LM1875 power amp problems

[MrStab]

hi,

i'm building my first solid-state amp - please bear in mind that i'm not too experienced in that area beyond general maintenance of tube amps - and over two different chips, i've had the same problem: the negative supply rail goes from -18V down to -1.6V whenever i hook up the negative supply to the power amp.

as i originally started out with an old TDA2030, i used the power amp section here: http://www.coldcity.com/index.php/hybrid-guitar-amp/, and the vero layout seems to check out.

for a preamp, i'm using a modified & expanded Xotic BB Preamp, which is essentially a glorified Tube Screamer. it's using a single supply, so when i do get things working i'll probably have to even out the load on both sides.

the sound is crackly with the uneven rails, and louder with just the positive rail connected. there's no continuity between the heatsink and ground, but there is direct continuity between pin 3 of the LM1875 and the heatsink. well, i say no direct continuity to ground, but i measure about 48K resistance on my DMM (and 1480 on my continuity setting) between the heatsink & ground. goes up as the heatsink cools, to be expected i guess.

i only have one left, please don't tell me i'll have to go through 3 in a day lol (i fried one by putting it in wrong altogether, stupid error).

any help appreciated, i'll be happy to provide as much info as i can.

[MrStab]

just noticed: with the negative rail disconnected, i'm getting about 5V on pin 3/ heatsink

[MrStab]

okay, so i went over the power amp board re-soldering everything and removing a 10uF cap which i had in parallel with another for 20uF (which may have been shorting), and now for whatever reason i have more or less the right voltage. +10 and -13 (bearing in mind the positive rail is loaded down by the preamp + an LED). now it's just noisy as hell. oscillation? doesn't sound like my typical idea of oscillation but hey.

some of the noise is undoubtedly from the power supply, as i left a bunch of 2W 100R resistors at my lady's house so i have to score more tomorrow, but there's something else afoot. or is there? also, the heatsink is becoming hot as hell now. i have a history of computer nerdery and overclocking, so that just sounds like a fun challenge. lol

the schem i'm using is in section 3/page 8 here: http://www.st.com/st-web-ui/static/active/en/resource/technical/document/datasheet/CD00000128.pdf (the one on the left). would adding the optional 100uF caps be of benefit?

i know this is outwith the realm of stompboxes, but it's a one-off (unless i get it working...!) and an amp is just a big pedal that can kill you, right...? lol

oh, and i'm using a 4 ohm speaker. i don't have any other speakers that i'm willing to ruin, but i may have some small non-guitar ones i could hook up to increase the load if needs be.

Edit: i think it was a loose connection with the power filter/rectifier section causing the low voltage problem, now i'm getting +/-10V. which is too low, i take it. wtf! also heatsink is getting ridiculously hot. i might go to that SSGuitar forum if you guys understandably feel that this isn't your jurisdiction.

[psychedelicfish]

 I've been down this road myself (in fact, it's how I got into stompboxing). I had problems with oscillation (I had no power supply caps on the vero with the chip) and I fried several chips. I was also buying all my components from jaycar, so I was spending NZ$7.50 each time I killed a chip. In the end, I gave up on the amp but by this point, I was hooked on breadboarding various distortion circuits and I got into stompbox building. I came away from this with lots of knowledge about amps, a 1.1c/w heatsink and a 100VA +/-20V supply.

Anyway, what heatsink are you using? Also, do you have any thermal grease or silicone pads between the heatsink and the chip? I think you've probably got too high of a thermal resistance from the chip to the open air which is causing your heat problems. As for the voltage issues, what power supply are you using? If you're pulling near its max current, the voltage from it will drop. Your supply should be capable of at least double the max power you expect to draw.

[MrStab]

glad to receive a response here, Edward - i went over to the SSGuitar forum and they were mean to me lol not really, they just weren't accustomed to my self-propagating confusing nature. they have offered some useful pointers so far. http://www.ssguitar.com/index.php?topic=3096.0 if you're interested.

intuitively (though that doesn't always stand for much), the problem seems to be with the power amp section itself as i've tried another chip and the transformer + preamp work fine independently.

the heatsink was from an old Park G25R amp i acquired, which used an LM1875 to begin with so i'd imagine the heatsink was suitable. i'm lacking in thermal paste at the moment (i may have some from my last overclocking phase but it's quite old), but the problem kicks in before the chip has even had a chance to heat up. there were some loose connection issues (which leads me to think the PS works fine now), but now it's just plain-ol' oscillation. well, intense ticking = oscillation, right?

luckily i came across 3 new LM1875's for free, but i fear they may all be screwed because of this issue. hopefully not. i have some LM3886's as well, but i'd rather not go that high for my first build.

what i don't get is that the layout i used seems true to the TDA2030 datasheet, and afaik people upgrade the TDA with the  LM1875 without any other part adjustments. so once i score more strip/veroboard at the weekend, i'll be making a layout based on the 1875 datasheet... but i'm unsure how to position things to prevent oscillation, as i hear "geography" is important in that regard.

cheers

[psychedelicfish]

Just make sure all capacitors are as close to the chip as possible, and the input and output traces and wires are kept separate, and you should be fine. You should also consider covering the power supply traces with a good thick layer of solder. Read this for more info on LM1875 amps.

[Kesh]

in most variants of this chip the chip's case is internally connected to its negative power pin.

chip case HAS to be insulated from chassis/heatsink, etc, unless you use a single rail power supply, or you aren't grounding your chassis.

else you are effectively grounding the negative power rail

[MrStab]

(edit: i wrote something about it possibly being my speaker wiring being on the same path to ground as the power amp, but forgot i'd already rectified that)

there's no continuity between the heatsink and ground

seems ok Kesh - but in case my DMM's a bare-faced liar, i measure V- over the heatsink, so that in itself should indicate no short, right? also no awesome sparks & subsequent electrical fire (yet). lol

would you recommend i make that schematic on the ESP article, Edward?

cheers

[Kesh]

sorry, missed where you said that.

probably obvious, but is your output overvoltage protection diode to the negative rail the right way around?

[MrStab]

both those diodes are the right way, unfortunately. good call though, it took me a while through the debug process to question that myself - i'm always paranoid about diodes when they're standing upright, as my mind somehow confuses which side the actual component is facing and which side the negative line is facing.

from reading around the place and the other forum i'm reading on (i'm still loyal to DIYSB, it's all good lol), it seems that to some people the 100uF caps i thought were optional may in fact be expected as mandatory. i'll be sure to include them once i pick up more solder & get to work tomorrow.

i'm gonna work on a layout tonight - i'll post it here if i remember to.

cheers