I'm working on a Nakamichi PA-7 amplifier that was found by an acquaintance recently at a thrift store. When he first started using it, the left channel would go into protection after a few hours. I got it on my bench and started to take a look; the first thing I noticed was that the bias was extremely high on both channels, over 150mV before I powered it off. I was able to adjust the bias to the nominal 40mV, and the offset to the nominal 100mV. In this state, the amp draws about 2-2.5A, which seems comparable to what others have seen.
I gave the amp back and it's still tripping protection on the left channel. He swapped speaker leads and it still trips the left channel only. Oddly, I've never seen it happen in my workshop, but perhaps that's a difference in speaker loads.
This particular amp has the additional small protection PCB on each channel, which I understand was a later revision by Nakamichi. I did diode tests on the two transistors on each board, and all pin combos measure the same between the two channels. Similarly, the resistors, diodes, and capacitors on the PCBs also measure the same. One potential clue is that, with the amp powered on, when I probe the voltage across the top two pins that connect the protection PCB to the main PCB, it measures 86v on the right channel, but on the left channel it will trip the protection, the relay will open, the voltage will drop to about 2v, then it will rise after a few seconds to 86v, at which point the protection relay closes again. The two pins I'm probing attach to one lead of D109 and R152 on the main PCB.
Does anyone have suggestions for other things I could look at to diagnose this problem?
I gave the amp back and it's still tripping protection on the left channel. He swapped speaker leads and it still trips the left channel only. Oddly, I've never seen it happen in my workshop, but perhaps that's a difference in speaker loads.
This particular amp has the additional small protection PCB on each channel, which I understand was a later revision by Nakamichi. I did diode tests on the two transistors on each board, and all pin combos measure the same between the two channels. Similarly, the resistors, diodes, and capacitors on the PCBs also measure the same. One potential clue is that, with the amp powered on, when I probe the voltage across the top two pins that connect the protection PCB to the main PCB, it measures 86v on the right channel, but on the left channel it will trip the protection, the relay will open, the voltage will drop to about 2v, then it will rise after a few seconds to 86v, at which point the protection relay closes again. The two pins I'm probing attach to one lead of D109 and R152 on the main PCB.
Does anyone have suggestions for other things I could look at to diagnose this problem?
Would you clarify a bit--- between which two schematic points are you placing the meter probes? Do you have to actively probe to trigger protection?
I suggest probing within the protection circuit to help identify the triggering mechanism. I gather you have confirmed low offset voltage at the left channel output before the relay, eg. on collector of Q121. Excessive amplifier offset voltage is detected at the emitter of Q132 or the base of Q133; ideally, the voltage appearing there will be 0V, and voltage near +/- 0.6V will trigger protection.
Two other triggering mechanisms are monitoring of emitter currents in Q121 and Q126. Q121 current is sensed across R131 and fed to Q138 on the "PROTECTOR L PCB." Similarly, Q126 current is sensed across R136 and fed to Q137. There's an unlikely chance R131 or R136 could have increased resistance, so confirm low DC drop across each. Then check DC drop across R157 and R153; voltage of about 0.6V would trigger protection and should be well below these values during no-audio testing.
Collector current Q137 is the eventual path that trips overcurrent protection when it discharges C110. Conduction of Q133 is the path for excessive offset voltage trigger.
Let us know if you need additional help/details.
Good luck!
Yes, I've never seen it go into protection unless I probe those two points joining the left channel protection PCB to the main PCB, specifically one lead of C114 and the point where R158 and R159 are tied together.
Thanks for the suggestions. Here are some measurements.
Q121 (2SA1294) collector to GND: 103.4mV
Q132 (2SC3333) emitter to GND: 0.5mV
These both seem ok.
R131 measures: 34.7mV (right channel measures 23.2mV)
R136 measures: 43mV
R157 measures: 5.7mV (right channel measures 8.1mV)
R153 measures: 0mV (right channel measures 22mV)
No current going through R153 seems like a clue. I measured the resistor and it's ~22k as it should be.
