Marshall MkII Woes

[walter]

Tube arced in a Marshall Master Model MkII 100W. I have just re-tubed and when I crank the volumes to max and hammer on the guitar, the screen grid resistors start to smoke. I have biased plenty of amps, but this is no longer my day job. In the past I would use the shunt the transformer method and read the current across the plate and center tap. Typically 35mA a side for 50Watt amplifiers and 50mA a side for 100Watt amplifiers. I now use a socket extender that breaks out the cathode so I can read the current through the tube. I haven’t used math in the past, but am curious about applying some electronic theory.
I found an equation on-line that states (Plate dissipation/plate voltage)*0.7 = bias. My amp has 457v on the plates, 6550s. The amp originally was designed for EL34 with plate dissipation rated at 25W, so (25/457)*.7=.038 or 38mA. That’s a bit hotter than I would set it as that’s 76mA a side. I have it set at 25mA for each tube. I looked at the screen grid resistors and at idle read 452v. I want actual values so I read them while blasting through and got 340v at the tube and 385v feeding them. This gives me 45v across 1kΩ, 45/1000=.045A and .045*45=2.025W These are 5W resistors, so why are they smoking?
I have never used an o-scope to check crossover, so I tried that and see some. It was actually quite easy to set up.
I re-biased, maxed out the bias pot to -54V and got about 30mA a tube. This put the plates at 439V, now the screen grid resistors drop 49V for 2.4Wdissipation. They don’t seem to be smoking as much now.
The crossover is gone too.
Now I do my guitar hammer at full volume, one tube arced and the 4A fuse blew. These are new Tung-Sol 6550 matched quad. Maybe the tube was poor quality? I have an old tube and new fuse, put it in there and power up. Seems o.k. I removed the quad and checked the transformer. 14.6Ω on one side and 16.2Ω on the side the tube went bad. About 1.5Ω different, this doesn’t seem too far out, does it?
I have seen some Marshalls with 4A fuse, some with 5A. This one is labeled 4A, but I have replaced the blown 4A fuse with a 5A, that’s what I have on hand. At this point I’ll take a close look at the tube socket for carbon. Put a used set in and try to get it to run without blowing. The original problem was…..Beer spilled in the amp and blew an output tube.  Any thoughts?

 

[mjrippe]

At this point I’ll take a close look at the tube socket for carbon. Put a used set in and try to get it to run without blowing. The original problem was…..Beer spilled in the amp and blew an output tube.  Any thoughts?

Definitely check the sockets.  Also, wash everything down with isopropyl alcohol.  Amps tend to like that better than IPAs 

 

[CJ]

the resistors are smoking because when you crank there is a large ac voltage across the resistor.

still, it should not smoke,

what kind of transformers, dagnall?    maybe the output xfmr is creating bad mojo due to something going on inside the windings,

check for hi freq osc which might be cooking your resistors and making the tubes short out.

 

[walter]

Thanks for the suggestions! Now I’m researching how to check for HF oscillations in amplifiers. I don’t have a spectrum analyzer. I do have an A.M. radio and an o-scope. I did notice a 1.2Ω resistor on the ceter tap, looks original, (can be seen in the picture I attached before), and the grid stoppers are 5.6kΩ. Components that suggest this design had stability issues.
I found a lengthy discussion on this at the diyaudio site. Some suggestions are: add smaller caps in parallel to the filter caps. Increase the negative feedback resistor. 100pF 1kv caps on the output, I have seen this on Fender amps. I’ll also try to separate the input and output wire routing, although I would like to identify the problem, so I can confirm a fix before imposing random mojo.
Here is a picture of the output transformer, looks original.

 

[Tubetec]

Hmmmm Mk 2 thats old , as previously said Isopropyl and an old tooth brush clean and re-tension sockets, often carbon deposits exist after arching.
Pay attention to the bias supply caps ,and power supply electros ,unless they have been replaced Id considder doing it as a matter of course in an amp of this age .Its easy enough to rejig the values in the bias circuit if more neg volts are required to keep the 6550's happy.  Do suspect high frequency osscillation also , there are a number of small adjustments in old marshalls that help in this regard, lifting the grid wires up and away from the other wires is a good one ,have them come directly down from the socket and across to the tag board . Maybe the feedback network is unhappy  with the new output tubes ,try it disconnected maybe ,although you will have a lot more gain .

