MJL21193 said:
I think you misinterpreted, slightly, what I was trying to say, not that it really matters.
Yes that circuit is the exact "print screen" if you like of what I have setup in Circuit maker, but it doesn't work in there. That's what I was trying to say before, it's impossible for me to have made an error as the schematic you used is the one the program is using for its simulations.
I have a copy of this somewhere, I think it's rather ancient, but I could try using that. However if you simulated and it worked, I have no reason to doubt. Slone uses the electronics workbench products himself, so the fact it works for him holds no surprises when it works for you too. I just find it strange when Circuit Maker doesn't seem to like it, and the built version acts odd too.
Hi,
I think this amp could get discussed forever at this rate. To isolate the protection entirely remove D6 and D7. If the problem is corrected it would appear to be the protection operating incorrectly. If the problem remains there is some underlying cause within the amp, either design or layout.
I think this amp could get discussed forever at this rate. To isolate the protection entirely remove D6 and D7. If the problem is corrected it would appear to be the protection operating incorrectly. If the problem remains there is some underlying cause within the amp, either design or layout.
Back in post 32 I said I thought the limiter was coming in early due to the ground referencing of the signal by R21 D5 etc.
Having had chance to look more closely I am wrong on this point, the emmiters of Q9 and 10 being tied to the output stop that happening. I still think C14/15 are compromising the limiter action.
Without test gear to actually see what is happening you have to eliminate by disconnecting it as in my previous post.
Having had chance to look more closely I am wrong on this point, the emmiters of Q9 and 10 being tied to the output stop that happening. I still think C14/15 are compromising the limiter action.
Without test gear to actually see what is happening you have to eliminate by disconnecting it as in my previous post.
Posted together again 
Capacitor type will not effect the operation of this circuit. All apart from C14 and 15 need to be adequately rated voltage wise, C18 needs to be at least 160V AC rated.
If the problem completely vanishes there's only so much you can do without test gear to pin it down. Without blowing it up, what AC voltage can you get across your 4 ohm load with your 55hz sine wave. If you can get 25 volts RMS thats 156 watts RMS.
Capacitor type will not effect the operation of this circuit. All apart from C14 and 15 need to be adequately rated voltage wise, C18 needs to be at least 160V AC rated. If the problem completely vanishes there's only so much you can do without test gear to pin it down. Without blowing it up, what AC voltage can you get across your 4 ohm load with your 55hz sine wave. If you can get 25 volts RMS thats 156 watts RMS.
Hi Jacco,
Nice link to Bob Cordells page. I know what you mean about the high transistion frequency, I use a 100Mhz scope and that sometimes struggles to show weird instability problems. The high current draws a give away though. I wish I could find the reference to the failure mode I mentioned, it simply refered to "A Failure mode" that appears to have been observed in HEXFET's under certain real world conditions, nothing to do with audio anyway, and that seemed similar to secondary breakdown.
What convinced me of the ruggedness of the " good old laterals" was many years ago building one of JLH's designs, his 80 watt MOSFET amp. I did a daft thing, shorted the output to the positive rail, there was a big flash and I thought, there goes a pair of very expensive FET's , but no they survived to fight another day and it went on to give many happy hours listening.
Regards Karl
Nice link to Bob Cordells page. I know what you mean about the high transistion frequency, I use a 100Mhz scope and that sometimes struggles to show weird instability problems. The high current draws a give away though. I wish I could find the reference to the failure mode I mentioned, it simply refered to "A Failure mode" that appears to have been observed in HEXFET's under certain real world conditions, nothing to do with audio anyway, and that seemed similar to secondary breakdown.
What convinced me of the ruggedness of the " good old laterals" was many years ago building one of JLH's designs, his 80 watt MOSFET amp. I did a daft thing, shorted the output to the positive rail, there was a big flash and I thought, there goes a pair of very expensive FET's , but no they survived to fight another day and it went on to give many happy hours listening.
Regards Karl
Interesting.
I can manage, well I'm not entirely sure, but I'll explain that.
First of all, I decided to check a few more of the components to make sure they were the right value, everything checked out, but for two diodes. Now these I couldn't see the value of when soldered in place, I had just assumed they were correct. However they were not, luckily I had replacements that were. D9 and 10 used to be 1n4001 instead of 1n4004.
With D6/7 removed something now has changed and the only thing I can think is responsible for this is the changing of those diodes.
Anyway. I can reach 25 Volts but only momentarily. There is some bizarre effect going on at the moment. It sounds, for lack of a better explanation, like a capacitor is charging, or something is getting hot and subsequently its parameters change.
See if I throw enough at the amp to deliver an instantaneous 25V at 55hz it appears to reproduce this fine, but very quickly you can hear the same telltale premature clipping noise in the background, then very quickly it gets louder and louder and louder then I assume boom, its limiting excursion. What with the power being thrown about I just back off at this point for fear of destroying something. The transition from clean to ugly depends on volume setting.
One thing that is cause for concern I think, is the FETs are getting extremely hot when just passing low voltage sines (we're talking 1Vrms), too hot to touch.
This very much IS the same result I used to get with the old protection circuitry in place, albeit it takes time to kick in. When using lower volumes (what would be 15V) it is quite clear this phenomena IS limiting excursion and relative output power. Albeit without the heat.
I am tempted to switch back to the 1n4001 and see what happens as I have never encountered the amps acting like this with the protection circuitry off. Just yesterday I had it disabled with the 1n4001 and I wasn't getting any clipping at all, just lots of clean sounding power.
