Marshall JCM900 bias issue

[Randall]

So where does this leave me with regard to what to do with this amp? Do I mod the bias circuit as some say, or leave it be and buy tubes until I find a pair that are compatible? Perhaps I will replace C15 and see what result it has.

[nsubulysses]

Can someone please explain how C15 could be possibly be thought to be"on it's way out" or causing an issue? AC comes from the transformer to C15 and then is rectified by D1. so the only way I see it that C15 could fail (it's an X cap, made to fail open, not short, right?) is if the Xcap goes open we have no bias supply and the tubes redplate badly. The issue here is we have plenty of bias, so much that the tubes are biased very cold. Are you suggesting that the capacitance of the Xcap has changed and it is affecting the bias supply voltage and shifting it more negative? I would tend to believe the dude and g1 more than myself but in this case I dont' understand.

If you replace C15 I would recommend using an Xcap or just check first to see how much AC is on the cap. Most caps that are 600VDC rated have an AC rating of 200V max. X caps are usually rated at 400VAC or 600VAC max, or more.

[g1]

My concern is the vast difference between a 56K or a 13K feeding the bias diode. It seems to me more than a 'tweak'.
C15 is basically dropping some AC voltage depending on it's impedance (which is determined by it's value). R7 is the other part of the voltage divider. If you have to drop R7 to 1/4 it's value, it means way too much AC is getting through C14. This could mean C14 is leaky, and even worse would be if it is defective and ends up open resulting in no bias. I would at least try subbing a .047 600V to test if there is any change in bias voltage levels (it's 47n not pf). If so, replace it. Class X is a special safety rated cap.
If no change in bias voltage, you can keep the original cap.
In this case I would tweak R28 and R29, not R7. That is what they changed for EL34 vs 6L6. With 6L6 R28 is 68K and R29 is 8K2. With EL34 R28 is 56K and R29 is 15K. For your requirement you would increase R29 and decrease R28.

(edit: nsub, yes class X is supposed to fail open but I'm not sure they can't go leaky first)

[Randall]

Thank you for asking this question^^^, because I don't understand either. The cap that is in question is 0.047/250v X2. I understand that if it goes open the tubes will lose all bias voltage, but how does this effect it now having too much bias voltage?

[g1]

Randall, I hope post #18 answered your question. I'm not sure C15 is your culprit but I believe something in the circuit is.
The adjustment you had to make to R7 just seems too great for a properly working circuit.

[Randall]

Well, perhaps we are getting somewhere. I pulled C15, and it measures 66nF, so there's that. And it is leaking a little. At 120 vdc it leaks a few volts until after about 20 seconds it settles down to 0.5vdc. Not much, but a replacement 0.047/630v mylar (I think) cap settles down to 0.09vdc under the same test. Is this a small enough amount to cause concern? I will change it anyway since the value is off by 40%.

Running it with the substitute cap gets me down to -41v, which is an improvement, but still these tubes will only bias to 24mA at about 10-11 watts.

Now where and how to find a 0.047/250v X2 cap that will fit? And once again, do I try to tweak this circuit or write it off as a cold set of EL34s?

[nsubulysses]

I use these http://www.mouser.com/new/Kemet-Elec...emet-900-caps/

When you replaced the x cap with another one did the bias change at all?

[Enzo]

SO it is biased cold, how does it sound now? Did you hear the problem the customer described, and has that been corrected?

If my bias adjust range was from -60 to -45v and it didn't get hot enough, then I'd probably change a part of two and extend the range from -60 to -40 or even -35. Then you still can adjust it to the original tubes but now have the range to suit other tubes out of the former range.

