Feeling the HEAT - on my lm3886 amp

I recently finished my chipamp.com lm3886 based kit and really like the sound it produces....a wonderful sounding little amp for the money. My issue is the heat produced by the chips - they are getting really freaking hot!

The amp uses all of the standard parts supplied with the kit, except the rail capacitors. I replaced them with Nichicon 2200uf 50v vx series caps. The reason for this was that when I was building the amp a new puppy came along and ran off with my original caps...so I used the Nichicons which I had on hand. Could this be the cause of the heat?

For the most part the amp is normal..I am using a 400va torroid with 2x22v secondaries...and I am getting around 30.6v dc going to each channel. wiring for the power supply is solid copper ribbon wire (4 conductor) and the input/output is silver wire. The PS is the CalrosFM version supplied with the kit. Heatsinks are 1/4 aluminum plate approx 3in tall and 4in wide.

I metered all resistors and checked to make sure that nothing was out of place or had a bad solder joint - but everything seems to be in order.

My test rig consists of a pair of BR1's from Parts Express wired up with some copper cat5 Plenum wire (about 4 ft on each side). The source is an old Philips 926(?) model transport hooked up to an Audio Alchemy DAC. These are run into a passive preamp which uses an Alps 100K pot (Radio Shack version)

Everything seems normal...grounds are good....etc....

Hopefully I have supplied enough info to solicit a little help. I suspect that it might be the caps...but not quite sure. An wanted to ask a for some help before I started tearing the thing down.

Thanks,

Wayne-o
 
What is the impedance of your speakers? For 4 ohm speakers at 30V, that heat sink is way too small. Take a look at the spreadsheet that National provide and see the size of the heatsink they recommend.

Also, Cat5 based speaker wires can (depending on construction) have high capacitance. Are you sure your amp isn't oscillating? Do you have access to an oscilloscope?
 
Hi cabbagerat

Thanks for the quick response...and for raising a few more questions.

The speakers are 8ohm...and I am using it as 2pairs per side (ie striped for neg and solid for pos). should I try some different wire or a different configuration?

The heatsinks that I am using are for the most part a temporary solution until the full chassis is complete.

On the oscillation question....I don't have quick access to one, but I can probably get use of one in few days or so.

What might be causing the oscillation?
 
The heatsinks you describe aren't heatsinks. Those are heat spreaders. They need to be attached to a chassis or really any large bit of metal will do.

Hey, while you're at Radio Shack, pick up some Artic Ceramique. They have it in ThermalTake's packaging, but you can see the print on the white tube in there. You'll want the thermal compound between chip and heat spreader plate, and between the heat spreader plate and the chassis.

Or, you could stop at a "mom-n-pop" computer store to see if they have any of the old Pentium 2 "slot" heatsinks. Those are nice and big, and usually almost free. Also, the goodwill store and salvation army stores are good spots to check for defunct electronics with very nice heatsinks clearly visible through the vents. ;)

Have fun!
 
(I'm not familiar with that kit, by the way.)

How thick is the power supply wiring that you mentioned? Was it supplied with the kit? What is its length?

What sort of small bypass caps are in parallel with the larger ones, for the chip's power pins?

You could try plain 'zip cord' (lamp wire) to the speakers, just in case your cable capacitance is causing the amp to oscillate. Or try a very short cable.

This might be a 'long shot', but, did you make all component leads as short as possible (i.e. fully inserted the components) before soldering them?

You should probably at least try some real heatsinks, before worrying about anything else being wrong. Surface area near the chip is probably what matters most, at steady state. Get some real heatsinks, with the proper thermal resistance rating or better.
 
How thick is the power supply wiring that you mentioned? Was it supplied with the kit? What is its length?

The wire that I am using for the power connections is 24AWG 4 Solid Conductor Intercom Wire. it just seemed like a very "tidy" solution to having a "rat's nest" of wire. I have several other types of wire in the electronics box - should I try a thicker stranded wire, say 10 or 12 AWG?

