Marshall JCM 2000 amps from around 1997 to 2003 have circuit boards that are prone to leakage when they warm up. This is due to the type of circuit board material used, not the humidity of the your music room or anything else. You can read an excellent description of the problem on the Valve Tube GuitarAmps page.
The problem is fixable by modifying the main circuit board in the Marshall to isolate the bias wiring to the four EL34 output tubes, and replace that wiring with a new part. Valve Tube GuitarAmps sells a kit for this. The kit includes a special hollow drill that is needed to physically isolate the control grid pins (pin 5) on the EL34 tube sockets.
Diagnosis of the bias drift problem can be difficult and time consuming. The amplifier needs to be hot, and you need to be able to monitor the bias voltage on the test pins at the back of the amp chassis with a voltmeter when there is no signal through the amp. The typical symptom is that as the amplifier heats up, the negative bias voltage increases (less negative voltage) and the current in the output tubes also increases. As the current in the output tubes rises, they get hotter which causes the bias voltage to increase more and more until the tubes glow red hot and eventually fail. A set of four JJ EL34s will cost you over $100, and the bad news is that replacing them will not fix the problem. The new tubes will heat up and fail too.
As I mentioned above, the problem is curable by modifying the amp. I did such a mod on a JCM 2000 DSL recently and the rest of this post shows how I did it in detail. If you think you have this problem and want to install the kit, you need to be very competent working with the high voltages used in these amps and also have enough mechanical skill to do the circuit board modifications. The Marshall is also a rather odd design and has many, many wire connectors between the circuit boards so you will also need to be very detailed in how you document where they all go before you start.
Let me just say again: THE HIGH VOLTAGES IN A TUBE GUITAR AMP CAN KILL YOU. If you are in any doubt about your ability to perform the work safely, find a repair center that can do the work for you. Here in Florida you can get in touch with Russ Kruse at Florida Tube Amp to get the work done.
Here is the Marshall that I fixed recently.
This Marshall came in for repair because it had stopped working. Opening the amp up showed a screen grid resistor that had fallen out of the board at one end. Another screen grid resistor was open circuit. The PCB had got so hot that the solder melted. The components are under the PCB when the amp is upright and so after the solder melted, gravity pulled the resistor out of the board.
Tube amps can get pretty hot in normal use and I’ve seen resistors fall out of boards in a variety of designs. After re-soldering the hanging resistor and replacing the dead one the amp came back to life … but not for long. Once it had got good and hot again it lost power and sounded awful.
Before we go on, here’s a general shot of the underside of the chassis showing all the connections between the different PCBs, the output transformer, and the input power supply.
Measuring the bias showed it was very high. These are all the symptoms for the bias leak problem, so the amp was disassembled and the PCB set up on the bench with a high voltage power supply. With a high voltage on the HT input and a suitable grid voltage on the C- input I could watch the bias voltage on the tube socket grid pins rise as the board was heated with an air gun.
A bias fix kit was ordered from Valve Tube GuitarAmps in the UK. It took a couple of weeks to arrive. It’s during that time that, if you dismantled the amp yourself, you’ll be starting to wonder if you can remember where all those connectors went :).
The bias drift fix kit is not very large. It consists of one PCB with a few wires attached to it, and one large multi-pin connector on a short cable. A small hollow drill also comes with the kit. We’ll see how that’s used next. The kit has well written printed instructions with a lot of pictures.
The drill bit is really more of a hollow grinder as you can see in the picture below.
Before you can drill out the PCB it’s necessary to do a little prep work. The instructions don’t mention this but a couple of the big 5W ceramic resistors on the PCB are very close to the tube socket pins that need to be drilled out. I removed those resistors before I got started so as to have a clear field to work in. The instructions also say to remove a couple of small electrolytic capacitors that will be replaced by ones on the kit PCB. You have to remove these so that the kit PCB has a place to sit on the main tube PCB.
In the picture above the two upright capacitors behind the notch in the board are the ones to be removed.
The final step before drilling out the grid pin holes is to remove as much solder as possible from around pin 5 of each tube socket. You can do this with a solder sucker. I prefer to do it with solder wick as it’s a bit less violent, and with some patience produces a clean result as you can see in the next picture.
In the picture above you can also see how close to the pin the outline of the large 5W ceramic resistor was. This is why I removed them. You may also notice that C46 has gotten very hot and fried the circuit board. This is unrelated to the bias leak problem. This 22 pF capacitor was added to just one output tube in later editions of this PCB. As part of this repair I replaced the capacitor (which was just a little piece of charred ash).
