Hi everyone,
I built my first GC, based on the LM3875T, and verified everything, but when I plugged it in with no load, and let it sit for 2 minutes, the 1000 uF caps blew! The fuses didn't go off though..
Can someone shed a light as to what I did wrong?
The 1000 uF caps were NTE, rated at 35V at 85 degrees.
Basically, here's what I did:
pin 1 -> + of the bridge (full wave, 100PRV 30 Amp), and + of the 1000uF cap. Then from the - of the cap to the ground.
pin 3 -> 20K into pin 8, also to speaker +
pin 4 -> - of the bridge, and - of the 1000uF cap. Then from the + of the cap to the ground.
pin 7 -> 1k resistor -> variable pot (50 K) -> input
pin 8 -> (20 K from pin 3), then to 1k, then to 22uF, then to ground.
Single heat sink for both chips, but not grounded. The transformer/bridge make 16.53Volts (plitron dual 15V) that I verified before hooking everything up.
Three of the the 4 1000uF caps blew up at the same time. My circuit must be screwed up somewhere.
BTW, do you think the chips are toast?
Thanks for the help!
Jam
I built my first GC, based on the LM3875T, and verified everything, but when I plugged it in with no load, and let it sit for 2 minutes, the 1000 uF caps blew! The fuses didn't go off though..
Can someone shed a light as to what I did wrong?
The 1000 uF caps were NTE, rated at 35V at 85 degrees.
Basically, here's what I did:
pin 1 -> + of the bridge (full wave, 100PRV 30 Amp), and + of the 1000uF cap. Then from the - of the cap to the ground.
pin 3 -> 20K into pin 8, also to speaker +
pin 4 -> - of the bridge, and - of the 1000uF cap. Then from the + of the cap to the ground.
pin 7 -> 1k resistor -> variable pot (50 K) -> input
pin 8 -> (20 K from pin 3), then to 1k, then to 22uF, then to ground.
Single heat sink for both chips, but not grounded. The transformer/bridge make 16.53Volts (plitron dual 15V) that I verified before hooking everything up.
Three of the the 4 1000uF caps blew up at the same time. My circuit must be screwed up somewhere.
BTW, do you think the chips are toast?
Thanks for the help!
Jam
Attachments
ya the same happened to me. I had 4800uf on the ps board of a 3886 amp, and a 330uf on the actual board. They were all rated at 35V, and none on the ps board blew, but the second i plugged it in, the lights dimmed, and the 330s started spraying out this horrible smelling smoke. I think they were in backwards.
-Mike
-Mike
My first thoughts were reversed caps too. It wasn't clearly marked on these caps, but they had a black band on one side, I assumed that was the positive (no other markings of any sorts).
Everything else looks right to everyone? In particular, was it ok to have the uninsulated LM chips on the same heat sink if it's not grounded? Or the mica is a must?
It broke my heart that it blew, the housing came out really nice.
Thanks much.
Everything else looks right to everyone? In particular, was it ok to have the uninsulated LM chips on the same heat sink if it's not grounded? Or the mica is a must?
It broke my heart that it blew, the housing came out really nice.
Thanks much.
Hi Jamh;
>>>...In particular, was it ok to have the uninsulated LM chips on the same heat sink if it's not grounded? Or the mica is a must?...<<<
I'm just guessing here, but could it be that the amp went into ultrasonic oscillation because of this? In which case, you'd have lots of current draw, even if it was not receiving an input signal.
If you did not allow a return path directly to the PS trafo, then the PS caps are the only return path - and all the extra current of oscillation might have been enough to smoke 'em. I made the same mistake with a virtual ground on my first GC - it worked fine, but as was pointed out by another contributor the PS caps were the return path for all current drawn, which offers some real possibilities for disaster.
Good luck and all the best,
Morse
>>>...In particular, was it ok to have the uninsulated LM chips on the same heat sink if it's not grounded? Or the mica is a must?...<<<
I'm just guessing here, but could it be that the amp went into ultrasonic oscillation because of this? In which case, you'd have lots of current draw, even if it was not receiving an input signal.
If you did not allow a return path directly to the PS trafo, then the PS caps are the only return path - and all the extra current of oscillation might have been enough to smoke 'em. I made the same mistake with a virtual ground on my first GC - it worked fine, but as was pointed out by another contributor the PS caps were the return path for all current drawn, which offers some real possibilities for disaster.
Good luck and all the best,
Morse
Ok, the problem is definitely with the heat sinks, as I double verified the caps and all. So here are some questions:
1. When you put the mica behind the chips, won't the screw make contact with the heat sink and the chip anyways? Or you have to put 2 layers, one in front, one in back, and make sure the screw doesn't touch the sides? That seems pretty hard.
2. How thick should it be, and where can I get them.
3. What is the purpose of the LM uninsulated chips if they're not meant to be used with ungrounded heat sinks?
Again, thank you all for helping out.