Diode measurements of Q137 compare similarly to the right channel. What else should I check?
Thanks for the suggestions. Here are some measurements.
Q121 (2SA1294) collector to GND: 103.4mV
Q132 (2SC3333) emitter to GND: 0.5mV
These both seem ok.
R131 measures: 34.7mV (right channel measures 23.2mV)
R136 measures: 43mV
R157 measures: 5.7mV (right channel measures 8.1mV)
R153 measures: 0mV (right channel measures 22mV)
No current going through R153 seems like a clue. I measured the resistor and it's ~22k as it should be.
Diode measurements of Q137 compare similarly to the right channel. What else should I check?
Thanks, tenelson. Sorry for delayed reply.
Hard to quarrel with any of your measurements--- they all look reasonable.
I wouldn't expect much voltage across R153, since there's little voltage drop across R136, plus resistors R154, R155 and D104. But I do share your suspicion about R153. With power off, would you measure R153 resistance by probing at the base and emitter leads (or their pads) to confirm intact PC traces; if you probe with black lead on the base, the base-emitter junction shouldn't disturb the resistance measurement.
It's curious that only the left channel seems susceptible to the connecting the voltmeter. If I interpret correctly, your probing voltage from the R158,R159 junction and the C114 lead that connects at collector of Q137; if so, the meter is connected betweethe base and collector of Q137--- right? This begs the question, what's the voltmeter input resistance and is it sufficient to make Q137 conduct, thus triggering protection? Do you know meter (load) resistance? Is it a DVM or an analog meter? What's the displayed voltage when it triggers protection? Is the reading the same when probing the working channel?
Hard to quarrel with any of your measurements--- they all look reasonable.
I wouldn't expect much voltage across R153, since there's little voltage drop across R136, plus resistors R154, R155 and D104. But I do share your suspicion about R153. With power off, would you measure R153 resistance by probing at the base and emitter leads (or their pads) to confirm intact PC traces; if you probe with black lead on the base, the base-emitter junction shouldn't disturb the resistance measurement.
It's curious that only the left channel seems susceptible to the connecting the voltmeter. If I interpret correctly, your probing voltage from the R158,R159 junction and the C114 lead that connects at collector of Q137; if so, the meter is connected betweethe base and collector of Q137--- right? This begs the question, what's the voltmeter input resistance and is it sufficient to make Q137 conduct, thus triggering protection? Do you know meter (load) resistance? Is it a DVM or an analog meter? What's the displayed voltage when it triggers protection? Is the reading the same when probing the working channel?
black lead on base of Q137 (pin 3): to pad 1 of R153: 0.2 ohms
black lead on emitter of Q137 (pin 1) to pad 2 of R153: 0.2 ohms
I also measured from C114 on the protection PCB to the Q137 collector: 0.3 ohms
I'm using a Fluke 115 multimeter. I tried to look up input resistance, but didn't see a spec listed.
Correct, when I measure the collector of Q137 to the R158/R159 junction, protection trips as I've described.
When probing those points on the left channel, it trips so quickly that it goes from 0v (meter disconnected) to about 2v. On the right channel it measures a solid 86.4v. Interestingly, the left channel will also sometimes trip with only one probe connected to either the Q137 collector or the R158/R159 junction, which I find very strange!
I also checked TP101 and TP102 to ground on both channels, and in both cases they're around -77v.
black lead on emitter of Q137 (pin 1) to pad 2 of R153: 0.2 ohms
I also measured from C114 on the protection PCB to the Q137 collector: 0.3 ohms
I'm using a Fluke 115 multimeter. I tried to look up input resistance, but didn't see a spec listed.
Correct, when I measure the collector of Q137 to the R158/R159 junction, protection trips as I've described.
When probing those points on the left channel, it trips so quickly that it goes from 0v (meter disconnected) to about 2v. On the right channel it measures a solid 86.4v. Interestingly, the left channel will also sometimes trip with only one probe connected to either the Q137 collector or the R158/R159 junction, which I find very strange!
