Real TrainWreck mojo!

[Chuck H]

I'm satisfied. That is, with Ginger's explanation of high NFB and his reasoning. Some speakers that are much happier in a closed box will behave better if damping is controlled at the amp. So all understood there I think.

As to the capacitor polarity. One amp was configured with the outer foil of each film cap oriented towards the lowest impedance. The other had all caps oriented opposite. I've built identical amps before with all caps oriented the same and heard the same amount of difference between the two. This is just a busted myth as far as I'm concerned. It's said to be much more important for high frequency applications, not so much audio, and I'm prone to believe that now. My example is the only one I'm aware of where someone actually troubled to do this. To say this won't do is a clear invitation for me to ask you for a significant contribution. I no longer have both amps together and I wouldn't trouble with changing out all the film caps anyway based on my findings so far. What's your researched contribution?

[Gingertube]

The cap polarity thing is more of a background noise thing (and so more important for HiFi but still worth thinking about for guitar), the outer foil should go to the lowest impedance to signal ground circuit point. For coupling caps that will usually be the driving tube anode connection but depending upon interstage attenuators etc. that will not always be the case.
It is one of those "last few percent" things, which is backward way of saying that in 98% or so of cases you will have more pressing problems to worry about.
Cheers,
Ian

[Shadrock2]

I suppose that's fair enough. I do want to build one of these beasts though and I am quite willing to experiment. Whether or not I can get the materials together to make two that are identical is an open question. I have put together 1 kit amp and built exactly one amp of my own design which took me almost a year. I learned a tonne from that experience but that should be a different thread. So I'm pretty much an amateur compared to you guys though my dad owned and operated a TV repair shop in the 60's so Ive been around electronics all my life.

[Gingertube]

Guys,
I reviewed this over the weekend and decided that we may be over thinking this.
The Express front end is largely a Fender preamp with an extra gain stage tacked on the end.
So what stands out from an examination of the Express Preamp:

1)The 150K "Gain Dumping" resistor as the grid leak on the input to the third stage
The RL of this 2nd stage would be 90K (100K parallel 1M) ordinarily but with a 150K grid leak following instead of a 1M then RL becomes (100K parallel 150K) = 60K.
Gain = mu.(RL/ (RL + ra) so the actual gain of the 2nd stage is dropped marginally from 58 to 48.
However the output signal is developed across RL so with 60K instead of 90K we are swinging 1/3 again as much current in the tube to get the same output voltage.
That means that the ra variation with signal current is larger.
AND
since RL is much lower then Delta ra vs RL is a much bigger ratio - that is important because delta ra vs RL is the distortion level.
So we have modified the 2nd stage to drop the gain by a bit less than 20% but are likely to have increased distortion by much more, at least 50% and possibly as much as 100%

2nd - the extra gain stage
Well it has a couple of things working against each other.
Gain is low - you could predict that just by looking at the schematic and noting that even if the triode had infinite gain then feedback theory would tell you gain would be RL/Rk or (100K parallel 1M)/10K = 9.
In fact if you do the math the gain is actually more like 7 to 8.
We are working at the low current end (estimated 300uA) of the triode characteristics which means high ra and low gm and also that when considering the ra vs Ia curve we are on a section of that curve with very high slope indicating large changes of ra for small changes of Ia.
Normally we would associate that with VERY high distortion but in this case the unbypassed cathode resistor will act to oppose the distortion - even so distortion is likely to be high.
The other thing this does is set the anode voltage of this stage close to its B+ (300uA across 100K means 30 volt drop) so the positive going signal swing will be limited to 30V peak. That 30 V is likely to be increasingly comperessed as we approach the B+ but clipping is unlikely or in any case will occur only at extreme volume settings. Conversely the swing in the other direction is not only larger but is more linear, the asymmetry will be significant, introducing primarily even order distortion.
If you draw a transfer function graph - that is volts out (anode) for volts in (grid) then the departure from a straight line indicates distortion and the sharpness of the curvature indicates the order of the distortion, sharp curves indicate high order harmonics.
Now my view is that we will have fairly gently curving transfer function for those positive going signal swings and that means EVEN and LOW order distortion products, mostly 2nd harmonic.

This can be "sharpened up" and some harder clipping introduced and for that you need to look no further than the final preamp stage of a high gain Soldano, same circuit but with a 39K instead of the 10K.

This suggests to me that the poster above who mentioned varying that 10K until he found what suited him (6K8 I think it was) was onto something.

So thats my erratic "pot shots at the barn door" thinking.
Sound/Tone is primary formed in that final cold bias stage with the 10K cathode bias resistor which is large enough to cause significant compression on one side of the signal but not quite enough to cause clipping.
There is some additional distortion being introduced in the second stage by that 150K "Gain Dumping" resistor which is also the grid leak for the 3rd stage.

