New Express with VVR....and issues!

[ampgeek]

I was pondering my situation during lunch and realized that there is a discrepancy in the reference documents.

UR12's Kelly 90 schematic shows V5 control grid fed by the 100K loaded PI half

Ron Worley's layout shows V5 control grid fed by the 82K loaded PI half.

I subscribe to Ron Worley's convention in this thread FWIW.

Cheers,
Dave O.

[Richie]

I think the confusion is, change v4 and v5 labeling on the kelly 90.
I don't think UR12 drew up the kelly 90 schematic. . Look at the layout or any of the pics, they are correct.

[ampgeek]

Thanks Richie!

Only trying to make sure that anyone reading along and using the schematic from the Trainwreck Files section knew my naming convention.

I have been chipping away on the Aiken site's LTPI theory section but still can't understand what might be occuring in my case. I also spent a couple of minutes this evening poking and prodding specifically in the PI socket and board componants to no avail. I am convinced that there is something wrong with a componant but it isn't clear which one(s).

Attached is a picture with comments that briefly describes my observations.

My primary questions are:

What could cause an ugly, distorted signal on V3/pin 2 when there is such a strong, lovely signal on V3/pin 7?

If the PI bias condition explains it, what componant(s) are suspect (see voltage chart)?

What could be causing the V3/pin 6 signal to get so nasty when it's grid signal is so nice? Bad 82 K plate load resistor? Too much voltage swing across that resistor? Leakage of the ugly signal across C12? I need to pull C12 to check that theory.

Could the stronger signal off of V3B drive V5 to such a high current draw condition?

Thanks as always
Dave O.

[Cliff Schecht]

So you've measured each cap right? How about after you put them in circuit? Many of the components can be measured "in-circuit" if the tubes are pulled. Make sure that the leads inside of the caps are still strongly connected to the body, a soft horizontal tug on the cap bodies should quickly reveal if any leads have been weakened (don't use enough strength to actually bend the leads too badly though!).

Are both of your grid stoppers the correct value? Go through the circuit with the tubes pulled a re-measure all of the passives around the PI and PA. It looks like you are using a CC resistor on the left socket and a metal film on the right although that could just be the angle of the picture skewing what I'm seeing. Take some more gutshots at different angles for us if you could, maybe we'll see something that stands out..

[Richie]

simple thing to try first is tube swaps. if you look back at the info about express amps, picking the right tubes is important,and will have different effects on how the amp works or sounds.

Once you do that, and you have the best tube selection in the amp, try a few things like, place a shield, or lay a piece of thin plate for a cover over the amp. See if it helps. other things to do would be use a .001 cap, start at the first 100k plate resistor, just sub it in, see if it kills the oscilation, then move to the next plate resistor. Sometimes in some amps they may be used across the plate resistor of the PI. this is more for a test, once you see it stopped the problem, you'll know where to look to fix it.
The amp should work without problems,without useing the snubbers.

Also try a shielded cable coming back into the second part of V1.
Try these simple things first, it may help narrow it down.

If you look at some express amp pics, Ken used a grid stopper on the wire coming back from the volume pot to the other section of V1..820ohm or 1k see if it helps.

[marcoloco961]

Have you checked c10, c13 and c14 for leaking DC voltage. After looking at your last pic, I think I would start there. I believe a cap can measure the proper value, but still leak DC voltage. This might explain why the signal would distort past that cap. I think I would start by replacing all 3 with new ones, just to be safe. I would think that it would almost have to be there, or in the resistors (or soldering) directly prior to that if it is not a bad tube in v3.

[ampgeek]

Thanks mates!

Just got home ready and rare'in to have another go at it.

Just for clarification purposes, I have two dozen NOS 12 AX7's (Mullard, Amperex, Telefunken, Tungsram and EI) that I have fairly meticulously screened (in all similar Wreck positions) through my 64 Black Face Deluxe (low'sh gain), a Dumble'sque build (medium'ish gain) and a Mesa Boogie DC3 (high'ish gain). And they are all VERY good!

Mind you, I realize that this is a significantly different playing field...but...I am confident that I don't have a gross tube problem.

