I think I know what I'm doing.....

[G._Hoffman]

Actually, I think I've got just enough knowledge to do serious harm, without actually knowing enough to do things right.  Far and away the worst place to be.


So, I just "finished" wiring up a new amp last weekend.  I found a few idiots during the idiot check (a wire I forgot to do, and one or two in the wrong places), but when I powered it up it passed signal, and did more or less what it was supposed to do...except for one thing.  120hz buzz.  A lot of it.  I tried a bit of a modification to the ground scheme, and it made a small difference (maybe 2-3 dB).  I tried mucking about with the lead dress, and got nothing (which is actually kind of satisfying, since it means I wasn't a complete idiot when I laid everything out, which I was afraid of).  I put a 10nF cap across the main power supply caps (two 100uF 450V caps in series, with a couple 220k resistors to keep the voltage even across the two caps and act as bleeders, etc.), and THAT made a huge difference - a good 10-12dB or so.  But it is still there, and now I can hear a 60hz hum too, though that isn't all that loud and I assume can be fixed by elevating the center tap for the heaters.

My question is - 120hz has got to be in the PS, right?  There is no way I'm getting 120hz out of the heater circuit, right?  That could only give me the 60hz?

The ground scheme is what Merlin suggests in his book and on his website, and everything that can be shielded is shielded.  All of this has worked for me before. 


Gabriel

[whittele]

i would bet a guitar pick your 120 is heater

does you PT have a filament centertap and did you ground that ct lead

if the PT did not have a CT did you create an articficial center tap with two 100R res going to ground from either side of the filament supply

did you float your centertap on a dc voltage

[jjasilli]

My question is - 120hz has got to be in the PS, right?  YES (based on your expert guitar knowledge, I trust your ear that it's 120Hz.)

There is no way I'm getting 120hz out of the heater circuit, right?  RIGHT         That could only give me the 60hz? YES

One simple test is to use jumper wires to parallel ea filter cap in turn with another cap of the same value.  Make sure the test cap has solid connections to the B+ rail on the (+) side; and GRD on the (-) side.   If the hum goes away you've isolated the problem. 

[G._Hoffman]

i would bet a guitar pick your 120 is heater

does you PT have a filament centertap and did you ground that ct lead

if the PT did not have a CT did you create an articficial center tap with two 100R res going to ground from either side of the filament supply

did you float your centertap on a dc voltage

Heater CT is currently tied to ground.  I tried floating it on one of the power amp cathode resistors (the power amp is Kevin O'Connor's AC30) which is at about 8-9Vs (If I remember correctly - it might be as high as 12V, I don't remember), and it made a small difference in the 60hz.  I put it back to the main star ground just because I want to try and get one thing at a time.  It seems odd to me that the heater could create a stronger 120hz than 60hz, since it is only going through the transformer and so shouldn't be at a different frequency from the wall.  But hey, maybe I'm wrong.

Oh, and this is only a problem when V1 is in - without it there is only a small 60hz hum.  V1 is two different input stages, which can be used separately or in parallel (another thing out of Merlin's book).  If it will help, I can post the schematic when I get home, though it will likely be a while as I have a meeting I need to go to tonight.

My question is - 120hz has got to be in the PS, right?  YES (based on your expert guitar knowledge, I trust your ear that it's 120Hz.)

I don't know about the guitar work, but the years of audio ear training when I was at Berklee make me pretty confident in that one.  Also, years of running down 60 cycle hum in live audio rigs!  I'll try that test with the caps tonight.  Are you thinking I've got a bad cap, or that I've got a bad ground/V+ connection?


Gabriel

[Fresh_Start]

Start with JJ's suggestions.

If you post a schematic, could you show each ground point with an identifying letter or number?  Or did you do an actual, strict star ground?

Aiken on grounding:  http://www.aikenamps.com/StarGround.html

Merlin:  http://valvewizard2.webs.com/Grounding.pdf

Doug Hoffman:  http://www.el34world.com/charts/grounds.htm
http://www.el34world.com/charts/currentflow.htm

No, they don't agree.

Have you checked continuity to ground for every point in the circuit which is supposed to be grounded?  In this case, the input jack and area around V1.  A loose tube socket connection could be the culprit.

Does your amp have negative feedback?  If so, where is the speaker jack grounded in relation to the filter cap for the stage the NFB returns to?

