Hey everyone,
I've recently completed my first Mosfet-Max, and as I turned it on for the first time (all trim pots except RB12 set to max. resistance, RB12 set to min. resistance on both sides, I hope that is correct?), I noticed the voltage the power supply produced started out at 15V, slowly climbed up to 20V, stayed there for a few seconds and then went back down. I also noticed that the left QB9 was turning very hot and so, after turning it off and on again (everything repeated itself as I wrote above, I tried a couple of times) measured the Mosfet bias. It was waaaaay too high at >2000mV and I quickly turned it off again. The right QB9 doesn't have any of those problems. I haven't dared to turn it on again since then, does anyone have an idea what the problem could be? Should I get some new FETs for that position?
 
EDIT: OK, so I figured that one out myself. The trim pots were turned in the wrong direction, I'm still kinda confused about which direction increases and which decreases the resistance on which trim pot, but that's not much of an issue. However, I think I might have broken something (I was trying to measure the bias voltage on the right tube with the DMM still in high-current-measuring mode and accidentally shorted the measurement point with ground, incredibly dumb mistake I know), because it's become impossible to bias the right tube at a sensible voltage; It's constantly at 20+V and the right DB bias stays at 0, no matter what I do. Also, the right QB9 stays completely cold, whereas the left one warms up a bit. Might it be broken? I'll try to replace it tomorrow (still have some left for a second build I'm planning, and I'll have to order a ton of stuff for that anyway as I broke some other stuff and didn't order enough of yet some other things.), but if anyone could confirm that suspicion or maybe hint at something else that's likely the problem, I'd be very grateful.
 
EDIT: Forgot to say this, the left side can be biased without any problems, I have the tube running at a perfect 13.5 and the DB at 250mV. It's just the right side that's making trouble.
 
Edit: I replaced QB9R, everything's still the same. I'd really appreciate any help as I'm at a loss right now.

How about some pictures of the top and especially bottom of the board?

Lots of light and as much focus as you can muster.

I'll go take some right now, my cam is decent so hopefully you'll be able to see something. There are some contacts that look really bad because I only had a very, very (I mean REALLY) crappy soldering iron and some weird, presumably lead-free, solder at the beginning. And then there are some brownish contacts, that's because I had to get solder out of there with desoldering braid when I replaced some parts, but they should conduct fine, the solder flows nicely when heated. I've also checked all of the resistors on the right side, they all seemed fine (good contacts and correct resistance values)
 
OK, here you go:



Left side- everything normal

Right side, with 22-24V tube bias voltage.
 
I'm guessing one of the small TO-92 transistors on the right side is blown, but how would I figure out which one? I really don't want to desolder them all and replace them since I don't have a desoldering gun and doing it with desoldering braid is kinda messy. It also seems that this guy http://www.head-fi.org/t/566846/millet-max-problem had the same problem, fixing it by just letting it sit for a while, but I'm kinda hesitant to do that since 10V too much can't be good for those tubes for a longer time...

Sorry I didn't see this sooner.  Yep, it's a good thing you got on top of those RB12 trimmers.  You see, minimum resistance means you have maximum current going through the buffer.  The controlling current loop for the buffer is formed around the 2N5486 FET and the RB12 trimmer.  RB12 controls the current through the loop by setting its resistance.  If you set RB12 at a minimum, then you have maximum current through the loop and you will blow up the buffer if it's left on long enough.  Thank goodness you kept your head and continued to make measurements and realized the situation.
 
As for the tube bias, you are lucky.  The reason being is that you have one tube biased, so all you really have to do is compare the voltages across the resistors from one channel to the other to find out what's wrong.
 
That said, it may still be that you haven't worked the tube trimmer enough.  Because the amp is designed to take three different tubes of varying gain, the tube trimmers are a compromise.  Some tubes/tube types may end up adjusting in 2-3 turns somewhere within the 20-25 turns of the trimmer.  The problem is finding out where those 2-3 turns are.  They could be at the end, the beginning, or somewhere in the middle - and the tube may give no indication until you hit that sweet spot.  If you're satisfied that you've racked the trimmer in both directions and see no change whatsoever in the tube, then continue on with measuring the voltage across the resistors and comparing them from the bad side to the good side.  It should become clear to you if something is out of place.
 
