Hot LM3875

[Ropie]

Hi,
Does anyone have an idea what could be causing my IGC to be running so hot. A few weeks ago both channels suddenly started getting hot even when they were just sitting quietly, without playing any music. I have checked numerous times but can find no errors in the internal wiring. Surely it shouldn't get hot doing nothing
Supply is +-37V, feedback resistors 2 x 432k. I tried a zobel on the output which didn't help.

Grateful for any suggestions...

 

[theChris]

checked for high frequency ossilation? that seems the logical thing.

 

[Ropie]

Hi theChris,

I already tried a zobel - wouldn't that cure any hf oscillations (I used 220nF cap and 4.7ohm r)?

 

[Andypairo]

432k seems a bit high, what's the value of the other resistor (for gain setting) ?

Cheers

Andrea

 

[jackinnj]

Hi,
Does anyone have an idea what could be causing my IGC to be running so hot. A few weeks ago both channels suddenly started getting hot even when they were just sitting quietly, without playing any music. I have checked numerous times but can find no errors in the internal wiring. Surely it shouldn't get hot doing nothing
Supply is +-37V, feedback resistors 2 x 432k. I tried a zobel on the output which didn't help.

Grateful for any suggestions...

What is the rating of the heat-sink you are using? Check this link for the proper C/W:
http://www.national.com/appinfo/audio/files/Overture_Design_Guide13.xls

And the gain-setting resistors -- the ratio should be 20:1 or thereabouts -- if the gain is set too high the chip clips, producing a highly distorted signal. Did you parallel the 432K's? (that should be OK for IGC)

If it is oscillating even with a RC Zoebel network, instead try an LR network as described in the Nat Semi PDF, OR a 0.22 ohm series resistor in the output.

 

[Ropie]

Yes, the 432Ks are paralleled. I'm a novice at all this but is the gain setting resistor on the input? If so it is 10K.

I'm fairly sure the heatsinks are OK as the amp ran fairly cool for a few months but just changed one day. I think I may try the .22r resistor on the output as you suggest, jackinnj, thanks.

 

[GregGC]

Yes, the 432Ks are paralleled. I'm a novice at all this but is the gain setting resistor on the input? If so it is 10K.

I'm fairly sure the heatsinks are OK as the amp ran fairly cool for a few months but just changed one day. I think I may try the .22r resistor on the output as you suggest, jackinnj, thanks.

The gain may depend on your source impedance also. Can you show us the sch. Have you changed any of your equipment?

 

[Ropie]

Greg,

Funny, it happened when I built a new set of shelves for the stereo to sit on. 😕 I've not changed any equipment with the exception of the speaker cables which are now Cat5 braids.

The schematic is the one on Nuuk's Decibel Dungeon
http://www.decdun.fsnet.co.uk/gaincloneFAQ.html

 

[jackinnj]

well, the L, C or R of the new cables is weighing heavily on the analysis -- try running the amp into a dummy load --

 

[jan.didden]

Indeed. He said, it ran fine, then suddenly changed. So SOMETHING must have changed the operating mode. I would gamble on the cables also. Try with the original cables.

Jan Didden

 

[GregGC]

Indeed. He said, it ran fine, then suddenly changed. So SOMETHING must have changed the operating mode. I would gamble on the cables also. Try with the original cables.

Jan Didden

I think you have it spot on.

 

[Ropie]

Yes, though I have tried several different 'blends' of Cat 5, I'll try it with some shop-bought cable tonight.

Having said that my amp is hot when it's on but not playing (even if left for hours to 'cool down') - would speaker cables have an effect even then?

BTW, jackinnj, you said try into a dummy load - presumably this means attaching a resistor across one of the inputs?

 

[GregGC]

Yes, though I have tried several different 'blends' of Cat 5, I'll try it with some shop-bought cable tonight.

Having said that my amp is hot when it's on but not playing (even if left for hours to 'cool down') - would speaker cables have an effect even then?

BTW, jackinnj, you said try into a dummy load - presumably this means attaching a resistor across one of the inputs?

It must be the cable capacitance. I bet when you disconnect the cables from the amp it'll cool off. Some people report very good results with single twisted pair of the CAT5 cable (you have to pull it out of the cable though).

Don't forget, this amp at gain of 20 has frequency response up to 300kHz. So, handle with care.

 

[Ropie]

Greg,
I'll try taking the cables out and let you know..

>> Don't forget, this amp at gain of 20 has frequency response up to 300kHz. So, handle with care. <<

This last sentence sounds interesting but can you put it into novice English for me?

 

[jackinnj]

BTW, jackinnj, you said try into a dummy load - presumably this means attaching a resistor across one of the inputs?

I use 5 -- 40 ohm, 15 watt -- dale resistors in parallel -- works just fine.

 

[GregGC]

Greg,
>> Don't forget, this amp at gain of 20 has frequency response up to 300kHz. So, handle with care. <<

This last sentence sounds interesting but can you put it into novice English for me?

It's just a fairly wide frequency band amplifier, so you need to take the proper care of it: Proper bypassing (if the PS electrolytic are crappy type you'd need to use 100nF in parallel for example and you have to place them very close to the chip), extremely short connections and very good shielding. If HF noise gets in will most likely be amplified and destroy the sound even if you don't hear the actual interference signal.

 

[jan.didden]

Yes, though I have tried several different 'blends' of Cat 5, I'll try it with some shop-bought cable tonight.
[snip]

Try logic for a change: It ran fine. Then you changed something. What was it? Undo the change. Fine now? OK you have found the problem.

Just blindly trying all sorts of cable is like groping around in the dark. Which is fine, if you are the adventurous type. Maybe you have other uses for your brain.

Jan Didden

 

[Ropie]

Unfortunately changing the speaker cable didn't have any effect, neither did unplugging the cable altogether , nor unplugging all the cables and just leaving the amp on but isolated. So I assume the problem is either rfi or somewhere in the circuits!

 

[richwalters]

I've run a 3875 load off bell wire without instability. Since you checked the the Zobel network on the o/p of th IC, are you sure that an IC further down the amp line isn't the culprit ? You say the same symptom occurs on Both channels.

more anon
rich

 

[macboy]

Sounds like a high frequency problem to me. Your amp is either picking up stray RF signals or is oscillating.

Try adding a small capacitor, like 50 pF to 100 pF (0.0001 uF), in parallel with the feedback resistor(s). If you have a high frequency problem this may help, by reducing gain at very high frequencies.

Make sure that you have at minimum, 20 uF of power supply capacitance close to the amp. Again, due to inductance in the power supply leads, you can run into problems if this capacitance is not there. National usually recommends 1000 uF, and this is a common value in practice. In addition to these make sure also that you have 0.1 uF capacitors from the V+ pin to ground and and V- pin to ground. Place these as close as possible to the actual pins, to avoid parasitic inductance in their leads. You simply cannot get away without these caps!


Another potential cause of the problem is a relatively high DC offset at the output. With no signal (but a speaker connected) measure the output voltage with a DC voltmeter. There should be less than 10 mV (0.01 V). If there is more, then this could account for the heating when idle.