Tube Replacement JCM 900

[shortfuse]

Did my 1st tube replacement and bias adj tonight on my JCM 900 50w model 4500 w/ 6l6GC's.  I have a few questions.
What I did.
Changed all 3 pre amp tubes and the 2 power tubes ECC83S and 6l6GC back with the same.
I 1st checked plate voltage 476.  25 / 476 = .0525  70% = .0367 Setting to use
Using my Hoffman bias adapters I went to set bias.  One side I can get to .037 the other will not go higher than .027.  Is that normal or is there something off, bad resistor, or other issue?  I think I did this right but i thought you were supposed to get the same reading for both tubes?  Plate was 473 when done.
Should I have filled out a voltage chat for all 5 valves for this question?
Now the beginning of the story 2 weeks ago when I fired up the amp it did a snap crackle pop when i flipped the stand by switch on.  Played a while and noticed I had to set the treble down to 2 or 3 to play and the amp just did not sound right.  When i went to shut it down I noticed one tube glowing much brighter than the other so I ordered the new tubes and did not use it again until tonight.  Checking plate on the old tube's 476 one side 190 the other 33.  the brighter tube was the 190 reading.  Pulled them and replace with the new.

[eleventeen]

...one side 190 the other 33.  the brighter tube was the 190 reading

Understatement of 2010, but the year is not over yet.....!

The bad-tube episode may well have "altered" the value of a screen-grid resistor or one of the other grid resistors on the 6L6's. I would check those. 

[shortfuse]

Please send the award to my home address for the understatement of the year I will be unable to attend the ceremony.   Hey at least I won something substantial this year.

The only thing I have found thus far is R28 which is supposed to be 47K is actually 69K.  Could this cause the .010 difference between the two?
Diagram I am working off of http://www.drtube.com/schematics/marshall/4500-60-02-2.gif

[eleventeen]

Doesn't seem to me like the value difference you cite (69K vs 47K) is that drastic; plus, the other "output" side of that biasing network feeds both tubes equally.

Are you saying R28 is identified/striped as a 47K but measures 69K? That's actually typical for a resistor to "bake up" in value after an overload. (Not all resistor types; I forget which ones do this, but it is most of them) Replace w/correct value, it may even do you well to go to say a 33K on a temporary basis, which will have the tendency to give a little too much negative bias to the tubes....which is a good situation...run them cool while you're futzing around. Just tack that resistor in.

I would perform a general surveillance on the components that touch that -50 volt "split" point...the 1500 ohms, the 150Ks...and take a really close look at the tube sockets and maybe give them a scrub with some alcohol if there's any gunk on them. Be darn sure you don't have an arc'ed contact, meaning a carbonized path between two contacts. Same if you have wires running across a PC board. Wouldn't hurt to replace those 22 pf caps on the off-chance that one is leaky and letting some + volts through. That would ruin your day, that could well produce exactly the effect you described with one tube going nuts. Certainly measure those. On replacement, I'd definitely be looking to find some higher voltage rating caps in that position, too. 400 is just barely there. I haven't worked on any Marshalls, but I find the "hi-power-low-power switch a little bit drastic in terms of what it could do if switched under power. Just my .02.

[shortfuse]

Quote
"Are you saying R28 is identified/striped as a 47K but measures 69K?"
No i am saying the schematic says it should be 47k and it measures 69K.

Hear are the actual readings and the values that are shown on the schematic (schematic link 2 posts up).  I put ** next to R28 and R17 as those are the only ones that seem out of line to me.  The bias resistor VR2 on the schematic says 47K pre is this pot measurable with the meter or does that mean it should swing 47k in range from one extreme to the other?  Now I switched the tubes around twice tonight and the bias values (.037 - .027) move with the tubes and do not stay with the sockets does that mean the matched pair I bought is not matched and I am chasing my tail?
Again will filling out the voltage info help with this type of issue? 


Location   Value     Actual
R35            100     99
R30            10K     9.7K

R31            470   470
R34            470   470

R9            100K   100.2K
R7            56K   54.8K
R29            8K2   7.9K
R28            47K   67.3K**
R25            1K5   1.4K
R26           150K   148.9K
R27           150K   148.2K
R24           1K5   1.4K
R23           100K   94.8K
R22           82K   77.5K
R19           1K    .9K
R20           1M   .97M
R21           1M   1.0M
R18           15K   14.7K
R6           47K   4.4K
           47K   4.5K
R17           47K   26K**
VR2           ?   ?

[Danskman]

I would come back to your reseller with the "" matched "" pair and ask for a replacement set. Those are obviously not a matched pair... What brand are they?

"No i am saying the schematic says it should be 47k and it measures 69K"
Are you saying it is marked 69K and you measure it 69K but it should be 47K following the schematic or it is marked 47K and you measure it 69K? In the second case, you have to replace it, for sure; it has drifted in value. Sometime, amp maker change components value during production time.
R6 is marked 47k on the schematic and you really measure it 4,7K?? Shouldn't it be an error in the schematic?

Forget it, I think you have to replace your 6L6 pair FIRST, IMHO and see and hear what happens.
Good chase and best regards,
Danskman

[RicharD]

Swap the tube placement.  If the current difference follows the tube, then the tubes aren't matched, although 10mA is not that mismatched.  If the current difference stays with the socket, then something is amiss.  Don't sweat R28.  Measure R31 & R34 with the tubes removed.  They should be fairly closely matched.