Good luck finding the gremlin(s)!

Chip

[jjasilli]

Oh, and this is only a problem when V1 is in - without it there is only a small 60hz hum. That would be my next suggetion, pull tubes.  Because the 120 Hz hum goes away with V1 pulled, then the problem is most likely: a) that tube; or b) in the power supply @ the last filter stage -- could be a bad cap or bad physical connections in the B+ rail, or to ground.

[G._Hoffman]

Oh, and this is only a problem when V1 is in - without it there is only a small 60hz hum. That would be my next suggetion, pull tubes.  Because the 120 Hz hum goes away with V1 pulled, then the problem is most likely: a) that tube; or b) in the power supply @ the last filter stage -- could be a bad cap or bad physical connections in the B+ rail, or to ground.

I've tried several different tubes, including one known good from my other amp, so I'll try the PS issues when I (finally) get home. 

Now I've got to go to this bored meeting...err, make that Board Meeting.


Gabriel

[eleventeen]

Maybe the simple answer is to add a filter cap very much closer to your V1. Something like a 20-40 ufd. In case it isn't obvious, make sure you don't connect the + of any added caps right to the plates; you want to connect on the high side of the plate resistors.

[G._Hoffman]

Maybe the simple answer is to add a filter cap very much closer to your V1. Something like a 20-40 ufd. In case it isn't obvious, make sure you don't connect the + of any added caps right to the plates; you want to connect on the high side of the plate resistors.

The filter caps are distributed through the amp, so it's as close as I can get it already.  Good point about the top of the cathode resistor, though.  Unfortunately, it didn't help, and I'm really not sure what to try next. 

I'll post the schematic, though.  There are some odd things that I wanted to try, most of which you'll see if you look.  Most of the ideas for the preamp were out of Merlin's book, and the values of most things are just a starting point, and I plan to play around with them a lot, once I've got everything working right. 

One other thing, which isn't too surprising - the noise is much worse when the local NFB on the second gain stage is turned off.


Gabriel

[bigdaddy]

Before I saw the schematic I said to myself "I wonder if there's something wrong at the cathode of the first tube". Then I saw the schematic and I said well maybe it is the cathode of the first tube.

Try changing out the diode and those caps and use a pair of known to be good 25MFD caps. Also separate the 2 channels, in other words redo and simplify the input section and see if that works. If it does then rebuild it slowly checking everything as you go. 

I would also move your standby switch to right after the diodes. I had some issues with the standby switch in other places, even on some old Fender amps. If I am not mistaken you used a voltage splitter and decoupled the preamp stages. Maybe your grounding scheme needs to be revised because of that. But I think I would look into that first stage and the way it's setup.

[jjasilli]

I'm suspiscious of the 22nF bypass caps across the rectifier diodes.  Usuual value is .01 = 10nF.  For diagnostic purposes I would remove them. 

If quick diagnostic checks, per above posts, don't work, then I would get more radical as to isolation of circuitry.  Disconnect: a) the B+ rail from the preamp (everything after ther PI); b) disconnect signal from the preamp into the PI.  Does the power amp hum with no signal?  Input signal from a signal generator or other source intro the PI.  Does the power amp pass signal & not hum.  Fix any issues. 

Then Re-build B+ & signal to the last preamp stage near the PI.  Re-check for hum.  MKeep going like that 1 stage at a time until you get to the input.  BTW: speaking of quick diagnostics, is your input jack properly grounded???

[G._Hoffman]

I'm suspiscious of the 22nF bypass caps across the rectifier diodes.  Usuual value is .01 = 10nF.  For diagnostic purposes I would remove them.  

I just added them last night to see if they helped - they didn't.  I used what I had on hand that was within Merlin's suggested range (10-47 nF).  No better, no worse.  I guess I'll try mucking about with the V1 cathode stuff tonight.  I won't hurt anything running one of them with the cathode open, will I?  Because it occurs to me that if I do that I can see which half of the tube is causing the problem.


Gabriel

[jjasilli]

Cathode disconnected from ground is OK.  It turns off the tube, while keeping the heaters warmed-up.  Some standby circuits work that way.

[eleventeen]

I really do not like that LED on the cathode! Not one bit.

