JTM45 rectifier question

[SPeter]

Got it, so there is a fundamental issue with the position of the standby switch as it is implemented on the Fender Bassman layout and the JTM45 layout just copied that mistake same as the Metro Layout. Our modification to make that first cap a part of a CRC filter did not make it worse.
The 'essay' you linked makes sense, I just have to figure out how to wire this thing neatly. Can I put the standby AFTER the choke?

So now Pin 8 of the V6 would go directly to the first 32 cap, one lead from the choke would be connected to the 2nd 32uF cap the other would go to the standby switch one side (instead of the 16uF cap). The other side of the standby switch would be connected to the 16uF cap. Or this would put stress on the 16uF cap now?

[martin manning]

I just have to figure out how to wire this thing neatly...

Easy. Right now you have the the two leads from the standby switch going to the rectifier cathode (pin 8) and the first cap in the CRC, correct? Disconnect those and put a short lead from the rectifier pin 8 directly to the first cap in the CRC. On the other side of the CRC you should have one of the choke leads and the OT's CT. Move those to one side of the standby switch, and run a lead from the other side of the standby to the second cap in the CRC (maybe run that on the board side of the PT with rest of the DC). Exactly like the sketch I posted earlier.

[SPeter]

Easy. Right now you have the the two leads from the standby switch going to the rectifier cathode (pin and the first cap in the CRC, correct? Disconnect those and put a short lead from the rectifier pin 8 directly to the first cap in the CRC. On the other side of the CRC you should have one of the choke leads and the OT's CT. Move those to one side of the standby switch, and run a lead from the other side of the standby to the second cap in the CRC (maybe run that on the board side of the PT with rest of the DC). Exactly like the sketch I posted earlier.

Thanks, I'll see how that works out. I will have to extend both the choke lead and the OT's CT to reach the standby. Should I twist these two leads together? How about the lead coming back? should that be twisted with the incoming two?

If I put the standby after the choke there would be only one lead going to the standby (choke's) and one jumper coming back to the 16 cap, this would create a shorter total lead length.

[martin manning]

I will have to extend both the choke lead and the OT's CT to reach the standby. Should I twist these two leads together? How about the lead coming back? should that be twisted with the incoming two?

I envisioned the choke and OT CT leads twisted together as they are now, making their way over to the standby switch more or less directly from their grommets. The lead from the second 32u in the CRC would go by itself passing between the PT and the HT fuse.

If I put the standby after the choke there would be only one lead going to the standby (choke's) and one jumper coming back to the 16 cap, this would create a shorter total lead length.

That would silence the amp, but nobody does it that way. I suspect because you would get a big pop (bang?) in the speaker when you switch to play.

[SPeter]

Ok, done. I would not call it neat lead dress wise but there is no hum or any other noise which would be lead dress related. I made the mistake to put in a bunch of NOS carbon resistors which added some hiss but my guitar is still more noisy than the amp. Once I put it in its cabinet it won't bother me ever again LOL.

20190426_174616.jpg

I measured 168 F temp on the 470R resistors after playing it out on 7 for about 30 minutes. Is that concerning?

[martin manning]

The interaction causing the ghosts is at the plates. Martin suggested it is actually parasitic oscillation, not intermodulation. I need to think about that.

You may be right about this, Tony. The sawtooth being fed into the CT has all odd and even harmonics. Knocking down the B+ ripple and in the process filtering out higher harmonics may be eliminating intermodulation.

[martin manning]

Ok, well this has been interesting. We seem to have identified the root cause of the ghosting in these amps, that being the large ripple voltage at the OT CT. Roe suggested way back that more filtering would fix it, but the question was where and how to add it. The 32u-235Ω-32u CRC reservoir seems like a good solution since it has more ripple reduction with less risk to the vacuum rectifier than simply doubling the reservoir. It's simple, involving only electrically rearranging the existing caps and adding a couple of inexpensive resistors. I'm not sure what the effect on tone and feel is. Some A/B comparison tests between the CRC reservoir and the original 32u would be required to assess that. It'd be interesting to try a single 100Ω 5W resistor to see if that would be enough to exorcise the ghosts with potentially less deviation from the original.

I measured 168 F temp on the 470R resistors after playing it out on 7 for about 30 minutes. Is that concerning?

To find out for sure you'd have to get the data sheet for those resistors and look into the temperature derate. Operating at 75C doesn't sound scary to me, though.

Edit: added a couple more comments to summarize this epic ;^)

[SPeter]

So if we agreed that the ripple voltage is the root cause then there are some additional things to consider:
- as Martin pointed out this should be the same for all jtm / bassman style amps and as such widely documented. But it is not the case. So we can conclude that the solution implemented is a good band aid but does not address the root cause which may be the PT / OT pair.
- If so then what is the parameter in the OT which makes this better or worse
- Would replacing the OT with a different brand would fix this? (marstran, heyboer, mercury magnetics... others?)
- Would replacing the PT would fix this? If so what parameter/feature would improve this condition?