I am also hearing the glorious Wreck'ish tone up to the first PI grid (volume at 4:00, VVR dimed and everything else at noon) through my audio probe feeding an all octal Bassman (5F6A) build of mine. Very clean, round sounding high'ish fidelity tone from that plan. The power tubes are removed from the Wreck while doing the audio probing.

The Wreck tone is clean and strong and just on the edge of break-up on the first PI grid. No whining, squeeling or oscillating. Very little hum as well (other than that from my soldering iron and neon lit magnifying glass which I can recognize from experience in the signal). It gets weak, distorted and ugly from that point on. Even on the cathode of the PI.

Has anybody else done this/similar with their builds? I *ass*ume that this is not a normal condition.

I have swapped six of the select tubes into the PI with, literally, no change in the condition whatsoever. Probably ought to try some more including the lower gain variant (5751) to see if that sheds some light on it all.

I plan on running through as many of the suggestions and a couple that I thought of as I can before it gets too late. Darned neighbors and kids that THINK they need to sleep Then...some more study on PI theory as I really think the answer is there and in the PI voltage readings.

But...what the hell do I know! Fortunately, it's all about having a hobby that you can learn something from.

Please keep'em coming so we can all learn together. Forgive me for being so verbose here. But...I am hoping that a little extra "documentation" will help those others of us in the novice camp doing/considering building one of these amps.

Thanks as always,
Dave O.

[Cliff Schecht]

Now I really think it's a cap.. As I recall, all of your PI voltages look in spec so you could have a bad part put in there (too much heat/stress during building). This would cause the nasty distortion past a certain point without affecting the previous circuit..

[paulster]

I think it's likely to be a cap too.

If it were a parasitic oscillation, which is very common on these, you'd expect it to be affecting both tubes as it's likely to have got in there at V1 (probably on the grid wire from the volume control to the second triode) and worked its way through the amp.

A leaking cap would shift the bias point on that tube dynamically, so it could still show a normal bias at idle.

Even with a scope this can be difficult to see because you're looking for changes to the DC offset (bias) whilst you've got a healthy AC output signal on there. Make sure also if you get a scope that you use a x10 probe on it, regardless of the voltages you are planning to measure. A regular x1 probe has a 1M input impedance usually (fine for solid state stuff) but is a significant enough load in high impedance tube circuits to present a load itself and affect your readings.

[ampgeek]

Thanks as always!

So...I slapped in another 6 known good tubes in PI with no change in condition. Even a lower gain 5751 showed no perceptable difference. Now I am absolutely, positively convinced it isn't a tube problem.

I squinted a little longer at the voltage chart and, as I had recalled, the bias condition on the PI is outside of my normal comfort range of +/- 10% for anything tube amp related. If anyone takes a look at that chart, note that I calculate the bias condition and compare it to standard conditions.

I disconnected the lead to the presence pot and slapped in a 4.7K resistor to ground and saw no change. So...I think that I can eliminate that part of the circuit as problematic.

I measured the voltage drop across each PI plate load resistor and calculated the current draw on each half and see a fairly significant difference. 1.2mA across the 100 K and 0.79 mA across the 82 K. According to the reference voltage chart(s) they should be much closer at 93 mA and 87 mA.

So....that seems like an un-acceptable condition for a PI. Yes?

I also measured the combined current flow across the 470 ohm cathode resistor and get 2.47 mA. Odd....the combined plate current is only 1.99 mA.

Where is the extra ~0.5 mA of current coming from? Through a grid maybe?

I replaced C10 (0.1 mf), r10 & r12 (1M ohm), R11 (10K ohm tail) and R13 (470 ohm) with known good parts. And....no change in condition.

I also replaced the PI plate loads (100 K and 82 K) thinking that there may be some type of dynamic problem there. Again...absolutely no change.

I had previously replaced the PI tube socket (from a Belden to a ceramic) as I thought that I may have damaged it and felt it may have been the source of the problem early on (mangled one of the pin sockets doing a re-tensioning after all of the tube swapping early on in the trouble shooting). So...that doesn't feel like a culprit.

Any thoughts on what to check next? It isn't obvious to me what is causing such a large quiescent mis-match in the PI side-to-side current draw and where the "extra" cathode current is coming from.

Thanks much,
Dave O.

