Tired old Marshall JCM900 4100 problem

[Dave Mac Farlane]

I have leafed through threads on this one on this forum, and I must say, maybe you have a better class of people, but everyone seems friendly, helpful, not conceited, and not sarcastic. I wonder if we've all floated over here from some famous name forums.

I have a Marshall JCM900 4100 1999 (signed version (Jim Marshall)). I got it with Two similar EL34's one different make, and an E34L.
Ordered quad matched 5881's (Watford valves) as recommended by Marshall, as well as their Biasing procedure, which is to aim for -49V on 5.
Copy as
"Thank you for your e-mail regarding your question the JCM 900 should be fitted with 5881 we recommend you purchase these from your local marshall dealer. Regarding the bias we adjust this between -48 to -50 volts on pin 5 of one of the power valves."

Found this to sound simplistic, so did research, and incidentally noticed that someone on this forum quoted a Marshall tech as having said - take the current in the fuseholder, and this'll give you the bias for 2 valves. Hmmm.
Being a lazy person, (You know, I just like to play music), I checked the voltage. Spot on 49V.
Less than a year later, one blows up and takes out the 1.5K grid resistor.
I replaced this, and replaced the 5881 with an old EL34 - Ahh, sound.

Ordered new valves from a different company, who sent me 6l6's. Now, I know they are supposed to be identical, but they do have subtle different characteristics. In addition to this, I had accepted that this company don't think Cryo makes as much difference as 'burning in'.
Being a trusting sort of soul, (or perhaps bone lazy, just wanting to get on with playing) I installed a complete set, and again checked for my -49V.
Ok, now it seems a bit transformer hummy, but it was low enough to live with, and if I pressed the chassis, it lessened, so definitely a mechanical resonance, although I hoped no damage had been done to the tranny.
Hey presto, a few weeks later, and not a lotta playing, for I had been away, valve goes pop.
Now replaced with old 5881 - hmm, sound, but what's happening.

453 Volts on Anode (Plate), interrupt Ground to Cathode with Ma meter, could not get the current to go below 49Ma - this was the bias pot at it's lowest.
I am afraid for my amp, and my valve theory from my college days is rusty.
I am worried that someone might have changed a component in order to run this amp with EL34's, and left me with superheated valves.
Any suggestions?
Dave.

[Enzo]

Any tube can fail at any time, never forget that. Even new right out of the box. Just because one set of one brand lasts five years doesn't mean the next set of the same exact tube won't die in a week. Never forget that.

You note that the 6L6 and the 5881 have subtle differences in tone. Well so do each different brand and variant on 6L6. Hard to say that the differences you hear between one 5881 and another 6L6 are due to the number on the label rather than finding the same amount of difference by switching from SOvteks to JJs, both 6L6.

6L6 - of which any current production 5881 is one - has a different bias need than the EL34. Now each type has a range, so it is possible one EL34 needing more bias will work with a 6L6 that needs less, but as a rule, mixing them is not kosher.

Bias won;t adjust down far enough? Change a couple resistors to widen the adjustment range.

I have a question. YOu said you inserted a current meter in the cathode leg. was your 49ma per tube? as in 98ma reading?


49ma tube current at 453 volts is about 22 watts disipation. A bit hot for my taste.. I'd raise the bias voltage some to bring that down.

A word on 5881. If you want to consider specs, look up the data sheet on the actual 5881 you are buying (or considering at least). Don;t look up 5881 in the old RCA guide. Unless you are buying NOS tubes, the 5881 today is not the same tube as the one in the old tube manuals.

