Marshall 1959HW 100W Plexi Oscillation

[nevetslab]

A client brought over a near-new Marshall 1959HW 100W Plexi, with the complaint that it has very low output and smells, like something burnt. They had the inside pair of EL34 power tubes fail, and replaced them. I hadn't yet checked the output level....having opened it up to have a look inside for anything obvious. All spic and span clean, no burnt parts. Two different brands of power tubes. Checked each pair with my bias probes, finding a 25% mis-match in plate current. So, left one pair out, then looked to see what the output looked like, knowing it would be low with only one pair of tubes, and impedance off. I only got 1V RMS with a very distorted waveform. Output Transformer partially shorted...secondary side.

So, hunted for a replacement. I didn't find any Dagnall TXOP 0002 replacements, and after seeing the range of prices fall between $78 and $300, all with different core sizes and mounting hole issues, I wasn't looking to have to remove the perf board and loads of wires to drill new mounting holes. Ended up with a Magnetic Components 40-18026 O/T...1.7k Pri/16-8-4 ohm Sec, and near identical core size. When it arrived, I found I wasn't going to be able to re-use the mounting clamps Marshall used on the Dagnall xfmr, as that was held together with two core bolts in the middle edges, while this one used four core bolts, one per corner. And, the hole spacing was still incorrect. So, had to alter the mounting tabs to mechanically fit it. Left the original Xfmr wiring in, cutting them as the exited the end bells, and made in-line splices, after first verifying the phasing of the Dagnall and that of the replacement. All mated up, nicely mounted, biased up and had full output....though I still had the mismatch of plate currents to deal with. Found a different pair of tubes that matched, and went to use those.

Under drive conditions on the Bright Channel, I was getting a HF static-like distortion, that sort of sounded like a capacitor breaking down. Chased that, and found it went away when I removed the selected alternate power tubes to match one set it came with. With just the one set, that problem was gone.

New matched quad set of JJ/Tesla tubes arrived. Selected the best plate current match, and put them in, had full output, and put the clamps back on, buttoned it up. Checked it again, now ready to go. Coming out of Standby, there was an odd turn-on hissy-thump that I've never heard before. And, turning either volume controls, it was making a odd noises. Then, switching in and out of S/B, I was greeted with a low pitched groan while seeing the amp drawing 330W instead of 140W. Something has changed!

Pulled the power tubes out, put one of the original pair in, all back to normal. Then tried one pair at a time of the new tubes, all fine. Got all four tubes back in, still fine. Odd. No noises or misbehavior as I had just before. I put the hold-down clamps back on, switched out of Standby, and once again, got the low groan and high current draw. Huh?? Released the clamps, and it seemed back to normal again. What do the clamps have to do with this?? The stray capacitance between the inside workings of the tubes and the tightly drawn Grounded springs outside the glass bottle can't amount to much at all.

I found by bending the clamps' spring flange out a bit, so the spring isn't in contact with the glass, I didn't get that low groan and high current draw. But, as soon as I tapped on top of one of the tubes, it started right back up. Removed the clamps, and problem stopped. I've never run into this one before!

Its' back on the bench, and with all four new matched tubes installed, I found I could get a 100kHz low level oscillation to sustain, driving burst pink noise thru it....Bright Ch, boosted the Treble control up......in the off time, a low level steady oscillation is sitting there. It's not doing the higher current oscillation thing yet....just starting to look at the situation.

I did notice only one pair of 1.5k grid stoppers were installed, NOT both pair. Lead wires come to the inside tube sockets, feeding the input grids directly, then 1.5k resistors feed from there to the outside tube sockets' input grids. I added some insulated terminals, and revised that so each tube has its' own grid stoppers, as we're used to seeing. But, I was still getting this low level oscillation. I don't get it with only one pair of tubes in place.

47k feedback resistor off the 8 ohm tap to the NFB side of the LTPI drive stage. No feedback cap, as per normal. No compensation caps on the power tube sockets. Have seen the 22pF cap on JCM2000's on V8, between screen grid and plate.

That's as far as I've gotten so far. Now makes me wonder just what caused the original Dagnall O/T to fail? Did the tubes failing cause that?? Client was no help on that.

I hadn't planned on having to work out re-compensating the power amp stage, dealing with a different Output Transformer. Still not sure if this new set of tubes is ok. Have another set to try, lower transconductance than what I've installed.

Anyone ever run into instability triggered by hold-down tube clamps?? That's SO bizzare!

[Chuck H]

I only buy power tubes as needed for individual repairs and builds. Even so, I get a disproportionate amount of microphonic power tubes anymore. It sounds like yours are grossly microphonic. I've somewhat solved for spring clamp noises by taking silicone ring tube dampers and looping them in a sort of "figure 8" over each clamp spring. Lots of guys report problems with spring clamp noise for el34's even with head cabinets.