That LED has no (series resistor) current limiting on it, which as far as I'm concerned is not kosher all by itself, regardless of function or placement; tube, SS, nuclear, or steam. Secondly, your typical 12AX7 preamp stage is what, a 1-2 ma affair? That isn't enough to fire an LED which to my recollection wants 10-20 ma. Thirdly, I REALLY don't like the idea of diode switching (it IS a diode, after all) glitches going into the tube, especially in the V1 position. When B+ is turned on, that cathode, has no DC connection to ground below the LED turn-on threshold (it IS a diode, after all) On one hand my sense is that it would act as some sort of relaxation oscillator with the parallel cap governing period, which for sure argues against 60 Hz (seems like it would be much higher freq) ...and then you have heater power which is MUCH more likely, volts and amps, to fire the LED.

I just do not understand why that LED is there in the first place and what it's supposed to do.

[tubesornothing]

Scared of new things?  Don't let your fears grip you - try it before judging.

The LED is just fine on the cathode, I have used them on lots of amps and no noise, no worry, no explosions. 

Lots of books explain the theory: TUT and I think valve wizard come to mind.  I am sure sluckey, PRR or Merlin will chime in.

[eleventeen]

It's not at all that it is new, I've used literally tens of thousands of LEDs and worked with hundreds of digital and analog television circuits (which have or I should say had blanking intervals and embedded sync and subcarrier signals in them..far more friendly to switching than analog audio) in the 80's and 90's before returning to tube stuff.

Generically, why would I want a non current-limited diode turn-on switching transient in my audio path at the point of maximum sensitivity, an amplifier V1?

[jjasilli]

If you search the Forum or Merlin's site, you'll find LED's used in place of cathode resistors.  They actually limit voltage, if that's the right term, to just what's needed.  Different color LED's yield different voltages on the cathode.  With an LED in place of the cathode resistor on the trem tube, and mounted on the amp's panel, you can watch it blink in time with the trem setting!

[PRR]

> That LED has no (series resistor) current limiting on it

The tube and plate-load IS the "series limiting resistor", as you say:

> your typical 12AX7 preamp stage is what, a 1-2 ma affair?

Assume 400V supply, 100K plate resistor, tube has gone dead-short. You get 3.98mA in the LED and 1.6 Watts in a 0.5W-1W resistor. Which dies first?

> That isn't enough to fire an LED which to my recollection wants 10-20 ma.

No. Above 20mA it melts and dies. Below 1mA (old surplus) or 0.1mA (modern) it leaks, will not develop 1.5+V and will not light.

You can run an LED anywhere between leakage and meltdown. 1mA-2mA is very common for indicator light on stage pedals.... with modern LEDs, anything more leaves after-dots on your retina.

> diode switching

The tube is a diode too. (Try working with plate negative of cathode.) The trick is to keep the tube in the FORWARD direction, and not too close to the edge of that quadrant. If we manage that (we do), then the LED will also be forward bias and semi-constant impedance.

> act as some sort of relaxation oscillator with the parallel cap

You are showing your age. Yes, a NEON bulb against a cap will oscillate because plasma has a Negative Resistance in its characteristic. LEDs don't, very monotonic.

However the cap should not be needed, should do nothing. The forward-bias LED's impedance is dozens of ohms. The cathode impedance is 600 ohms or more. Reducing LED impedance with a bypass cap has nearly no effect. In this case, C12 100nF gives a quarter dB more paper-gain above 83KHz. Waste of a dime.

> I just do not understand why that LED is there in the first place and what it's supposed to do.

We need the cathode standing 1V-2V higher than grid. The LED does that.

For the record: I do NOT like it either. This is FIXED bias. Use self-bias whenever possible.

[G._Hoffman]

I really do not like that LED on the cathode! Not one bit.

Others have done a fine job of explaining why it works, but I just wanted to try it.  I'm still dealing with the buzz and hum, but I had a chance to open it up and play for a few minutes tonight and it sounds pretty damn good.  Not what I was expecting, and I'm going to spend a lot of time playing with different cathode setups (other colors, actual resistors, bypass caps or not, etc.)  (And for what its worth, PRR, in Merlin's book he says you must bypass LED's on cathode biased tubes.  Not knowing, one way or the other, I did what the book said.  At the very least, that tends to be a good starting point.)

It's not the LED input that is causing the problem, though.  I disconnected the cathodes in turn, and the problem was with the "Low" input, which has the resistor.  I've run that straight back to the main "star" ground.  That helped the 120hz buzz a fair bit, though its not perfect.  The 60hz is the loudest at this point.  Not there yet, but it's a step in the right direction.