Cheers,
Ian

[Chuck H]

Absolutely capitol. And some considerations that certainly deserve further discussion, but... The circuit has been dissected before in this manor. Evaluating the amp isn't the issue. At least apparently, since clones are made to the exact same circuit and often with considerable care taken to duplicate layout and lead dress, yet, original KF amps still sound best. And it's not tom foolery or mojo thinking. In fact it's precisely BECAUSE the circuit is so easy to evaluate that the difference between KF built amps and clones is so hard to pin down. Think whatever you like about it. It's true.

One note on the third stage... You mentioned that because of it's gain and bias structure that it wasn't likely to contribute much clipping. With a nod to it's asymmetrical nature making any clipping 2nd order harmonic. But in reality there is plenty of drive at that grid and on a scope the clipped lower portion of the signal is quite broad. In other words, there's no mistake about the third stage contributing to the distortion. Others having observed this have taken to calling it, and other high resistance cathode circuits in guitar amps (other than cathode followers) a "cold clipping circuit" because of it's primarily cutoff distortion... For your consideration.

[Gingertube]

There is one other thought that re-reading what I wrote above triggered - so more of Ian's ramblings/random shots:

That third gain stage: looking down into the anode you will see ra + (mu+1)x 10K -- say a bit over 1Meg paralleled by the very high impedance into the phase splitter, looking up to the rail you see the 100K. That means that there is very little division of any residual power supply ripple. A clean power supply node is essential. A significant amount of any power supply ripple for that stage will appear at is output and will then intermodulate with the real signal in the power amp. That intermodulation tends to make the sound "mushsy" and "vielled" and is most noticable on the lower frequency notes. It could certainly affect the persceived atttack of lower notes. This intermodulation effect is also dependent to some degree on the AC balance of the power amp.

Is there any indication/reports of Ken obsessing over matching of output tubes or phase splitter balance?

Should onyone think this affect might be important then there is a fairly easy way to conduct an experiment. You can significantly reduce intermodulation in the output stage by adding in some common mode feedback. Various ways to do that but the simplest is probably to have some common resistance (unbypassed) in the output tube cathodes.
I have never tried this in a Guitar Amp but I do run this scheme in my push pull 6V6 HiFi Amp (39R common unbypassed in the cathodes).

Phase Splitter AC balance (which will also affect the intermodulation) can be adjusted by trimming the tail resistor in the PI.

You might note that if you are looking for "Iconic" Amp designs where the designer has obsessed over power amp balance then there is no need to look further than a Dumble.

Aside: That HiFi Amp I spoke of above used balanced shunt feedback from the output tube anodes to FORCE AC balance.

I keep looking for some significant schematic thing as any "Mojo" should be able to be attributable to something specific.

Is'nt it strange that we insist there being a rational explaination for these things but at the same time maintain these sneaky beliefs in "Mojo" or "Magic" or whatever. In my case its "sympathetic magic", any of the major components in the Laser Airborne Depth Sounder I design/build in the day job, over which I have shed blood from a scraped knuckle or whatever, I know I don't need to worry about as it will run beautifully for years. Do I have definitive evidence for that - of-course not, but then I haven't looked for it either, being too embarassed to admit to belief in anything so obviously "wrong".

Cheers,
Ian

[loudthud]

The trim pot in the Dumble PI is set to deliberately introduce even order distortion. Start with 6V difference at the PI plates. I got this from a guy out of Phoenix that makes Dumble clones at a recent Amp Show. He also uses lead dress to introduce a small amount of positive feedback in the preamp.

I got my Express clone working, it has a very clean output as far as visible power supply ripple on the output goes. The Express also has a strong tendancy of the output to "shark fin". That's where the bottom side of the output squares up and then the trailing edge collapses as drive is increased. Many Marshall amps also do this. If you are a member at TAG, you can see examples here: The Amp Garage :: View topic - harmonic distortion=

Positive feedback might also be at play in the Express from the output of the clipper stage to the wire from the Volume pot that goes under the board to the second stage grid.

The tone stack in the Express is kind of dark sounding, but the high pass at the input to the clipper stage cuts the muddyness. Overall the tone controls work well in the amp. Very enjoyable to play.

[echoride]

The one idea I've seen speculated is pretty tough to refute (or prove I guess), that is using lead dress as "invisible" capacitors.

I never met KF but talked with him on several occasions via phone. He was affable but I thought a trifle neurotic. We had a couple of mutual acquaintences, one of whom has also passed on since. They were in his shop many times.

For a guy who did not have an RF background, he had figured out the importance of layout and lead dress and parasitic inductances and capacitances pretty well. He was mostly self taught except for a correspondence course, and those as Don Lancaster says were "thinly disguised programs for veterans to get free color TV sets from the government".

