I got a HK980 for cheap today. I have a modified HK670 ( https://www.diyaudio.com/community/threads/hk-670-integrated-amp-sim-mods-etc.384245/ ) that I like, and this has the same power amp so I could not resist.
Owner said it goes into overheat protection randomly, but works fine other than that. I tested it abt 10min when I picked it up, and it worked fine. Did not heat up, and owner claims it's not hot when it goes into protection either. Based on this I guess there is a problem with the detection circuit, and not bias. Seems to be a NC temp switch that is connected to an input on a processor. Pictures below.
I have not opened the amp yet, thought I would check if someone here has fixed the same problem before, because it seems to be a common issue with these? If not, I will post what I find when I get to it.
Owner said it goes into overheat protection randomly, but works fine other than that. I tested it abt 10min when I picked it up, and it worked fine. Did not heat up, and owner claims it's not hot when it goes into protection either. Based on this I guess there is a problem with the detection circuit, and not bias. Seems to be a NC temp switch that is connected to an input on a processor. Pictures below.
I have not opened the amp yet, thought I would check if someone here has fixed the same problem before, because it seems to be a common issue with these? If not, I will post what I find when I get to it.
I have been fault tracing this amp a bit, but the overheat fault is hard to reproduce. Moving wires and tapping stuff did not seem to do anything. Played it for hours without a glitch. I ended up re-soldering all joints related to the temp switches, replacing caps in the low voltage supplies etc, but less than 1min after first start I got the overheat shutdown again. After that if did not happen while playing music for more than 1h.
I then decided to mount a not installed SMD pull-up resistor for the overheat input on the processor, and disconnect the circuit to the switches. I have several DIY amps without overheat protection, so I don't worry about that. The plan is to see if the fault is actually gone while the input is definitely high. If that works, the plan is to replace that resistor with a cap, since I'm thinking there might be some ripple on the voltage in the loop causing intermittent shutdowns. No filtering as is, and abt 1s filtering on the temp should not be a problem.
Possibly there could be some internal corrosion on the temps switches causing a bad connection..? I have never hear about something like that, but who knows..
I have also modified this amp with the nested feedback, as mentioned in the above link. Same positive results as before. I did use higher resistor values (27 Ohm) for the LTP emitter resistors, because that is what I happened to have. I simulated it, and it did not do much to the loop gain, and I thought I could try a slightly different 'tune' on this one.
Bias was close to 0 as I got it, so I adjusted it according to manual for now, and plan to adjust while measuring distortion later.
I then decided to mount a not installed SMD pull-up resistor for the overheat input on the processor, and disconnect the circuit to the switches. I have several DIY amps without overheat protection, so I don't worry about that. The plan is to see if the fault is actually gone while the input is definitely high. If that works, the plan is to replace that resistor with a cap, since I'm thinking there might be some ripple on the voltage in the loop causing intermittent shutdowns. No filtering as is, and abt 1s filtering on the temp should not be a problem.
Possibly there could be some internal corrosion on the temps switches causing a bad connection..? I have never hear about something like that, but who knows..
I have also modified this amp with the nested feedback, as mentioned in the above link. Same positive results as before. I did use higher resistor values (27 Ohm) for the LTP emitter resistors, because that is what I happened to have. I simulated it, and it did not do much to the loop gain, and I thought I could try a slightly different 'tune' on this one.
Bias was close to 0 as I got it, so I adjusted it according to manual for now, and plan to adjust while measuring distortion later.
A small update on this. No overheat protection with the pull-up. I found the temp switches online (China), so I plan to replace them when they arrive and I pull the amp apart again.
I will decrease the LTP emitter resistors to 10ohm, add the possibility to bypass the pre-amp (direct signal to the amp boards).
The loop gain with 27ohm resistors was not enough to get the last of the 'tightness' in the bass, so I that's why I want to try 10ohm. I still want to experiment a bit, so not going all the way to 1ohm as on the HK670. I now have 5.7p compensation caps, that should be fine with the 10ohm resistors, so I will change those too. Possibly gain some 'clarity' in the treble.
I measured distortion (FFT) and tuned bias, and lowest distortion at normal listening levels @ 4-8 ohm load was around 20mV bias. At full blast, factory setting around 14mV gave less high order distortion components, but I found 20mV to be a better compromise for normal use. Heat should not be a problem either. Distortion levels are around -90dB, which is perfectly fine IMO. Running multitone, the 'grass floor' is at least -80db, and flat over the audio band. However, there seems to be some 'loudness' even in the direct mode. At low volume settings I saw a boost close to 2dB in the upper treble. This was depending on the volume setting on the amp, not the output volume or load. I have no clue where this is coming from, volume, tone etc is all handled by rotary encoders and chips, not potentiometers. This is one reason I want the possibility to bypass the pre-amp.
