distortion on leading edge of note

[pula58]

When you tried both amps, were you using the same speaker for both?

Hi Tom, yes, same speaker for both.

[fiftywatt]

You really ought to take the advice of those above who have recommended that you wire the PT centertap directly to the neg. terminal of your main filter. That is a big (perhaps biggest) part of the problem.

I haven't looked at your chassis photos, but from the sounds of the discussion, your PT centertap is wired to the same star that your large signal AND small signal grounds are wired to AND this star is grounded to the chassis along with the incoming AC green wire. That's a recipe for ... well, exactly the problem you're having (big noisy currents modulating smaller ones).

Here's what I would do to cure the problem: (1) AC green wire bolts to chassis all by itself physically near where it enters the chassis; (2) connections in the PT secondary/rectifier/filter circuit are short and direct as possible, and centertap is wired directly to the neg. of first filter; (3) PS and PA grounds can all star together (but insulate from chassis); (4) small signal grounds get their own star and tie that (and only that star) to the far end (i.e., near the input jack) of the chassis. Wire big star and small star together.

[LeftyStrat]

Just another angle. If you put the speaker cab as far away as possible from the head, does it affect things?

Higher volumes can cause mechanical issues.

[pula58]

You really ought to take the advice of those above who have recommended that you wire the PT centertap directly to the neg. terminal of your main filter. That is a big (perhaps biggest) part of the problem.

I haven't looked at your chassis photos, but from the sounds of the discussion, your PT centertap is wired to the same star that your large signal AND small signal grounds are wired to AND this star is grounded to the chassis along with the incoming AC green wire. That's a recipe for ... well, exactly the problem you're having (big noisy currents modulating smaller ones).

Here's what I would do to cure the problem: (1) AC green wire bolts to chassis all by itself physically near where it enters the chassis; (2) connections in the PT secondary/rectifier/filter circuit are short and direct as possible, and centertap is wired directly to the neg. of first filter; (3) PS and PA grounds can all star together (but insulate from chassis); (4) small signal grounds get their own star and tie that (and only that star) to the far end (i.e., near the input jack) of the chassis. Wire big star and small star together.

Here is the update on my grounding scheme as it is currently implemented (see photo in this post):

there are three ground points

gnd "1"

PT CT,
"-" terminal of 1st filter cap
cathodes of the output tubes.

All three things star connect at the chassis. Very short wire (maybe 2") from PT CT to that point. Similarly, the length of wire from the cap "-" terminal is short, perhap 3". The output tube cathodes (highlighted in the photo on green) have one cathode wired to the other, then, off to the start point.

gnd "2"

green (ground/earth) of the AC power entry
center tap of PT heater winding
PT hum shield
Ground of the bias circuit (BTW I fiddled with adding a 47uF cap between the bias (where it enters the PA at the 220K grid R's) nad the cathodes of the output tubes (at the "1" gnd") and it made no difference in this porblem that I am trying to solve.

gnd "3"
wire bus, conncted to the chassis about 1" from the (isolated) input jack. The input jack gnd is connected to the bus where it anchors to the chassis. The bus is only connected to the chassis at that one point. All the preamp cathodes connect to the buss, as do any other things that "need" gnd (pots, feedback network of the PA, etc).

The screen filter cap, phase inverter supply filter cap, and preamp supply filter cap "-" terminals share a wire that terminates at the input jack chassis gnd tie-in.

** I need to go down to the lab and take some photos of the chassis, as it is today, and post......

[pula58]

today I tried the following experiments

1) new recitifer diodes - made no change
2) put a 100nF poly cap acorss the 1st filter cap (made no difference)
3) Gave the AC power entry its own dedicated connection to the chassis,
4) tried moving around the OT (moved it near, and father from the choke, also, changed it angle (wrt the choke, from parallel, to 45 degrees) - made no difference
5) The PT CT has a 2" wire to the #1 star gnd. The "-" terminal of the 1st power supply cap uses a ~ 3" length of wire (AWG 20) to the same point, Do you guys think that this amount of wire could be causin gthe porblem I am having. I keep thinking that it can't be that..


Of course, I have spent quite some time on this build, trying to see what is wrong by looking at what could likely cause the problem (sidebands). Perhaps it is time to look at things that couldn't possibly cause this to happen?!


The weird thing about this is that the sidebands are not there at all under normal operation. But, as soon as the output of the amp starts to clip (and this is when the grids of the output tubes go into conduction) the sideband tones start to appear. Normally, I don't play the amp @ full clipping. But, when plucking a string hard, the amp momentarily clips, and you can hear this non-harmonic distortion -yuch!

[martin manning]

That second cap from the left is your screen node cap, yes? I'd try disconnecting its negative lead from the caps to its right, and ground it with the power tube cathode ground.