[Dave Mac Farlane]

Thank you Enzo, for that helpful reply, and I agree with everything you say.
The differences were really what I was seeing, but I am one step further in prognosis.
Last night, after resting away travel to work fatigue, I attacked the beast again.
I put in known good 5881's in the right, and the new 6L6's in the left.
I patiently disconnected each cathode and inserted my meter - I am an electrical engineer, so I am more able to look into a circuit with care than warranted the 1ohm resistors - one has to break into the circuit to do that anyway, but:-
on the right side, the meter crept up to 29 mA on the right for V4 and V5, but on the left, I couldn't get it below 50 (and climbing) via the bias pot.
Ran out of time.
Next step, which you may consider futile, will be to swap V4, V5 with 6 and 7. If the same side is high with the 5881's I would presume to have a fault in the circuit, perhaps caused by the original failure, but if the sides (current) change with the valves, well, it looks as if all of the new valves are not what they seem to be!
I am still open to suggestions and help, for having been involved in the service industry as well, I am never too proud when someone comes up and says "Have you tried..........", for I know that experience is sometimes stranger than fiction.
At the moment, I am less concerned about eventual sound quality than getting the darned thing to work - 'cos when it works, it's great.
Dave.
PS I believe you posted something ages ago about hijacking posts - I did do my searches, but didn't come up with anything pertaining to this particular problem, and it's sometimes difficult to navigate forum pages anyway, so I hope I haven't disturbed any protocol.
Dave.

[Enzo]

I don't worry about hijacking threads so much as I like to suggest people start new threads for their amps rather than tacking onto the end of a year old three pager from someone else. It simply will get better attention and will be less confusing that way.

Here is a brute force technique I sometimes suggest. Pick a socket, one you can measure at. Take one of your tubes and install it there. Power up. Take whatever readings you care about. Now remove that tube and stick another in the same socket. Take readings. Number your tubes somehow or otherwise keep track of them. By doing this, we know that the exact same conditions will exist testing each tube. Any differences in readings should be due to tube differences. We should then find any that are prone to running away or are way off from the others in some way.

That was half the process. Now take a tube you feel OK about, and stick it in that first socket, retake your readings, now CHANGE NOTHING, but move the same tube to the next socket and take readings, then the third socket, then the fourth. That way, we know the same tube was used for all measurements, and any reading differences will be due to the circuitry.

Probably wouldn;t hurt, before installing any power tubes, to verify DC voltages in the sockets. Is B+ present in all four at pins 3 and 4, and is bias present at the same level in all four at pin 5. Measure the resistance of the screen grid resistors. A simple way is with power off, stick your meter intp in 4 and measure resistance to the other pins 4. In each case you are reading two screen resistors in series. You probably have that step covered.

Measure the DC resistance of eirher half of the output transformer primary - from center tap to the plates of each side. Ideally they would be the same, but usually they are not. And in some transformers that are simple winds, they can be substantially different. The difference in resistance doesn;t much upset the audio (which relies on turns ratios), but that resistance is in series with the tube and can cause one side to consistently read alittle diferent from the other in idle current. It may or may not be a factor, but it is something one should be aware of.

[Enzo]

I don't worry about hijacking threads so much as I like to suggest people start new threads for their amps rather than tacking onto the end of a year old three pager from someone else. It simply will get better attention and will be less confusing that way.

Here is a brute force technique I sometimes suggest. Pick a socket, one you can measure at. Take one of your tubes and install it there. Power up. Take whatever readings you care about. Now remove that tube and stick another in the same socket. Take readings. Number your tubes somehow or otherwise keep track of them. By doing this, we know that the exact same conditions will exist testing each tube. Any differences in readings should be due to tube differences. We should then find any that are prone to running away or are way off from the others in some way.

That was half the process. Now take a tube you feel OK about, and stick it in that first socket, retake your readings, now CHANGE NOTHING, but move the same tube to the next socket and take readings, then the third socket, then the fourth. That way, we know the same tube was used for all measurements, and any reading differences will be due to the circuitry.

Probably wouldn;t hurt, before installing any power tubes, to verify DC voltages in the sockets. Is B+ present in all four at pins 3 and 4, and is bias present at the same level in all four at pin 5. Measure the resistance of the screen grid resistors. A simple way is with power off, stick your meter intp in 4 and measure resistance to the other pins 4. In each case you are reading two screen resistors in series. You probably have that step covered.

