With the input shorted, it seems to go away. The probe still picks up something, but whatever it is seems to be the same as what's picked up if I measure the output (+28v DC / -28v DC / 0v) of the power supply boards, or even just the case / safety ground....
I don't know if it's worth mentioning, but I do have a wireless access point in my room, perhaps something to do with this?
I don't know if it's worth mentioning, but I do have a wireless access point in my room, perhaps something to do with this?
Here's something quite funny (I found it mildly amusing)
If I touch the input wire from the RCA socket to the PCB (which is unshielded), with the input unconnected and not shorted, the wave I was describing earlier becomes much larger in amplitude, to the point where it's almost off the screen at 10mv/cm. If I wave my hand about above said signal wire, it still raises in amplitude, though not as much! Magic hands
With the input shorted it changes slightly. Nothing seems to raise in amplitude, it just remains a fuzzy line.
Is it possible that at least most of the problem here is that I am picking up RF from the unshielded input cable (cat 5 and I twisted the ground around the signal wire)? I use a star ground scheme, with a disconnecting network of a bridge rectifier (wired as on the ESP site, I think both AC terminals joined and both 'DC' terminals joined, but I forget) with a 10R power resistor joining the AC to DC terminals. I haven't got a cap here, though I read it was supposed to act as a path for RF interferance to ground. Would this help, perhaps?
Also, if I disconnect the input wires from the PCB, and hold a 100nF film cap across the +ve and input ground, the 'fuzzy line' becomes smaller.
If I touch the input wire from the RCA socket to the PCB (which is unshielded), with the input unconnected and not shorted, the wave I was describing earlier becomes much larger in amplitude, to the point where it's almost off the screen at 10mv/cm. If I wave my hand about above said signal wire, it still raises in amplitude, though not as much! Magic hands
With the input shorted it changes slightly. Nothing seems to raise in amplitude, it just remains a fuzzy line.
Is it possible that at least most of the problem here is that I am picking up RF from the unshielded input cable (cat 5 and I twisted the ground around the signal wire)? I use a star ground scheme, with a disconnecting network of a bridge rectifier (wired as on the ESP site, I think both AC terminals joined and both 'DC' terminals joined, but I forget) with a 10R power resistor joining the AC to DC terminals. I haven't got a cap here, though I read it was supposed to act as a path for RF interferance to ground. Would this help, perhaps?
Also, if I disconnect the input wires from the PCB, and hold a 100nF film cap across the +ve and input ground, the 'fuzzy line' becomes smaller.
A smaller crappy 10R 1/4W (I think) carbon film resistor across the speaker terminals doesn't seem to get hot at all, either with no input, shorted input, or connected to my hissy preamp with the volume all the way up, at which point the hiss becomes audible, but not playing any music. I expect if I were to play music it would get rather hot, but that's not really the point of the exercise is it?
Ok, I have just been reading on these forums and came across a thread explaining that some noise seen on a oscilloscope may actually be... noise that isn't part of the circuit you're measuring, picked up from somewhere else... maybe noise from the scope itself or whatever.
I'm not sure I did it right, but it was suggested that the probe be 'shorted', then probe something. So what I did, was connect the groun of the probe to the chassis ground of the amp, and probed the crocodile clip. I saw what looked exactly like the noise I was seeing at the power supply outputs. Is it safe to assume then, that this noise isn't actually really there?
The noise I am seeing at the speaker outputs looks rather like an amplified version of this noise though, so maybe it does really exist? Or maybe when I am probing, the noise is getting in to the feedback and being amplified or something?
Mind you, this then raises the question of why one channel is still getting hot while idle. The power supply board it is connected to also gets much hotter than the other, but I have switched the amp boards so that the cool running one is connected instead, and it still gets hotter, and that amp still runs hot.
It's really so confusing...
I'm not sure I did it right, but it was suggested that the probe be 'shorted', then probe something. So what I did, was connect the groun of the probe to the chassis ground of the amp, and probed the crocodile clip. I saw what looked exactly like the noise I was seeing at the power supply outputs. Is it safe to assume then, that this noise isn't actually really there?
The noise I am seeing at the speaker outputs looks rather like an amplified version of this noise though, so maybe it does really exist? Or maybe when I am probing, the noise is getting in to the feedback and being amplified or something?
