| Lucas_G |
Hi,
I had just finished my LM3886 chipamp using the pcbs of chipamp.com
I thought this was an easy job compared to other projects that I finished succesfully.
However, when I switsched on power there was a huge flash at the LM3886 chip, followed by some hefty smoke.
I checked the LM3886 chip and pin 4 an 5 (looking at the front from left to right) seem to have had the shortage.
I doublechecked the connections, and also the resistorvalues, but I still have no clue. I hope someone can help me in tracing the problem.
I enclose some pictures of the boards.
There are some things I thought of:
1. I did not attach the CHG from the ampboard. It should eventually be connected to the chassis, but for now I had everything running on a piece of wood.
2. I used Welwyn resistors at the amp board. They are a bit longer so I had to mount them a bit upward. I checked whether they made a shortage but this is not the case. However they present a longer path. Could it be oscillation? Could oscillation bring the hole thing to shortage in such a short moment?
3. Could it be that the LM3886 is faulty? Has anyone encountered such failures?
I am still reluctant to check whether the second mono amp is alright. I would like to pin the problem down in order to avoid the chance of frying my second LM3886...
Any suggestions are highly appreciated.
Lucas |
|
|
| Nordic |
So you connected the + and - rails but not the ground rail to the PSU?
If so, thats your problem right there.... connect the PSU ground the the amp board's ground. Then use an earth decoupling network between the PSU ground and the case.wich in turns goes to mains earth if you have 3 prong plugs. |
|
|
| Lucas_G |
| quote: | Originally posted by Nordic
So you connected the + and - rails but not the ground rail to the PSU?
If so, thats your problem right there.... connect the PSU ground the the amp board's ground. Then use an earth decoupling network between the PSU ground and the case.wich in turns goes to mains earth if you have 3 prong plugs. |
Hi Nordic,
Thanks for looking.
I checked this too.
Indeed I did not connect the CHG to chassis (since I still don't have a chassis)
However, from the PSU board there are two PG (power ground) connections that meet at the ampboard. So there is a positve voltage, a common ground and a negative voltage supplied to the amp. This is alright, isn't it?
I am still wondering whether oscillation/instabillity might cause this.
I opted for the standard configuration with just R3 (680 Ohm) and no Ci (47 uF) feedback capacitor as described on page 8 of this pdf: http://www.chipamp.com/docs/lm3886-manual.pdf
Regards,
Lucas |
|
|
| sayang001 |
| quote: | | However, from the PSU board there are two PG (power ground) connections that meet at the ampboard. So there is a positve voltage, a common ground and a negative voltage supplied to the amp. This is alright, isn't it? |
Yes it is alright. What are your voltages ? |
|
|
| Nordic |
I don't THINK oscillation would kill off straight on the flip of a switch like that....
If the power rails shorted with either the input or outpu pins it would kill it like that right away, also if you connected power the wrong way around (I have certainly done that... nowadays I keep an LED in series with a 20K resistor between the positive and negative rail to show properly working PSU. |
|
|
| Lucas_G |
| quote: | Originally posted by sayang001
Yes it is alright. What are your voltages ? |
Hi Sayang,
I will check them tomorrow.
As you can see on the photos, I was so sure that nothing could go wrong with this "simple" propject, that I connected the the PSU and Ampboard with soldered wires, and no connectors in between. So I will have to deconnect them tomorrow.
I did check the other (100% identical) PSU board. I measured 25 Volt AC and 34 Volt DC without any load. I guess that is alright?
Regards,
Lucas |
|
|
| Lucas_G |
| quote: | Originally posted by Nordic
I don't THINK oscillation would kill off straight on the flip of a switch like that....
If the power rails shorted with either the input or outpu pins it would kill it like that right away, also if you connected power the wrong way around (I have certainly done that... nowadays I keep an LED in series with a 20K resistor between the positive and negative rail to show properly working PSU. |
Hi Nordic,
Yes I know. We all make such ridiculous mistakes from time to time. With this board design, things are very straightforward. All the connections on both boards are facing each other...
Still no clue, but as I said, I will check voltages of THIS Psu tomorrow...
Regards,
Lucas |
|
|
| peranders |
| Do you have any fuses? If yes, which values? |
|
|
| Lucas_G |
| quote: | Originally posted by peranders
Do you have any fuses? If yes, which values? |
Hi Peranders,
I don't have any fuses other than at the iec 230Volt inlet.
I know that some people like to add fuses at their DC rails, and maybe it would have been a life saver for my LM3886 chip, but I also know that many people think they are too slow to be a lifesaver for chips... Do you think I should add one?