BTW - are you sure the tube is lit?  Look for an orange glow in the very middle at the top - that will be the heater.  It's sometimes hard to see through the LEDs, but you should be able to recognize whether the tube is lit or not - just a thought.

Hi tomb,
thanks very much for your answer. The tubes definitely both glow and get warm. Also, I've turned the trim pot all the way in both directions (I checked by measuring the resistance across it) and it doesn't make any difference whatsoever that I could make out. And I've already compared some voltages across resistors on both sides, they're pretty much all different, and I'm not quite sure what to do with that information... How would I figure out which part (as I said, I'm assuming it's a TO-92 transistor) is busted?

Quote:

hitman47 said:

Hi tomb,
thanks very much for your answer. The tubes definitely both glow and get warm. Also, I've turned the trim pot all the way in both directions (I checked by measuring the resistance across it) and it doesn't make any difference whatsoever that I could make out. And I've already compared some voltages across resistors on both sides, they're pretty much all different, and I'm not quite sure what to do with that information... How would I figure out which part (as I said, I'm assuming it's a TO-92 transistor) is busted?

Click to expand...

Well, the TO-92 transistors are difficult to measure once they're in the PCB.  That's why I suggested the resistors.  There's a resistor connected to every one of the transistors somewhere. 
 
Anyway - make a list of the resistors in two columns - one column for the R channel and one for the L channel.  Carefully but methodically measure the voltage across the leads of each resistor and write them down.  You should be able to see the differences.  If the buffer is OK in both channels, then just focus on the resistors in the tube section.  Measuring RA8 and RA9 may be particularly interesting.
 
 

OK, I'm doing that now. I might still need some help with the interpretation of the results though
 
Edit: OK, I'm done. These are my results:
 

 
However, I'm still not quite sure what to make of them. Is it one of the transistors between RB2 and RB3? The discrepancy between RB2R and RB2L sticks out most to me, but I don't really understand enough of the circuit to judge it...

So, as far as the pics go, it would be far better if you got closer with better focus. The distance and lack of focus makes it real hard to check the simple stuff, like orientation of the transistors, unsoldered joints, etc.

As it stands now, you have all of the adjustment pots oriented backwards. This should not be an electrical problem, but, might present a problem if you use the casing template for the top panel as the adjustment holes will be in the wrong place. I would not change anything, but, realize the adjustments will be backwards of what you would expect.

Have you verified the transistors on the problem side are all correct in part number and orientation?

hitman47 said:

However, I'm still not quite sure what to make of them. Is it one of the transistors between RB2 and RB3? The discrepancy between RB2R and RB2L sticks out most to me, but I don't really understand enough of the circuit to judge it...

Click to expand...

I would verify the part number, orientation and solder job on QB4R and QB6R.

 
Quote:

hitman47 said:

OK, I'm doing that now. I might still need some help with the interpretation of the results though

 
Edit: OK, I'm done. These are my results:
 

 
However, I'm still not quite sure what to make of them. Is it one of the transistors between RB2 and RB3? The discrepancy between RB2R and RB2L sticks out most to me, but I don't really understand enough of the circuit to judge it...

Click to expand...

There are way too many resistors with zero as your measurement.  That's not possible if the the power is on.  Go back and turn the sensitivity of your meter up - measure in mV if you need to.  After that, turn the power off and measure all the same resistors for ohms.  There are very few resistors in parallel with anything, so you should get fairly consistent results.
 
As for RB2L and RB2R, check to make sure that they're both 10R.
 
While you're at it ... since you can't bias the Right channel tube, verify that the trimmer (RA1R) is working and is not shot.  Measure ohms (with the power off) of the one pin connected to a trace - underneath the PCB - with one of the other pins without a trace (two of the pins are connected to ground).  Turn the screw about 10 times and measure again.  If no change in ohms, turn the trimmer 10 turns in the other direction.  If you still can't measure any change, then the trimmer is shot and should be replaced.
 
EDIT:  Looks like you said you already checked the trimmer, so never mind that last paragraph if that is true.
 

