Relays And Popping

[Paul Marossy]

My first guess would be collateral damage from the pot changeout.

I've changed the master pot before without incident. All I did was made it the original value again (from my 250K to the factory 100K). I checked all the solder joints, checked for continuity etc. before putting the PCB back in. If it were the master pot, then I expect that the drive channel only would be having a problem since the master has no affect on anything until the drive channel is engaged.

Obviously, there shouldn't be DC on those pots, so something is definitely awry. If there was an internal failure of the relay coil, couldn't that leak DC into the audio path since it's a DC operated coil? (I have no idea LOL) It still switches, and there is still a pop that you can hear in the reverb circuit but the signal level is almost non-existent.

I'll check to see if maybe the preamp tube on the input may be the cause of some of this when I get home from work tonight. But I am still feeling that the relay coil just failed somehow. I've had relay coils fail on me in other things, one time it was in a dishwasher. Somehow the contacts inside get messed up and I was able to take it apart and fix it. In this case, that's not possible since it's so small and it's potted anyway.

EDIT: It's hard to say exactly what is going on since these relays are a little hard to follow. I might have to measure voltages at these test points on the schematic to find the problem. I prefer the shotgun technique however. 

[R.G.]

I've changed the master pot before without incident. All I did was made it the original value again (from my 250K to the factory 100K). I checked all the solder joints, checked for continuity etc. before putting the PCB back in. If it were the master pot, then I expect that the drive channel only would be having a problem since the master has no affect on anything until the drive channel is engaged.

Side effects have a very wide range of applicability. I know this from experience. 

Obviously, there shouldn't be DC on those pots, so something is definitely awry. If there was an internal failure of the relay coil, couldn't that leak DC into the audio path since it's a DC operated coil?

It's possible, since there is both DC and signal inside that one part. However, relays are mostly designed specifically to prevent that. Most relays have insulation between coil and contact in the thousands of volts, for safety reasons in many of their common applications. My gut feeling is that it's more likely to be board contamination, soldering issues, broken/shorting wires or leaking caps than a relay suddenly leaking coil into contact, unless you know of the relay being tinkered with or damaged. But it's possible.

I prefer the shotgun technique however. 

I do too, in general, as it makes a hit more likely. But just like with firearms, the ammunition is more expensive per round with shotguns.

It occurs to me to wonder if anyone has ever made a Gatling-style shotgun.   

[Paul Marossy]

They make full auto assualt shotguns. 

Well, I'll pull it apart again tonight and see if maybe a via on the PCB got screwed up or something.

[Lurco]

You shorted C29. Does the Presence-pot crackle  OK now?

[Paul Marossy]

You shorted C29. Does the Presence-pot crackle  OK now?

I actually made that a switch a while ago. It doesn't affect any part of how it operates in either clean or drive mode except that I can't switch it while the amp is on or it makes a big pop. The presence control makes no scratchy noises when that cap is switched out of the circuit, it just gets defeated.

[Mike Burgundy]

While you're in there be sure and check for cracked solders, I can't even begin to explain how much warranty work I used to have to do on the whole Hot Rod series when I worked in repair and it was almost always cracked solders, mostly on the PCB where the tubes mount, but sometimes other places.  They're great amps, but I'm not exactly sure why Fender thought it was a great idea to mount heat sources under solder joints with no other mechanical reinforcement in a high vibration environment. 

It probably comes from the changeover from lead-bearing to lead-free solders. That was a period of pure hell for manufacturers all over the world.

My experience exactly. Our guitarist has two HR deluxes, which started to get noisy quickly after a recap. I recalled having a bit of a hard time soldering in the new caps - really had to clean the pads with wick, reflow, clean again and only then it would take nicely - so I checked for poor ROHS solder joints - bingo. ALL power tube connections had circular hairline cracks - practically invisible, but responded to the poke-it-with-a-chopstick-test, sometimes with pretty sparks. Yikes. Same solder routine as with the caps solved it.
Be careful though, these PCB's aren't the best quality, and a lot of soldering might lift traces. Take care, and manage your iron's heat (hot enough to flow *quickly* but no hotter) and make damn sure the powercaps are discharged properly.

[Paul Marossy]

I recalled having a bit of a hard time soldering in the new caps - really had to clean the pads with wick, reflow, clean again and only then it would take nicely - so I checked for poor ROHS solder joints - bingo. ALL power tube connections had circular hairline cracks - practically invisible, but responded to the poke-it-with-a-chopstick-test, sometimes with pretty sparks. Yikes. Same solder routine as with the caps solved it.

I've looked for things like that but haven't found anything conclusive yet.

Be careful though, these PCB's aren't the best quality, and a lot of soldering might lift traces. Take care, and manage your iron's heat (hot enough to flow *quickly* but no hotter) and make damn sure the powercaps are discharged properly.

