A few seconds thought will tell you that any amplifier which works fine for some years and then begins to show increasing
overheating with deteriorating sound quality, has developed a fault - most likely a component or latent assembly fault.
That fault will be fixable and the condition effectively eliminated, whether it has thermal instability, damaged output devices,
oscillation or any typical fault. The 640 may not be one of D. Self's designs but we can assume it was sufficiently competent
to invest large sums in global commercial distribution. I see no reason to scrap what was said to be a satisfying, somewhat
expensive product, only because it no longer works properly.
The essence of DIY is, of course, to provide that fix for yourself even though depending on others for the methodology.
That's good that the speakers are OK
Well I must admit this doesn't fit the normal fault patterns seen. My instincts at the moment say this is still a problem with those compound output devices. If one of the internal resistors went OC you could get weird effects.
I can see there is doubt over the bias pot. Whether faulty or not, with C47 shorted the amp should work but lets try and prove that. If you short C15 on the good channel, is the sound still OK (there will be some distortion at low levels but it should not be terrible)
The constant current sink of the driver stage should be checked as well. Measure and compare readings between channels of the voltage across R55, R56 and D9.
D9 should have around 3.3 volts across it and R55 around 2.6 volts. R56 I would guess would have around 0.4 of a volt (but compare with the other channel)
Not sure if this has been mentioned before but also check that R52 and R69 (the 0.33 ohms) have not gone high in value. Check with the amp OFF of course. Also check that with the amp ON, that the correct DC voltages (approx. plus and minus 42 ish volts) are present on the left hand sides of R96 and R98. First rule... always check the supplies are correct
Well I must admit this doesn't fit the normal fault patterns seen. My instincts at the moment say this is still a problem with those compound output devices. If one of the internal resistors went OC you could get weird effects.
I can see there is doubt over the bias pot. Whether faulty or not, with C47 shorted the amp should work but lets try and prove that. If you short C15 on the good channel, is the sound still OK (there will be some distortion at low levels but it should not be terrible)
The constant current sink of the driver stage should be checked as well. Measure and compare readings between channels of the voltage across R55, R56 and D9.
D9 should have around 3.3 volts across it and R55 around 2.6 volts. R56 I would guess would have around 0.4 of a volt (but compare with the other channel)
Not sure if this has been mentioned before but also check that R52 and R69 (the 0.33 ohms) have not gone high in value. Check with the amp OFF of course. Also check that with the amp ON, that the correct DC voltages (approx. plus and minus 42 ish volts) are present on the left hand sides of R96 and R98. First rule... always check the supplies are correct
This doesn't surprise me re. being made in China - have gone for a Denon this time around, not quite as pretty but seems a lot better build quality on the face of it (not sure how the internal components compare).
Thank you, will get a replacement ordered based on yours and jaycee's advice earlier. Will do some more research before i go ahead to make sure i do it correctly.
Thank you for your support
There is a sense of satisfaction when you can get something working yourself so i am eager to continue with the repair and hopefully, some day i will be able to provide help to others.Hello again
. I think i must have 2 separate problems - first being the pot which is causing the over-temp and an issue with the distortion - its just seems strange that the audio was working fine until i adjusted the pot (bar the occasional cutting out of the left channel when it got hot). So maybe they are related, maybe i am just thinking wishfully .The constant current sink of the driver stage should be checked as well. Measure and compare readings between channels of the voltage across R55, R56 and D9.
D9 should have around 3.3 volts across it and R55 around 2.6 volts. R56 I would guess would have around 0.4 of a volt (but compare with the other channel)
Not sure if this has been mentioned before but also check that R52 and R69 (the 0.33 ohms) have not gone high in value. Check with the amp OFF of course. Also check that with the amp ON, that the correct DC voltages (approx. plus and minus 42 ish volts) are present on the left hand sides of R96 and R98. First rule... always check the supplies are correct
Will get the results and post back. Thank you again. Everyone here has been fantastic.
Thought i would also quickly upload a recording of the audio changing from right channel to the left channel. Same files just of different sites if anyone has a preference.
Upload Files, MP3, Music. Free File Hosting To Share Files
http://www.filedropper.com/ca640-distortion
Think this post is long enough now so time I actually get testing things again
. Thank you. Nick
The increased temperature could have been caused by a faulty pot, that then, over time, caused something else to fail because of the excessive heat. This could be a pot, a capacitor or even a dry solder joint etc that gets worse from larger expansion/contraction forces experienced from the increased temperature.
I too am skeptical towards the bias pot causing the distortion, as shorted out, it isn't part of the circuit any more, but we shall see. If it is a bad solder joint, replacing the pot might fix it, if only because everything got moved around a bit physically. You can always try tapping components with a plastic pen, when music is playing, and see if that makes any difference.
