All credit to JC Maillet!  The Marshall TSL122 JCM2000 Repair/Mods Page Photos/Text JC Maillet (c) 2007/2008/2009   A little background check doesn't hurt, go to Harmony Central and read what players have to say about these amps. I've worked on a bunch of JCM2000's now and I've seen two sets of problems crop up. In this one TSL122 I'm about to describe the mother board was found to be populated with negative Temp-coefficient resistors. This sent the bias voltage for a loop when the amp got hot - I'm sure it must be an accident at the factory or something. In a DSL100 I didn't find that specific problem but in all JCM2000's so far (TSL's, DSL's) I've found 220k grid block resistors on the power tubes instead of the customary 5k6 as indicated in all schematics - converting them to stock values definitely makes an audible change !   Hot Rat's Nest - The TSL122 Thermal Runaway Saga.  This amplifier first came in with a pair of crisped-out power tubes on one side of its output stage and some nagging microphonic problem. Having never seen one of these before I assumed that the amp had no major design or manufacturing flaws and likely had faulty power tubes put in and/or the bias hadn't been adjusted properly previously ... I biased in a new set of EL34's to around 35mA per tube (B+ just below 450vdc) ... found/fixed a cold solder around the B+ connector (W5 blue) on the mother-board ... rebiased, let the amp idle for a while as I normally do, double checked bias a few times more using the bias pins (which I usually don't use but at 1.3 ohms they give a conservative reading anyway) ... played the a bit more and sent the amp back to the owner ... The amp resurfaced a week later with a pair of tubes fried on the other side of the output stage, and the HT fuse again did not blow - something I couldn't figure out. All this prompted a serious in-depth study of the overall design. I began to notice strange things happening while the amp was idling - basically a form of runaway bias drift was taking place over the course of an hour or so, something I hadn't seen before. The mother-board showed severe temperature stress on all four power tube sockets which are directly mounted to the mother board - a potential source of untold and incurable problems on the one hand. Had the amp been under warranty I would have turned back at this point and had the customer send his amp back to the factory or ask for an exchange. But, that wasn't the case ... so trudging on ... The bias circuit was obviously a first place to look for leading clues in the mishaps. Being of the dual type (not a good idea to begin with) in which two trim-pots mounted on an external board connected to the mother board via a multi-wire connector (another not-great idea) adjust bias on each pair/side of the output stage - this off-board arrangement invites lots of trouble. My original beef with general dual-biasing is outlined in my Tube Amp book (IF&MTA), but the beef I have with mutually interactive versions of that idea is way worse. Siding with reason and caution I decided to hard wire a standard/classic single 10k bias pot directly to the mother board (all other bias circuit component values remain the same). I went to single control because having two degrees of adjustability made it difficult to isolate the source of "long term" bias drifting I was observing on the bench - basically I wasn't sure if grid current draw on one pair of power tubes was affecting the bias on the other side of the output stage. Installing it as such eliminates that 2-way variable and, more importantly, renders the bias circuit as a whole more immune to circuit "opens" via vibrationary action and faulty/intermittent connectors which I also had to rule out as a potentially likely possibility for frying tubes. Who knows about these connectors, I guess they're ok now that I've tugged at them while the amp is working with no apparent sign of weakness - still, the fact that the bias lines are going through two sets of connectors before making it to the main circuit is pretty ballsy, almost contradictory to some other overkill safety features in the amp. Scraping the enamel off the traces corresponding to pins 2 and 6 on the bias board connector and mounting a support connector to connect both traces together into a common bias feed to the power tubes. This doubles the total bias feed filter cap value to 100uF which lowers output stage hum at the expense of slightly longer charging times - not worth fussing over. The new bias pot (10k - 1/4watt) is mounted via two standoffs on pins 1 and 7 of the bias board connector. The old bias pots are removed from the board and only the bias measuring pins remain active in the circuit "just in case" they're needed in the future by someone. Or I'll remove the bias board altogether - I don't use the bias pins for biasing anyway. I reconnected everything together and was ready to do some more testing. One may wonder if the location for the new bias pot is optimal - personally I don't like the idea of snaking lines around in a high gain amplifier ... potential for added radiation-prone interference increases rapidly with distance ... so the safest thing is to install the trimmer right over the connector, in effect shortening the wiring distances (always a good thing if anything) ... I'm not worried about heat too much here since the circuit is shielded by a metal plate and the board lies well below the tube sockets ... about access, a long-neck screw driver will do the trick ... but I agree it's a PITA to have the control on the opposite side to tranny access but that's how it is for now based on the other considerations mentioned before ... suggestions are welcomed. Wrong Value resistors + Thermally Weird Components on Motherboard. This is where the weirdness begins. Having gone back to the usual tranny-shunting for measuring plate currents I discovered that a matched quad of tubes would creep to arbitrarily high currents over a period of an hour or more ... this occurred regardless of initial bias setting and then a spread between sides would develop as the increase took place. Two things I noticed were of