[shortfuse]

Richard thanks for the info in your post.
In my previous post it does say that I did swap tubes several times and the 10ma does stay with the tube.  I also cleaned the tub sockets again and same results.  So reading further on i guess that is here to stay.  Also reading in the library or other documentation i think it said to stay about 2ma apart.  But if 10ma is OK to use and will not damage anything I will leave it alone I think the amp sound great now.  I set the bias up at about 38ma and got rid of a lot of harsh noises and the clean channel is crystal clear sounds better than when I bought it.   I would assume you would use the value on the tube with the highest ma of the 2 while setting which is what I did. 
R31 & R34 are dead on 470 ohm.
I had never changed the original tubes it came with in 1993.  What the heck did I know I turned it on and it played fine so why mess with it until i heard the snapping 2 weeks ago.

I would come back to your reseller with the "" matched "" pair and ask for a replacement set. Those are obviously not a matched pair... What brand are they?

They are a full set of JJ's from JJ.  But if 10ma is not out of line then what do you do.  Although it says matched set on the web.  What should you get when you ask for a matched set?

To clear up the resistor value questions.  Yes it says 47K in the schematic and it stripes out to should be 47K but R28 is 67K which is +20K (i dropped the decimal and rounded) and R17 is 26K which is -21K from it original value.  I picked up some 47K resistors this morning and will change them out as everything else seems fine.

[eleventeen]

I'd ask for a replacement of those allegedly matched tubes. I don't call 10 ma variance matched at all. No way. Y'know, tubes can be pretty sloppy things in terms of nominal precision, but 10 ma difference in a range of zero to say 45 ma is huge. No way would I call that matched. In fact, I'd turn the challenge around. Go buy 100 power tubes of one type & mfr and check their currents. If 10 ma variance is an acceptable "match" then I'd challenge you to find me any UNmatched pairs. I'd bet that 85% would match.

Reason I say this is, I am doing this, right now, with my tube matcher. I am trying to find tube pairs that match up within 1-2 ma. I have fair numbers of NOS JAN 6L6GA from the 1940's, 6L6M, and some 5881. I find pairs of 6L6GA with the same date code, both NOS, with 10 ma variance, easy. To get matched pairs at what I call "matched" I really have to weed out substantial numbers of tubes. To do this, as a business, it's clear (to me) that you'd have to intake many hundreds of tubes. A dozen or two would not do it, not for the smallish premium suppliers charge for matched pairs. The 6L6Ms I happen to have are simply ludicrous, they are incredibly all over the place. (some are used, some are used A LOT, a few are JAN NOS, most are commecial grade...no two are the "same", per se, and I have zero expectation of finding any MPs in that motley group. Nor do I really care; these are for power supply pass elements)

Strictly my opinion, SWAG: If you really want Gm to match within say 5% and cathode-measured idle current to match within say 2 ma, IMO you can't count on more than 3-4 matched pairs per 20 incoming tubes. Obviously, if your standards are looser, you can find more matched pairs. BTW, I consider the current test to be a LOT more important than Gm, in power tubes.

[shortfuse]

Well I thought this was done.  Replaced R17 & R28 back to 47K played the amp for a few hours this afternoon went to recheck bias after I had been playing a while because I noticed that the gain was a bit less now then when i started.  I set bias at 38ma and now it was at 32ma the bias adjustment resistor VR2 will only allow me to adjust 27ma to 34ma so a total range of 7ma.  Where do you order a Marshall PCB bias adj resistor?

[RicharD]

R17 is a NFB resistor.  Changing it from 68k to 47k will reduce gain.

R28 is a voltage divider resistor in the bias supply.  Changing it from 68k to 47 k will raise voltage thus lower bias current.  Put both the 68k resistors back.  The schematic probably changed somewhere between rev1 and rev?.

[shortfuse]

I got to thinking about it and doing the math biasing them at 36ma and them being 10ma off is almost 30% off so I called JJ Monday, they apologized and said 10ma was to far off, they should be within 2ma and are sending me a new set.  Thanks for all the other info thus far it has helped.

[eleventeen]

Interesting.  I wonder what their "match" rate is, eg; how many tubes they have to go through to get a matched pair. Keep in mind that a "matched pair" could occur over a range of currents. By that I mean, given an arbitrary 320 volts on the plates and an arbitrary -25 on the control grid, some tubes could show 28 ma, some could show 30, some maybe 33 ma, all while being within spec. While these are "pulled out of the air" numbers, these numbers are based on running through a pile of 6L6 and 6L6G and 6L6GA, some NOS, some used, some WELL used.

I would consider any of those corresponding pairs to be matched. Alas, they would not form similarly-spec'ed "matched quads".

It would take a serious pile of tubes and time to come up with matched quads.

[shortfuse]

Returned home this weekend and new tubes were here replaced them 1.8ma apart so the miss matched ones are packed up and going back.

R17 is a NFB resistor.  Changing it from 68k to 47k will reduce gain.

R28 is a voltage divider resistor in the bias supply.  Changing it from 68k to 47 k will raise voltage thus lower bias current.  Put both the 68k resistors back.  The schematic probably changed somewhere between rev1 and rev?.

The R17 is color coded as 47K measuring 26K So that one i did change back to 47K and measures accordingly now.
As for R28 I looked at the color bands with a magnifier glass and it is coded to be 68K, so i took the 47K i replaced it with and put 68K back as recommended.  But bias is still low read below.

Here is my math again 25/462=.054 X 70%=.037ma bias setting????????  Or should I be using 75% for 6l6GC if that's the case i would have to be at 40ma and cant get there by adjustment.  The bias adj pot will only allow me to go from 22ma to 33.1ma.  So i set it a 32 and let it go.  Should the VR2 bias pot be replaced or should the R28 value be altered?  Or should I just leave it alone an play it.