This is my first post. Thank you for any help given.
I am rebuilding my Kenwood KA-7002 after an output transistor failure. I have a MJ21193/4G transistors in the output stage for the one side I have working now.
At idle (no input signal) everything is fine; voltages are correct and total power draw is 60 Watts. At low power levels 8-10 W the outputs heat up to ~65C., and seem to continue to heat up rather than stabilize. The unit will produce full 50 W output but the outputs heat rapidly.
They are properly mounted, bias and idling current are correct. (comments on this are welcome as the service manual is a little difficult to understand.)
I have replaced all the driver board electrolytic caps but not the pf bypass units.
Anyone have any thoughts about what appears to be excess heat.
Thanks again for any helpful advice.
Dave
I am rebuilding my Kenwood KA-7002 after an output transistor failure. I have a MJ21193/4G transistors in the output stage for the one side I have working now.
At idle (no input signal) everything is fine; voltages are correct and total power draw is 60 Watts. At low power levels 8-10 W the outputs heat up to ~65C., and seem to continue to heat up rather than stabilize. The unit will produce full 50 W output but the outputs heat rapidly.
They are properly mounted, bias and idling current are correct. (comments on this are welcome as the service manual is a little difficult to understand.)
I have replaced all the driver board electrolytic caps but not the pf bypass units.
Anyone have any thoughts about what appears to be excess heat.
Thanks again for any helpful advice.
Dave
60 W in idle seems a tad high for a 50 wpc amp.
Do the trim pots operate smoothly? In this restoration job, they turned out to be heavily corroded and were replaced, along with a bunch of electrolytics (Japanese caps had yet to reach their current quality levels at the time):
Kenwood KA-7002 und KT-7001 – highend von 1971 | good-old-hifi
Do the bias diodes (STV-3) still make good contact to the heatsink? I would imagine that any thermal paste would long since have dried up. If needed, use plain white silicone heatsink compound.
Good point about the infamous 2SC458 - stocking up on some 2SC945s may be a good idea.
Beyond mere restoration, I see plenty of tweaking potential in this amp:
Power amp section:
No Zobel oder Thiele networks present - it was the early days. Would try adding those. (1.5-2 mH self-wound || 4.7 ohms, 10 ohms 3 W + 100 nF or thereabouts. Needs identifying a good output-side grounding point to connect to.)
The drivers are relatively beefy TO-66 case jobs, yet run at a paltry 2 mA. I'd drop Re29-32 to 120-150 ohms each. Improve cooling if necessary, but we're talking <250 mW idle max.
Output emitter resistors are still 0.47 ohms as found on many lower-powered classic amps. Add 0.47R 3 W in parallel or replace by 0.22R 5 W cement, adjust bias to compensate.
Power supply:
If you have a serial number 410501 or higher unit, these seem to feature a more advanced voltage regulator than earlier ones. That one looks pretty good as-is, though I would still want to add at least 100 µF across zener Dk2 (noisy buggers). Arguably, voltage rating on Ck6 and Ck5 should be upped a bit.
An earlier pre-410501 unit features a simple 1-transistor capacitance multiplier of relatively limited regulating capabilities; I would aim to replace it by a more capable circuit along the lines of what's found in the power version, or something based on an IC like the TL783 which can be used much like a higher-voltage version of the LM317 (capacitor on ADJ pin included - I am just not sure whether the preamp circuitry actually draws the minimum 15 mA supposedly required to maintain regulation).
Preamp section:
I would give Ci23 (NF Tone board) and possibly Cp17 as well a dedicated power ground return to avoid signal ground pollution.
-hang on, my keyboard driver just died...-
Do the trim pots operate smoothly? In this restoration job, they turned out to be heavily corroded and were replaced, along with a bunch of electrolytics (Japanese caps had yet to reach their current quality levels at the time):
Kenwood KA-7002 und KT-7001 – highend von 1971 | good-old-hifi
Do the bias diodes (STV-3) still make good contact to the heatsink? I would imagine that any thermal paste would long since have dried up. If needed, use plain white silicone heatsink compound.
Good point about the infamous 2SC458 - stocking up on some 2SC945s may be a good idea.
Beyond mere restoration, I see plenty of tweaking potential in this amp:
Power amp section:
No Zobel oder Thiele networks present - it was the early days. Would try adding those. (1.5-2 mH self-wound || 4.7 ohms, 10 ohms 3 W + 100 nF or thereabouts. Needs identifying a good output-side grounding point to connect to.)