I too am skeptical towards the bias pot causing the distortion, as shorted out, it isn't part of the circuit any more, but we shall see. If it is a bad solder joint, replacing the pot might fix it, if only because everything got moved around a bit physically. You can always try tapping components with a plastic pen, when music is playing, and see if that makes any difference.
Haha, i have already tried this method last night
- not wanting to damage anything further though i just gave various components a slight wiggle with the back end of a pencil - i assumed that if it was a bad solder joint, i might get something cutting out.I think your theory regarding the overheating causing a secondary problem is quite plausible and i understand that bypassing the pot should get it working providing this is the sole problem - as it 100% fixes the heat problem, i can safely assume that the bypass is working and therefore i require a new pot. But yes, you are completely right in that there must be a secondary issue somewhere.
Is there any other tools that i could buy to help test this further or is just a matter now of comparing left vs right channel voltages until i stumble upon something that doesn't quite look right?
Here are some further figures;
R55 : 2.69v
R56 : 0.42v
D9 : 3.25v
R96 : 2.74v
R98 : 2.3v
R52 : 0v
R69 : 0v
Mooly - i am impressed, the figures are as near as identical to what you mentioned - i assume these are relatively universal figures?
The other channel show the exact voltages (give or take a few mv).
Please could i ask you to clarify the following;
"Also check that with the amp ON, that the correct DC voltages (approx. plus and minus 42 ish volts) are present on the left hand sides of R96 and R98"
I wasn't quite sure what was meant by measuring the left hand side of R96 and R98 so i measured across each one individually but as the above results show, neither are anywhere near 42v so i have come to the conclusion i must have done this bit wrong
.Thank you
The distortion recording is strange in that it cycles between good and bad... very odd.
The DC voltages are fine (which makes this all the more unusual). The voltages are pretty easy to calculate given that the amp is made up of "basic building blocks".
Voltages on R96 and 98 will be correct, those readings sound like the voltage across them. I really meant the voltage measured from ground to those resistors. There will be full supply on one end (R/H) and slightly reduced supply on the other end of each.
So where does that leave us ? Ideally a scope should be used to just confirm that there is nothing really weird going on but without we have to guess a little bit.
I still think the output devices could be suspect and should be replaced, possibly along with Q32 and Q25.
Did R52 and R69 read OK on the ohms range on your meter ? (with the amp OFF)
The DC voltages are fine (which makes this all the more unusual). The voltages are pretty easy to calculate given that the amp is made up of "basic building blocks".
Voltages on R96 and 98 will be correct, those readings sound like the voltage across them. I really meant the voltage measured from ground to those resistors. There will be full supply on one end (R/H) and slightly reduced supply on the other end of each.
So where does that leave us ? Ideally a scope should be used to just confirm that there is nothing really weird going on but without we have to guess a little bit.
I still think the output devices could be suspect and should be replaced, possibly along with Q32 and Q25.
Did R52 and R69 read OK on the ohms range on your meter ? (with the amp OFF)
Sorry, think I should clarify, this was me changing the balance dial on the front of the amp between the working right channel and the faulty left channel to demonstrate the issue
sorry if this caused confusion.& sorry again, that's my lack of knowledge/understanding here giving you the wrong information. Just got back from work so will go test again
. I will post back asap.Well this will teach me to work with electronics when I am tired - I think I just well and truly killed it (I hope not but feel for the worst). Whilst trying to measure R96 to Ground i *think* I have bridged R96 and LK10 which resulted in a pop and the circuit protection kicking in. I disconnected power, powered back on and it instant loops into protection mode and now getting a huge amount of heat to the smaller heatsink to the left near C58 and C57. Unplugged immediately after I noticed (5/6 seconds) but made some very strange noises through the speakers during this.
So, clearly I have made things a lot worst and now in a position where I can test very little which i feel really bad for as everyone has spent a lot of time with me which i now feel i have wasted.
What is the likelihood this can be 'easily' solved and get me back to the previous state?
There are no obvious components blown on the board but i assume i have blown a resistor somewhere close to either R96 or LK10.
So, clearly I have made things a lot worst and now in a position where I can test very little which i feel really bad for as everyone has spent a lot of time with me which i now feel i have wasted.
What is the likelihood this can be 'easily' solved and get me back to the previous state?
There are no obvious components blown on the board but i assume i have blown a resistor somewhere close to either R96 or LK10.
Do not worry! you are not the first one that made this mistake
, I know for sure many people that are "technicians" have made mistakes like this. so do not feel bad about it. What I would do know since now it is more difficult to fix it for you, I would use it for parts and I would "build" one from scratch like something not too complicated like a Chipamp LM3875/LM3886 or another amp kits. it sounds complicated, but if so many newbies have done it...so can you!!!
Oh heck
This doesn't sound promising. LK10 looks like it goes to the PIC (the microprocessor or uP) so applying -45 volts to a PIC input is probably going to zap it.