Mind you, this then raises the question of why one channel is still getting hot while idle. The power supply board it is connected to also gets much hotter than the other, but I have switched the amp boards so that the cool running one is connected instead, and it still gets hotter, and that amp still runs hot.
It's really so confusing...
I might also add that as far as I can tell, the on-chip SPIKE protection has never kicked in due to getting overly hot (either that or I have fake LM3886 chips that actually work rather well?! Got them off ebay, but they look genuine to me). I don't hear any 'clicking' noises from the chip clipping at full volume.
Also, for all bypassing I have used "Polypropylene" (I think EDIT: they're actually "Metallized Polyester" supposedly) caps. Upon reading, perhaps these should actually be changed to Ceramic!
Anyone care to take a guess at whether it all could just be because of the bypass caps that I have used?
Can you tell I am quite new to all of this?
Also, for all bypassing I have used "Polypropylene" (I think EDIT: they're actually "Metallized Polyester" supposedly) caps. Upon reading, perhaps these should actually be changed to Ceramic!
Anyone care to take a guess at whether it all could just be because of the bypass caps that I have used?
Can you tell I am quite new to all of this?
This seems to suggest interference of some sort.markiemrboo said:
What I think you're describing, is normal.
Inductive or electro/magnetic pickup?If I wave my hand about above said signal wire, it still raises in amplitude, though not as much! Magic hands
You're amp is picking up interference beyond the inputs?
That makes sense.
A scope probe can have noise, it can get amplified.
Is your amp in a shielded, non-magnetic enclosure? You might also, just for good measure, check the mains input for RFI filtering (some ferrite and mains rated caps).
Put the lid on the amp, short the inputs, look at the noise from the oscilloscope and then look at the noise from the amp (turned on). Then connect a source (turned off) and use a good interconnect. Look at the output. If there is RF there, you might want to put in an RF filter at your inputs, or improve your cable or your source. You can also turn on your source to see if that is giving more noise.
lndm said:
I've confused myself now. Seems I have said that with the input shorted it disappears, but later say it only changes slightly. I'll have to retest I think.
A good start
It picks up that nasty mobile phone interferance. I was told that I probably need the... Thiel(SP) network or a ferrite bead around the speaker outputs to remove this.
When I "zoom out" as it were (change the horizonal time from say 1us to 5ms), the wave on the output does looks suspiciously similar to that of the wave I was seeing when I probe the probes ground clip, while it's connected to chassis ground. To me it just seems like an amplified version of that pattern.
I do believe, after some reading, that I probably shouldn't have been trying to use a 1X probe to measure this. I figured that it would allow me to "see better", as it's not attenuated, but apparently it does introduce noise and has restricted bandwidth.
I can try and get pictures of the speaker output noise and 'probe shorted' noise with the 1X probe if you like. The quality probably won't be great, but you should be able to see what I mean.
I should note that just connecting the probes ground clip to the probe tip does result in a flat line on the scope. It's only when the probe ground clip is clipped to chassis ground, and I then probe the ground clip that I pick up this noise.
I believe it's steel, but has an aluminium front. If by shielding you mean something like a tinfoil(?) lining, it hasn't got any of that, no. I was originally going to put in a IEC socket ripped out of an old computer PSU in there, which had two blue jacketed ceramic caps from live to safety and neutral to safety. Is this RFI filtering?
Ok, so.. with the amp off.. connect the probe ground to the speaker output -ve, and probe the speaker output +ve, yes? And observe the noise on the screen just from this? Should I have the mains plug on the amp unplugged at this point? This would eliminate any noise which may be coming from anything else?
Rather than connecting a source, should I just short the inputs instead? My current pre-amp (Cambridge Audio A5, using the pre-out) does give noise in the form of a quiet white noise type hiss. I figure measuring with this connected would just make things harder to read because of the hissing?
I have at this point led myself to believe that it's actually not oscillating, it was just the noise from the X1 probe being amplified and it wasn't actually really there. This does leave the heat unexplained.
A possible explanation here might just be the fact that the heatsinks on the two LM3886's are different. Perhaps the one getting very hot just isn't as good. Why does the regulator get hot on one channel and not the other? I have used slightly different thermal insulators here as it goes, the hot one is a thin mica sheet, and the cool one is, well, I don't know what it is. It's some sort of grey dotted sheet, quite thick too. I think tomorrow I may replace this grey stuff with the same mica as on the other channel and see if that starts to run hotter.