OK, I re-measured the voltages as you said, here are the results:

All numbers are in volts unless otherwise specified. I also seem to have measured RA8 and RB2R wrong the first time, and I forgot RB12-14 for some reason.
 
Here's another pic of the top of the board, not sure if this helps you jdkJake.. I know the orientation of the transistors is correct, and I can't really make out part numbers on most of them but I managed to check QB4R and QB6R and they're both correct and the solder jobs on all of them look OK to me as well.

 
Thank you very much for your help, I really hope I can fix this... Everything already looks so pretty with the LEDs

Yes, that helps VERY much! As they say, a good picture is worth a thousand words.
 
So, are you using 2SJ74/2SK170 JFETs? If so, then unfortunately, you did not see this page on the build site:
 
   http://www.diyforums.org/MOSFET-MAX/MOSFET-MAXmosfets.php
 
Most importantly, this image:
 

 
If you are using the 2SJ74/2SK170 JFETs, then your orientations on QB2 and QB3 are reversed. Yes, I know the artworks says otherwise, but, the correct orientation for those parts is as detailed above. While this may not be the only issue, this does need to get fixed provided you are using those parts. You are also missing your compensation caps for the MOSFETs.
 
Can you confirm the JFET parts you are using?
 
I will keep looking. Nice focus on this picture. Can you provide a shot of the backside as well?
 
 

Several things here -
 
1. If you used the recommended Toshiba JFETs, they have to be installed backwards from the PCB silkscreen - that's QB2 and QB3 in both channels.  If you simply used the 2N5087/2N5088's in those positions, then it's OK.
 
2. There's something very wrong with RB3R - the voltage drop across the resistor is way too high.  Check to make sure it's 10R and not something larger.  If it's not 10R, then it got switched with another and you'll have two in the wrong place.
 
3. You have no voltage across the RB10 and RB11 resistors on the Right channel.  I thought you had biased both channels in the buffer.  This says no.  It could be because QB8R and QB9R are blown, but hopefully not.
 
4. From the pic, it looks like some of the transistors have some lack of solder on the ground connections.  You may want to try re-flowing the solder joints across the PCB, especially the resistors and transistors. 
 
EDIT:  So many distractions tonight ... I see that jdkjake posted while I was still working on this one and ended up posting it later.  Item #1 says we think alike.

Wow, I really didn't see that those two have to be reversed, I did that just now but it didn't help. also, I was never able to get any kind of bias voltage on the right buffer ever since I fixed the first issue, I'm starting to think that I've blown the right mosfets. Also, about the compensation caps, I somehow forgot to order those and will do so when I order the other parts I'm still missing for the second build. Anyway, should I replace the mosfets and see if that helps? And if so, should I replace the ones on the left side too (I have another matched pair) or do they only have to be matched on the same channel? Again, thank you very much for your help, I don't think I'd ever have come this far without you guys

EDIT: Ok, so I figured out what the problem with the buffer was. I somehow managed to leave an entire mosfet completely unsoldered (happened when I replaced it to see if that was the problem) and that of course messed with things pretty badly. I can now bias the right buffer OK, but I still get the abnormally high voltages for the tube bias. And as I said, QB2 and 3 are now facing the proper side on both channels.
Also, here's a pic of the back side of the PCB. There's one bridge which I can't get away because I accidentally scraped some of the PCB's plastic covering off, exposing the trace, but it's between two contacts that are connected anyway (via the trace I exposed) so it doesn't matter.

Okay. What is abnormally high? Can you repost your readings now that you have soldered in the MOSFETs correctly and biased the buffers?
 
Also, a couple of construction tips.
 
1) Go buy yourself a decent pair of diagonal cutters (diagonal cutting pliers). You need to trim the excess leads from the bottom of the board once the parts are solder on. I like the small versions for PCB work, but, YMMV. As it stands now, some of those leads *might* be shorting hazards, especially when you case this baby up. Trim them close, but do not cut into the solder joint.
 
2) Go buy yourself some 91% alcohol (drugstores, grocery stores, walmart, etc...) and a nylon toothbrush and wash/gently scrub off some of that flux residue from the back side of the board.
 
Once finished, measure it, post the measurements and post another picture of that sparkling clean, neatly trimmed back panel.