I also checked TP101 and TP102 to ground on both channels, and in both cases they're around -77v.
You're observing -77V on TP101 is quite a bit higher than the nominal -60V depicted in the schematic. I don't know if that's relevant, but it's 28% and is surprising. Might be worth exploring. Assuming symmetry, there'd be +77V on the opposite rail, 154V rail-to-rail. Some versions of this amp have line-voltage adjustment options; you should examine this issue in your amp. Maybe line voltage is even higher at your friend's house, thus aggravating protection. Obviously, large lethal opportunities present within. Be careful!
I found a spec on the Fluke 115 indicating greater than 10M meter resistance on DC measurement ranges. I'm betting it's nominally 11M, but a guess. When you apply the Fluke where you indicate, I believe the meter sees ~ 150V. Would you confirm? That would mean 13.6uA if all guesswork is correct, and would deliver about 0.3V across R253. That shouldn't be enough to trigger Q237 and protection, but it's close. Transient currents when you connect might do the trick. Also, the meter may be acting as an antenna and coupling hum/noise sufficient for trigger, as you described above.
In short, I'm growing suspicious that the VM experiment may be an unduly provocative test. But how to troubleshoot the problem when you can't reproduce on your bench?
I can think of some further experiments, but they become more tedious. I suggest you explore the possible line voltage adjustment issues. I'll contemplate next tests till I hear back.
Good luck!
I found a spec on the Fluke 115 indicating greater than 10M meter resistance on DC measurement ranges. I'm betting it's nominally 11M, but a guess. When you apply the Fluke where you indicate, I believe the meter sees ~ 150V. Would you confirm? That would mean 13.6uA if all guesswork is correct, and would deliver about 0.3V across R253. That shouldn't be enough to trigger Q237 and protection, but it's close. Transient currents when you connect might do the trick. Also, the meter may be acting as an antenna and coupling hum/noise sufficient for trigger, as you described above.
In short, I'm growing suspicious that the VM experiment may be an unduly provocative test. But how to troubleshoot the problem when you can't reproduce on your bench?
I can think of some further experiments, but they become more tedious. I suggest you explore the possible line voltage adjustment issues. I'll contemplate next tests till I hear back.
Good luck!
Thanks again for the suggestions. I double checked the voltages on the main caps themselves, and they're at +/-77v for both channels. I guess that would be driven by the line voltage and the the rectifiers, D401 and D402, the Toshiba 10B4B41. Since both devices are behaving the same, I guess I have to conclude that the +/-77v is fine, even though I agree it the difference between the schematic and reality is much higher than I'm used to.
I appreciate the reminder about the lethal voltages. Working on amps like this always makes me a little nervous! I take all of the precautions I can.
I can see no line voltage adjustment capability, so I guess this is the standard US version.
The place where I probe on the protection PCB is about 86v on the right channel, and 2v when it trips on the left channel, ramping up to 86v after a few seconds. So, not 150v. Unless I misunderstood your suggestion.
I used my variac set set the AC at about 100v, which resulted in a main cap voltage of +/-64v. Left channel protection still trips when I probe the usual spot. Do you have any other ideas of what to look at?
I appreciate the reminder about the lethal voltages. Working on amps like this always makes me a little nervous! I take all of the precautions I can.
I can see no line voltage adjustment capability, so I guess this is the standard US version.
The place where I probe on the protection PCB is about 86v on the right channel, and 2v when it trips on the left channel, ramping up to 86v after a few seconds. So, not 150v. Unless I misunderstood your suggestion.
I used my variac set set the AC at about 100v, which resulted in a main cap voltage of +/-64v. Left channel protection still trips when I probe the usual spot. Do you have any other ideas of what to look at?
Let's brain storm a bit. Radom thoughts.
I loathe replacing components until I determine they're likely defective--- makes me feel like an inept mechanic who keeps installing new parts until the problem goes away, if he's lucky. But maybe I'm being obsessive. I think I might replace C110, C112, C114; I have the impression that the protection is "twitchy" and defective filter caps could exacerbate susceptibility, plus replacement is easy and cheap. If susceptibility to your VM probing then improved to resemble the good channel, that might constitute success.