[rooster]

Wait a minute? Am I missing something? Did you already change the .02 cap? What I said so long ago was to check the bias supply wire to V5 and then replace the .02 cap if you still had a problem. Did you do this already? Sorry if I missed this news in which case because this was where I thought your trouble was. Fill me in?

[paulster]

I measured the voltage drop across each PI plate load resistor and calculated the current draw on each half and see a fairly significant difference. 1.2mA across the 100 K and 0.79 mA across the 82 K. According to the reference voltage chart(s) they should be much closer at 93 mA and 87 mA.

So....that seems like an un-acceptable condition for a PI. Yes?

I also measured the combined current flow across the 470 ohm cathode resistor and get 2.47 mA. Odd....the combined plate current is only 1.99 mA.

Where is the extra ~0.5 mA of current coming from? Through a grid maybe?

Try removing V1 only and repeating this test. If you do have a parasitic oscillation problem then it will almost certainly be getting in at V1 (probably in the grid wire from the volume pot to the second triode) and this will take that out of the equation and leave you with an (effectively) grounded input to V2.

I replaced C10 (0.1 mf), r10 & r12 (1M ohm), R11 (10K ohm tail) and R13 (470 ohm) with known good parts. And....no change in condition.

I think more likely if it's a PI condition the problem will be the couplers to the output tube grids, since these can allow the PI plate to skew the bias on the output tubes or, conversely, the bias voltage on the output tubes to skew the balance of the plates on the PI.

[passfan]

Can the grid wires be swapped on v3 as a sort of troubleshooting curveball ? Swap pin 2 for pin 7 and see what the pi does? Could this eliminate or point out a problem in the bias section for v3?

[marcoloco961]

Wait a minute? Am I missing something? Did you already change the .02 cap? What I said so long ago was to check the bias supply wire to V5 and then replace the .02 cap if you still had a problem. Did you do this already? Sorry if I missed this news in which case because this was where I thought your trouble was. Fill me in?

I am suspecting leaking DC voltage on c13 or c14. I think I would replace both and try again.

[ampgeek]

Thanks as always guys!

Rooster: Nope. You didn't miss anything. I failed to mention back then that I didn't have suitable 0.022 mf replacements and that I switched them on the board with no observable change. At that time, it was a gross mismatch on power tube current draw when hammering out a power chord and a singing OT that were the primary indicators of a problem.

I have since acquired a handful of NOS Vit Qs and more current production Mallorys (all white body) to play with.

I lifted the bias supply and the current draw across the PI plate resistors evened out nicely with the existing PVC OD couplers. The sound quality at the PI plates improved dramatically. GREAT Wreck'ish feel/response and "the" tone is just below the surface (if that makes any sense) But....there is still an ugly'ish distortion over everything.

The Vit'Qs were installed with no observable change in condition. I loaded up the power tubes, biased them to 44 ma firing into a dead load. Hit it with a Les Paul deluxe with Duncans power chord and still hear the OT singing and see the mismatched power tube current draw. But...not quite as dramatic as previously (VVR dimed, volume at ~4:00 and everything else at high noon).

Using the audio probe, the sound quality at the PI plates up to the power tube grids is nearly identical (Wreck'ish but with a distortion over it all). Power tube plates and 8 ohm OT tap are way over distorted and "lifeless". Ugly!

With all tubes loaded and firing into a dead load I see the following:

PI plate load resistor current draw: 0.94 ma across the 100K and 1.11 ma across the 82K.

PI cathode resistor indicates: 2.3 ma across the 470R (cathode voltage = 35.6)

So...still some extra current at the cathode but not quite as much as before.

I then pulled V1 and measured the following

PI plate load resistor current draw: 1.03 ma draw across the 100K and 1.46 ma across the 82K

PI cathode resistor indicates: 2.6 ma across the 470R (cathode voltage =38.6V)

Now...almost a perfect match between combined plate current and the cathode current.

Not sure what is going on now. I would have expected just the opposite if the PI were being driven by an inaudible oscillation!

So....maybe throw in a gridstopper and/or shielded line to the second amplification stage? What would be the best value? I seem to recall 33K to 45K directly on the pin?

I will likely throw in the Mallory couplers just for poops-n-grins and I now have another OT from RJ to play with too!

TIA,
Dave O.