Yes, I am well aware of all that stuff. These are some of the cheapest PCBs I've ever had to work on. I would expect that the tube PCBs would be very sturdy, but they are in fact some of the flimsiest PCBs I have ever had to work with. Way too thin and fragile for the application IMO.

[DavidM]

Try this

http://relays.te.com/kilovac/appnotes/fig48.asp

[Paul Marossy]

So I got my relays yesterday. Switched them out and used some 16-pin IC sockets for them.


And while I was at it, I also socketed the opamps. I dissected one of the old relays just for my own edification to see how it's constructed and it look like it's pretty much not possible for it to fail in such a way that DC could appear in the signal path.

The relay(s) appears to have been part of the problem, as now I have sound out of the normal channel at a normal volume level. It appears that the DC on the pots is gone now (at least I can't hear anything). The drive channel however is still malfunctioning. It's only like 10% as loud as it should be and when I switch out of the drive channel it pops, even when volume controls are all turned down completely. So that tells me that the problem is with the drive channel somewhere. Now I am back to where I was when I started this topic. At this point I am kinda stumped. I've checked the master volume control many times, it works and there is continuity in all the right places. I've checked the B+ voltages, relay coil voltages, the voltages on the ICs, reflowed several solder joints, etc. and I can't find anything that's obviously wrong. I tried it with the footswitch which overrides the onboard switches, nothing. I did a tube dance, can't find a problem there either. So what the heck is it?!?!?!?!

The only active components I haven't monkeyed with are the transistors in the switching section. What if one of those went bad? Would it still switch channels but maybe somehow kill most of the signal on the drive side? It does seem like the gain/more gain relay isn't working as I don't hear it click when I switch it like I do the other relay (but it does change the sound like it's supposed to). I'm thinking it's got to be maybe the 2N4401 or the 2N4403. (or a bad zener diode?) The LED switches as it's supposed to, however. I can't really measure the voltages on the transistors because they are stuffed so far into the PCB that I can't even access the pins with a DMM. I am convinced that Fender wanted to make the Hot Rod DeVille the hardest amp ever to work on. Geez.

Any ideas?

[Paul Marossy]

Well it's not Q3 or Q4, tried something else in their place, same end result. I don't have any J111s to try, so i can't rule them out just yet, although I swapped them with no chsnge in behavior (all active components are in sockets now).

I'm not measuring any voltages on any of the pins other than 1 & 8 on the opamp in the preamp out / power amp in section. I would think that if this is were the problem is that both channels would be affected, but I guess not necessarily?

I'm beginning to wonder if a Zener diode somewhere went bad. Is there an easy way to check those?

[R.G.]

What are the DC voltages on the gate, source, and drain of the J111s in both states, and also the voltages on both sides of CR16?

[Paul Marossy]

What are the DC voltages on the gate, source, and drain of the J111s in both states, and also the voltages on both sides of CR16?

Q1 -- G= -13.45, S=0, D=0
Q2 -- G= -13.6, S=0, D=0

The voltages stay the same whether in normal or drive mode.

CR-16 has -13.7V on the banded end and -13.4V on the other end.

[R.G.]

"More drv" signal not getting to the JFET gates. Check pin 1/u3 for what happens there with both conditions. If pin 1 is changing and JFET gates are not driven, problem is the series resistor between them. If pin 1/u3 is not changing, it's with the opamp or circuit around it.

[Paul Marossy]

"More drv" signal not getting to the JFET gates. Check pin 1/u3 for what happens there with both conditions. If pin 1 is changing and JFET gates are not driven, problem is the series resistor between them. If pin 1/u3 is not changing, it's with the opamp or circuit around it.

Yes, I follow, this would make sense.

I measured 1/U3 in both conditions, it's -15.4 volts either way.
Series resistor R87 measures 100K and it appears to be OK.
Different opamp yields same results.

Looks like I'll have to poke around some more, but this certainly helps to narrow it down!!!!! Thanks RG.   

[R.G.]

Looks like I'll have to poke around some more, but this certainly helps to narrow it down!!!!

At this point, measure the DC volts at test points 31 through 36 for the different switch positions.  You've found that it's not likely to be the opamp bad. That means that the opamp that's in there is not being told to operate. The test point voltages should tell why.

And one suspects that's why they were made check points. The design is a bit too clever, running AC into a footswitch with both positive and negative rectifier diodes and then messing about with what's higher and lower than what else in a mess of diodes and zeners. That's the hardware equivalent of beginners writing FORTRAN - it's almost 100% side effects and bandaids.

[Paul Marossy]

At this point, measure the DC volts at test points 31 through 36 for the different switch positions.  You've found that it's not likely to be the opamp bad. That means that the opamp that's in there is not being told to operate. The test point voltages should tell why.