What sort of small bypass caps are in parallel with the larger ones, for the chip's power pins?

0.1 uf BC polypropylene cap - these may eventually get replaced with some WIMA caps when I get around to it.

danielwritesbac:

I will head to the basement and go through my "box-o-heatsinks" and find something a little more suitable for the job. I think that I used the heat spreaders because they seemed to fit very well in my chassis. I probably have several of the pentium slot 2 heatsinks as well as several old mosfet heatsinks...some are marked Halfler....some have motorola marking on them...etc. So I will try something else.

Thanks for everyone's help so far...

Wayne-o

BTW...if anyone is interested I will be having a "get rid of some junk firesale" soon...I have tons of old lab equipment, tubes, caps, hammond organ amps, solid state amps, transformers, etc. I will post a list in the next several weeks...or when I get s chance to go through some things.

Thanks
 
wboyd said:

The wire that I am using for the power connections is 24AWG 4 Solid Conductor Intercom Wire. it just seemed like a very "tidy" solution to having a "rat's nest" of wire. I have several other types of wire in the electronics box - should I try a thicker stranded wire, say 10 or 12 AWG?

0.1 uf BC polypropylene cap - these may eventually get replaced with some WIMA caps when I get around to it.

<snipped>

Thanks

I think you might want to try at least 18AWG. Twist them tightly together.

I would try some 0.1uF ceramic (not NPO or C0G; probably X7R or worse), instead, in case the polypropylene's low ESR is causing an oscillation. This might be a pretty good bet, since you changed the electrolytics' type.
 
Wayne,
Your heatsinks may be a bit small, but hot should really be defined. As I read National's specs, operating temp is to 150 degC = 300degF. Do you have a means of checking the temp? Your amp may be fine as-is...

BTW: I believe the term "heatsink" is defined by function rather than shape. In other words, your aluminum plates are indeed "heatsinks".


7/10
 
wboyd said:
Tom,

Thanks, I will try your suggestions this evening. Could you recommend a brand of capacitor? I will probably need to make a parts order soon anyway.

It shouldn't matter very much, what brand the 0.1 uF ceramic caps are. I think I usually buy the X7R ones in rolls of 1000. I think the last batch was from Digikey, probably BC Components brand, although I usually get the bulk of my components from mouser.com. I usually buy Kemet's "Golden Max" ceramic caps, and get the ones with the straight leads with 0.2" lead-spacing. For the X7R, that would be mouser.com's part number 80-C320C104K5R, as far as I can tell, which is 0.1 uF X7R 50V, 10%, for $0.16 qty 1 or $0.11 ea qty 10.
 
wboyd said:


The wire that I am using for the power connections is 24AWG 4 Solid Conductor Intercom Wire. it just seemed like a very "tidy" solution to having a "rat's nest" of wire. I have several other types of wire in the electronics box - should I try a thicker stranded wire, say 10 or 12 AWG?


0.1 uf BC polypropylene cap - these may eventually get replaced with some WIMA caps when I get around to it.

danielwritesbac:

I will head to the basement and go through my "box-o-heatsinks" and find something a little more suitable for the job. I think that I used the heat spreaders because they seemed to fit very well in my chassis. I probably have several of the pentium slot 2 heatsinks as well as several old mosfet heatsinks...some are marked Halfler....some have motorola marking on them...etc. So I will try something else.

Thanks for everyone's help so far...

Wayne-o

BTW...if anyone is interested I will be having a "get rid of some junk firesale" soon...I have tons of old lab equipment, tubes, caps, hammond organ amps, solid state amps, transformers, etc. I will post a list in the next several weeks...or when I get s chance to go through some things.

Thanks

Ceramic caps are even lower ESR than polypro.
For higher ESR than polypro, mylar is higher, and polyester is higher yet. These are all good options to have.

The braided AC cables are great. All that I know about the DC cables are that they should be short, same length, and that the ground should be never farther than 1/4" away (point blank range is better).