Gabriel

[Fresh_Start]

Gabriel - did you implement a true star ground? 

IOW is the cathode of each triode grounded at the negative end of its respective filter cap or is it grounded at a universal star point?

Try grounding the cathodes and grid return resistors (R8 & R34) of the first two triodes directly on the negative end of cap 20 instead of at the star ground.

If you are running cathode (and other) grounds directly back to a single star ground point, I believe that's a problem.  IMHO there are pretty convincing arguments for what Kevin O'Connor describes as a "galactic" grounding system.  Filter caps are located as close to their respective circuitry as possible and their respective circuitry is grounded at the negative end of the filter caps.  Then you ground the filter caps, either with a star or with a ground bus.  This minimizes the loops current travels through.  KOC describes it much better than I can in The Ultimate Tone Vol. 3. 

Look at Fig 13.9(a) in Merlin's article linked to above.  Exact same idea. 

Look at Hoffman's current flow diagram:  http://www.el34world.com/charts/images/CurrentDiagram2.gif  If you study this and Hoffman's grounding scheme, you'll see that he comes very close to the "galactic" ground concept too.

I hope this doesn't come off as lecturing.  Other amp builders have good success with a variety of approaches to grounding.  I just know what has worked for me (and what hasn't).  It took me a long time to realize that the experts I respect the most were actually saying virtually the same thing - it just sounds/looks different until you study it carefully.

Hope this helps,

Chip

[jjasilli]

PRR:  For the record: I do NOT like it either. This is FIXED bias. Use self-bias whenever possible.

Fixed bias -- it becames obvious after you point it out!  Can you explain your preference for self bias here:  operational reasons? subjective tone preference?

[Fresh_Start]

PRR:  For the record: I do NOT like it either. This is FIXED bias. Use self-bias whenever possible.

Fixed bias -- it becames obvious after you point it out!  Can you explain your preference for self bias here:  operational reasons? subjective tone preference?

I hadn't thought of it terms of fixed bias either but that's exactly what an LED on the cathode does - provides a fixed voltage between the cathode and the grid. 

I think it wouldn't reduce sag because our preamp triodes are running Class A anyway and plate current doesn't vary much.  However, it doesn't allow each triode to reach its own happy bias point.  With a cathode resistor, a tube which naturally draws more current would settle at a point where the cathode voltage is higher (I think) and vice versa.  With fixed bias, the tube can't self-balance. 

Does that make its response "stiffer"?  "Colder"?  Don't know.  Does it mean there would be even more difference between how different tubes sound?  My guess is "yes".

Sorry for rambling,

Chip

[G._Hoffman]

Gabriel - did you implement a true star ground? 

IOW is the cathode of each triode grounded at the negative end of its respective filter cap or is it grounded at a universal star point?

Everything is grounded to its respective filter cap.  I tried moving the cathode of V1b to the central star point and it got better.  I'm not fond of it, but it seems to be working.  I may try to alter that if I can find a way to do so, but for the moment it is a kind of progress.


Gabriel

[PRR]

How is PT HV CT connected to first filter cap?

This has to be short and DIRECT.

If PT HV CT goes to chassis, and first cap goes to chassis somewhere else, the chassis is full of crap-current.

[G._Hoffman]

How is PT HV CT connected to first filter cap?

This has to be short and DIRECT.

If PT HV CT goes to chassis, and first cap goes to chassis somewhere else, the chassis is full of crap-current.

They go to the main star point, which is a turret on a separate board, so I can have a signal ground lift.  I'll try taking the CT directly to the first filter cap, though.


Gabriel

[Fresh_Start]

How is PT HV CT connected to first filter cap?

This has to be short and DIRECT.

If PT HV CT goes to chassis, and first cap goes to chassis somewhere else, the chassis is full of crap-current.

IIRC the importance of PRR's observations is that ALL of the current in the amp goes through the loop from the PT center tap => negative end of reservoir cap => rectifier => PT secondaries.  See Fig. 13.5 in Merlin's article linked to above (Reply #4).