Measure the DC resistance of eirher half of the output transformer primary - from center tap to the plates of each side. Ideally they would be the same, but usually they are not. And in some transformers that are simple winds, they can be substantially different. The difference in resistance doesn;t much upset the audio (which relies on turns ratios), but that resistance is in series with the tube and can cause one side to consistently read alittle diferent from the other in idle current. It may or may not be a factor, but it is something one should be aware of.

[Dave Mac Farlane]

Brilliant Enzo - 10 out of 10 for obeying the old Maxim - "localize the fault".
I went home last night without catching your e mail with something the same in mind.
I swapped the 5881's side to side, and found the same high readings from the 6L6's. still 28 mA from the 5881's and 49 up for the 6L6's - for me, proved the amp was ok.
Put in the old el34's and the current was in the same ball park - I know they run cold put straight in as substitute for 5881's, but that was worth the risk to the sound to get it going - hey presto, balance - I biased the 5881's to 38mA, and the EL34's were a bit low, but I can handle that.
This means, of course, that I assumed the 6L6's would plug straight in - ASSUME - makes an ASS out of U and ME!
There are two disturbing things, however. One is that I now have a plethora of umatched valves - not a big worry, but the other thing is that the current would not reduce below 49mA for the 6L6's.
No matter what I did, it looks as if I would have to effect a component change in order to get these to work - or have I been sold a complete bum steer with a matched quad 6L6's wot aren't really?
I am running now with EL34's right and 5881's left, and thankful I got in a bit of playing time with my LP standard.
I am awaiting new 5881's and will thoroughly bias said valves.
I should have said, plate voltage 455V, pin 3 -49V, and I know that 38 mA is a bit low, but I wasn't taking any chances for now - just wanted to play!
I checked the assocaited resistors and although some looked a bit dark, and as if they had suffered, they were ok.
I really appreciate the level of this discussion Enzo -thanks for that - any further ideas on this conundrum are welcome.
Dave.

[g-one]

Dave, I think it is important for you to find out if this amp is set up for 5881 (6L6), or EL34. I think there were different versions of this model. Marshall told you 5881, I'm not sure if they were going by the serial number or what?
Anyway, even if it was the 5881 version, it is possible someone changed it over for EL34's.
Schematics for both versions are available via google. You should find the version for the power tubes you want to use and make sure the bias circuit is correct. It seems to me there are a few other changes, such as screen resistors and maybe some B+ dropping resistors?

[Enzo]

For 6L6 type tubes, I'd kinda want up to maybe 55 volts of bias available. You should be able to change a couple resistors in the bias supply to get a few more volts at the cold end. A g-one suggests, if the amp is set up for EL34, that may be why you can;t bias the 6L6 where you want it.

As your tube currents decrease, expect the B+ voltage to rise some in response. Expect some interaction.


I don't think there is anything in the world wrong with 38ma. What matters is how the amp sounds. More than that what matters is how the amp sounds TO YOU. The bias needs to be cold enough the tubes don;t red-plate. And they need to be hot enough you don;t have way too much crossover distortion. But between those extremes, if it sounds good to you, then it is set right. Don't worry about overly precise internet recommended adjustment standards.

[Dave Mac Farlane]

Wow!
G one, Enzo, I can't believe how helpful you guys are being. I am going to stop being old and cynical.
Dammit, when I first bought the amp, it had EL34's in it - but of course, I took Marshall's techs at their word.
I am sure you have hit the nail on the head - and I have been gullible enough to buy a matched quad Harma set - destroy one of them, and be left with a mismatched set, then 6L6's, same again, and have just ordered 4 5881's as replacements.
Wow!
For one who assumes rightness on the basis of one's technocal ability, I see now, Education is what you're left with after you have forgotten what you have learned.
HA ha ha ha ha ha......
The joke's on me.
If you're interested, I shall post here, the results of my labours and your advice.
Thank you again, gentlemen.
Dave MacFarlane

[jmaf]

It's easy to ID which tube it's made for, screen resistors for 6L6 are 470R, for EL34 are 1K. If you inserted EL34's on the 6L6 version, you might have damaged the screens. But I don't know that that would have been enough to destroy the half ampere fuse.