I haven't been able to get Ruby (Shuguang) el34's that aren't microphonic. The last time I bought JJ's was over a year ago, but I did read two occasions, one here and one on another forum where, more recently, a new quad of JJ el34's were microphonic such that they weren't even suitable for use in head cabinets.

I think the HF oscillation may have been due to the mismatched power tubes. Since that amp was originally the design that uses a single pair of grid stoppers the mismatched tubes may have reduced the induced Miller capacitance from the single pair of grid stoppers. So, in this case you have solved for that by installing individual grid stoppers.

I've had about all I can stand with problematic tubes, brand new out of the box.

[pdf64]

What a nightmare!
A long shot but maybe try swapping the OT primary wires.
I'd consider 5k6 control grid stoppers on all the EL34, mounted at the socket terminal.
I assume that the 1k screen grid resistors are mounted on the tube sockets?

[nevetslab]

Yup...the screen resistors are right on the socket terminals. I haven't yet tried increasing the grid stoppers value. Having verified the phasing of the old and new output transformers, and wired accordingly, swapping the primary wires doesn't sound appropriate..

The Hold-Down Clamps have been fitted with high temp silicon rubber grommets & sleeving over the spring/clamp joints and wings to center the clamps, being oversized for the narrow tube bottles.

Further findings after yesterday's post was upon powering up from Standby, with the hold-down clamps in place, and this time scope connected to the output, the LF groan I heard from the speaker and seeing more than twice the mains power being drawn (330W vs 140W @ idle), it looked a little like power supply ripple, but with a lower fundamental (neither 120Hz nor 60Hz. I didn't get a screen shot of it, and, after several repeats of that start-up anomaly, that stopped, and hasn't occurred since, nor could I induce it by tapping on the power tubes, jarring the chassis, etc. It made me think about the first set or second set of power supply caps.(dual section 50uF-50uF/500V), series-connected that perhaps took a hit with the OT & tube failures. I haven't yet disconnected them to investigate further on the different GenRad bridges I have, nor look at the charging current thru them.



Images above of the chassis sitting in the service cradle, view of the revised grid stopper resistors added, in-line splices made on the primary wiring (routing still the same as before, only bundled the short service loop in making the splices), and the power tubes with hold-down clamps in place.

I also tried a different new quad set of tubes, same brand, and got a somewhat different LF groan, those tubes also held with the clamps in position. Mains power draw was less...270W, still twice nominal idle. I put the bias probes back into place and checked the two pair of tubes, finding one of the four measuring significantly higher plate current (30mA, 30mA, 30mA, 45mA). So much for their being matched tubes. Numbers were lower from the distributor than the set I have been working with.

I haven't yet tried one of those pairs with the tubes I got with the amp.

This morning, it powered up without any issues. Oscillation is still present when going thru bright channel, and with the treble control up full, presence near full up.

[pdf64]

A good way to verify that a feedback circuit is actually negative is to lift it and monitor whether the output increases or decreases.
Good practice for grid stoppers is to dress them so that the grid side resistor lead is as short as feasible, with the resistor body right up close to the socket terminal.
And having those OT crimped splices so close to signal circuits may be faciliting unwanted coupling; can the splices be moved outside the chassis?
I suggest twisting the blue and green LTP output wires, so as to reduce any radiated field; same for any wiring carrying balanced signals, eg OT primary.

[nevetslab]

I was concerned about the lead dress on both the Primary and Secondary leads of the new O/T. Not wanting to upset the existing wiring bundles, I cut them so they just do come thru the chassis grommet openings. I looked at doing tack-solder & shrink-tubing in-line splices. Not seen is the four underneath the component board...made those in similar fashion, and had better luck with keeping the lead dress as it was routed, though they still added a bit more lead length, not having full access without major work in removing one side of all the wiring to lift that board up.

There just wasn't enough lead length available even after cutting the old O/T leads off as they exited the end bells. I may be able to re-do the primary lead joints, soldering them instead. Tight quarters, and now, if I can 'un-crimp those terminals, I'll have a shot at shorter in-line splices. The path of the original primary leads is still where you see the in-line splices. Parallel lead dress in their harness. I will be able to un-solder and re-dress the grid leads to some extent...their lead dress is still as done from the factory, just now terminating on the terminal posts I had available.

More to follow....thanks for the input.

[Tom Phillips]

I'm having a hard time making out some of the details in the posted photos.
Is anyone else seeing grainy pics? Or...is it something going on at my end?
Tom

[nevetslab]

Un-soldered the Grid Drive leads from the terminals I installed, replaced the 1.5k grid stoppers with 5.6k, snaked the two leads back underneath the component board enough to re-dress and twist them together. All of Marshall's board wiring connects on the bottom side of that board, which isn't coming up without major labor. So, it felt like I was still hampered by some of their criss-crossing of wires. Still, improved that area. Powered it back up, and, having found before I could get the HF oscillation to occur with both the treble and presence controls cranked fully CW, that wasn't happening now.