importance here, (1) if I swapped tubes the spread tendency remained the same socket-wise, pointing to the circuit and not the tubes, and (2) if I turned off the amp once it was well warmed up and measured the bias feed resistor values I found that one measured around 110k instead of its nominal 220k and the other laid around 180k ... something funny going on (!) I pulled both resistors out and tested them under a blow-drier and sure enough both these critters dropped their values when heated. that's Negative Temp Co shit. Could it be that all resistors on the mother-board had negative Temp Co's ??? I wondered how they would affect the bias circuit, and other things not to mention ... To answer part of that question I replaced all resistors on the mother board that were part of the bias circuit (R68=33k, R77=10k, R67=220k, R69=220k) by 1watt film resistors - they hardly budge in value when placed over the flame of a lighter. This made a big difference - the tendency towards drifting was much slowed down and at some point it peaked to a semi-stable value unlike before. The only thing to explain the residual bias drift and divergence towards a semi-stable value was the burnt sockets and oxidized contact with the mother-board. My advice here is to give the amp about an hour to settle following this mod to make 100% sure all is fine ... The other thing I noticed was grid blockers (R7, R10, R66, R70) lying at 220k instead of the usual 5k6 ... this finally explained the choked clean sound and a muddy/grainy power distortion tone. By replacing them with 6k8's (closest I had) the classic Marshall sound came out more, it was slightly more alive and clearer in clean mode and the distortion tones were less grainy - to my ears anyway.   Scoping the High-Tension Circuit. One good thing about 1 ohm resistors in the cathode circuits is in giving the ability to scope the pentode plate currents indirectly through them. A 500mV peak cathode voltage swing translates into a total 500mA peak current sink per push-pull side, this justifies the standard use of a 1A fast-blo fuse in the B+ circuit of 100watt EL34 amps, right in series with the output transformer feed ... bring a scope to your gig and hook up :)~ I think this mother board could have suffered damage from several previous harsh malfunctions around the power tube sockets and I guess it's only a matter of time to see if these corrections will allow the amp to hold up over time (unlikely I'm afraid - that's PCB based amps for ya). Not a job I could guarantee, so I warned the customer about it and he was ok about it.     DSL-100 check-up. This DSL100 mother board had 220k grid blockers, I piggy-back them with 6k8's as I did in the TSL122 ... and the bias circuit resistors were changed to 1% metal film ... yes, I'm just about to change that 10k bias circuit resistor in the back ... the bias resistors on this amp also exhibited NTC behavior which again was confirmed OFF-BOARD using a multimeter and 1200w hair drier ...   Note: A few guys I talked with were thinking of drawing up replacement boards - for anyone nuts enough to tackle this re-design I would suggest doing a board with a cut-out around the power tube section and mount the power tube sockets to a metal bracket instead of the PCB board - I got an email from a dude who says he did it successfully ... the only resistors I think need changing are in the bias circuit and the grid-blockers - they should be 5k6 and not 220k ... all other resistors, incl. screen block power resistors, I wouldn't touch unless there's a reason ...   \\\ CONCLUSION /// I got in touch with the top dawg at JAM industries in Montreal and he had a few random ideas to offer: C46 the small enigmatic ceramic cap next to the last power tube circuit is said to short out - of course, this wouldn't explain bias drift at all for a shorting cap on the B+ line would drop tube current if anything and eventually blow a fuse (assuming fuse circuit would work here)... recommended advice on C46 is to snip or replace with 1500v ceramic cap (if you feel you must have a cap there) - why they put one in the first place is beyond me since the amp doesn't oscillate as far as I can see on a scope ... I tried the snip on the TSL122 I have here and it did nothing to fix bias drift ... I'm not sure how much more I can quote here, but bad Russian tubes were said to be the cause of much malfunction in these amps in his opinion (I don't think so myself) ... Winged "C" were recommended by him ... aside from mentioning the usual output jack connection issues that's typical of all gear nowadays, they said the mother boards shouldn't be replaced on these amps except if the mother board was fried badly ... (???) ... when I asked about bias drift issues it came up as "outside of C46 shorting out that doesn't happen" ... the price for a new mother board from JAM was around $168CDN ... [* if you do buy one please send in pictures and details *] Marshall wouldn't get back to me on this so I can't answer anybody's questions with any degree of certainty - I can only echo whatever info you dudes are discovering ... at some point I will try to hook up the tubes remotely on this TSL122 since there are some reports of that fixing the problem - but we'll see ... it's not on my list of priorities to re-design this bird ... at this point, unless you're a DIY freak, the best option is to buy a new mother board or move on to another amp ... bottom line is you've got a double sided PCB board in a hot environment that has very tight tolerances - that's asking for trouble just by the very nature of the assembly, never mind NTC resistors showing up in the mix (which I'm sure was a mistake at the factory) ... As far as I'm hearing TSL/DSL users are getting new boards from the vendors mentioned above and things are working out ... I hope this helps anyone concerned - I have heard friends play the new Vintage-Modern and it seems Marshall finally got the tone private techs were obtaining for decades by modding original Marshall amps (eg., Rick Onslow comes to mind) - hopefully the Vintage-Modern is sensibly designed ;)

JC Maillet Marshall TSL122 Article