I also checked the voltage rails under load and they droop by a small amount but nothing unexpected.
I can manage, well I'm not entirely sure, but I'll explain that.
First of all, I decided to check a few more of the components to make sure they were the right value, everything checked out, but for two diodes. Now these I couldn't see the value of when soldered in place, I had just assumed they were correct. However they were not, luckily I had replacements that were. D9 and 10 used to be 1n4001 instead of 1n4004.
With D6/7 removed something now has changed and the only thing I can think is responsible for this is the changing of those diodes.
Anyway. I can reach 25 Volts but only momentarily. There is some bizarre effect going on at the moment. It sounds, for lack of a better explanation, like a capacitor is charging, or something is getting hot and subsequently its parameters change.
See if I throw enough at the amp to deliver an instantaneous 25V at 55hz it appears to reproduce this fine, but very quickly you can hear the same telltale premature clipping noise in the background, then very quickly it gets louder and louder and louder then I assume boom, its limiting excursion. What with the power being thrown about I just back off at this point for fear of destroying something. The transition from clean to ugly depends on volume setting.
One thing that is cause for concern I think, is the FETs are getting extremely hot when just passing low voltage sines (we're talking 1Vrms), too hot to touch.
This very much IS the same result I used to get with the old protection circuitry in place, albeit it takes time to kick in. When using lower volumes (what would be 15V) it is quite clear this phenomena IS limiting excursion and relative output power. Albeit without the heat.
I am tempted to switch back to the 1n4001 and see what happens as I have never encountered the amps acting like this with the protection circuitry off. Just yesterday I had it disabled with the 1n4001 and I wasn't getting any clipping at all, just lots of clean sounding power.
I also checked the voltage rails under load and they droop by a small amount but nothing unexpected.
D9 and 10 are protection against back EMF from highly reactive loads, and it's questionable whether they serve much purpose under normal use. These can be ommited, and the In4001 is only rated at 50 volts P.I.V. from memory-- not enough.
With D6 and 7 out and D9 and 10 out what A.C. voltage can you get before distortion sets in.
With D6 and 7 out and D9 and 10 out what A.C. voltage can you get before distortion sets in.
Not a great deal at the moment, I guess its the FET self protecting itself from over heating? They are getting very hot, which is strange as they are attached to two rather large sinks each rated at 0.8C/watt, the sinks are still easily touchable, I slowly increased the drive level to let them heat up. I get around 11 Volts.
Something here is not right.
They are running cool if nothing is passing through them.
Something here is not right.
They are running cool if nothing is passing through them.
Are you using proper thermal heat pads for the fet's to conduct the heat to the heatsink. There is also the possibilty the amp is unstable, and without a 'scope it's imposible to see what's going on.
Are the FET's soldered directly to a PCB or are they wired. Either way the resistors that go to the gates need to be as close as poss to the FET for stability. If you can connect an ammeter (on a high range-10 amps) in series with one of the rails and gradually increase the drive any instability MAY show by the current suddenly increasing at a certain level of drive. You also need to be sure that the problem is not the bias voltage increasing for some reason, and it may be worth turning the preset pot for minimum current in the outputs and repeating tests. This will cause some crossover distortion but see if the outputs are much cooler.
Are the FET's soldered directly to a PCB or are they wired. Either way the resistors that go to the gates need to be as close as poss to the FET for stability. If you can connect an ammeter (on a high range-10 amps) in series with one of the rails and gradually increase the drive any instability MAY show by the current suddenly increasing at a certain level of drive. You also need to be sure that the problem is not the bias voltage increasing for some reason, and it may be worth turning the preset pot for minimum current in the outputs and repeating tests. This will cause some crossover distortion but see if the outputs are much cooler.
Mooly said:
Yup, the heatsinks do warm up, but there's just a bit of lag time involved. I get the impression the FETs are getting REALLY hot very quickly, too quickly.
For instability checks im guessing I need wide bandwidth, I can probably rig a scope on the computer, but obviously that has rubbish bandwidth.
Directly soldered.
There is about 2cm of track between them.
I can do this if necessary but it would be rather irritating to implement, so will try that last.
Ill try that now.
Edit - Just FYI, P1 and P2 are directly related in as much that if P1 is set so that the wiper connects directly to the emitter of Q4 I cannot get any bias at all. The further P1 rotates towards Q3 the more bias I can create.
I am not entirely convinced here that the heat is being conducted away quickly enough. If I press my finger to the FET it's too hot to touch, yet if I place my finger on the heatsink directly next to the FET it's a totally different story.
When I purchased the pads, they were sold as being thermal pads providing electrical isolation. I do however have some other pads I purchased a couple of months back, for another amplifier, which are slightly different, I am interested in trying these instead.
http://uk.farnell.com/jsp/search/productdetail.jsp?SKU=681090
Are the other ones I've got, taken directly from my order history with Farnell.
When I purchased the pads, they were sold as being thermal pads providing electrical isolation. I do however have some other pads I purchased a couple of months back, for another amplifier, which are slightly different, I am interested in trying these instead.
http://uk.farnell.com/jsp/search/productdetail.jsp?SKU=681090
Are the other ones I've got, taken directly from my order history with Farnell.
IMO, regardless of how difficult it is, having a current meter in series with the rails is essential. It will tell you instantly when to stop doing what you're doing, and it will also tell you if something is suddenly going amiss at a certain level. Better than nothing is a current meter on the mains feeding the unit.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.