Well darn! I didn't even SEE those--so tiny for my aged eyes. I just measured 'em and every single one is shorted.
Unfortunately, the closest stock part I have is 1N4742A. I'll have to order them.
Unfortunately, the closest stock part I have is 1N4742A. I'll have to order them.
Just ordered a selection of 600 1W zeners, 20 of each type, including this one. It's the only seller on Amazon that it's 2 weeks out for delivery date.
The zener diodes arrived today, so I set up four back to back pairs and installed them in place of the originals. Nearly every diode was shorted. One was open and only one of the 8 was good.
I powered up the unit, waiting two minutes for the delay timer and then measured the voltages at the gates. I was able to get them to balance with positive and negative values this time, but they top out at 0.94 volts. Can't get to 1.2 volts.
Figured I'd try and see if I at least get a cold sounding amplifier and installed the MOSFETs. Powered up, but no signal coming through. Traced back to the 6DJ8s and no signals on any of their plates. Turned out there was no B+. I had B+ at the rectifier, but it turned out yet ANOTHER bad trace under the board, so I jumpered the neg bridge to the ground point on the PCB. Powered up, now have signal right up to the MOSFETs but only a millivolt or so on the speaker terminals.
Eventually found out the MOSFETs have no rail voltages. The fuses were both blown in this channel. Odd, because earlier I checked them and they were good. I put new fuses in and powered up and one fuse promptly blew. Long story short, the SJ55s both read about 8 ohms from gate to source. All of the zeners are shorted again. Replaced them again. Checked the bias with new fuses and no MOSFETs installed. Still at +/- 0.9V. Can adjust down from there, but not up.
I have maybe 20+ hours into this quagmire and almost $600 in parts out of pocket so far. Everything from rotted capacitors, rotted PCB traces, blown MOSFETs, shorted HV winding, bad traces, no grounds in LV and HV power supplies, shorted zeners... and so on. It's been one goddamned thing after another (thinking of Tony Escott's wonderful full page cartoon in the April 1979 issue of OMNI Magazine). My back hurts from leaning over the bench. I'm going to set this aside for a while and think about it when my emotions settle down a bit.
I powered up the unit, waiting two minutes for the delay timer and then measured the voltages at the gates. I was able to get them to balance with positive and negative values this time, but they top out at 0.94 volts. Can't get to 1.2 volts.
Figured I'd try and see if I at least get a cold sounding amplifier and installed the MOSFETs. Powered up, but no signal coming through. Traced back to the 6DJ8s and no signals on any of their plates. Turned out there was no B+. I had B+ at the rectifier, but it turned out yet ANOTHER bad trace under the board, so I jumpered the neg bridge to the ground point on the PCB. Powered up, now have signal right up to the MOSFETs but only a millivolt or so on the speaker terminals.
Eventually found out the MOSFETs have no rail voltages. The fuses were both blown in this channel. Odd, because earlier I checked them and they were good. I put new fuses in and powered up and one fuse promptly blew. Long story short, the SJ55s both read about 8 ohms from gate to source. All of the zeners are shorted again. Replaced them again. Checked the bias with new fuses and no MOSFETs installed. Still at +/- 0.9V. Can adjust down from there, but not up.
I have maybe 20+ hours into this quagmire and almost $600 in parts out of pocket so far. Everything from rotted capacitors, rotted PCB traces, blown MOSFETs, shorted HV winding, bad traces, no grounds in LV and HV power supplies, shorted zeners... and so on. It's been one goddamned thing after another (thinking of Tony Escott's wonderful full page cartoon in the April 1979 issue of OMNI Magazine). My back hurts from leaning over the bench. I'm going to set this aside for a while and think about it when my emotions settle down a bit.
The valve section can be tested completely independently of the output stage. Remove the +/-70V rail fuses and proceed.
The output section can be tested completely independently. Best to start without the MOSFETs installed.
You can disable the relay timer by shorting out R70 or the diode that is connected parallel to it. Waiting 90s is too nerve wracking.
You need to get the bias and offset circuit functioning properly before adding the MOSFETs.
The Bias/Offset cct runs off +/-60V rails that depend on the two 60V zeners (D1/D2 on the schematic) being OK. If you haven't got +/-60V, figure out why.
Then check that Vr2 will swing the centre point - it is way too sensitive in practice.