Now it’s time to drill out the PCB around the control grid pins. The purpose of the exercise is to a) disconnect the grid pin from the original bias/drive circuit on the PCB, and b) to make sure we have some nice distance from the leaky PCB when we connect the new circuit wires.
Before I started this job I realized that it was going to be hard to keep the PCB level when I drilled out the pins, so I made a little wooden jig to go under the board while it was on the drill press table.
With the PCB on the wooden steady jig, line up the center of the hollow drill over the pin 5. Then drill down slowly until the bit is all the way through the PCB. When you lift up the drill again, you get left with a small piece of PCB fiberglass around the pin. You might also note that the drill bit tends to accumulate the fiberglass. I had to take mine out of the chuck twice to clear it out.
I used some very small pliers to crush the remaining bit of fiberglass and pull it out, leaving the hole clear around the tube base pin.
As it says in the kit instructions, it is VERY IMPORTANT to ensure that there is no swarf around the hole. That means no little pieces of copper from the circuit board which might form shorts between the tube base pins. I used some mega magnifying glasses and a bright light to check. I fished out any debris with a right-angled scriber tip. I also used an air line to blow around the tube bases from both sides to get it all really clear of any dust and debris.
I want to say again that it is very important to be sure the board is clean from any result of the drilling before continuing to install the kit itself.
Back in my electronics area I got the kit out of its packaging and inspected the board. These are very well made and should present no problems of their own. The wires were already stripped and tinned. I trimmed the tinned ends down to be a little shorter and also sleeved the wires with some heat shrink tube to slide over the tub pins once the wires were soldered in place. Follow the kit instructions to attach the grid pin wires. You’ll notice that the lengths of the wires are matched to the distance from the kit PCB to the appropriate tube base.
So, at this point you might be rethinking the idea that you can do this yourself. Hold that thought because we are just getting to the tricky bit.
The next step is to attach the two remaining wires to the plate (anode) circuits of the phase splitter. This is how the drive signals get from the ECC83 splitter tube to the output tube grids.
The first step is to find the two anode resistors on the tube PCB and lightly tin the ends that face the two ceramic coupling capacitors. You’re going to need a fine soldering iron tip and steady hands. Then attach the two wires from the kit. Be sure to follow the instructions and get them the right way round or you could alter the phase of the amp output, and we don’t want that.
Now it’s OK to breathe again. The final step is to plug the kit board into the main PCB and mount it with the supplied stand-offs.
The amplifier was then reassembled and tested. Reassembly is (as they say) the reverse of what you did several weeks ago to pull it apart. I take a lot of pictures when I pull anything apart as it’s just too easy to forget.
I always power up an amp after any kind of major surgery with great care. I run it up without the tubes using a variac and series light bulb in the power lead. Once I’m happy it doesn’t have any shorts, the tubes go back in and I repeat the power up watching the AC current consumption and monitoring the HT (B+) and bias (C-) voltages.
For this case I also set the bias pots to maximum bias voltage (as negative as it will go) to under-run the output tubes until they are good and warm. I then did a quick signal test to make sure we actually have an amplifier before removing the drive signal (a 1 kHz sine wave) and setting the output tube bias levels.
We could discuss bias levels for a couple of days here. The best source of information, in my view, is right here on the Robinette website. For this Marshall I used an HT voltage setting of 480 V (measured from the amp on the bench), and then set the bias for a fairly cold setting which runs the tubes at around 50% of max.
One of the nice things about these amps is that the bias is easily adjusted by just removing the back grille. So if the owner doesn’t like the sound of cool tubes it can be adjusted later.
Since the whole point of this was to fix a bias leakage problem that caused the bias to wander about, I ran the amp for quite a while with a meter attached to the bias monitoring pins. I played music through the amp as well as using a guitar to establish that it was indeed working OK, and then used a 1 kHz sinewave with the amp connected to a dummy load to do a full power test. After all of that, remove the signal and re-check the bias. (Remember to remove the signal before checking the bias reading on the voltmeter.
When I work on an amp for someone, I’m really pretty paranoid that the work I do is good, fixes whatever problem the amp arrived with, and that it stays fixed. That isn’t always what happens of course as sometimes I end up masking the symptoms of a problem and not fixing the root cause. It’s always disappointing to get an amp back for re-work. In this case, I think we have the amp fixed and ready to rock on.
November 2022