1972 Marshall JMP 50 Watt Evaluation and possible Modifications

[Platefire]

What? Don't anybody have any idea of the cause from variation/drifting in voltage from 445 to 437 in plate voltage? Even after I tried new tubes the drifting voltage continued. Is the original rectifier failing or could failing filter caps cause the drifting---the amp is pretty quiet. I tired it without the power tubes in and the voltage drifted from 468 to 460.

Also the great divided in mA of biasing even with new matched set(8mA dif)? Is it something in the amp causing the divide in the amp? or is all in the tubes? It's about 1.3 Ohms difference in OT CT resistance to plates (43.9/45.2) but I don't think that would make that much difference.

I am wanting to get the bottom of possible issues with this amp before I return it to the customer so he may have some trouble free operation. I have no explanation from my experience for the amp behavior and I was hoping some of you more experienced hands would chime in regarding these issue above.  Platefire

[sluckey]

What? Don't anybody have any idea of the cause from variation/drifting in voltage from 445 to 437 in plate voltage?

I don't consider that a drift. Certainly not a problem. The power supply is not regulated. Even if it was regulated, consider this... Suppose the voltage was supposed to be 440VDC. Your voltage only increased by 1.12% and only decreased by 0.68%. That's would be considered pretty good regulation.

Quit climbing the mole hill! Get a bag of 1Ω resistors so you can get some consistent and reliable bias current measurements. Once you try it and see how easy it is you'll never use that klutzy OT voltage drop method again. IMO, a service tech needs to consider these 1Ω resistors as standard bench items, just like solder, or a meter.

[Platefire]

OK, thanks! Good to know the voltage changes is not a problem.

Honestly I have avoided the 1 Ohm resistors because I was avoiding any additional mods. I have the 1 Ohms bias resistors on hand. I did contact the customer and he said yeah. I won't drill holes for outside contacts and ground but just remove the existing cathode ground wire and replace it with the 1 Ohms.

It may only confirm my previous readings regarding un-matched tubes, but I will then know what they are for sure and it will then be set for easy bias readings from now on. Platefire

BTW-Just for basic information, this type of MV with bias running through the pots does effect the bias voltage 1 negative volt from off to full up. Actual readings are: Off=-40.92, 50%=-40.70 and 100%=-39.93

[Platefire]

I installed the 1 Ohm bias resistors. Results is it pretty much confirmed what I had already determined by the other methods.

With his tubes last reading after good long warm up: V4 Plate 439.5V/41.8mA,  V5 Plate 440.1V/24.5mA,  -41.06 Bias voltage (These tubes are 17.3mA out of balance)

Using my tubes last reading was: V4 Plate 444.4V/33.4mA,  V5 Plate 444.6V/43.4mA, -40.35 bias voltage (These tubes are 10mA out of balance)

Platefire

[sluckey]

Very good. That's easy and safe. So, what are you gonna do about his tubes?

[Willabe]

Did you flip/flop the tubes? Left tube in right tube socket, right tube in left tube socket? Then take readings again. You have a 50/50 chance they will be closer in mA's. 

[Platefire]

He is going to bring by some more tube sets for testing. I offered to sell him my new set but it even checked 10mA off. I have tried swapping and it may have helped a hair but the divide pretty much followed the tube. Platefire

[sluckey]

I see a possible match...

his V4 (Plate 439.5V/41.8mA) and your V5 (Plate 444.6V/43.4mA).

[Platefire]

Well you have to realize, I wasn't using the same bias voltage on his set as mine. I was biasing each set to get the mA about 2 to 3 mA above the 39.5mA setting for the plate voltage in order to bring the low side up as high as possible without getting too far above safe on the high side. I would say probably just a volt and a half higher his than mine. That is probably enough to be 7.3mA dif than your recommended match at 1.6mA dif.

So I did try your recommended match. It came in at V4=438.4/42mA   V5=439.1/34.7mA @-40.93 bias voltage

Anyway, this is the closest match yet with what tubes I've got on hand. Platefire

[sluckey]

We all probably obsess a bit too much about perfect current matching. I know I did a few years back. I've relaxed quite a bit since I installed a dual bias circuit in my Sunn just to get a 'perfect' match. But what does that really mean? And who's to say how well that perfect match will be a month from now after the tubes have had some time to settle in?

The last fixed bias amp I worked on was my AO-63 conversion. The "matched set" of 7591s were running at 24.4mA and 18.2mA with plates at 470V. I called that good enough.

Don't let this matched tubes craze get to you. If you do, then prepare to keep a bucketful of tubes on hand. Just get it close. Then let your ears tell you when it's done.

