach for these high gain preamps.  I made a couple using the nice and clean looking fiberglass boards with turret lugs and all the components lined up. But when I started adding that extra triode in front of the 2204, noise , noise, noise started ruining things.

[nandrewjackson]

Terminal strips around the tube socket, I've found to best for low noise high gain preamps

[Fiat_cc]

I can't match what I see in the pics with the schematic.

I lost my layout diagram with my stolen laptop.

I am currently drawing up a fresh schematic, and a new layout, and I'm going to try building it again with a new board/better layout.