There is of course an oficial marshall change to 6550 for customers in the USa ,its only a handfull of component differences in any case ,but worth a look in case something was missed in the changeover.Just had a look ,the screen grids are 1k in either EL34 or  6550 100w ,what is different is the grid resistor on the 6550's ,82kohms being the marked value (220k-EL34),the series resistor on the bias winding is also reduced to 15k from 27k to give a bit more bias volts ,also feedback is taken from the 4 ohm winding on the amps with 6550 ,where normally with EL34's its off the 8 ohm tap. Hope you get it sorted soon .

 

[CJ]

cool that's a Drake transformer,

you can try a couple of .01 cap from plate to cathode,  make it high voltage, like 3 KV, ceramic,

 

[walter]

I get on this a little at a time, last night my 5yo son was interested, but then he started to balance on my tool case-kid standing on an upright suitcase behind dad in a powered 100W amp hanging off the edge of the workbench. I shut it down and we watched Yellow Submarine.
I did identify high frequency on the screen grid resistors, see picture. The HF remains with dummy load, so it is not from an interaction with the speaker. HF remains with negative feedback disconnected, so it is not from phase shift-positive feedback.
This amp is dated 1980 and is similar to the model 2203 schematic that I found on-line. Work was done in 1995 including a cap job, and it looks like the grounding sceme was changed. There is a ground rod near the pots, all the grounds have been moved from the pots to this thick wire. I suspect the HF is related to a grounding issue. The screen grid resistors are 1k, the grid resistors are 5.6k, resistors from bias supply are 150k, and the series resistor on the bias supply is 15k, The negative feedback is on the 4Ω tap as far as I can tell. It was almost impossible to measure the taps while still connected and the wire colors are not consistent throughout the years.
I do not suspect the filter caps, but I won’t rule them out. Next time I get some time, I will take a chopstick to the wires and see if moving a wire effects the HF. If not I’ll look at the grounds around the power supply, after that I’ll try to suppress the HF with capacitors across the output. I’ll look around for a spare output transformer to try…

 

[Tubetec]

Hmmm not affected by removal of feedback ? the level of oscillation  remains unchanged even ? ,

In the absence of signal input does all measure well with your amp ie no oscillations? as soon as signal is driven ,does it turn to mush ?

If its like that ,suspect your load on the cathode follower pre tone stage ,or the resistors around the phase splitter cathode/grids
Dc voltages can read good on a cathode follower sometimes yet resistors are multiples of their value ,its an easy one to overlook  and Ive seen it on a few Marshalls .

If those big psu caps are Daly in a translucent blue shrink wrap there original ,by 95 the Daly 50+50 was in light blue opaque wrap , look for buldges and cracks around the vents on the caps especially the ones closest to the rectifier ,strongly suspect your bias caps ,try a single pair of tubes on the output for a small time just to illiminate a bum tube in the batch maybe .

 

[Tubetec]

just had quick look back at the photo again , the output tube on the left ,do  I see a black deposit between the pins , close to the screen?
Washing with iso may not get rid of carbon thats fused with the phenolic ,scrape the areas affected  down with a blade or small file,'DISCHARGE CAPS FIRST',  I have used clear nail polish to seal up the surface of the socket where you hit it with the file , just so less chance of mildew or moisture ingress later.

Check sockets topside too as arc can deposit there as well.

 

[walter]

Mystery solved. It turns out that what I was seeing on the o-scope was caused by a poor connection of the probe to the scope. After correcting the connection, I determined that my guitar is the source of the HF. I see it at the input with the offending guitar plugged in, I can also see a snivet in the output signal. “ Snivets are thin parallel vertical lines which may appear at one or more positions of the reproduced image.” I used a different guitar and all is well. I did find an interesting website that describes how to analyze negative feedback loops. I’ll may start a discussion on this in “the Chamber” as I find it very interesting and a bit over my head. http://www.angelfire.com/electronic/funwithtubes/Amp-Compensation.html

 

[CJ]

do you have active pickups?

 

[walter]

Humbucker. and something? I made this guitar in highschool woodshop 30+ years ago. The guts are hanging, so that doesn't help. The o-scope shots show the signal at the input jack, 3 different resolutions. The guitar still needs some work, but it rips for punkrock. The fender with single coil pickups is more appropriate for amp repair testing.