[Platefire]

sluckey, I hear you on not being obsessed by a perfect match. Hardly any of my PP amps are any closer than 5 mA. Believe me it's the customer that's obsessed in this case. If it sounds good and not to far out of balance, it's good with me. I'm trying to be patient until this job is finished but I'm way ready to be done with this
Platefire

[Platefire]

Well the customer brought me several sets of EL34's and I tested 4 sets and picked the closest match(5mA dif) and biased them. So this is done. I also made an announcement that I won't be repairing other folks amps anymore. If I talk to myself just right, I might still repair my own Platefire

[HotBluePlates]

sluckey, I hear you on not being obsessed by a perfect match. Hardly any of my PP amps are any closer than 5 mA. Believe me it's the customer that's obsessed in this case. If it sounds good and not to far out of balance, it's good with me. ...

RDH4 addressed this topic, at least with regards to output triodes (and then noting the results with pentodes were similar).

An push-pull output stage was investigated which had two different tube types in the 2 sockets, with one triode having half the transconductance of the other side. This should equate to a 50% mismatch in 2 tubes of the same type, or maybe one new and one "dead" tube out of a pair of the same type.

The end result was ~5% THD (and IIRC, slightly lowered output power). The conclusion was tight matching was not critical unless the maximum clean output power is required.

[Platefire]

Thanks HBP

What I've been doing and don't know if it's correct but at the same time I haven't had any major burn outs yet, is in balancing the mismatch. Just for instance say two EL34's running plate voltage of 420V and your goal is to bias at 41mA. When you bias your hottest tube up to 41mA and lower one is running 32mA.

What I do is run the high side up a couple of mA above the safe zone to bring it up to 43mA that brings the lower up to around 34mA. I don't know if that good practice or not?

How do you handle adjusting the mismatch? Platefire 

[Platefire]

I did go read RDH4. It did say not over bias more than the tube recommended dispensation. So if you're already at a pretty hot 70%, I guess I shouldn't raise the high side above that in order get the low side up. According to them the divide is not a big deal in the overall sound other than being a little less clean. So I may discontinue my balancing method mentioned in previous post

So I would say within 5mA apart is OK and to shoot not to be more than 10mA if possible. Platefire

[HotBluePlates]

In the end, it's all a judgment call.

If you bias a bit over 70%, but play the amp full-tilt and see no redplating (may take a darkened room to see the onset), then you're fine.

I just re-looked the triode and pentode comments on mismatching in RDH4. It did note that "the effects of mismatching are more serious in pentodes than triodes". But if you know first that push-pull operation with perfectly balanced tubes, circuit & OT cancels all even harmonic distortion generated in the output stage, and that RDH4 is written from a perspective of engineering to high-fidelity standards, well... Then mismatched output tubes means you'll likely get more distortion in your output section.

Severe mismatching (60mA in one tube, 10mA in the other) could temporarily erode the OT's ability to transfer power from the primary to the secondary. That's because the mismatched idle and presumably signal) currents will tend to magnetize one half of the primary more than the other, and looks like d.c. through the OT in the direction of the higher-current tube. Push-pull OT's use smaller cores for the same power through-put than single-ended amps precisely because equally-balanced idle currents looks like 0mA d.c. to the transformer. The push-pull OT core then is only big enough to transfer the full power of the output stage, rather than full power plus some extra core material to support the magnetization due to idle direct current.

So even severe mismatching idle currents might only lead to OT core saturation, and another form of distortion (as well as some shaving of bass response at max power).

Mismatching could be a problem if it's bas enough to let hum through the OT to the speaker. I don't know if you've had a tube try to burn up while you were biasing, but other than the red plate, what usually happens is you hear hum in the speaker as the tube current is getting really out-of-hand (I heard it once when a tube lost bias, right about the point it hit 110mA or so). Hum at the 1st filter cap is usually cancelled by the balanced nature of the push-pull OT and because hum from the B+ is applied common-mode to an OT which transfers differential power from primary to secondary, but greatly imbalanced tube currents seem to overcome this somehow.

Assuming the amp's power supply/heaters would support it, your customer could put a KT88 on one side and an EL84 on the other side (keeping each tube biased within its limits). The amp wouldn't be able to develop clean output power in excess of using a pair of EL84's, but neither the amp nor tubes would suffer as a result.

[darryl]

BTW-Just for basic information, this type of MV with bias running through the pots does effect the bias voltage 1 negative volt from off to full up. Actual readings are: Off=-40.92, 50%=-40.70 and 100%=-39.93

The bias circuit you are working with has a relatively high impedance, so the loading caused by your multimeter may be sufficient to create the difference you're measuring.

I've just tested the Lar-Mar master volume in one of my amplifiers, and that is exactly what happened. When the meter was used to measure bias voltage on one side of the push-pull pair, current through the matching 1Ω sense resistor rose slightly. ( I did use two multimeters for simultaneous measurement... )

[Platefire]

Well bias is something I seem to get hung up on. I would like to get to a place where it's easy as falling off a log. Seems the more I learn, the more I need to learn. On one hand it seems simple and the other, a deep dark hole. I know I don't like cold or cool biased sound. So the trick it get up to the edge of the abiss with your slightly out of balance tubes without sliding off the cliff.  Platefire

[Platefire]

Well the customer has got it and reports it sounds great. Maybe I should send him a note to keep an eye on the plates.