I also want to try measuring output impedance, I will try to post something on that, if I succeed.
The plan is to feed one channel (with shorted input) from the other via a resistor and use ARTA/LIMP to plot impedance. This is basically what I have done to sim output impedance, and I think I read something about it here some time ago, but don't remember seeing actual measurements.
I will decrease the LTP emitter resistors to 10ohm, add the possibility to bypass the pre-amp (direct signal to the amp boards).
The loop gain with 27ohm resistors was not enough to get the last of the 'tightness' in the bass, so I that's why I want to try 10ohm. I still want to experiment a bit, so not going all the way to 1ohm as on the HK670. I now have 5.7p compensation caps, that should be fine with the 10ohm resistors, so I will change those too. Possibly gain some 'clarity' in the treble.
I measured distortion (FFT) and tuned bias, and lowest distortion at normal listening levels @ 4-8 ohm load was around 20mV bias. At full blast, factory setting around 14mV gave less high order distortion components, but I found 20mV to be a better compromise for normal use. Heat should not be a problem either. Distortion levels are around -90dB, which is perfectly fine IMO. Running multitone, the 'grass floor' is at least -80db, and flat over the audio band. However, there seems to be some 'loudness' even in the direct mode. At low volume settings I saw a boost close to 2dB in the upper treble. This was depending on the volume setting on the amp, not the output volume or load. I have no clue where this is coming from, volume, tone etc is all handled by rotary encoders and chips, not potentiometers. This is one reason I want the possibility to bypass the pre-amp.
I also want to try measuring output impedance, I will try to post something on that, if I succeed.
The plan is to feed one channel (with shorted input) from the other via a resistor and use ARTA/LIMP to plot impedance. This is basically what I have done to sim output impedance, and I think I read something about it here some time ago, but don't remember seeing actual measurements.
I decided to try some output impedance measurements with LIMP (ARTA) today. Not all plain sailing.. The version I had installed actually added cable compensation to the measurement, even if it should have subtracted, so after measuring one channel, I had to update to the latest version. There is also trouble measuring really low impedances, the result is like there is inductance, but with higher impedances, the phenomena disappears. I tried using different reference resistors, changing sample rate and bit depth output level etc, but nothing seemed to cure it. Ok, LIMP is not meant to measure mOhms so I can't blame it, and its also free (demo).It could be my setup too. High frequencies seem sensitive when I move the cables around while I measure a short. I'm guessing the reason is that the signal on the DUT-side of the reference resistor is basically zero when measuring shorts/low impedances.
The setup is a normal impedance measurement setup as described in the ARTA/LIMP manual, I just reduced the reference resistor to 2ohms, and measure the speaker terminals instead of a speaker. So, one channel is feeding the other in the same amp. In this case there is no common power ground for the L&R channels, because the transformer has separate secondary windings for each channel. The only common ground is the signal ground. This complicates things a bit, because I need to link the Speaker grounds between channels, or the current would pass through he thin coaxial signal cable to the amp boards. To avoid ground loops I disconnected the input signal to the channel measured (input left open, I did not have a connector that fit to short it) . There is some 50Hz pickup, but not too bad.
To be able to visualize the impedance of the amp terminals, I did one measurement with a short, and used is as overlay, then measured the amp. So the output impedance is basically the diff between the two lines, except in the top end where the measurement is not behaving very well at mOhms.
Something wrong
Good IMO
Also ok
Pretty bad!
Surprisingly there are massive differences between the A/B output terminals, so this is definitely something I will have to look at when I take it apart. While measuring one of the 'bad' output terminals, I tried pushing the PCB around the output relays, output terminals and an angled connector between terminal PCB and amp PCB, and I could se the output impedance change. I don't think I would have investigated that without doing these measurements. I think on the two measurements in the middle, the effect of the nested feedback is visible, with output impedance going down slightly to abt 200Hz where it levels off. Maybe measurement #2 is the one where it's easiest to see.
This was an interesting thing to measure, I have simulated it before, but now I did it in reality
I was actually surprised the measured impedance is actually this low, I was expecting it to be higher with all the traces, output inductors, relays, connectors, terminals etc in the output circuit.