Another annoying question:  how and where is the speaker jack grounded?  Not having read Merlin's book, I would guess that he recommends an insulated speaker jack.  There is a lot of current in that ground and it needs to be kept away from any sensitive signal wires.  If your circuit had NFB, grounding the speaker jack at the filter cap for the stage where NFB is re-inserted seems to work well.  Otherwise, the base of the reservoir cap might be a good point to try for the speaker jack ground as well.

Cheers,

Chip

[G._Hoffman]

Well, it's all going to have to wait for a week or so, because the little tab for the CT on the PT broke while I was replacing the wire with a longer one (its a reused PT, and when I first used it I was a bit heavy handed with the soldering iron.  I've gotten better since then).  There is no way for me to feel I can make this safe right now. (The CT is still connected to the coil, but those coil wires are an awfully small gauge, so I'm not using it.)  And because I need a PT that will fit the existing chassis, its an expensive one.      I could probably make it safe enough to keep testing, but I'm just going to get a new PT and work on the cabinet for the next week.  The nice thing about making the cabinet is that I actually DO know what I'm doing, so I can get a lot accomplished in a short period of time.


Gabriel

[G._Hoffman]

On the other hand, as frustrating as the whole PT thing is, it does have the great advantage of allowing me to spend a night doing woodworking, where I don't have to deal with the vagaries of any sub-atomic particles, and can just worry about good old fashion macro physics, and actually know what I'm doing.  I spend about 10 hours at the shop tonight, and nothing went wrong, nothing is going to have to be redone, and nothing is going to create problems down the road. 

It really is nice to, at least occasionally, work on something you actually understand, and to actually get something done!!!!


Gabriel

[G._Hoffman]

So, to update anyone who is interested.....


I got the new PT on Friday, and put it in on Saturday (I decided that I should spend Friday at the shop, since the 20 inches of snow we got Friday night/Saturday meant I wasn't going anywhere on Saturday - or today, for that matter!).  While I was at it, I moved all of the grounds to the reservoir/first filter cap.  I fired it up, and hey presto! it is just about perfectly quiet.  At least, it's the quietest amp I've made.  So, as I said, I know just enough to make trouble for myself.  So, just a bit of tweaking and it'll be done.  The amp is a bit bright, and I'm going to have to try to tame that a bit!

And for the record, it turns out I"m not fond of the sound of the LEDs on the cathodes, and I've already changed the one on the second stage (which was not able to keep up with the amp when the volume was above about 8).  I'll be changing out the other one whenever I can next turn the amp up and decide what I actually want to use.  The LED channel is excessively bright, and has a very harsh glassy sound (kind of like the ting from a microphonic tube, but following the note you are playing).  Not a good sound.  Not the nice smooth glassy, but a nasty harsh glassy.  Still, at $0.08 for an LED, it was worth trying for myself.  Other colors or different bypass caps might make it work better - I don't know.

At any rate, thank you all for your help. 


Gabriel

[G._Hoffman]

It's done!  Not the greatest Tolex job, and the grill cloth is a bit loose in places, but it looks OK for what is, after all, my second Tolex job.  There are a bunch of other pictures on my Flickr account.











(Of course, are any of these things ever REALLY done?  I'm going to be doing plenty of playing around with it, and I'm going to add a line out taking a tap off of the output transformer.  But for the moment, it works just great, and it sounds wonderful.  Its a bit loud before it starts to distort, but since I built it primarily as a clean amp, I'm good with this.)


Gabriel

[CraigB]

Nice!  The only unattractive thing I see in the pics is the grill cloth.  The rest looks killer, especially inside!  Pull the speaker and use a hair dryer to heat the cloth (I prefer a ConAir over a heat gun), and it might snap tighter and get the wrinkles out, or at least subdued.  Careful, if you get too hot and too close, it'll melt.

[G._Hoffman]

Nice!  The only unattractive thing I see in the pics is the grill cloth.  The rest looks killer, especially inside!  Pull the speaker and use a hair dryer to heat the cloth (I prefer a ConAir over a heat gun), and it might snap tighter and get the wrinkles out, or at least subdued.  Careful, if you get too hot and too close, it'll melt.

The grill cloth is paper,so I don't think it will shrink much, but it was cheap and I may replace it at some point.  But thanks!  I'm reasonably happy with everything but the corners on the Tolex, which I just covered with the nickle corners and they look fine.  There is a bump in the purple bit around the control panel, but as soon as I get around to it I can iron that out.  I really like that about hide glue!


Gabriel