The drivers are relatively beefy TO-66 case jobs, yet run at a paltry 2 mA. I'd drop Re29-32 to 120-150 ohms each. Improve cooling if necessary, but we're talking <250 mW idle max.
Output emitter resistors are still 0.47 ohms as found on many lower-powered classic amps. Add 0.47R 3 W in parallel or replace by 0.22R 5 W cement, adjust bias to compensate.
Power supply:
If you have a serial number 410501 or higher unit, these seem to feature a more advanced voltage regulator than earlier ones. That one looks pretty good as-is, though I would still want to add at least 100 µF across zener Dk2 (noisy buggers). Arguably, voltage rating on Ck6 and Ck5 should be upped a bit.
An earlier pre-410501 unit features a simple 1-transistor capacitance multiplier of relatively limited regulating capabilities; I would aim to replace it by a more capable circuit along the lines of what's found in the power version, or something based on an IC like the TL783 which can be used much like a higher-voltage version of the LM317 (capacitor on ADJ pin included - I am just not sure whether the preamp circuitry actually draws the minimum 15 mA supposedly required to maintain regulation).
Preamp section:
I would give Ci23 (NF Tone board) and possibly Cp17 as well a dedicated power ground return to avoid signal ground pollution.
-hang on, my keyboard driver just died...-
OK, back with a working keyboard.
The preamp circuitry obviously is not the cat's meow by modern standards but should be serviceable (maybe 0.03% THD at full output). The part that's most suspicious looking to me right now is the tone amp, which has to sustain the highest levels and involves a voltage feedback common emitter circuit. I would have to simulate that to know for sure. (Speaking of simulation, the voltage levels given around Qp1 in the schematic seem dodgy to me. The 2SC458 is supposed to be a low noise, high beta transistor - yet it would need to have a beta of less than 100 at ~2 mA for those to work out as shown. Not even a BC547A will get me there.)
Missed one earlier - the aforementioned zener diode on the power amp board, Dq1/De1, should get some parallel capacitance (100+ µF). Its ground return may also appreciate some beefing up on the board at least - it seems to be connected to output-side ground already, which is good.
Also note that Ce18 on the last version is 220µ, while Cq18 was 100µ before.
The preamp circuitry obviously is not the cat's meow by modern standards but should be serviceable (maybe 0.03% THD at full output). The part that's most suspicious looking to me right now is the tone amp, which has to sustain the highest levels and involves a voltage feedback common emitter circuit. I would have to simulate that to know for sure. (Speaking of simulation, the voltage levels given around Qp1 in the schematic seem dodgy to me. The 2SC458 is supposed to be a low noise, high beta transistor - yet it would need to have a beta of less than 100 at ~2 mA for those to work out as shown. Not even a BC547A will get me there.)
Missed one earlier - the aforementioned zener diode on the power amp board, Dq1/De1, should get some parallel capacitance (100+ µF). Its ground return may also appreciate some beefing up on the board at least - it seems to be connected to output-side ground already, which is good.
Also note that Ce18 on the last version is 220µ, while Cq18 was 100µ before.
PS:
Speaking of adjusting bias, while you could do it with a small current clamp or even an ammeter in series as described in the service manual, I would suggest the now-common voltage drop over one 0.47R and Ohm's law method instead. The suggested 50 mA would give about 25 mV across, which is in the optimum range of ca. 13-26 mV. 0.33R would place you right in the middle of that range.
I forgot to mention one other possible power amp board tweak: swapping the 5.6k Rq4 (Re4) for 2.2k would improve dynamic impedance balance and hence distortion. Well, maybe. This amp has sort of a differential amplifier for a VAS that hasn't been used much in that role since the 1970s, would have to sim that to see how it performs and how input pair current balance does. Wait, I actually did years ago, after seeing it in a Sanyo JCX-2600 receiver. Think it did OK, but resulting input common-mode nonlinearity was pretty bad, so it's probably a good thing that we've got about 28.5 dB of voltage gain going. Maximum input amplitude for full nominal output is less than 0.8 Vrms.
I did end up simulating the entire preamp (HIGH FIL) circuit. As expected, it basically checks out OK performance wise. Obviously the power supply needs to be kept clean as PSRR is on the minimal side of things, but that's why both onboard and off-board RC filtering are present.