I used the http://www.classictone.net/40-18033.html PT and the http://www.classictone.net/40-18039.html OT transformers as these came in a kit from Triode Amps ($200 including the choke). I know if I ever built another JTM45 I would not use the same kit but I have no scientific proof why. Classictone has an optional PT with variable B+ voltage http://www.classictone.net/40-18054.html but no options for other OT's.

What would be the recommendation for people who are just getting around to build their own JTM45 when it comes to PT/OT pair. And we can't just say 'hey buy the Mercury Magnetics' as those are insanely overpriced (close to 600 for the PT - OT set is crazzzzzy) and some reports are out there claiming that their OT is nothing to write home about.

[martin manning]

I don't think the PT has much to do with it since it is just producing the roughly sinusoidal HT voltage, the same as any other. Ripple voltage is a function of reservoir capacitance and current draw, and it's true that a 5F6-A Bassman with its standard power supply would produce a similar amount of ripple at the OT CT. With the Metro standard 32uF reservoir the JTM should be around 30V p-p. Doubling the reservoir would cut that in half. If you measure it on your amp now (with 32u-235Ω-32u), I believe it will be around 4V p-p. Original JTM45's had a CLC reservoir, 32u-20H-32u, which would reduce the ripple down to tenths of a volt (see below).

Marshall reissues are running 50uF for the reservoir. 50's Bassman schematics show 40uF for the reservoir, and actual parts manufactured at that time may have been higher. Reissue schematics show 50uF. It seems that the reservoir is generally higher than the 32uF used in the Metro JTM, and that should help reduce ghosting based on your findings here.

It's hard to imagine that the Magnetic Components OT is the issue either. AFAIK they are as good as any, including Mercury. I'm sure there are some subtle differences, though, and perhaps some of those could contribute to the ghost note problem, as could some of the other things Roe mentioned like PI balance.

Besides the CLC reservoir in the early JTM45, two things stand out wrt the Bassman: One is its 8k or 6k6 primary impedance and the resulting position of the load line; and the other is that the JTM has a 2.8x more NFB voltage due to feeding the same 27k FB resistor and 5k pot from a 16 ohm tap instead of the Bassman's 2 ohm output. I suspected that the high primary impedance might be a contributor, but you tried miss matching the speaker load to get a 4k primary, and that didn't seem to make any difference in the ghosting. I've not thought about how the NFB might play.

[SPeter]

The layout you posted shows the OT CT AFTER the choke, the Metro Layout shows it BEFORE the choke... I had it set up like the Metro layout shows below from the beginning. So it got a less filtered signal from the PT in the Metro configuration and none of the filtering of the choke and the second cap. I don't recall trying it like that original layout... Not until Tony recommended the double choke test. When I did the double choke setup I moved the OT CT to the second 32uF cap positive side per Tony's recommendation and it cleaned up the ghost notes. And when I moved it front of the first choke the ghosting came back. Maybe all I have to do is to move the OT CT lead to the second cap.... I just checked a bunch of Metro gutshots and some of them definitely shows the OT CT connected to the first cap but I found some where it is connected to the second cap.

Metro layout:

Metro JTM45 schematic HI RES.jpg

[Tony Bones]

I think it was only the very first schematics that showed the choke in that position. Were any amps actually made that way, or was it just a mistake in the drawing?

[martin manning]

Maybe all I have to do is to move the OT CT lead to the second cap....

Your 20H 70mA choke can't handle the current to do that. That was the reason for using the 3H in front of the plate node.

I think it was only the very first schematics that showed the choke in that position. Were any amps actually made that way, or was it just a mistake in the drawing?

The guitar trem amp schematic (the one I posted) and the PA schematic both show it like that, so I think it's probably accurate. The 100W 1959T (JTM100) has 4x 32u series/parallel for 32u total on the plate, followed by a 20H choke. That seems really light, and probably explains its reputation for ghosting.

[SPeter]

So Mercury has a 7H 150mA or I can just try the 3H as it can handle 250mA

[Roe]

150ma isn't sufficient. 250-300mA is better

[martin manning]

So Mercury has a 7H 150mA or I can just try the 3H as it can handle 250mA

You could try the 3H choke in place of the 2x 470Ω, and remove the 20H choke, but you might get a big transient voltage spike on start-up due to the low DC resistance of the 3H. If the reservoir caps are 500V rated, they may be ok.

Note that the early JTM schematic feeds the pale and screen from the same capacitor, which is generally not a good idea, but there is very clean DC there due to the large choke. All later amps are arranged Reservoir cap <plate> Choke, Screen cap <screen> Resistor, PI cap <PI>. Notice also that the "flying" resistor discussed earlier appears there. Instead of feeding the screen from the same node as the plate, you could decouple it with a resistor and the unused 16u.

I think the bottom line here is that at 32u the Metro JTM just doesn't have enough capacitance in the reservoir to eliminate the bad behavior. You did try paralleling the unused 16u without success, though, right? Anybody know if the reissues ghost with 50u?