But, next, using my 'chopstick' to go probing, I began plucking the springs of each hold-down clamp, and found one tube (V4)the most sensitive to that, and you could hear the modulating ground wire of the spring coming thru the amp...and finally set off that instability factor that yielded the LF groan with accompanying 3.4A Mains current/340W power draw.. I wasn't able to stop it, with repeated S/B switching, and turned it off for the time being. Each tube could pick that vibrating spring noise, with V4 reacting to it the most. So....yes, indeed, the clamps are playing havoc with these tubes and the borderline stability of the amp in it's present condition.

I need to see this again with the clamps removed all together. At present, the amp being near new, the tube socket terminals are tight fitting, so the tubes fit very snug into the sockets. I could get away without the clamps for now. No bear trap clamps on hand to replace them with at present.

Is the primary wiring spacing area (from the in-line insulated splices) playing a role here? Don't know yet. I wasn't able to cause any changes while probing that cluster of 3 primary leads where they join the wiring bundle that runs along the front/chassis floor corner. Going back for more abuse.

[nevetslab]

I'm having a hard time making out some of the details in the posted photos.
Is anyone else seeing grainy pics? Or...is it something going on at my end?
Tom

My appologies....still stuck with a low resolution digital camera...other one tanked, haven't spent the $$ to replace it.

[nevetslab]

I found after being able to excite the tube installed in V4 tube socket with the hold-down clamps' springs, that tube appears to have suffered some internal change, which also seems to have something to do with that tube socket. With that tube installed, it pulls current now in any socket. I moved another tube into that place, and even without the clamps in position, I was able to excite something in the interface (tube pins & socket contacts) that started up a high-current condition. The second tube survived, but now left with 3 matched tubes! And a seemingly physically healthy socket that wants to perform otherwise, in spite having solid, tight fitting contacts. Sigh............. Dealing with lead-free solder throughout, from the way the solder behaves in de-soldering operations on it.

My thoughts on the clamp springs....getting rid of them, and will replace them with shock cord, spaced far enough away from the tube glass so they survive. I haven't yet made the change.

[Tom Phillips]

My appologies....still stuck with a low resolution digital camera...other one tanked, haven't spent the $$ to replace it.

No apologies necessary. Your posts are very well written and illustrated. I wouldn't want to imply anything that would discourage you from posting photos. I just thought that there might be something happening to your photos during the upload/download process.
Cheers,
Tom

[Chuck H]

I've dealt with new, but failure prone tubes such that I would be looking into the possibility of coincidence. Too many bad tubes in play. Be they new or otherwise. And no wat to know until you have some known good tubes to plug in there. And because we can't trust new to mean good you'd basically need to pull tubes from an amp that has been working fine and plug them into that amp. I know it seems like crazy speak. I promise that after some recent experiences I don't like how loony it is either.

[nevetslab]

I've dealt with new, but failure prone tubes such that I would be looking into the possibility of coincidence. Too many bad tubes in play. Be they new or otherwise. And no wat to know until you have some known good tubes to plug in there. And because we can't trust new to mean good you'd basically need to pull tubes from an amp that has been working fine and plug them into that amp. I know it seems like crazy speak. I promise that after some recent experiences I don't like how loony it is either.

Not crazy speak at all. Fortunately, there's a large collection of Marshall heads in our rental inventory that I haven't had this issue with. Tried a fresh new set, only to be hoodwinked again. Took the day off to rehearse bass parts, and clear my head from the craziness. So, that's good sound advise. Not the first time I've gone to fetch an amp that works right to and borrow parts from. More to come............Steven

[nevetslab]

I made a set of tension bands using 1/8" dia shock cord (small dia elastic bunji cord material), with hooks fashioned out of large paper clips, fitted into aluminum ferrules.



After those were installed, I went and fetched a JCM 900 from our inventory, it having a trouble-free set of J/J EL-34 tubes in it....gave it a listen first to verify nothing unusual or odd about it. I moved those tubes on into this 1959HW Plexi, put the revised clamps into place, let it warm up, then switched out of Standby (8 ohm speaker attached, AC mains sourced via Variac and AC Power Analyzer). It immediately groaned, drawing 350W, and I switched right back to S/B! This is getting annoying.

Pulled the outside pair of tubes, remembering my question about the V4 tube socket status. Inside pair of tubes only, works fine. I moved that one tube from V5 to V4 socket.....drew current immediately! So, that clinches it, as far as this high current oddity goes....something about the tube socket.

Turned the amp over to address the contacts and see what it is. Forked-style Cliff socket fashioned after the brown Cinch socket (think it was Cinch). two-point contacts. I absolutely HATE those! Cleaning and re-tensioning them made no difference. So, replaced it, and started again.

That cured the high-current problem. It's running with the non-microphonic tubes from the JCM 900 at the moment. I then looked to see if the HF oscillation is there. Yup. In fact, that appears to NOT be a function of the output amp. I ran out of time today, heading to catch some of the MLB division series playoffs, so more to follow tomorrow.

[TimmyP1955]

The grid stoppers (four) are connected directly to the sockets, yes?