Determining if 2203 is set up for 6550 or KT88

Discussion in 'Amps/Cabs Tech Corner: Amplifier, Cab & Speakers' started by Tron Pesto, Jan 12, 2019.

  1. Tron Pesto

    Tron Pesto Member

    Messages:
    296
    Joined:
    Feb 18, 2016
    I just found myself in a very interesting situation. I got a great deal on an ~1984 JCM800 2203. Plugged it in - sounded ok - a little flat and lacking punch/depth so I went happily along replacing original filter caps, and checked the rest of the circuit - the usual. Looks like the amp had been retubed at one point - EL34s, but one had a Marshall logo on it, so I figured that one was original.

    In recapping, I did note a thing or two - the first that jumped out to me was the bias pot was a 10K, not the typical 22K, but I didn't really give it much thought at the time, it was just more of a passing thought of "hmm, it's definitely original, I guess it's just one of those anomalies in this silly amp world". (Spoiler alert, I should have looked closer at the rest of the bias circuit, it all would have been obvious if I did that.)

    When I got the caps in and put in a brand new set of tubes, the amp was better but still lacking "fullness" so FINALLY, I decided to check output tube current and at the very hottest end of the adjustment, the tubes were each drawing less than 10ma on a plate voltage of 468V - and BANG!!! it hit me, this amp was designed for 6550s or KT88s!!! In finally taking a closer look at the bias circuit, it's clear. Below is a snapshot of a typical EL34 circuit where I have pencilled in the bias circuit values in this amp.

    MY QUESTION: Is there a simple way to determine if this amp should use 6550 or KT88 tubes? Is an indication the bias feed resistors? The 150K in this amp make me think 6550 as a lot of KT88 amps will have 100K bias feed resistors... I don't see any difference anywhere else in the amp compared to the "standard" schematic.

    Or would either work? While 6550s require a higher negative bias than EL34s, but my gut tells me the bias at its hottest setting would still not allow quite enough current draw. I'm thinking the bias is actually high enough for the even higher negative bias requirements of a KT88. The thing is, I don't have quads of either laying around to test the theory. The negative bias sweep is about -53.4V down to 48.6V which is pretty narrow, but that's what it is. More than likely after I figure out which is the best tubes, I may modify the bias to give a wider range.

    [​IMG]
     
  2. Tron Pesto

    Tron Pesto Member

    Messages:
    296
    Joined:
    Feb 18, 2016
    The more I poke around, I'm actually leaning towards 6550s - it doesn't seem that U.S. bound Marshalls shipped with KT88s. The 10K adjustment pot still is bothering me. I can't find any evidence that talks about that value, but it is definitely stock - solder joints are original.
     
  3. Tron Pesto

    Tron Pesto Member

    Messages:
    296
    Joined:
    Feb 18, 2016
    One thing I need to check is where the NFB loop is connected - I just read that the NFB is connected to the 4ohm tap for 6550s and the 8ohm tap for EL34s. Not that helps with determining between 6550s and KT88, but it's just another point of confirmation it's all original (not in front of amp right now - eager to check)

    Also, looks like I got it backwards on the bias feed resistors - KT88s can handle a high resistance than 6550s. It seems to be common to go to 100K or even 82K for 6550s - although it seems Marshall used 150Ks in stock 6550s.

    If nothing else, this has been a good exercise in detective work and a little deeper look into Marshall configurations.
     
  4. gldtp99

    gldtp99 Member

    Messages:
    2,594
    Joined:
    Dec 23, 2005
    Location:
    Texas
    I picked up a beat-up, inop 1980 JMP 2203 that's currently on my bench.

    It's a factory 6550 amp---- 150k bias splitters, etc.

    I replaced Filter and Bias caps last night----- also Screen Grid resistors as one had signs of thermal damage.

    A Quad of JJ 6550's biased up with an almost mid-pot position.

    A very good sounding amp with a good once over (needed a bit more work than just what I described above, but nothing major).

    Is your 1984 2203 a vertical or horizontal input model ?

    How many Filter Cap cans does it use ? (my 1980 uses 6 50/50 cans)
     
    Last edited: Jan 12, 2019
    Badstrat likes this.
  5. Janus Alfador

    Janus Alfador Member

    Messages:
    423
    Joined:
    Feb 18, 2009
    KT88s can be installed directly into a stock 6550 JCM 2203 without changing anything. You only have to set the bias for the new tubes, and bias them as though they were 25 watt tubes.

    This thread has all the conversion information for switching a 2203 from EL23 to 6550 or KT88: http://www.marshallheads.com/viewtopic.php?f=32&t=346

    A 6550 2203 should have a 15K bias range resistor and 150K tube grid leak resistors.
    An EL34 2203 should have a 27K bias range resistor and 220K tube grid leak resistors.

    That Marshall Heads forum page suggests that for KT88s use 100K grid leak resistors, and possibly a 10K bias range resistor in the rare case that a 15K resistor doesn't work. However, using 100K grid leak resistors isn't needed for using KT88s in a 2203, and could also impact the tone (according to what I've heard) by reducing some of the top-end.


    If your goal is to emulate Billy Corgan's 1984 2203 with KT88s in it, then you want to use 150K and not 100K grid leak resistors because Billy's amp tech from the early days, Mike Rubano, said that he didn't change anything in Billy's 6550 amp when installing KT88s in it. The whole amp is stock apart from having KT88s in it rather than 6550s.


    The 10K bias range resistor sounds like it was changed at some point, so it might have previously been running KT88s. The 150K grid leak resistors mean that it is not set up for running EL34s, but either 6550s or KT88s.
     
    Steppin' Wolfe, Tron Pesto and pdf64 like this.
  6. gldtp99

    gldtp99 Member

    Messages:
    2,594
    Joined:
    Dec 23, 2005
    Location:
    Texas
    Yes, I want to echo an important point in made in Janus Alfador's post.

    Bias the 6550's or KT88's as if they were tubes with 25 watt plate dissipation ratings in these amps.

    Yes, they will be running cold------ that's ok.
     
    Tron Pesto, pdf64 and Badstrat like this.
  7. Tron Pesto

    Tron Pesto Member

    Messages:
    296
    Joined:
    Feb 18, 2016
    Thanks for the insight and the stressing to set up the biasing as if they are 25W plate dissipation tubes! I'm going to keep it "original" and install 6550s. It really is amazing that someone just dropped in EL34s in this amp and used it - it sounds like crap - it has no depth or character - the output tubes were running at about 8W each.

    Maybe I didn't explain clearly - the amp has a 15K bias range resistor (if we're talking about the same thing). The adjustment pot is 10K versus what I believe should be a 22K pot - and it's definitely original to the amp. We'll see if that will enable a usable range to bias the 6550s when I get them. I'm going to target about 17W per tube for an effective ~70% on the theoretical 25W.

    If it can't get there without maxing out either end of the range, I'll adjust the circuit by modifying R26.
     
    Janus Alfador likes this.
  8. gldtp99

    gldtp99 Member

    Messages:
    2,594
    Joined:
    Dec 23, 2005
    Location:
    Texas
    The 10k bias trim pot (RV1) is unusual.

    It is usually 22k or 25k.

    So the resistor in series (as referenced by Ground) with the Bias pot (R26) may not be as printed on any schematic.

    Or the value of the Bias input resistor (R30) may be different on your amp.

    What I'm saying is, because they built your bias circuit with a 10k Bias pot they may have compensated with a different (larger) value R26 than can be found on the schematic.

    Both R30 and R27 should be 15k on a 6550 amp's bias circuit.

    The only value resistor in the Bias circuit that shouldn't be changed/messed with is the 15k R27 which isolates the two Bias Capacitors.

    The values of R30 and R26 can be adjusted to give the range of Bias Voltage that you need to work with the particular set of output tubes you choose to run.

    The R24/R25 Bias Splitter/Grid Leak resistors are 150k for 6550's (or KT88's) and 220k for EL34's.

    These are the only values Marshall used here.

    The 100k value for R24/R25 would be a modification/someone's opinion.

    I've built amps, running KT88's, with 150k R24/R25 and they work just fine and the output tubes have a normal lifespan.

    I've also built amps with effectively 100k R24/R25 (by using a dual 100k MV pot in a Lar/Mar type PPIMV)----- these amps also perform in a normal manner.

    I would consider changing the 10k Bias pot to the usual 22k/25k value to give a wider range of adjustment.

    I tried two different sets of output tubes in the JMP 2203 I'm fixing up----- the Chinese KT88's I used first were higher drawing and I had to max out the Bias pot----- the JJ 6550's I installed the other day settled in with a mid-range position on the Bias pot.

    I don't think it's a case of KT88's vs 6550's but instead that the set of matched KT88's were higher drawing tubes vs the set of 6550's I ended up using.

    If I was dead set on using that quad of KT88's I would change the value of R26 to compensate/adjust the amount of Neg Bias voltage so that I had a center-ish position on the Bias pot when those tubes were properly biased.
     
  9. Tron Pesto

    Tron Pesto Member

    Messages:
    296
    Joined:
    Feb 18, 2016
    Yeah - that pot is weird. The rest of the bias circuit is spot on to they typical 6550 bias circuit - you can see the values I pencilled in on a EL34 bias circuit schematic.
     
  10. Tron Pesto

    Tron Pesto Member

    Messages:
    296
    Joined:
    Feb 18, 2016
    2 Horizontal PCB input - PCB for circuit, full sized PCB mounted pots, chassis mounted sockets.

    It's got 5 Filter cap cans - all 50/50. There's an inspection sticker on it dated May 10, 1984.

    I replaced all the filter caps, and the Pre-Amp Volume which was intermittent internally (the pot itself was bad, not the solder joints). Bias caps tested well so those stayed and I added 1ohm bias check resistors on the pots. No one had ever been inside the amp before - it was 100% original.
     
    gldtp99 likes this.
  11. Tron Pesto

    Tron Pesto Member

    Messages:
    296
    Joined:
    Feb 18, 2016
    Update for those who are interested:

    Received and installed new matched quad of 6550s. Dropped them in and dialed it up with the bias set at coldest. (I was concerned what the odd 10K adjustment pot would give me.) That came in cold, but not crazy cold.

    Dialing in the bias ended up almost exactly at the center of the pot range to give a really steady ~37ma on a plate voltage of 455V for a nice just under 17W or so per tube - pretty close on all 4 (between 35.8 ma and 37.1) awesome!

    So all my concerns that the adjustment pot would be too narrow was for naught - the low-end of the range was about 10W per tube, and there is plenty of room on the upper end (nearly half of the pot) for tubes that run colder.

    And it sounds great to boot. Happy Day.
     
    Janus Alfador, gldtp99 and pdf64 like this.

Share This Page