As for inserting 6L6's in the EL34 version without a bias adjustment, that would be more likely to blow those cathode fuses - though half ampere at 500 VDC is 250 Watts - a little too hot for a pair of 6L6's

Dammit, when I first bought the amp, it had EL34's in it - but of course, I took Marshall's techs at their word.

From the above, there's a big chance you have the EL34 version.

[Dave Mac Farlane]

Hmm, Ok, here's the confusing part of the anomaly - 22k pot, 56k for R28, 15k R29.
V4/5 or two LH ones, EL34, V6/7 5881.
5881 = 37mA, EL34's 24mA - seems logical when extrapolated from the aforesaid data, but when I replace the EL34's with the 6L6's, it reads 79mA - I take cold feet and switch off - it didn't go much below that originally.
I have committed to 5881's, so is there a quick fix? I am awaiting a matched quad of sovteks.
Any takers?
Dave

[jmaf]

5881 = 37mA, EL34's 24mA - seems logical when extrapolated from the aforesaid data, but when I replace the EL34's with the 6L6's, it reads 79mA - I take cold feet and switch off - it didn't go much below that originally.
I have committed to 5881's, so is there a quick fix? I am awaiting a matched quad of sovteks.

5881's @ 37mA PER TUBE seems fine for 100Watts, biased at ~70%.

If I were you I'd remove all power tubes, set bias to maximum and check the voltage on pin 5 of all 4 power tubes. They should be close to -70 VDC or within the -60's. That voltage is enough to cut off both 6L6's and EL34's.

Then reinsert the tubes and check that indeed they're all cut off, no current. From there decrease bias to allow for 35 to 40 mA per tube, or around 70 to 80mA per pair if measuring from the fuse holder.

[Dave Mac Farlane]

Many Thanks jmaf,
And screen resistors are 470k, dammit!
I shall follow your recommendation, still treading cautiously with this strange mix match of valves.
Originally, I bought it with 2 EL34's one with an eroded number which looked like these, and an E34L - thought I couldn't do any worse!!!
Just to clarify, when I insert any of the 6L6's they suck 79mA at the lowest setting, yet my 5881's are running at 37.
I can't trust anything till I put 4 same valves in, and I'll try to run with the 1k's and 5881's.
Dave.

[jmaf]

And screen resistors are 470k, dammit!

Should be 470 ohms, not 470K, but if so(470 ohms) and this Marshall is unmodified, then it's set up for 5881's and my previous post's conclusion is wrong. Your Marshall tech was right in that case. We are assuming based on assumptions here, that is risky

I shall follow your recommendation, still treading cautiously with this strange mix match of valves.
Originally, I bought it with 2 EL34's one with an eroded number which looked like these, and an E34L - thought I couldn't do any worse!!!
Just to clarify, when I insert any of the 6L6's they suck 79mA at the lowest setting, yet my 5881's are running at 37.
I can't trust anything till I put 4 same valves in, and I'll try to run with the 1k's and 5881's.
Dave.

Just to avoid mixup: On unmodified original Marshalls 1K = EL34 and 470 ohms = 6L6 or 5881

What voltage do you find on pin 5 of those 6L6's when they are pulling 79mA? That is obviously heinously wrong, 6L6's are red plating violently when at 79mA.