Today I disconnected the PSU and Amp board and measured the voltages on the particular PSU that was connected with the blown LM3886. I measured 25 AC and 33 DC on the rails. Polarities were all OK.
So it remains a mystery to me. I doublechecked all soldering points, values etc.
There remain two possibilities:
1. Might it still be oscillations that killed the chip so fast? I did not install the feedback capacitor, only the feedback resistor (22,1K) and its loop is a bit long-legged...
2. Are there any fake LM3886 around? I bought these from some source in Asia (don't remember which) and it is labelled PM14BV above the LM3886TF inscription. Or did I just catch a faulty one?
If I do not find the clue to this, I guess I will have to try the second ampboard and see whether it will smoke too. If I would insert fuses, what values would do?
Regards,
Lucas
P.S. I hope BrianGT or Peter Daniel might look into this, since both must have a lot of experience with these pcbs and LM3886 chips... |
|
|
| sayang001 |
| quote: | | Are there any fake LM3886 around? I bought these from some source in Asia (don't remember which) and it is labelled PM14BV above the LM3886TF inscription. Or did I just catch a faulty one? |
Can you make a clear picture of the front of the chip. I have experience with fake transistors from Asia ....
http://www.sayang.demon.nl/projects/sanken/sanken.html
I have some real (directly from National) LM3886TF as well to compare yours with.
http://www.sayang.demon.nl/projects...y/kimberly.html
| quote: | | If I do not find the clue to this, I guess I will have to try the second ampboard and see whether it will smoke too. If I would insert fuses, what values would do? |
Use small 24V 2w bulbs in series with + and - rails, in case of high current these will light up and limit the current. LM3886 with no load will not light the bulbs. I used this when testing with Kimberly her amplifier.
|
|
|
| Lucas_G |
| quote: | Originally posted by sayang001
Can you make a clear picture of the front of the chip. I have experience with fake transistors from Asia ....
http://www.sayang.demon.nl/projects/sanken/sanken.html
I have some real (directly from National) LM3886TF as well to compare yours with.
http://www.sayang.demon.nl/projects...y/kimberly.html
Use small 24V 2w bulbs in series with + and - rails, in case of high current these will light up and limit the current. LM3886 with no load will not light the bulbs. I used this when testing with Kimberly her amplifier.
|
Hi Sayang,
I cannot come that close wityh my camera to the chipamp.
I enclose one photo from above and one made with my scanner of the backside of the chip where one can see the damage.
Regards,
Lucas |
|
|
| Lucas_G |
| Picture of backside of LM3886 |
|
|
| sayang001 |
| quote: | I cannot come that close wityh my camera to the chipamp.
|
Desolder the chip from the pcb and put it on the scanner with it's front, it is broken anyway :smash: |
|
|
| MikeB |
| quote: | Originally posted by Lucas_G
1. Might it still be oscillations that killed the chip so fast? I did not install the feedback capacitor, only the feedback resistor (22,1K) and its loop is a bit long-legged...
|
OMG, i am not familiar with the board, but it looks like you skipped the dc-blocking cap instead of shorting it. This makes the amp unity gain and oscillate like hell.
:RIP:
Why do you simply skip parts ? :scratch:
Mike |
|
|
| Lucas_G |
| quote: | Originally posted by sayang001
Desolder the chip from the pcb and put it on the scanner with it's front, it is broken anyway :smash: |
You are right. I will do this this evening... |
|
|
| Lucas_G |
| quote: | Originally posted by MikeB
OMG, i am not familiar with the board, but it looks like you skipped the dc-blocking cap instead of shorting it. This makes the amp unity gain and oscillate like hell.
:RIP:
Why do you simply skip parts ? :scratch:
Mike |
Hi Mike,
The reason can be found here:
I opted for the standard configuration with just R3 (680 Ohm) and no Ci (47 uF) feedback capacitor as described on page 8 of this pdf: http://www.chipamp.com/docs/lm3886-manual.pdf
I checked the pcb. The 22,1 kOhm resistor is connected!
Regards,
Lucas |
|
|
| MikeB |
| quote: | Originally posted by Lucas_G
Hi Mike,
The reason can be found here:
I opted for the standard configuration with just R3 (680 Ohm) and no Ci (47 uF) feedback capacitor as described on page 8 of this pdf: http://www.chipamp.com/docs/lm3886-manual.pdf
I checked the pcb. The 22,1 kOhm resistor is connected!
Regards,
Lucas |
Sorry Lucas, my fault. I should not post not knowing the pcb...