Then check the bias adjustment - normal operation is +/- 3V to 5V. You are, I think, using different MOSFETs, but you should still be able to get this voltage range before you start adding the new MOSFETs (just remember to set this to a minimum BEFREO you add the new MOSFETs). Likely failure points are Q1 and the potentiometers - not for any particular reason apart from the fact that this amp is now well worked over and its 30+ years old.
If all that works, then set bias to minimum, install MOSFETs.
I assume that you are using a VARIAC to slowly increase the rail voltage instead of just plugging it in and turning it on?
The output section can be tested completely independently. Best to start without the MOSFETs installed.
You can disable the relay timer by shorting out R70 or the diode that is connected parallel to it. Waiting 90s is too nerve wracking.
You need to get the bias and offset circuit functioning properly before adding the MOSFETs.
The Bias/Offset cct runs off +/-60V rails that depend on the two 60V zeners (D1/D2 on the schematic) being OK. If you haven't got +/-60V, figure out why.
Then check that Vr2 will swing the centre point - it is way too sensitive in practice.
Then check the bias adjustment - normal operation is +/- 3V to 5V. You are, I think, using different MOSFETs, but you should still be able to get this voltage range before you start adding the new MOSFETs (just remember to set this to a minimum BEFREO you add the new MOSFETs). Likely failure points are Q1 and the potentiometers - not for any particular reason apart from the fact that this amp is now well worked over and its 30+ years old.
If all that works, then set bias to minimum, install MOSFETs.
I assume that you are using a VARIAC to slowly increase the rail voltage instead of just plugging it in and turning it on?
The valve section has been restored to functional.
The last remaining issue is the bias circuit. I do have rail voltages at R14 and R15, so D60/61 are at least not shorted.
The only component I have not replaced is the stud-mounted device on the heatsink. It is covered in layers of shrink tubing and none of my documentation mentions what it is. The reverse engineered schematic I have indicates an NPN device. Last I checked it with a component tester and it produced the typical "L" shaped curves that normal transistors do. However, it's the only component not yet replaced. The problem is equal in BOTH channels.
The last remaining issue is the bias circuit. I do have rail voltages at R14 and R15, so D60/61 are at least not shorted.
The only component I have not replaced is the stud-mounted device on the heatsink. It is covered in layers of shrink tubing and none of my documentation mentions what it is. The reverse engineered schematic I have indicates an NPN device. Last I checked it with a component tester and it produced the typical "L" shaped curves that normal transistors do. However, it's the only component not yet replaced. The problem is equal in BOTH channels.
That would be a 2N2222. Any NPN will do. I used the PN2222 which was also used for the relay driver transistor - in later SA220s. I insert in a carefully drilled (i.e. tight) hole in the heatsink.
Today I physically removed the bias transistor and tested it on a Tek 576 curve tracer and got beautiful curves. The bias transistor is fine.
I have the appropriate rail voltages at D60/61.
There is no R70/D71 as called out on schematic on this particular board. The parts around the 555 have numbers in the teens.
I have positive and negative .33 volts on either end of the bias pot.
I took some readings and marked in green. This problem is the same in both channels.
I have the appropriate rail voltages at D60/61.
There is no R70/D71 as called out on schematic on this particular board. The parts around the 555 have numbers in the teens.
I have positive and negative .33 volts on either end of the bias pot.
I took some readings and marked in green. This problem is the same in both channels.
Sorry brain fade, I was looking at the SA220. The timing resistor is R29 with D31 in parallel. They are next to the 555 timer chip. So short out D31 to bypass the timing delay. At least on the revC SA100 board
I'm thinking that the value of R60 is too low, hence the differential voltage is low. Even so, I would have thought the lateral MOSFETs would conduct at least a little bit with this low bias voltage. Even more perplexing is why two of the SJ55s failed with a gate/source short.
Over 20 hours, $590 in parts into this repair since August of last year. Countless hours finding a power transformer. I'm getting closer, but not there yet.
Over 20 hours, $590 in parts into this repair since August of last year. Countless hours finding a power transformer. I'm getting closer, but not there yet.
Did you notice that in case the bias trimpot wiper disconnects due to bad contact - all output power MOSFETs will blow instantly. By far not the only amp of that era with that design flaw.