The flaw to my current (probably quite idiotic) theory, assuming the amp is running hot because of an inadequate heatsink rather than oscillation, though: why are the regulators still running fairly hot when idle? This would surely point back to oscillation, which is something I can't seem to measure with a 10X probe.
When you put a resistor on the output and it stays cool you can be pretty sure the amp is NOT oscillating. That still leaves us with the problem of why one channel is running hot. I think you better check your wiring for errors.
It is possible you damaged one of the chips when you soldered the whole thing together. If you wiring is good, and the thing isn't oscillating (and it seems it isn't), then the problem is with the chip.
I_F
It is possible you damaged one of the chips when you soldered the whole thing together. If you wiring is good, and the thing isn't oscillating (and it seems it isn't), then the problem is with the chip.
I_F
I_Forgot said:
On the one hand I am glad it sounds as if it's not oscillating, on the other hand i'm annoyed that there still seems to be a problem with heat
As in the previous post, I do have a theory. I will change the insulation on the other regulator board to mica tomorrow and see if that channel starts to heat up the same.
Maybe that I just have rather sensitive hands when it comes to heat or something. I don't know if that's possible. One of those laser type thermometers would probably be very useful indeed around about now!
Anyway, thanks for bearing with me. I'll let you know if the insulation swap does change anything with regards to heat.
When I said interference, that is normal, we all need to keep it out. It can get in through your inputs, or directly to your circuitry if it is not shielded, or through your speaker cables and the feedback loop.markiemrboo said:
BTW, I don't necessarily think this is causing your heating issue, however it appears to be at the forefront obscuring your investigations.
If your brain tells you it is the same signal, it should be the same signal. We humans can recognise these shapes so trust your instincts. That said, there might be more than one source and you can usually find out by changing the time base as this tends to focus the shapes around a band of frequencies and blur the rest (not literally though).
Is your signal ground lifted from the chassis with some components? You should measure from your signal ground and not the chassis.
That sounds satisfactory.
Basically, it ought to be satisfactory. Have you checked your DC rails for cleanliness and compared what you see to the outputs?
You might try this, but might be more interested in what it looks like with the amp on.
You may want to try both. The cable shielding might come under scrutiny this way. Interference from that could be of a greater level than the hiss.
You may need to check your components and joints and you may have a bad chip. (This is not very helpful but it may need to be checked)
lndm said:
Oh right. Well I have heard tinfoil going to ground wrapped around the wire acts as a good shield. Seems kind of difficult to do and quite fragile though. Might give it a go.
As for ferrite bead on the speaker cable, I assume I would be able to put this around the internal speaker cable runs instead of the run from the amp to the speaker (which is what i've heard people saying).
I did change the insulator to mica today but it made no difference... of course
I did attempt some calculations to explain the rather large difference in DC offset between the two chips. The calculations seemed to suggest the... bias current (I think) on the chip with the higher DC offset was about twice that of the other chip. I don't quite know what bias current is yet, but might this mean that one chip would be drawing more power at idle than the other? (apologies for my non understanding & usual guessing)
Ok.
Yes it is, sorry. What I meant was that I was measuring from the star ground side of the disconnecting network, which connects to the chassis. Bad wording
If I understand correctly, yes. The display I was seeing at the output looked very much like an amplified version of the display I saw when measuring the power rails, even ground to the "0v". This is what made me think that it was maybe just external noise / non amp related noise bought in by the 1X probe and amplified.
Ok
I managed to get a temperature probe to *just* reach the amp chip heatsinks. The 'hot' one is 43C, and the 'cold' one is 38C. I can't measure the regulator heatsinks as the probe lead doesn't reach. But by touch, it seems like the difference in temperature is probably about the same as with the chip's.
Perhaps I just shouldn't be worrying
Perhaps I just shouldn't be worrying
Nothing useful to say other than I feel that such a difference is probably not a problem per se, but possibly a mismatch. For one thing, are your chips mounted identically in orientation, componentry, and the amount of thermal grease (not too much)? You mention different capacitors and to a small degree these may affect the level of the rails.
I would probe the components on the boards and compare the voltages between the two in an attempt to find a lead.
I would probe the components on the boards and compare the voltages between the two in an attempt to find a lead.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.