But, you might check the soldering next to r3, the "SG" point. It looks like the pad on the bottom is broken. In this case you would get the same result, oscillation.
It just looks suspect on the photo...
Mike |
|
|
| SpittinLLama |
Oscillation will not kill an Overture part (LM3886 is one) out right. Instead it will get hot and then Spike protection will come on followed by thermal shutdown. Shorting leads to different supplies will also not necessarily kill it either. If you ever attached it to a heat sink it is possible to over tighten the mounting screw and crack the die but you have to crank it pretty good. Power up with out a LM3886 in the board and check all the voltages at the pins. Then try another one.
-SL |
|
|
| Lucas_G |
| quote: | Originally posted by sayang001
Desolder the chip from the pcb and put it on the scanner with it's front, it is broken anyway :smash: |
Here is the picture of the frontside. |
|
|
| Lucas_G |
| quote: | Originally posted by SpittinLLama
Oscillation will not kill an Overture part (LM3886 is one) out right. Instead it will get hot and then Spike protection will come on followed by thermal shutdown. Shorting leads to different supplies will also not necessarily kill it either. If you ever attached it to a heat sink it is possible to over tighten the mounting screw and crack the die but you have to crank it pretty good. Power up with out a LM3886 in the board and check all the voltages at the pins. Then try another one.
-SL |
Hi Spittinlama,
I will check the voltages tomorrow and report back.
Thanks for your advice.
Regards,
Lucas |
|
|
| sayang001 |
Lucas_G,
Your LM3886TF is fake :mad: The one I have at home are the same as the picture below:
Compare the National Logo , Text Font and casing.
Also try to remove the plastic casing and see what is undernead. |
|
|
| Lucas_G |
| quote: | Originally posted by SpittinLLama
Oscillation will not kill an Overture part (LM3886 is one) out right. Instead it will get hot and then Spike protection will come on followed by thermal shutdown. Shorting leads to different supplies will also not necessarily kill it either. If you ever attached it to a heat sink it is possible to over tighten the mounting screw and crack the die but you have to crank it pretty good. Power up with out a LM3886 in the board and check all the voltages at the pins. Then try another one.
-SL |
Hi Spittinlama,
Today I checked voltages on the board:
Pins from left to right:
1= +33
2= 0
3= 0
4= -33
5= +33
6= 0
7= 0
8= -33
9= 0
10= 0
11= 0
I think all voltages are right, except maybe for pin 8?
The diagram states that pin 8 is for mute....?
Any comment?
Regards,
Lucas |
|
|
| pinkmouse |
| Pin 8 should be connected to the negative supply, but via a current limiting resistor. Check your PCB. |
|
|
| Lucas_G |
| quote: | Originally posted by sayang001
Lucas_G,
Your LM3886TF is fake :mad: The one I have at home are the same as the picture below:
Compare the National Logo , Text Font and casing.
Also try to remove the plastic casing and see what is undernead. |
Hi Sayang001,
Do you really thinks so?
I have no idea how to open this chip. The housing seems heremtically sealed...
Lucas |
|
|
| pinkmouse |
| quote: | Originally posted by sayang001
Your LM3886TF is fake :mad: The one I have at home are the same as the picture below: |
Unlikely, I can't see the market being big enough to support fakes. |
|
|
| Lucas_G |
| quote: | Originally posted by pinkmouse
Pin 8 should be connected to the negative supply, but via a current limiting resistor. Check your PCB. |
Hi Pinkmouse,
I just measured a resistance of 10k Ohm from pin 8 to the V- entrance. This should be OK then?
Lucas |
|
|
| pinkmouse |
| Yup, that's fine for those rails. |
|
|
| Lucas_G |
| quote: | Originally posted by pinkmouse
Yup, that's fine for those rails. |
Hi Pinkmouse,
Well that's that then.
This means that everything was ok with the PSU and Amp board, and that the LM3886 was either fake or just a faulty one.
So I will just connect the other channel and see whether "l 'histoire se repête".
I will order new LM3886 through RS-Components.
This all adds to a new Murphy law: "If you choose the simpliest of simple DIY projects, your main component will be rotten...
Thanks all for your advice BTW.
Lucas |
|
|
| pinkmouse |
Simple projects are sometimes the most difficult to debug! I built a 4780 that had a really strange distortion on the scope, the top and bottom thirds of the sine wave were perfect, but the middle had about 10% white noise superimposed over the top. In the end, I tracked it down to a dodgy solder joint on the mute pin that looked fine, though when resoldered it fixed the problem, though I still don't know how that affected the waveform in that way.