As others have said, guitar amplifiers of the Dumble and Trainwreck styles are built deliberately on the edge of stability so when they are overdriven all sorts of harmonics are engendered. That part doesn't amaze me but that these things hold up as well as they do under road use does.

Fischer used oversize transformers and very light aluminum chassis. I don't think the Wrecks are as roadworthy as the dumble, but others with real experience may have different opinions.

[jazbo8]

Bumping an old thread... katopan over at TAG has done extensive testing and documentation on the Express and summarized his finding (see bottom of page 12) on the Trainwreck mojo - he believes the 1k screen resistor on the output tubes and the way the PI is clipped are the primary contributors. Anyway, just some more data to consider.

[Alan0354]

I happened to read up the TW schematics and find this thread. I read through about 1/2 of the thread. My question is whether anyone actually have a real TW side by side with a clone and use a A/B pedal to play through them and A/B compare? Or the comments are based on recording only? If someone did compare, are both using the same brand and model of speaker?

Reason I asked is because the amp can sound very different through a mic or listen to it raw. And different recording environment produce totally different sound characteristics. I mentioned in my Dumble thread that I happened to listen to Carlos Santana at the front row that I heard the raw amp sound and it sound totally different when I went to the bathroom where I walk towards the back and heard the guitar through the PA.

Different speaker make the amp sound totally different.

I do believe in mojo, the wire coupling that produce different sound. But if people have pictures of the real TW and wire exact the same way, using wire with the same insulation thickness AND the wires are same distance to the grounded chassis. I cannot see that you cannot reproduce the coupling effect.

Just wiring drawing is not good enough, you really need a real picture of the real TW.

I don't think I have very good ears, I never find a few % variation of component values make much difference in the sound. I do know the wiring can make a big difference. I remember when I progressively develop my first amp, I added in circuits at the later days that result in crossing input and output wires, the amp never sound right. When I build the exact circuit into the KMD with preplanned layout, it just sound quite different from the Bassman and sounds right. My guess will be the wiring effect the sound a lot more.

[jazbo8]

I happened to read up the TW schematics and find this thread. I read through about 1/2 of the thread. My question is whether anyone actually have a real TW side by side with a clone and use a A/B pedal to play through them and A/B compare? Or the comments are based on recording only? If someone did compare, are both using the same brand and model of speaker?

Yes.

[g1]

But if people have pictures of the real TW and wire exact the same way, using wire with the same insulation thickness AND the wires are same distance to the grounded chassis. I cannot see that you cannot reproduce the coupling effect.

No. The amps would have to be otherwise identical. Tolerance differences in components alone would make each amp individual. Also you could not get precise measurements from looking at a picture. If a circuit is operating at the edge of stability, even a millimeter (or less) change in lead dress could make a difference. You will not get that kind of accuracy from a picture.

[Alan0354]

I can't help but be skeptical that out of hundreds of clones, nobody comes close to the original. I believe in mojo or even magic. But I still don't believe in miracle!!!! Before even getting deeper, did anyone done total double blind test? That is you have a few people that are not necessary hell bent on Trainwreck. Doing A/B comparison without knowing which one is the real thing and which one is clone. They can only know amp A and amp B. Have someone else adjust the volume, T,M,B and presence on his command. Then they rate them according to their liking.

This is audio frequency, not RF. Yes, due to high impedance of the circuit, coupling is definitely noticeable. but copy direct by looking at the real thing cannot make it sound the same? You can measure all the components one by one to duplicate the original.

Measuring and hand pick every component before putting in is a lot easier than it sounds. I measure all the components before I build my amps. It really does not take long. that way, I know exactly what I am getting into. But of cause in my case, I look for the components that are closest to the stated values.



Another question. People said the amp is not stable and can oscillate. I want to confirm they are not talking about guitar easily feedback, that the amp can burst into uncontrollable squealing when you turn the volume up without guitar plugging in?

[Dave H]

I don't think I have very good ears, I never find a few % variation of component values make much difference in the sound.

That's because you are listening like an engineer. You have to be like Mulder you have to "want to believe"

[Alan0354]

That's because you are listening like an engineer. You have to be like Mulder you have to "want to believe"

Yes, you have to have a clear head to do this, or else there is no point of this thread. That's why Mulder is in X-File the tv series!!! That's why I am not challenging anything else, mainly "Are you sure?". If you truly believe Only the original guy can make the amp like this, then there will be no discussion anymore. This then becomes more psychosomatic than technical discussion.

I still believe everything can be explained by electronic theory, just how hard you want to do it. This is still electronics and science. I can buy in the mojo of wire placements, BUT you should be able to calculate the muture capacitance, inductance of a parallel wires coupling. I chuck those into mojo because it is not easy to simulate, you need EM simulations. But I am not going to just go to miracle.