With the drivers being off the main heatsink as they are here, this may reduce thermal stability. Some extra thought on adding heatsinking to those may be required. Maybe focus on the bigger issues first.
Speaking of adjusting bias, while you could do it with a small current clamp or even an ammeter in series as described in the service manual, I would suggest the now-common voltage drop over one 0.47R and Ohm's law method instead. The suggested 50 mA would give about 25 mV across, which is in the optimum range of ca. 13-26 mV. 0.33R would place you right in the middle of that range.
I forgot to mention one other possible power amp board tweak: swapping the 5.6k Rq4 (Re4) for 2.2k would improve dynamic impedance balance and hence distortion. Well, maybe. This amp has sort of a differential amplifier for a VAS that hasn't been used much in that role since the 1970s, would have to sim that to see how it performs and how input pair current balance does. Wait, I actually did years ago, after seeing it in a Sanyo JCX-2600 receiver. Think it did OK, but resulting input common-mode nonlinearity was pretty bad, so it's probably a good thing that we've got about 28.5 dB of voltage gain going. Maximum input amplitude for full nominal output is less than 0.8 Vrms.
I did end up simulating the entire preamp (HIGH FIL) circuit. As expected, it basically checks out OK performance wise. Obviously the power supply needs to be kept clean as PSRR is on the minimal side of things, but that's why both onboard and off-board RC filtering are present.
Last edited:
PS:
This is definitely something to be verified, especially if you end up having to swap out those known-problematic 2SC458s (or if it has been done already). With a proper high-beta transistor (and nice new input coupling caps of low leakage), collector voltage may end up way low. A BC550 didn't get in the right ballpark until I reduced the 82k to 62k, and I needed 56k for 12 V. Similar results with 2N5089 and 2SC4102 models. A plain 2N3904, by contrast, could use bumping it up to 100k.
I did, however, have to vary biasing (Rp3/4) depending on transistor type used. There is a tradeoff for Qp1/2 collector voltage - while a lower voltage will decrease distortion across the board due to increased current, eventually high-frequency performance will be degraded due to increased input capacitance and associated nonlinearity. The 12 V indicated get reasonably close, the optimum seems to be around 10-11 V.
This is definitely something to be verified, especially if you end up having to swap out those known-problematic 2SC458s (or if it has been done already). With a proper high-beta transistor (and nice new input coupling caps of low leakage), collector voltage may end up way low. A BC550 didn't get in the right ballpark until I reduced the 82k to 62k, and I needed 56k for 12 V. Similar results with 2N5089 and 2SC4102 models. A plain 2N3904, by contrast, could use bumping it up to 100k.
Last edited:
Thanks for all the help with my KA-7002, either directly or indirectly throuogh some other member's question.
Correctly adjusted bias (idling current?) has solved the heat problem. I now measure just slightly over 50ma both by voltage drop and clamp-on ammeter. I just ran the amp at full power (20vac across 8 ohm dummy loads) for a little over 10 minutes. The highest temperature I read was 50C.
I replaced the H/S compound on the STV-3's (dry as forecast). The diff amps are near zero at the dummies with no signal input. The values waver some, =/-5 mv. Is that expected behavior.
Someone mentioned the diff pair needing to be thermally close. I will do a little careful lead bending to achieve that.
In spite of the trimmer pot order fiasco; I ordered the wrong item, way to small, still I am a happy guy. The amp works. And I have lots of suggestions from the folk here to make it better.
Any thought on inrush limiting resistors? Seems like cheap insurance to me.
Thanks again.
Correctly adjusted bias (idling current?) has solved the heat problem. I now measure just slightly over 50ma both by voltage drop and clamp-on ammeter. I just ran the amp at full power (20vac across 8 ohm dummy loads) for a little over 10 minutes. The highest temperature I read was 50C.
I replaced the H/S compound on the STV-3's (dry as forecast). The diff amps are near zero at the dummies with no signal input. The values waver some, =/-5 mv. Is that expected behavior.
Someone mentioned the diff pair needing to be thermally close. I will do a little careful lead bending to achieve that.
In spite of the trimmer pot order fiasco; I ordered the wrong item, way to small, still I am a happy guy. The amp works. And I have lots of suggestions from the folk here to make it better.
Any thought on inrush limiting resistors? Seems like cheap insurance to me.
Thanks again.
- Status
- Not open for further replies.