Per the modifications to make the bias circuit operate at the required 1.2V for the MOSFET conversion to lateral MOSFETs, I changed the pots. The pots are new.
And most curiously, only the SJ55s are damaged. The SK175s still test good.
I have the sickening thought that there are more problems with the PCB itself causing these issues, but I'm about out of ideas.
And most curiously, only the SJ55s are damaged. The SK175s still test good.
I have the sickening thought that there are more problems with the PCB itself causing these issues, but I'm about out of ideas.
Earlier in this thread there were some who claimed that the pinouts for the SK/SJ parts are different from the 10N parts. I have been over the spec sheets multiple times and am convinced that pin 1 is Gate on both versions and pin 2 is Drain, putting case as Source. Let's do a 'sanity check' here and make sure this is the case. Top are the 10N part pinouts and bottom is the SK/SJ part pinouts. They look the same to me. Am I mistaken?
In the amplifier, the physical lead labeled "D" is connected to the case of the output devices. In both data sheets, that's the Source terminal.
Something is not adding up. I wish one of the 10N12s were good so I could compare them to the Hitachi devices.
On the Hitachi part, Source is case. But on the amplifier, it's wired as if Drain were case, but the 10N12 data sheet shows Source to be case. Now I'm not sure about this data sheet.
Something is not adding up. I wish one of the 10N12s were good so I could compare them to the Hitachi devices.
On the Hitachi part, Source is case. But on the amplifier, it's wired as if Drain were case, but the 10N12 data sheet shows Source to be case. Now I'm not sure about this data sheet.
I decided NOT to trust the 10N12 data sheet because I realize that B+ should not be going to the speaker terminal. The way the data sheet has it, the case is Source, but in reality, it must have been Drain. So I swapped the D and S wires at the PCB and I moved the other end of the diode from the end terminal to the case terminal.
Brought the amp up, but even with the jumper there was no output until I reached 105 VAC on the variac. Fuses did not blow. I waited. That one channel operating alone had a total line draw idle consumption of 95W.
I was able to get signal through the channel, but the distortion levels were unimpressive. Mostly around 1%, and at 500mW it was around 2.3% with some diagonal flattening of the negative peaks at that power level. Much lower than 500mW was getting lost in noise. Lackluster performance.
Now to get the other channel going, I'm going to need two more 2SJ55 MOSFETs, now that I learned the data sheet pinout is wrongly showing case to be Source and not Drain which it actually is on the original devices.
I still have to look into the bias circuit. The low 0.9V bias could be the reason for high distortion, but am also perplexed that the channel is drawing so much current with that small bias voltage.
Some major progress once I stopped trusting data sheets, but still a ways to go!
Brought the amp up, but even with the jumper there was no output until I reached 105 VAC on the variac. Fuses did not blow. I waited. That one channel operating alone had a total line draw idle consumption of 95W.
I was able to get signal through the channel, but the distortion levels were unimpressive. Mostly around 1%, and at 500mW it was around 2.3% with some diagonal flattening of the negative peaks at that power level. Much lower than 500mW was getting lost in noise. Lackluster performance.
Now to get the other channel going, I'm going to need two more 2SJ55 MOSFETs, now that I learned the data sheet pinout is wrongly showing case to be Source and not Drain which it actually is on the original devices.
I still have to look into the bias circuit. The low 0.9V bias could be the reason for high distortion, but am also perplexed that the channel is drawing so much current with that small bias voltage.
Some major progress once I stopped trusting data sheets, but still a ways to go!
Egads! All the 2SK55s are from China and $35 for one piece with shipping.
I'm thinking about BUZ905 as a substitute for the J55. Not sure if mixing 2SK176 with BUZ905 is a good idea, but I trust even less the parts from China.
I'm thinking about BUZ905 as a substitute for the J55. Not sure if mixing 2SK176 with BUZ905 is a good idea, but I trust even less the parts from China.
Digging through my parts bins, I have some 2SJ56 parts. According to the data sheet, every parameter is the same except the SJ56 had a 200V rating vs 180V for the SJ55. So I'm going to use those parts on the other channel.
Yes, you are wrong. This is the original pinout
which you can see is quite different from the Hitachi lateral MOSFET pinout
which you can see is quite different from the Hitachi lateral MOSFET pinout
Last edited: