Hey guys, I need some help. I may have a disaster on my hands. After testing the thermal system of a P101 (rod elliot mosfet based amp) for more than 2 hrs and the heatsinks never exceeding 100F, I decided to finally get around to setting the bias as I never formally did. I put some 100Ohm 1/4W carbon (all I had in the way of 100Ohm) resistors in series with the supply lines, turned the amp on, and took some readings. I got around 1/2V across the resistor which wasn't what I expected, and then the resistor (or at least I think it was) smoked a bit. Hastily I turned the amp off, figuring that the 1/4W resistor just wasn't going to cut it being so small and all. I hooked the supply line back up w/out the resistor, hooked some speakers back up, and turned the amp on. I got some oscillation type noise instead of nothing. I hooked up an input and repeated. I got mostly oscillation type noises and in the background a very faint signal that was lacking all bass. Upon further inspection, the negative supply line is supplying 3amps or so to the board but the positive supply line is supplying none, with no signal hooked up. The scope on the output shows no whacky signals from the outputs. I tested the mosfets in the circuit and none are shorted but they do warm up quickly and the power resistors in the circuit next to them get toasty too.
Any ideas what on earth could be happening? I changed nothing except the series resistors and then back again.
Thanks so much guys.
Any ideas what on earth could be happening? I changed nothing except the series resistors and then back again.
Thanks so much guys.
I believe I have figured it out but I am very confused.
It appears as though the parallel R/C combo that connects the signal ground to the power ground was cooked. The 1W resistor appears to be an open. I put in a new resistor of the correct resistance with some alligator clips and all is well again, well except one of the woofers in the speaker I was using it is toast as in voice coil = roasted
The question is, what caused this? I will absolutely need to stop this from happening again and potentially ruining some of my much more precious speakers than my bench testing speakers. Maybe a DC protection circuit is in order?
It appears as though the parallel R/C combo that connects the signal ground to the power ground was cooked. The 1W resistor appears to be an open. I put in a new resistor of the correct resistance with some alligator clips and all is well again, well except one of the woofers in the speaker I was using it is toast as in voice coil = roasted
The question is, what caused this? I will absolutely need to stop this from happening again and potentially ruining some of my much more precious speakers than my bench testing speakers. Maybe a DC protection circuit is in order?
The odd thing is, I didn't see any noticeable DC output from the amp module when I looked at the output on the scope. How much DC does it take to ruin a speaker, electrically or mechanically?
Hi dswiston,
It takes very little DC to destroy a loudspeaker. 10 VDC will do it easily as the voice coil gets pushed out of the gap. Normally the heat gets radiated to the magnet structure and dissipated from there. A little gets cooled if using a vented pole piece. The part of the voice coil outside the gap will be burned much worse. Normally, the glue will fail, the former may warp as well.
So, it looks like you lost your ground reference for the circuit. Make sure the ground return goes to the supply common. There should never be any current flowing through the reference ground connection (for the input stage and input jacks).
I feel it is irresponsible to design or sell an amplifier without DC offset protection. It is a fact of life that devices will fail, that means that some amps will fail with a DC offset across the output terminals.
Try 1 ohm resistors or less next time for bias measurement. Do not hook up speakers until after you measure the amp under test for DC offset.
-Chris
It takes very little DC to destroy a loudspeaker. 10 VDC will do it easily as the voice coil gets pushed out of the gap. Normally the heat gets radiated to the magnet structure and dissipated from there. A little gets cooled if using a vented pole piece. The part of the voice coil outside the gap will be burned much worse. Normally, the glue will fail, the former may warp as well.
So, it looks like you lost your ground reference for the circuit. Make sure the ground return goes to the supply common. There should never be any current flowing through the reference ground connection (for the input stage and input jacks).
I feel it is irresponsible to design or sell an amplifier without DC offset protection. It is a fact of life that devices will fail, that means that some amps will fail with a DC offset across the output terminals.
Try 1 ohm resistors or less next time for bias measurement. Do not hook up speakers until after you measure the amp under test for DC offset.
-Chris
"well except one of the woofers in the speaker I was using it is toast as in voice coil = roasted"
Sounds like we know where that -3 amps was going. A disrupted ground can do all sorts of strange things. Just taking a guess from what you've said, your scope was probably referenced to a point at a different potential then the speaker's ground, so you wouldn't have realized there was a DC offset flowing current through it.
Sounds like we know where that -3 amps was going. A disrupted ground can do all sorts of strange things. Just taking a guess from what you've said, your scope was probably referenced to a point at a different potential then the speaker's ground, so you wouldn't have realized there was a DC offset flowing current through it.
Good to know, I wasn't aware that just 10v DC could do it, never thought about the voice coil moving beyond the magnet like that.
Any ideas why the resistor between signal and power ground went bad?
When speaking of the ground return, are you making reference to the power ground? The power ground is connected directly to the supply common ground. I'm thinking what may have happened is while I was playing around with the wires one of the psu caps discharged somehow by some random event, not sure how.
Any ideas for DC protection outside the amp and inside the speakers/crossover, or a VERY small DC protection circuit? With 10V DC causing problems a simple fuse will not help. I don't have much room left in the amp and did not plan ahead as well as I should have. I do however have 12v DC avaliable in the amp and would not be opposed to using a wallwart to supply a circuit outside the amp with 12v DC.
Thanks so much guys.
Any ideas why the resistor between signal and power ground went bad?
When speaking of the ground return, are you making reference to the power ground? The power ground is connected directly to the supply common ground. I'm thinking what may have happened is while I was playing around with the wires one of the psu caps discharged somehow by some random event, not sure how.
Any ideas for DC protection outside the amp and inside the speakers/crossover, or a VERY small DC protection circuit? With 10V DC causing problems a simple fuse will not help. I don't have much room left in the amp and did not plan ahead as well as I should have. I do however have 12v DC avaliable in the amp and would not be opposed to using a wallwart to supply a circuit outside the amp with 12v DC.
Thanks so much guys.
"Any ideas why the resistor between signal and power ground went bad?"
It's tough to say without actually being there and looking things over. Sometimes it can be as simple as a hidden component defect, or mechanical stress. There maybe some electrical issues during testing also.
As for protecting speakers from offsets, just for bench testing like you were doing it can be as simple as a coupling capacitor between the amp and speaker. For enjoyment listening in the easy chair with the "real" speakers, an active protection scheme is a better way to go. Since you have little room in the chassis, building one in an outboard chassis might be the better way to go. Placing a relay in the wires between your amp and speakers will, in all likelyhood, effect the sound to some extent. With that possibility in mind, it's probably better to have the protection circuit cut-off the AC power to the amp, as opposed to just disconncting the speakers from the amp. You may want to look around on the web for some design ideas.
It's tough to say without actually being there and looking things over. Sometimes it can be as simple as a hidden component defect, or mechanical stress. There maybe some electrical issues during testing also.
As for protecting speakers from offsets, just for bench testing like you were doing it can be as simple as a coupling capacitor between the amp and speaker. For enjoyment listening in the easy chair with the "real" speakers, an active protection scheme is a better way to go. Since you have little room in the chassis, building one in an outboard chassis might be the better way to go. Placing a relay in the wires between your amp and speakers will, in all likelyhood, effect the sound to some extent. With that possibility in mind, it's probably better to have the protection circuit cut-off the AC power to the amp, as opposed to just disconncting the speakers from the amp. You may want to look around on the web for some design ideas.
Well cutting off the AC leaves the caps to discharge, and the caps have enough juice to keep the amp running at decent levels for maybe 10 seconds so that may still not cut it, but I'll search online. I read some mosfet based scheme last night but it didn't make too much sense to me as I was too tired at the time.
BTW, both amp modules (left and right) had the same 10Ohm resistor go to resistor heaven, which is interesting.
BTW, both amp modules (left and right) had the same 10Ohm resistor go to resistor heaven, which is interesting.
Well dswiston,
Many kits and circuits exist for that purpose. Should be easy to find. Some of these are fairly small and 12V is a popular voltage.
As a note, a relay contact will affect the sound less than a fuse element. I wouldn't cut the AC with the relay, but you could shut down the current sources (if you have). That will effectively shut down the amp. If you are worried about damage to the amp, don't. Done is done. There is enough charged energy in the power supply caps normally to finish blowing stuff up.
-Chris
Many kits and circuits exist for that purpose. Should be easy to find. Some of these are fairly small and 12V is a popular voltage.
As a note, a relay contact will affect the sound less than a fuse element. I wouldn't cut the AC with the relay, but you could shut down the current sources (if you have). That will effectively shut down the amp. If you are worried about damage to the amp, don't. Done is done. There is enough charged energy in the power supply caps normally to finish blowing stuff up.
-Chris
Hi dswiston,
Seems I posted while you were at it. Do you have a schematic of your amp to show us?
-Chris
Seems I posted while you were at it. Do you have a schematic of your amp to show us?
-Chris
Well the amp is a Rod Elliot design, the P101. There is a schematic on the net of it, but without values for components (for this project Rod asks that you buy the PCBs to get the info on the component values). Here is a link http://sound.westhost.com/project101.htm
"BTW, both amp modules (left and right) had the same 10Ohm resistor go to resistor heaven, which is interesting."
Since the amp apparently was functioning before the bench work, and both resistors were dead afterwards, it's fairly obvious that some mistake in test equipment configuration occured during testing. You're probably the one best in a position to solve that mystery. About the only comment I might make is that while AC powered meters are isolated (or at least they're supposed to be), most O'scopes are not. The probe gnd leads also connect to the earth ground through the power cord (at least on most scopes, there are exceptions). If the total equipment configuration also has another connection to the earth gnd somewhere else, and the scope gnd is connected to a different potential, it's possible to get a current flowing, sometimes quite high if you try referencing the scope to a power potential. In other words, if your amp's signal gnd were connected to earth gnd through some path, and you tried referencing the scope to some other potential, say to take a reading across a part, you could get a current flowing through the 10 ohm resistors via the earth gnd path. This is just a guess in the dark, who knows ...
As far as the caps knocking out your woofer, that's something that needs to be considered on a case by case basis. A modest sized woofer, or really huge caps (e.g. 100,000uf or so) charged up to high voltages may very much be an issue. In fact if taken to an extreme, a protection circuit may not even be of much help if an output device were to fail. I don't believe you've indicated the size of the caps you're using, the voltage they're operating at, or the capability of your woofers, so it's tough to even make a guess, but interrupting the path leaving the amp to the speakers may in fact be the best way to go for you. I used to do it that way myself until about 10 years ago. At that point I tried changing over to cutting the AC power, and I've had no regrets since.
Since the amp apparently was functioning before the bench work, and both resistors were dead afterwards, it's fairly obvious that some mistake in test equipment configuration occured during testing. You're probably the one best in a position to solve that mystery. About the only comment I might make is that while AC powered meters are isolated (or at least they're supposed to be), most O'scopes are not. The probe gnd leads also connect to the earth ground through the power cord (at least on most scopes, there are exceptions). If the total equipment configuration also has another connection to the earth gnd somewhere else, and the scope gnd is connected to a different potential, it's possible to get a current flowing, sometimes quite high if you try referencing the scope to a power potential. In other words, if your amp's signal gnd were connected to earth gnd through some path, and you tried referencing the scope to some other potential, say to take a reading across a part, you could get a current flowing through the 10 ohm resistors via the earth gnd path. This is just a guess in the dark, who knows ...
As far as the caps knocking out your woofer, that's something that needs to be considered on a case by case basis. A modest sized woofer, or really huge caps (e.g. 100,000uf or so) charged up to high voltages may very much be an issue. In fact if taken to an extreme, a protection circuit may not even be of much help if an output device were to fail. I don't believe you've indicated the size of the caps you're using, the voltage they're operating at, or the capability of your woofers, so it's tough to even make a guess, but interrupting the path leaving the amp to the speakers may in fact be the best way to go for you. I used to do it that way myself until about 10 years ago. At that point I tried changing over to cutting the AC power, and I've had no regrets since.
Hi dswiston,
I am in agreement with hughmon as far as troubleshooting is concerned. The amp used to work. Heed my previous comment regarding current sense resistors, yours were too high.
You can not shut down the channel by turning off Q1 I see, as the output stage would get pulled negative. Not nice.
So, rebuild the amp and test it. Do not set the bias current with speakers attached.
I noticed that the source resistors are wire wound (inductive). The amp may become unstable at some signal / load conditions. Try some non-inductive ones some time (ceramic plate type). Given that the web page says the amp works down to +/- 5VDC, try using a variac to bring up the power.
-Chris
I am in agreement with hughmon as far as troubleshooting is concerned. The amp used to work. Heed my previous comment regarding current sense resistors, yours were too high.
You can not shut down the channel by turning off Q1 I see, as the output stage would get pulled negative. Not nice.
So, rebuild the amp and test it. Do not set the bias current with speakers attached.
I noticed that the source resistors are wire wound (inductive). The amp may become unstable at some signal / load conditions. Try some non-inductive ones some time (ceramic plate type). Given that the web page says the amp works down to +/- 5VDC, try using a variac to bring up the power.
-Chris
Well I bought new resistors today to replace the cooked one and all appears well. DC output is 18mv and 21mv for respective channels.
The power supply has 19,000 uF per rail, and each rail is +/- 42V. Not an enormous level of stored power but enough to worry me.
I have been pondering a simple circuit w/out needing a 12v DC supply that I could use in the speaker/crossover enclosure so as to take the circuit with the speakers (so the speakers are not at the mercy of someone's amp if I every bring them to a friends) and I think I may have come up with one. Someone tell me if this makes any bit of sense.
Using a simple RC circuit tuned appropriately, a mosfet or transistor is hooked up to be biased by the voltage across the cap. The mosfet/transistor source is hookd to the amp output through a quick blow fuse rated at a very low current. The gate of the mosfet/tranny is connected to the amp ground. The rational here is that if enough DC is detected, the mosfet/tranny effectively shorts the amp output and the fuse blows. I know its a "risky" method, but the amp I would use this on is using some very rugged mosfets itself and I am willing to chance the fuses life on the output mosfets life, not to mention the mosfet/tranny responsible for shorting the circuit, of which I care nothing about as it is easily replaceable.
Let me know what you think of the idea, is it plausible or did I overlook something? Otherwise large DC blocking caps paralleled with a few assorted smaller value caps looks like the easiest way to go (but not technically sonically best).
**EDIT** I realize that if DC is on the output, the amp is most likely already broken in some sorts but its not always the output devices that cause the DC as I found out through my problem.
The power supply has 19,000 uF per rail, and each rail is +/- 42V. Not an enormous level of stored power but enough to worry me.
I have been pondering a simple circuit w/out needing a 12v DC supply that I could use in the speaker/crossover enclosure so as to take the circuit with the speakers (so the speakers are not at the mercy of someone's amp if I every bring them to a friends) and I think I may have come up with one. Someone tell me if this makes any bit of sense.
Using a simple RC circuit tuned appropriately, a mosfet or transistor is hooked up to be biased by the voltage across the cap. The mosfet/transistor source is hookd to the amp output through a quick blow fuse rated at a very low current. The gate of the mosfet/tranny is connected to the amp ground. The rational here is that if enough DC is detected, the mosfet/tranny effectively shorts the amp output and the fuse blows. I know its a "risky" method, but the amp I would use this on is using some very rugged mosfets itself and I am willing to chance the fuses life on the output mosfets life, not to mention the mosfet/tranny responsible for shorting the circuit, of which I care nothing about as it is easily replaceable.
Let me know what you think of the idea, is it plausible or did I overlook something? Otherwise large DC blocking caps paralleled with a few assorted smaller value caps looks like the easiest way to go (but not technically sonically best).
**EDIT** I realize that if DC is on the output, the amp is most likely already broken in some sorts but its not always the output devices that cause the DC as I found out through my problem.
Oh for the love of me, scratch that idea. It makes no sense. I should stop staying up so late and conjuring up ideas that, to the normal non-sleep comatose person, obviously wouldn't work yet seem so obvious and simple to me. If the fuse was moved to the positive/negative rail and before/after the mosfet/tranny and changed to a current value higher than one would ever expect to see in normal listening conditions this might be a different story.
I'll check back in this thread tomorrow when I can think.
I'll check back in this thread tomorrow when I can think.
- Status
- Not open for further replies.