[Dave Mac Farlane]

drat!
My excuse is that I am home with a chest infection - I don't even smoke!
NOT 470K, or even 470 ohms - got the number wrong, not the power - they are 1K - sorry, typo in haste. Hence statement - "I can't trust anything till I put 4 same valves in, and I'll try to run with the 1k's and 5881's."
Anyway, 79mA = -46V, 467 plate.
Remember I am mixing the 2nd lot 5881's (2) with new 6L6's (2) 'cos I aint got a matched quad till tomorrow- but they will be 5881's!
Dave

[jmaf]

drat!
My excuse is that I am home with a chest infection - I don't even smoke!
NOT 470K, or even 470 ohms - got the number wrong, not the power - they are 1K - sorry, typo in haste. Hence statement - "I can't trust anything till I put 4 same valves in, and I'll try to run with the 1k's and 5881's."
Anyway, 79mA = -46V, 467 plate.
Remember I am mixing the 2nd lot 5881's (2) with new 6L6's (2) 'cos I aint got a matched quad till tomorrow- but they will be 5881's!
Dave

I have a hunch your matched quad's gonna work a dandy

The 1k resistor is ok for 5881's. What happens is the screen sags more and the tone is more compressed. If you feel it lacks punch, then you'll need to reduce it to 470 ohms.

Get well soon from the chest infection.

[Dave Mac Farlane]

I must be tired.
-51V is the lowest I can get - 512V plate.
I already have 56K, think I may change it for 47K, see how it does - I am so out of practice with audio - I'm gonna have to dig out all my old college books.
Dave.

[Dave Mac Farlane]

Thank you. You are most encouraging - I shall keep you posted.
This forum is tops.
Dave.

[Dave Mac Farlane]

OK, sovtek 5881's arrived.
Place in holders, knowing I have plate V - 510V no load, max available -52V on 5.
First Valve - Mains transformer side of amp - 35.9 mA. Looks good.
Reconnect - move on to next in line - wow! 45.7 mA.
Next, same again, but 47 mA.
Hmmm, any takers?
I'm going to alter the divider circuit, but I don't have a 47K pot - any suggestions on a configuration to get me - 55 up volts
Dave

[Dave Mac Farlane]

This isn't interim if it continues to work - I haven't thought it all the way through, but I strapped in a 100K in place of a 56K on the top of the bias pot - I've now got a healthy -68V with no tubes in.
Need a second brain on this one, for mine feels as if it's malfunctioning - stress of having no amp all week - my Peavey envoy is no substitute.
Help guys....... I need to put the tubes in now - am I doing anything stupidly dastardly wrong (shakes and sweats, reaches for the bottle, willing the screen to offer him a reply...........

[Dave Mac Farlane]

Told you I was getting desperate - couldn't get the bias current up to 20 mA - Hee hee!
Replaced that with three 22K in series - got good vibes so far.
Dave.

[g-one]

First Valve - Mains transformer side of amp - 35.9 mA. Looks good.
Reconnect - move on to next in line - wow! 45.7 mA.
Next, same again, but 47 mA.

Not sure here whether you are trying one tube in different sockets or you have all the tubes installed and are checking each tube?
A 10mA difference between tubes in a matched set is not acceptable. Please refer to Enzo's post #4. Try all your new tubes in the same socket to see if they are unmatched, then try one of the tubes through the 4 sockets to see if there is a mismatch in the circuit.

[Dave Mac Farlane]

Many thanks, G one, I checked my notes - 39 and 45 were the differences - and what you say is wisdom indeed. checked 'em just like you said - perhaps I didn't hang about too long at that current!
I have altered the resistor above the bias pot to 66K (yeah, three 22k's in series - it was all I had here) and I am now running all new 5881's at 34.9 to 35.1 as it heats up (checked all four) - just had a play and am delighted.
Was concerned I might have gone to the wrong part of the voltage divider, but it seems to be working without too much heat.
This now means I have a spare set.
Thanks to you and all of our friends for your help and support.
Dave

[Dave Mac Farlane]

Better than that guys, its going and sounding a treat!
Dave.