Finally a picture of the mess:
The setup is a normal impedance measurement setup as described in the ARTA/LIMP manual, I just reduced the reference resistor to 2ohms, and measure the speaker terminals instead of a speaker. So, one channel is feeding the other in the same amp. In this case there is no common power ground for the L&R channels, because the transformer has separate secondary windings for each channel. The only common ground is the signal ground. This complicates things a bit, because I need to link the Speaker grounds between channels, or the current would pass through he thin coaxial signal cable to the amp boards. To avoid ground loops I disconnected the input signal to the channel measured (input left open, I did not have a connector that fit to short it) . There is some 50Hz pickup, but not too bad.
To be able to visualize the impedance of the amp terminals, I did one measurement with a short, and used is as overlay, then measured the amp. So the output impedance is basically the diff between the two lines, except in the top end where the measurement is not behaving very well at mOhms.
Something wrong
Good IMO
Also ok
Pretty bad!
Surprisingly there are massive differences between the A/B output terminals, so this is definitely something I will have to look at when I take it apart. While measuring one of the 'bad' output terminals, I tried pushing the PCB around the output relays, output terminals and an angled connector between terminal PCB and amp PCB, and I could se the output impedance change. I don't think I would have investigated that without doing these measurements. I think on the two measurements in the middle, the effect of the nested feedback is visible, with output impedance going down slightly to abt 200Hz where it levels off. Maybe measurement #2 is the one where it's easiest to see.
This was an interesting thing to measure, I have simulated it before, but now I did it in reality
I was actually surprised the measured impedance is actually this low, I was expecting it to be higher with all the traces, output inductors, relays, connectors, terminals etc in the output circuit.
Finally a picture of the mess:
Another measurement while it's connected on the bench. Frequency response vs volume setting this time, since I saw something funny on low volume settings before. Did sweeps in steps of 10dB on the volume (-60 to 0dB), and adjusted output volume on sound card to get the response curves amplitude closer in the graph. Direct mode is on, so tone control is bypassed. The slight bass boost is most likely due to the nested feedback I added. The treble boost at -60db gets worse at lower settings, but I don't see even -60dB as a plausible setting when listening, so it's not as bad as I initially suspected.
Measurements are only done with pink noise, no accuracy is not so high. I don't have the patience for stepped sine!
ht.
Measurements are only done with pink noise, no accuracy is not so high. I don't have the patience for stepped sine!
You are on the right track with inductive component. It would be insignificant for short wires and measuring in the 100 mΩ range. But, you have long wires. Considering that 1.5 – 2.5 mm2 wires have inductance about 5-4 nH/cm there is even difference while measuring ‘short circuit’ consisting of several cm long wire. This is also the answer to my problem described in another thread. (slapping my forehead
).I’ve made impedance measurement of amplifier output shorted with 10 cm long wire. Measured impedance ratio from 1 kHz to 20 kHz was 4x. Ideally it should be 1. So, we can’t even make a proper short circuit for the whole audio band when several mΩ is measurement target.
That 10 cm wire made 3 mΩ short circuit at 1 kHz and 10 mΩ at 20 kHz.
Yes, I get it, but I don't get that the phenomena is not linear. I entered cable compensation parameters including inductance to try to make it a straight line, but I hit the limit of the inductance value and the line was still straight, but as soon as impedance went up, inductance was overcompensated and the top end took a dive in impedance. These tendencies can be seen in the graphs I posted earlier.
I did some work on this amp today. LTP emitter resistors are now 10ohm and compensation cap C6 is 5.7p. Square wave is perfect with no load/4ohm/8ohm. I also made it possible to feed signal directly to the power amps. I used rec out as input, and a switch to select between that and the pre-amp. Output is very clean when feeding the amps directly, SNR is at least 90dB 20-20kHz with multitone test signal, not even PS noise is above -90dB. There is a lot more PS noise when the pre-amp is in use. The transformer is making some noise and vibration, so I suspect it's not happy with 240VAC in. Maybe I will try a 'bucking transformer' on it to see if it makes it quiet. Could be DC-offset on AC supply too.
Bias voltage is set to 20mV.
I replaced temp-switches, and added a 1uF cap in the position of R73. This should filter out any noise/ripple on the overheat input to the processor.
I listened to the amp for a while, and the impression is positive. It's a keeper, -for me or somebody else
Bias voltage is set to 20mV.
I replaced temp-switches, and added a 1uF cap in the position of R73. This should filter out any noise/ripple on the overheat input to the processor.