[Randall]

nsub, yes the bias improved. The voltage dropped 4 volts, and the dissipation raised by about 8mA. And thanks for Mouser tip, but that series only goes to 10,000pf, I need 47,000pf. I think I found what I need here:

http://www.mouser.com/ProductDetail/...Fc1ZhgodGZMFuA

Enzo, the original complaint was due to it having a non-working tube. One of them lit up, but did nothing else. Replacing them led me to this bias issue. How does it sound now? Like a cold biased Marshall, and I'd like to get it to were I like to put them - 60-65%. To my ears that's a good Marshall sound and a safe place for the tubes.

[The Dude]

...... class X is supposed to fail open but I'm not sure they can't go leaky first......

My experience is that is exactly what happens, which is why I commented on checking the cap. I should have probably given more detail. They usually start out leaky and eventually fail open. Often the amp will start out working for a few minutes and then the cap stops "capping" and bias is lost. Red plate and blown fuse to follow.

[nsubulysses]

Thanks for explaning dude and g1. This is why I know to tend to trust you more than myself.

[g1]

They usually start out leaky and eventually fail open. Often the amp will start out working for a few minutes and then the cap stops "capping" and bias is lost. Red plate and blown fuse to follow.

Thanks for clearing that up. Sometimes that is not the end, tin-foiled fuses and cooked transformers may follow.
Randall, if the bias trimpot and R28 are the correct 47K and R29 checks out at 8K2, then I would suggest you put the stock value at R7 and reduce R28 to whatever gives you an acceptable bias range.

For anyone interested in the math of my objection to that R7 change, I'll try the following explanation.
C15 an R7 form a voltage divider for the AC feed to the bias rectifier diode. I'll ignore the rest of the bias circuit for simplicity. The impedance of .047 cap at 60hz is about 56K [ XC = 1 / (2 × π × f × C) ]. So you have 56K in series with the AC source (bias winding) and 56K to ground (R7). This means half the source voltage is dropped across C15. For example, if source is 100VAC, we end up with 50VAC at the top of R7. If we put 17K in parallel with R7 we change it to about 13K. Now instead of 50V at the top of R7, we only have 20V there, less than half of the stock voltage. This is just too big of a change to be a minor circuit variation.
The only plausible explanation I can think of (aside from a circuit fault) is that the bias winding is way out of spec, and this unit has been wrong and biased extremely cold since it was made.

[The Dude]

Tin foil fuse at your own risk.

Possibly too late if the cap was already ordered, but FWIW, I usually replace with a higher voltage rated cap for safety sake. Later schematics called for a 630V cap (I believe/if memory serves).

[Randall]

G1, you cite R specs for 6L6's I believe, as taken from previous post. This amp runs EL34's. They are actually 15k and 56K, and I have the original R7 in place. I had a math party today and came up with values of 22K instead of 15K, and 47K instead of 56K, but it ends up going a little farther in the direction that I was shooting for at -35v at the extreme with about 10-11v range on the trim pot. Turns out -49V is what this cold pair wants to see to put them at 33-34mA, which currently is at about 1/3 on the pot. I'd like to back off a little. I wonder which R I should adjust now? I'm leaning toward R28, the one that is now 47K. I first changed R29 from 15K to 22K, and it only made a 1.8v difference. When I went to 47K from 56K on R28, it made more of a difference.

Dude, yeah C15 is already ordered, but it did come in as a 275v, where the original is 250v. It'll have to do.

[Randall]

So I changed R29 to 27K and it was right on the edge, so I changed R28 to 52K and now it gets me where I want to go. However, when I set the bias to 34mA, it was fluctuating wildly on both tubes. I notice that the HT fuse is a link from ground to both cathodes, so all current must pass thru that fuse holder. Sure enough, I took a close look and it was a bit oxidized (like most else down here in SW Florida). A DeOxit and a swab cleared that up, and I put a drop on the trim pot to wick inside, but I'm not sure it was that because I monitored the bias voltage and didn't see any correlation when the current was moving. Now it sits pretty solid.

So, I guess now it a wrap. I enjoyed this one, and I learned something. So once again, thanks guys, you were very helpful as usual.