It is possible to defeat the spike protection on the LM chips though, I managed to blow up another 4780 on sustained testing on a 4ohm sub driver... :) |
|
|
| Lucas_G |
Good news,
I connected the second channel. It DID NOT smoke!
I checked DC at the output (with no load on either output or input) and measured -80 mV DC.
Is this alright? Can I safely connect my speaker?
Regards,
Lucas |
|
|
| pinkmouse |
| Do you have an old speaker you can use for testing, and is that 80mV measured with the input shorted? |
|
|
| Lucas_G |
| quote: | Originally posted by pinkmouse
Do you have an old speaker you can use for testing, and is that 80mV measured with the input shorted? |
Yes, good idea. I will first use an old speaker.
The 80mV is with the ipit not shorted. Should I also check soldered? |
|
|
| pinkmouse |
| I have a phono plug on my bench shorted pin to ground that I can just plug in for testing, but if you haven't soldered up the inputs yet then you can just tack a link in from input to ground for the moment. |
|
|
| JesseG |
Hi all
One thing that hasn't been mentioned yet, and that I learned by letting the smoke out of more than one chip...
The heat sink tab is NOT at ground potential on any of these chips. It is at V- potential. If your chip mounting tab makes electrical contact with the heatsink, and the heatsink is grounded, poof!! :cannotbe:
I notice from one of the previous pix that the inside of the mounting hole appears to be exposed metal. If you did not use insulating washers then that's why the chip blew up.
This won't apply of it wasn't mounted to a heatsink when you first powered up, but if it was...
Jess |
|
|
| Nordic |
Many expensive lessons to learn, burning chips, breaking pins off, dropping soldering iron, the value of good silver based ointment for finger burns etc...
With internal heatsinks this can be an advantage... I connect my chips to the heatsink directlywithout any insulation for better thermal transfer...
The whole heatsink is then at V- but it still doesn't shock you if you touch it... but most importantly it is isolated from the case
If you want to go nuts you can lap the back of the chip and the heatsink using some fine waterpaper glued to a mirror |
|
|
| Lucas_G |
| quote: | Originally posted by JesseG
Hi all
One thing that hasn't been mentioned yet, and that I learned by letting the smoke out of more than one chip...
The heat sink tab is NOT at ground potential on any of these chips. It is at V- potential. If your chip mounting tab makes electrical contact with the heatsink, and the heatsink is grounded, poof!! :cannotbe:
I notice from one of the previous pix that the inside of the mounting hole appears to be exposed metal. If you did not use insulating washers then that's why the chip blew up.
This won't apply of it wasn't mounted to a heatsink when you first powered up, but if it was...
Jess |
Hi Jess,
In my case the heatsink was not connected to ground anyway, so that could not have been the case.
My second channel worked right away. With the input shorted I had a DV of only 5 mV.
I then desoldered a LM3886 from one of my "audiophile 3886" amps (Mauro) and this way I could also get a perfectly working second channel. I have been listening now for about 2 days, and I must say (much to my regret, because it kills our hobby) that these amps sound damn good!
I will get back on this later.
Regards, and thanks to all fro their help,
Lucas |
|
|
| JesseG |
| quote: | | I must say (much to my regret, because it kills our hobby) that these amps sound damn good! |
I completely agree about chipamps sounding 'damn good'.
I'm not sure that it kills the hobby - I get a big kick out of getting one of these things working. Then, I usually give it away to a friend or school or whatever. Kind of 'spreading the word' if you know what I mean.
You might have some fun taking a look at tsome of the other National Overture series chips, the LM4731, LM4732 or the LM4766. I have been playing around with these and have found that the lower power chips can sound even better that the LM3886 or LM4780.
My LM4766 has been praised as "the best sounding chipamp I have ever heard" by one of the most critical audiophiles I know.
A Different Chipamp - LM4766
Jess |
|
|
| Stuey |
| quote: | Originally posted by JesseG
Hi all
One thing that hasn't been mentioned yet, and that I learned by letting the smoke out of more than one chip...
The heat sink tab is NOT at ground potential on any of these chips. It is at V- potential. If your chip mounting tab makes electrical contact with the heatsink, and the heatsink is grounded, poof!! :cannotbe:
I notice from one of the previous pix that the inside of the mounting hole appears to be exposed metal. If you did not use insulating washers then that's why the chip blew up.
This won't apply of it wasn't mounted to a heatsink when you first powered up, but if it was...
Jess |
I'm pretty sure this only applies to the LM3886T and not the LM3886TF. The TF is insulated. |
|
|
|
