Weber PT in new Marshall 2203 build behaving badly

[loudthud]

It looks like you have 1 ohm resistors on each power tube socket. Do you get very similar waveforms at each socket as the output approaches clipping? Does the phase inverter produce a nice big signal on both sides if the power tubes are removed?

[jamesmafyew]

I do have 1-ohm resistors between pins 1 and 8 on each socket. I haven't scoped each plate individually but will once I'm home after the new year. Are you thinking that one of the resistors could have gone high or be intermittent?

The phase inverter is completely solid. When I disconnected the NFB, it produced nice big clean sine waves all the way through to all pin 5s, with power tubes in. With no power tubes, it's completely clean.

[dai h.]

I don't if this applies to your PT, but my understanding is that some (all?) original Marshall PTs sag a lot under load (and in cases where the B+ is high this actually helps the output tubes survive whent the output is cranked). Some clone PTs do apparently have this characteristic copied while others do not. There was a long thread where Randall Aiken analyzes and comments on this (on the metroamp bbs IIRC).

[jamesmafyew]

^ That seems plausible, since this Weber PT is the only Marshall 100W clone PT that they make, and it's used in their Plexi kits. When I've worked on 800-series Marshalls, it seems like the supply is usually quite stiff. I probably have a Hammond 290HX around in the shop somewhere; I'm going to try that after the new year and see if it makes a difference. If this Weber PT is ok and intended to do this, then I may just use it in a Plexi-style build later on, and put a Hammond in this one.

Though that still leaves the question of why my output is asymmetric.

[g1]

Though that still leaves the question of why my output is asymmetric.

Maybe if you try with just one pair of tubes you may be able to find out if it is related to one particular socket.

[jamesmafyew]

Ran some more tests yesterday while I was at home for a few hours.

Checked things out with only 2 power tubes in:
B+ is 447 @ idle, tubes biased to 66% max dissipation
B+ drops to 379V @ full power
Full power is 27.5vac @ 16 ohms, 47 watts
Clipping is not very asymmetric -- just clips on one side (blue wire side of OT primary, negative going side) a little earlier than the other side. Doesn't matter which tubes are in which sockets, results always exactly the same.

Crossover notch appears gradually/steadily as B+ sags, probably because bias only sags half as much, being connected from only one half of the B+ winding, so... I guess bias gets colder as power output goes up and B+ goes down?

1-ohm resistors are ok. Scoped and measured all of them to see what they were doing, they're fine, not intermittent or high-resistance.

Here are some scope shots, with all four power tubes in the amp:
421V idle B+ at 40mA per tube idle current, 16-ohm resistive load, 300Hz sine wave input to FX return jack, NFB connected as normal.

2V/div vertical, 1ms/div horizontal, x10 probe on dummy load.

1. 12VAC output, 406V B+
2. 20VAC output, 388V B+
3. 25VAC output, 371V B+
4. 31VAC output, 349V B+ (top half begins to clip)
5. 32.5VAC output, 341V B+ (bottom half begins to clip)







Clipping doesn't look too asymmetric here either. Yes, the negative-going half clips first, but not by much. This is with a much more rigorously-matched idle current between the tubes than before. I may have just been overly sensitive initially, just seeing the inherent slight-imbalance in gain of the two halves of the phase inverter, compounded by unmatched biasing. Though I'm open to thoughts on this. The phase inverter is driving a heavy load -- parallel 150k bias feed resistors on each side, so each plate is driving 75k. I have a 12AT7 in there right now.

Checked OT by feeding sine wave back through the secondary. The primaries looked symmetrical and perfectly undistorted and the primary Z came out to 1830 ohms (1/10.7 voltage ratio -- 10VAC applied to 16-ohm tap, 53.5VAC showed up on each half of the primary). Each half of the primary measures 17.8 ohms. I think the OT is fine, best I can tell.

Phase inverter can produce 54Vrms at each pin 5, with no power tubes, before it clips. Seems like enormously more headroom than needed to drive this thing cleanly to >100W.

So after all this, I'm left thinking that the PT is my main problem. I still can't imagine that this transformer is intended to produce a maximum idle B+ of 440V with 4 tubes biased to 70% from a "700VAC" center-tapped winding, which sags to 341V under full load.

Thanks for all the help so far -- very helpful in clarifying the problem.

[jamesmafyew]

Just a quick update on this situation:

Weber is being cool and helping me out. Nothing totally resolved yet, but I did find a Hammond 290HX and I went ahead and installed it, just to see what would happen. The amp works perfectly with the Hammond:

475V idle B+, 33mA per tube idle current, no signal applied

B+ drops to 417V at full power -- which is, in fact, 100 watts. This is exactly what I expected from this amp to begin with. I haven't made any changes to the circuit, even reused the same diodes when I changed the rectifier to a bridge for the 350V Hammond winding.

The main question on my mind right now is whether my Weber PT is bad, or if it's intended to be like that. I did notice that the B+ winding's resistance on the Hammond is 20 ohms, whereas it was 140 ohms on the Weber.

[Jazz P Bass]

Well, all things where leaning towards a funky PT.
I guess the only way to 'prove' it was to install a different one.
I sure hope Weber comes through with the answer (bad/ designed that way?).
I have purchased transformers from Weber in the past without any issues.

[jamesmafyew]

Another interesting development in this saga --- that ugly crossover distortion I was seeing before? It was still there with the new PT. I tried biasing hotter but it still appeared at a certain point regardless. I patched in a spare OT I had laying around (2K primary/100W deal) and the crossover distortion disappeared, and power output went up to 120W! Ugh, what is going on here...?