Ha ha, that's exactly the next thing I did, measured all the test points having to do with U3. To my consternation, I can not find anything that looks like it's way out of spec. Everything is fairly close to what is specified on the schematic. I have just found that 1/U3 does change, but only when the "more drive" option is used, otherwise no change between normal and drive. But the signal level is still only like 10% of what it should be. Guess I'm going to have to look at the PCB for a bad solder joint or a cracked via in that area.

And one suspects that's why they were made check points. The design is a bit too clever, running AC into a footswitch with both positive and negative rectifier diodes and then messing about with what's higher and lower than what else in a mess of diodes and zeners. That's the hardware equivalent of beginners writing FORTRAN - it's almost 100% side effects and bandaids.

I was thinking that this design seems overly complex or something, but I wasn't exactly sure why. It's very hard to follow what is going on in this case.

[R.G.]

I was thinking that this design seems overly complex or something, but I wasn't exactly sure why. It's very hard to follow what is going on in this case.

The primary requirement seems to be stuffing three or four different states onto one signal wire line that goes to the foot switch. Notice it only has signal and ground, but controls channel switch on/off and more drive on/off. It seems to do that by selective loading of both positive and negative sides of the AC voltage that's fed to it, so you get different positive and negative voltage peaks back at the sense circuits made out of dual opamp U3 (ab)used as a dual comparator. The diodes, zeners, transistor switches, etc. are to let the comparators get binary up/down levels out of the analog stuff happening back at the footswitch input.

I'm pretty sure that's what it does. I instinctively shy away from the digging, supposition, error band evaluations and tolerance evaluations that would be needed to ensure that it worked properly. I guess I'm getting lazy as I get old. This situation fairly screams for a pair of $0.35 six-pin PICs with one-wire interfaces to run power and signal down to the footswitch and back up to the panel in a clean way.

Of course, then they probably would have programmed the PICs in FORTRAN. See http://www.eng.uwaterloo.ca/~comp03a/misc/humour/shootfoot.html

[Paul Marossy]

Yes, I did notice what they are doing with the signal path. It gets hard to follow through the relays though.  

OK, so I found something that definitely is not right. I measured the voltage at Pin 8 on V2B, it measures +5.4V instead of the +1.93V shown on the schematic. I think maybe this is where my problem is, but I am not sure why. I did notice that the B+ voltage marked "X" measures 355V vs. the specified 388V. I don't think that would cause my problem though. Pin 3 on V2A measures about what the schematic says. So it looks to me like there is some problem with V2B.

[R.G.]

OK, so I found something that definitely is not right. I measured the voltage at Pin 8 on V2B, it measures +5.4V instead of the +1.93V shown on the schematic. I think maybe this is where my problem is, but I am not sure why. ... So it looks to me like there is some problem with V2B.

Well, it's for sure not right. The plate voltage can vary all over the place, but a 12AX7 with 5+ volts on the cathode and its grid at 0V can't pass signal. I think you found a better lead on the issue than I did.

1. Have you replaced V2 with a known good 12AX7? I have to ask. Swapping with known good tubes is always the first thing to do in a tube circuit.
2. Do the voltages on pins 6, 7, and 8 change when you change the channel and drive switches?
3. What *are* those voltages, anyway? Particularly is pin 7 at ground and does pin 8 vary when you flip the switches?

[PRR]

> a bit too clever, running AC into a footswitch

I think they also have LEDs in the pedal.

27V AC via 820 ohms (2W!) is up-to 33mA available. More than a logic-switch needs. But you could run 20mA into an LED and still have >5V for clean logic.

Not that it uses any 5V logic. Looks like TP32 TP35 clamp for 10V up and 10V down. Hmmmm... 27V-10V in 820r looks a lot like 20mA. (It isn't, because of wave-shape, but WTF.)

I've seen much worse in pedal-jackery. This one starts clean U3B U3A. But then Q4 tacks-on some AND/OR function. MORE_DRV, FET versus relays, and the 3-way LED add more ad-hoc clutter.

Rip it all out, poke a screwdriver in the top to engage frills.

_______________________________

> beginners writing FORTRAN

Or old guys maintaining legacy code.

> screams for a pair of $0.35 six-pin PICs

Plan is dated 1995. While PIC was 20 years old, it had a troubled history and didn't become wildly popular until 1993. Post-Leo Fender has rarely been early-adopter.

OTOH, their new cheap Mustang has a mechanical latching switch, plus a many-patch interface exposed via USB and your laptop.
-------------------------------

> Pin 8 on V2B, it measures +5.4V instead of the +1.93V

12AX7 with 100K+1.5K, the plate will sit near 70% of B+.

If observed B+ at X is 355V, plate should lie near 248V. 355V-248V is 100V, in 100K is 1mA, so cathode should be 1.5V. 1.6V without rounding. 1.75V with nominal 388V supply.

Check V2B Grid. It should be super-close to Zero. Much less than 0.1V for good bias. Much-much-much less than 0.1V for quiet switching. 10mV may be tolerable.