I listened to the amp for a while, and the impression is positive. It's a keeper, -for me or somebody else
I forgot to update my last findings, so here is a short summary. I seems the transformer is not supplying enough current for the 12V(?) supply going to relay coils, led in volume knob, and overheat detection. I measured the voltage before the regulator, and it dropped below the regulator voltage with all relays energized. I think this makes the regulator flip out from time to time. Bypassing the resistors before the rectifier helped a bit, but energizing all relays still cased too much voltage drop. I could have added another trafo for this circuit, but I gave the amp to my brother and sintructed him to not activate A and B speaker outputs, that seems to solve it. You could try to mesure the voltage on your amp to see if it has the same issue.
It did not help for me, problem was that it was so intermittant that it took long to veify. At least check the supply voltage first to see if that is low in your amp too.
Mounting a pull-up resistor in the r73 position and disconnecting the wire to the temp sensing will get rid of the error, but disable the overheat protection.
Mounting a pull-up resistor in the r73 position and disconnecting the wire to the temp sensing will get rid of the error, but disable the overheat protection.
Ok, I checked with my brother in law and he said he uses both A and B speakers - he only had 1 set for a long time but bought some JBLs and wasn't happy with them by themselves so now runs two sets at the same time. That would explain why he never had the problem before, and I guess it means there is nothing to be "fixed".
I've got a full set of caps ordered so will replace those anyway and will get the bits to do the mods you've suggested. I found the temp switch on Ali (Klixon 17AM027A5) but will hold off ordering that for now.
I've got a full set of caps ordered so will replace those anyway and will get the bits to do the mods you've suggested. I found the temp switch on Ali (Klixon 17AM027A5) but will hold off ordering that for now.
I've replaced all the caps and added the mods for nested feedback.
Starting up with a dim bulb tester, I can't get it out of standby mode, just a quick bright glow and then it shuts off (back to standby). I started to troubleshoot by disconnecting one amp module, and it started up, which made me think that module has a problem...I guess I can't do any active fault finding if the amp won't turn on with the module connected?
I have double checked all the caps I replaced and they are correct values and correct polarity, so it got me thinking. Is there a chance that the protection circuit is cutting in because of the dim bulb tester in series? I am away at the moment, but next step when I get back to it will be to reconnect the left and disconnnect the right and see if anything changes.
Starting up with a dim bulb tester, I can't get it out of standby mode, just a quick bright glow and then it shuts off (back to standby). I started to troubleshoot by disconnecting one amp module, and it started up, which made me think that module has a problem...I guess I can't do any active fault finding if the amp won't turn on with the module connected?
I have double checked all the caps I replaced and they are correct values and correct polarity, so it got me thinking. Is there a chance that the protection circuit is cutting in because of the dim bulb tester in series? I am away at the moment, but next step when I get back to it will be to reconnect the left and disconnnect the right and see if anything changes.
I don't think so, but one similar issue I had was a bad connection on one of the coaxial lead connectors signal ground to the amp board, that triggered the protection because it generated a DC offset. Those connectors are not the best IMO. You can measure resistance between the preamp board ground and the input ground on the amp board (no power connected).
Ok…I triple checked checked the boards and all the transistors are ok as well.
So what is interesting here is that with either the left or right amplifier boards disconnected from the power supply board, if I press the power on button, the globe flashes bright for a second, the power light changes from red to orange, and a second later a relay clicks and the power comes fully on (green light and display).
At this point there is a faint glow from the light bulb. I don’t know how many watts it is, most likely 60W.
If I connect both amp boards to the power supply board (transformer connected) but disconnect both connections to the IO board it will turn on with just a slight initial glow of the bulb and then the bulb dims completely, but after about 10 sec the amp shuts down again. I guess those are the coax cables you are referring to?
So what is interesting here is that with either the left or right amplifier boards disconnected from the power supply board, if I press the power on button, the globe flashes bright for a second, the power light changes from red to orange, and a second later a relay clicks and the power comes fully on (green light and display).
At this point there is a faint glow from the light bulb. I don’t know how many watts it is, most likely 60W.
If I connect both amp boards to the power supply board (transformer connected) but disconnect both connections to the IO board it will turn on with just a slight initial glow of the bulb and then the bulb dims completely, but after about 10 sec the amp shuts down again. I guess those are the coax cables you are referring to?
Further to this:
II get ~0.2 Ohms resistance across all the ground points, eg
The last test I think confirms there is a good earth connection to the amplifier board itself.
II get ~0.2 Ohms resistance across all the ground points, eg
- from the back of the coax black wire to the same on the other board
- from the back of the coax wire to the chassis
- from the back of the coax wire on the amp board to the black wire on the transformer connector on the amp board
The last test I think confirms there is a good earth connection to the amplifier board itself.