Your heat spreader plates probably work fine when bolted (and thermal compound) to a metal chassis. But, bigger heatsinks would be nice. ;)

Well, this is going great. I think you'll have some excellent results.

P.S. The sale sounds great!!
 
Daniel and gootee, Please explain how low ESR on a power supply bypass cap can cause an amp to oscillate. I don't see how that can happen.

wboyd: Has switching to a larger heatsink solved the problem? Remeber to use a thin layer (as thin as possible) of thermal paste. Standard white compound works just fine, but you can use one of the more weird and wonderful brands if you want.
 
cabbagerat said:
Daniel and gootee, Please explain how low ESR on a power supply bypass cap can cause an amp to oscillate. I don't see how that can happen.

I honestly don't think it can. But, I do know that you can use higher ESR instead of the additional resistor you're supposed to use. ;)

Power supply has a certain bandwidth of?--well, it must be a noise of some sort. I don't know how to calculate it, but different capacitance makes a different noise, and less capacitance is less likely to hetrodyne. I just don't have a firm understanding on exactly how to choose the right capacitance, so there's practically no way I could explain this. ;) The higher ESR (or adding a resistor) should have the effect of a wider bandwidth noise at much less strength, instead of a peak? Well, that's why you'd do it on an audio signal. I just don't know if it applies to the power supply. So that's all like 100% imagination on my part. ;)
Can someone fill in the blanks here?
 
Howdy Wayne-o,

Just for comparison purposes, I have a Chipamp kit (supplied with their basic stock components, though) running on 21VAC secondaries.

It runs remarkably cool; for example, if I play music loud enough to be close to unbearable (ie. the kids disappear into their rooms with the doors shut) the heatsinks get just slightly warm after about half an hour. In normal use you can barely detect a rise in the temperature of the sink.

These are substantial heatsinks (Conrad's MF 20-75 x2), but I think its relevant that the sink base plate is only just warm under very heavy use, so they're not radiating heat in a big way. I think yours possibly has a functional issue as suggested above, rather than it being the sink (although for sustained use I'd suggest one with more capacity).

Cheers

Stuey
 
Well...last night I was able to change the speaker cables and it did seem to help a bit....I will work on the power wiring over the weekend and I found a few ceramics in my parts box...I think I have about 20 or so.

So I have my weekend planned out as far as the amp work goes.....I am anxious to finish this project because I have a partially finished grounded grid based clone that I have been working on to mate with the gainclone. It uses Rifa caps and Electro-harmonix 12AU7's- but that is another topic altogether and I don't want moderators getting me for going off-topic (j/j).

Thanks to everyone so far for the excellent advice....having so many sharp minds together in one place really helps.

Wayne-o
 
So after a few changes to the amp things seem to be getting better....the heat has been reduced (although still a little hot) and it seems to be a little more dynamic. I still have a few changes to make but all seems to be good so far. I put a few photos up on flickr just incase your interested.

http://www.flickr.com/photos/[email protected]/2263112800/in/photostream/

Currently I need to finish the chassis and nail some things down. I will post in the chipamp photo gallery when I get a chance.

Thanks again for all the help so far...

Wayne
 
Tom,

Thanks alot for all the advice, etc. I have been reading the threads you suggested and I came across this....

http://stereophile.com/solidpoweram...147/index4.html

In which the reviewer exclaims -

After the usual one-hour preconditioning period at 1/3 power into 8 ohms, the tiny 47 Lab 4706 Gaincard amplifier was far too hot to touch. It was so hot, in fact, that I was alarmed that something might have broken. Fortunately, everything was fine, but owners should be sure to give the 4706 plenty of room for adequate ventilation.

Mine is right at about 240-250 now. After doing a little research etc, and reading the National Semi data sheet on the lm3886, I think that I am well within tolerances - but I am going to continue tweaking the little amp - that is still a little hot for me. I have read many threads were gainclone builders have stated that their amps run cool to the touch...

Thanks,

Wayne