[jamesmafyew]

Sorry for endlessly posting about this, but I want to try and understand a few things.

I'm considering this particular amp done. It has a Hammond 290HX PT and an aftermarket replacement OT intended for a 70s Sunn Model T (2K primary, huge thing). I put KT88s in it and it does 130W before clipping. It sounds absolutely killer and is probably the loudest guitar amp (perceptually) I have ever played through.

Now, I have these two Weber transformers that I'm not sure whether they're good or not. The PT had all that sag issue, but now it seems the OT was doing weird things. As in the scope shots above, the output wave would develop a nasty crossover notch regardless of bias at around half to 2/3 of maximum power (also regardless of tube type -- tried EL34s, 6L6s, 6550s). It did in fact reach 100W with EL34s with the stiffer Hammond PT installed. But what kind of OT failure, if any, would cause this behavior? I don't know enough about load lines to understand REALLY what happens at the transition from class A to B.

I suppose it's possible that the PT is a Plexi-style designed with high sag, and is actually fine, but that the OT was limiting me in some way. Point is, I'm a bit mystified because the circuit and the build work just fine. But most of the time I just practically think of transformers as little black boxes that magically turn one voltage/current/impedance into another.

[Jazz P Bass]

How about sending them back to Weber?

[J M Fahey]

1) WHAT "ugly crossover distortion"?
Crossover distortion (due to cold bias) appers at low levels, hugging the zero volts line, is noticeable at lowe leevels and much less so at high ones.
Your waveforms show NO crossover distortion, which would be strange anyway given the high 35mA idle bias per tube.
The first two (where it should be the worst offender) show perfect waveforms, so .....
The slight kink visible at higher levels (within 3 dB of clipping!!!) is not crossover distortion but shows bias shift and beginning of bias rectification on peaks, which is natural on tube amps, because of the grid-cathode diode rectification, and the charging of the coupling capacitors with negative peaks.
Trying to overbias to "correct" that is a mistake.
You'll correct nothing, will burn tubes (or at least wear them much faster) and, really, that "kink" is part of tube sound.
Peavey went as far as simulating it on Transtube amps, go figure.

2) that Weber transformer does not look "bad", just built to a (low) price, with as thin as possible wire.
Besides that, good design practice is to balance primary and secondary resistive losses; otherwise you are "wasting" the "better" one.
*Real* designers do it, of course, but there's a ton of "me too" makers out there who design nothing, just unwind some competitor's product and "clone" it using measurements they made, sometimes inaccurate.
Of course, as everybody knows, photocopies of photocopies of photocopies are progressively worse.
I have seen *terrible* transformers.
2.6 ohms primary reflected to HV winding amounts to 89 ohms.
Since it's not the only load, (we have filaments too), I'd expect to find around 100 ohms resistance, end to end.
120 ohms, although not perfect, is not unreasonable.
They probably went to the "next thinner" wire, either by availability, cost, or simply, as I said, poor measurement.
Or they copied a bad example
In fine wires, enamel thickness is an important component of measured diameter.

So I think *that* Weber is cheaply built, nothing against.
And many home builders will just use it, and be happy.

Now, the waveform problems, unsymmetry, etc. are usually *tube* related.
In fact, what you show (after swapping and matching) looks very good.
Congratulations.

[big_teee]

I second the sending them back to weber.
I would be asking for some shipping money also.
Classic Tone still gets my vote for inexpensive transformers.
ClassicTone Transformers By Magnetic Components, Inc.
As far as the thinner wire?
If Weber is going to rate the PT at 100 Watts, then it should deliver!

[jamesmafyew]

The slight kink visible at higher levels (within 3 dB of clipping!!!) is not crossover distortion but shows bias shift and beginning of bias rectification on peaks, which is natural on tube amps, because of the grid-cathode diode rectification, and the charging of the coupling capacitors with negative peaks.
Trying to overbias to "correct" that is a mistake.
You'll correct nothing, will burn tubes (or at least wear them much faster) and, really, that "kink" is part of tube sound.
Peavey went as far as simulating it on Transtube amps, go figure.

Ok, here we go! Thanks a ton. And so that's why the kink is more present with EL34s, with their softer vacuum, than on a 6550, because the grid will conduct earlier? But I still want to understand more about why that particular shape happens there near the zero crossing, and why changing the OT to a different one would have made it go away.

[J M Fahey]

Well, see that up to (and even beyond) 20VAC which is a lot of power, you have no trace of crossover .
The "real" one would be very annoying at, say , 1 W RMS, and this one is not.
Which is obvious when you think that this is an AB amp, and at low power you are still in pure class A, so no crossover by definition.
In fact, there are some people who worry about the high power "curve kink" and do a lot of things to avoid it, basically (as I see it) using a couple Zeners so peak drive does not reach forward conduction point.
Many find the resulting sound "cold".
No wonder.
There must be some reason why none of the Classics (Fender/Marshall/VOX/Ampeg/etc. ) ever cared about or corrected it.
The kink *will* appear near zero crossing because
a) when increasing negative voltage at high power, tube gets colder biased and slides towards class B. Never reaching it of course.
Or to be more precise, it might even cutoff for a short period, that's perceived a sputtering, choking, farting sound, so horrible that the amp goes to a bench to correct that.
Happens sometimes when amps use too big coupling caps, have no series grid resistors and are wildly overdriven.
b) being a pushpull, zero crossing is where one tube stops pulling the load, and handles that job to the other one, and viceversa.
Good point to notice discontinuities and non linearity.