To prove it all would involve making the output stages "safe" such that nothing gets hot and then seeing if all the input and logic works OK. Possibly also forcing the "protect" data lines to some known logic level to fool the uP that all is OK.
The power amps can be fixed easily no matter what you have done... its the other damage to the processing circuitry that will be the problem. The uP would have to come from Cambridge as it is programmed specifically for this model.
All that is assuming the worst of course...
This doesn't sound promising. LK10 looks like it goes to the PIC (the microprocessor or uP) so applying -45 volts to a PIC input is probably going to zap it.
To prove it all would involve making the output stages "safe" such that nothing gets hot and then seeing if all the input and logic works OK. Possibly also forcing the "protect" data lines to some known logic level to fool the uP that all is OK.
The power amps can be fixed easily no matter what you have done... its the other damage to the processing circuitry that will be the problem. The uP would have to come from Cambridge as it is programmed specifically for this model.
All that is assuming the worst of course...
Hi nickalmond, Try this step to see if pot and output devices defective.
Disconnect power in right channel +v -v
1. pull out the pot from working channel and put it to left channel. Then test.
2. pull out the output transistor from working put it to left channel. Then test.
3. pull out the driver. same procedure 1 and 2.
Regards,
Boyet
Disconnect power in right channel +v -v
1. pull out the pot from working channel and put it to left channel. Then test.
2. pull out the output transistor from working put it to left channel. Then test.
3. pull out the driver. same procedure 1 and 2.
Regards,
Boyet
From the schematic, the Miller input inclusive compensation loop is unstable; I have simulated it. The amplifier is probably oscillating. Change C8 to 2.2nF and R103 to 10Ohms; also reduce C9 to 22p; make this change in both channels.
Never buy or build an amp. with those wretched Thermal trak transistors.
Last edited:
Why, they are an excellent idea. I was looking at Cyrus schematics with triple follower outputs, so 6 Vbe drops to compensate. There is no way a transistor on the heatsink sensor is going to be able accurately compensate thatNever buy or build an amp. with those wretched Thermal trak transistors.
Here is a link to a transcription of Mission Cyrus1, Cyrus 2 is similar
http://sound.au.com/tcaas/cyrus1pwr.gif
Here is a link to Tiefbassuebertr's PDF transcription of Mission Cyrus 3 (quite different)
http://www.diyaudio.com/forums/soli...hematics-later-models-wanted.html#post1998327
Schematics for Cyrus models 5 onward are a tall order. They are not freely available on the net.
Also note that thermal coupling between Vbe multiplier and output stage is only by close proximity to the heatsink.
There is no attachment of the TO92 part to anything on 1&2. See service manuals.
http://sound.au.com/tcaas/cyrus1pwr.gif
Here is a link to Tiefbassuebertr's PDF transcription of Mission Cyrus 3 (quite different)
http://www.diyaudio.com/forums/soli...hematics-later-models-wanted.html#post1998327
Schematics for Cyrus models 5 onward are a tall order. They are not freely available on the net.
Also note that thermal coupling between Vbe multiplier and output stage is only by close proximity to the heatsink.
There is no attachment of the TO92 part to anything on 1&2. See service manuals.
Last edited:
At ~ 15mA bias current as per setup instructions, it is also problematic what the classification means in the Quasi-complementary output stage. Standard double EF/CFP quasis (Naim etc.) have often been arbitrarily biased to the EF (25 mA) level, yet this is a triple.
When I first saw the general quasi triple design in Elektor ca. 1980, it had a fixed Vbe mult. and a range of bias current from 25...100mA with only a vague hint about which part of range might give best results
When I first saw the general quasi triple design in Elektor ca. 1980, it had a fixed Vbe mult. and a range of bias current from 25...100mA with only a vague hint about which part of range might give best results
I have seen tests on the Naim configuration and the value chosen is not the theoretical ideal, but is a sensible compromise to avoid both Class B and run away.
The triple is harder to stabilise as the 3 stages are all at different die temperatures and with very different thermal time constants. There is no way that the O/P device current is going to STAY at 15 mA
The triple is harder to stabilise as the 3 stages are all at different die temperatures and with very different thermal time constants. There is no way that the O/P device current is going to STAY at 15 mA
Agreed. If you consider there are five forward junctions requiring Vbias ~ 3.3V, 3 devices are virtually in class A with constant dissipation and only one OP transistor and the CFP driver actually require dynamic compensation .
I would think the tempco required should allow much the same V/degree range of variation in the triple as the basic (4 transistor) quasi OP stage requires. It makes quite an interesting jumble - I wonder if anyone has cracked it with a more dependable and faster direct attachment of the sense transistor, diode etc.?
I would think the tempco required should allow much the same V/degree range of variation in the triple as the basic (4 transistor) quasi OP stage requires. It makes quite an interesting jumble - I wonder if anyone has cracked it with a more dependable and faster direct attachment of the sense transistor, diode etc.?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.