[pula58]

That second cap from the left is your screen node cap, yes? I'd try disconnecting its negative lead from the caps to its right, and ground it with the power tube cathode ground.

Thanks for your post.

Yes, that is the screen cap. The power tube cathodes are currently grounded at the PT CT/1st filter cap star-gnd (gnd # 1 in my photo a couple of posts above this one). You mean to gnd the screen filter cap there Martin?

[martin manning]

Yes. I think the way you have it the screen current is mingling with the preamp filter current.

[cxx]

[quote="pula58"


The weird thing about this is that the sidebands are not there at all under normal operation. But, as soon as the output of the amp starts to clip (and this is when the grids of the output tubes go into conduction) the sideband tones start to appear. Normally, I don't play the amp @ full clipping. But, when plucking a string hard, the amp momentarily clips, and you can hear this non-harmonic distortion -yuch![/quote]

Are you saying that you go into grid conduction as soon as you see the beginning of clipping? If so, that seems strange. I would think that you would start to distort well before that happened and would see conduction when severely overdriving the grids. What are your bias voltage/current?

Am I wrong about this, generally? Most of my amps have lower plate voltages.

[pula58]

Are you saying that you go into grid conduction as soon as you see the beginning of clipping? If so, that seems strange. I would think that you would start to distort well before that happened and would see conduction when severely overdriving the grids. What are your bias voltage/current?

Am I wrong about this, generally? Most of my amps have lower plate voltages.

Yes CXX, starting-out with a clean output (i.el, no clipping), as I increase the vol knob to exactly the point where I can see the output start to clip, if I then look at the output tube grids they are clipping at just about 0V.

[pula58]

Yes. I think the way you have it the screen current is mingling with the preamp filter current.

Hi Martin,
I moved the gnd connection of the screen filter cap to the star gnd where the PT CT and 1st filter cap +-" terminal connect to the chassis.
It made no difference.

[martin manning]

Well nuts. I was hoping at least for some change.

[pula58]

Well nuts. I was hoping at least for some change.

Me too!
I appreciate your help in any case.

Somethings gotta give soon!

Do all amps do this, and I just never noticed?


P.

[pdf64]

As per Martin's earlier suggestion, when all else has been eliminated, consider another OT.
Perhaps for some reason its low frequency extension is too good, and sub harmonics that would normally be filtered out are getting through?
Pete

[pula58]

As per Martin's earlier suggestion, when all else has been eliminated, consider another OT.
Perhaps for some reason its low frequency extension is too good, and sub harmonics that would normally be filtered out are getting through?
Pete

I have a different OT i nthe mail...expected-in within a week or so.

Meanwhile, I think I am on to something that I would like to discuss here.

Afterelimination almost everything (and then saome!) I could think of I am still left with the same problem.

I could only conclude that the problem is the output tubes themselves, even though I tried another new set of output tubes (different manufacturer..i.e., physically different tubes)

So, I have been staring at old GE 6L6GC data sheets. What I noticed is that as the plate voltage gets pulled low, lower than the "knee", that the screen current starts to increase rapidly. See attached PDF. I show two load lines, "red" and "purple". Both load lines are for a 4kohm p-p (primary) output tranny, the difference between the two lines being the B+ voltage. The red line (which represents the amp I have been working-on) shows that the load line passes through the knee at pt "A." dropping down from "A" to the screen current plot trace (for 0V G1 grid voltage) at point "B". See that the screen current is increasing quite steeply at "B" as the plate volatge drops. Compare this to the purple load line that intersects the Vg1=0 plate line at pt. "C"." For Vg1=0 the screen current (pt. "D") is not increasingly as steeply for decreasing plate volatge as it is at pt. "B". I have one amp that sounds fine (it has a 470V B+), and this amp, the one I have been hacking away at, has a 430V B+. The amps are nearly identical except for one (the one with 470V B+) has a beefier OT .

It seemed to me that if I could get this amp, the one I am working-on, to cross the Vg1=0 plate current line ABOVE the knee that I would have less of a problem with the escalating screen current. So, I thought that if I add some resistance in-between the screen supply and the common point of the two output tube tube screen R's that I could drive the screen voltages lower, to compress the plate current curves so that the Vg1=0 line (pt. "A") looks more like the Vg1=-4V line. Then, the 4K p-p OT load line intersects the plate curve above the knee.

So, I added 2.2K of Resistance between the screen supply and the place where the two output tubes get their screen voltage (through one 470ohm Resistor each). And, sidebands...on the o-scope..practically gone! I also see no more 120Hz on the negative peaks of the screen voltages (at each output tube socket).

I have not played the amp loudly yet to see how this translates to what it sounds like but the sidebands are greatly attenuated by adding some screen resistance!