Having a reliable, rationally justifiable method of triggering protection is quandary.
One idea: The protection circuit has three input paths--- excessive offset detection, over current on positive rail, over current on negative rail. We could disable one path at a time, leave the amp at your friend's home, and hope to identify the problematic path. Sounds like desperation.
If we've replaced the few caps that were present, the most likely defects are leaky transistors and resistors drifted out of spec. The data you reported in post 3 were mostly checks for leakage and didn't reveal anything suspicious. You can make in-circuit resistance measurements and often confirm resistors even when there are shunting paths. I advise against removing resistors to measure them unless unavoidable as there's often damage done, not to mention being slow and laborious.
Any ideas/thoughts? I recommend you respect your instincts and pursue your suspicions. That's how you develop experience and skills.
I'm convinced we can find the problem but it may be tedious, need patience and persistence.
Steve
I loathe replacing components until I determine they're likely defective--- makes me feel like an inept mechanic who keeps installing new parts until the problem goes away, if he's lucky. But maybe I'm being obsessive. I think I might replace C110, C112, C114; I have the impression that the protection is "twitchy" and defective filter caps could exacerbate susceptibility, plus replacement is easy and cheap. If susceptibility to your VM probing then improved to resemble the good channel, that might constitute success.
Having a reliable, rationally justifiable method of triggering protection is quandary.
One idea: The protection circuit has three input paths--- excessive offset detection, over current on positive rail, over current on negative rail. We could disable one path at a time, leave the amp at your friend's home, and hope to identify the problematic path. Sounds like desperation.
If we've replaced the few caps that were present, the most likely defects are leaky transistors and resistors drifted out of spec. The data you reported in post 3 were mostly checks for leakage and didn't reveal anything suspicious. You can make in-circuit resistance measurements and often confirm resistors even when there are shunting paths. I advise against removing resistors to measure them unless unavoidable as there's often damage done, not to mention being slow and laborious.
Any ideas/thoughts? I recommend you respect your instincts and pursue your suspicions. That's how you develop experience and skills.
I'm convinced we can find the problem but it may be tedious, need patience and persistence.
Steve
Madis64, thanks for the suggestion. The first thing I did when working on this amp was to replace the trimpots with new Bourns units because the originals were very hard to adjust correctly.
I don't have the bipolar caps needed for C109 and C114. I replaced C110 and C112, and C111 because I was in the area, and I can no longer get the amp to go into protection when I probe it in the manner previously described.
Before I declare complete success, I think I'll go ahead and order parts for C109 and C114. If there's some sort of capacitor aging issue, maybe those are about to go too.
I measured all 3 removed caps with my Fluke 115, and the two 10uF caps were low by less than 10%, which seems fine. The 1uF cap (C112) was low by about 15%. So, I'm a bit surprised by the results! Thanks for the help! I'll report back when I get the bipolar caps replaced too.
I don't have the bipolar caps needed for C109 and C114. I replaced C110 and C112, and C111 because I was in the area, and I can no longer get the amp to go into protection when I probe it in the manner previously described.
Before I declare complete success, I think I'll go ahead and order parts for C109 and C114. If there's some sort of capacitor aging issue, maybe those are about to go too.
I measured all 3 removed caps with my Fluke 115, and the two 10uF caps were low by less than 10%, which seems fine. The 1uF cap (C112) was low by about 15%. So, I'm a bit surprised by the results! Thanks for the help! I'll report back when I get the bipolar caps replaced too.
In general, if I am going to get into an older circuit, I just replace electrolytic capacitors regardless. Their time is coming if they are more than 20 years old. And I hope to maybe put in a better cap in some instances. I recently heard someone mention that when a meter is checking a cap, it is not the same as when the cap is under load. I don't know if that has any merit to it, but I hadn't thought of that before.