Jam
1. When you put the mica behind the chips, won't the screw make contact with the heat sink and the chip anyways? Or you have to put 2 layers, one in front, one in back, and make sure the screw doesn't touch the sides? That seems pretty hard.
2. How thick should it be, and where can I get them.
3. What is the purpose of the LM uninsulated chips if they're not meant to be used with ungrounded heat sinks?
Again, thank you all for helping out.
Jam
Hi Jamh;
>>>...When you put the mica behind the chips, won't the screw make contact with the heat sink and the chip anyways?...<<<
2 ways to do it.
One is to use nylon screws (usually #6) and nuts to attach the chip to the heatsink. You can find nylon screws and nuts at most hardware stores.
The other is to use a small nylon shoulder washer (usually a #4) to insulate the nut and screw from the chip. You can get the nylon shoulder washers and mounting hardware at Radio Shack. It's about $1.69USD for a TO220 mounting kit that has everything you need for a single TO220 (you may need a larger mica spacer for your chip, depending, but if you already have the right size mica spacer, just save the one in the RS kit).
Good luck and all the best!
Morse
>>>...When you put the mica behind the chips, won't the screw make contact with the heat sink and the chip anyways?...<<<
2 ways to do it.
One is to use nylon screws (usually #6) and nuts to attach the chip to the heatsink. You can find nylon screws and nuts at most hardware stores.
The other is to use a small nylon shoulder washer (usually a #4) to insulate the nut and screw from the chip. You can get the nylon shoulder washers and mounting hardware at Radio Shack. It's about $1.69USD for a TO220 mounting kit that has everything you need for a single TO220 (you may need a larger mica spacer for your chip, depending, but if you already have the right size mica spacer, just save the one in the RS kit).
Good luck and all the best!
Morse
digi01 said:
those are the exact values the datasheet gives.
as for the Non-Insulated package, the resaon it's available is because it has 1C/W better thermal dissipation allowing for much more efficient cooling. the TF packages all always rated for about 5-10% less output power as the cooling will always be that little bit worse. granted you have to add an insulating pad which coud negate the benefits but if you use a good type of insulator it should still be better then the TF package
cooling will always depend on the weakest link, if you can avoid adding an extra barrier to efficient heat transfer then why not
p.s regarding the screw insulation, normally one would use plastic insulating bushes. they're t shaped washers i.e a washer where the centre hole is lined and exteneds past the rest of it. alternately, you can cut the top off a mica (carefully) or silicon insulating pad and use that under the screw - the plastic washer is of course a better idea though
I originally thought about oscillation too, not necessarily from being on the same heatsink. But the chips would almost certainly get hot and go into protection before the caps actually blow up - these chips don't switch fast enough to draw significant current without dissipating huge amounts of power.
There is of course another more obvious possibility...
There is of course another more obvious possibility...
adx said:
Jamh,
Sorry, I might have confused myself there. I (mis)read it that you have been assuming the black band was positive all along and hadn't tried anything else (the "bar" always indicates negative on aluminium electrolytic caps).
The blowup still seems consistent with caps round backwards, and it's not unknown for NTE to get things wrong. You could take your remaining cap and test it - it should act like a sort of a diode (conduct very little current when hooked up right, and quite a lot when reversed). Or try briefly charging it up - round the right way it should hold the same voltage for minutes. You can run the amp without filter caps (or just with very small ones) just to see what's going on. You could also try running the amp without heatsinks, just very briefly at first until you can verify nothing is getting hot. The chips will get quite hot if run without heatsinks for a long time.
Sorry, I might have confused myself there. I (mis)read it that you have been assuming the black band was positive all along and hadn't tried anything else (the "bar" always indicates negative on aluminium electrolytic caps).
The blowup still seems consistent with caps round backwards, and it's not unknown for NTE to get things wrong. You could take your remaining cap and test it - it should act like a sort of a diode (conduct very little current when hooked up right, and quite a lot when reversed). Or try briefly charging it up - round the right way it should hold the same voltage for minutes. You can run the amp without filter caps (or just with very small ones) just to see what's going on. You could also try running the amp without heatsinks, just very briefly at first until you can verify nothing is getting hot. The chips will get quite hot if run without heatsinks for a long time.
As for the values I used, I just wanted to stick as much as possible to the specifications from National Semiconductor. So the values are mostly from their product sheet:
http://www.national.com/ds/LM/LM3875.pdf
And the parts I used are mostly generic. I mostly wanted to learn, and to build, rather than going for the perfect sound.
Jam
http://www.national.com/ds/LM/LM3875.pdf
And the parts I used are mostly generic. I mostly wanted to learn, and to build, rather than going for the perfect sound.
Jam
Nuuk said:
www.soundnerd.com/brain
yes, I experimented with different values, and finally settled on 100K. not too much where the amp picks up line noise, but not too little where the amp's output is distorted caused by distortion caused by the input device (my case, a cd player).'
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.