Oborous said:
Been thinking of this since yesterday, have some other ideas for you to ponder.
I'd suggest you go overly nuts with the fit & finish of your amp. With all those tubes, you'd actually have a fair amount of surface area, ..............
......... would be interested in seeing some photos
Thanks for the suggestions. The silentX fans look good. If i need some, those seem the ones to have. Maybe i can put one about halfway down each tube and have it draw air across one board and blow it out across the other...
I ripped apart an old 12v 1000ma wall wart i had laying about. The board in that looks perfect, 4 diodes and a cap. If i wind some more wire onto one of the trannys and use this board do you think it will do the job ? I could house it right in the rear tube, as far away from the chips as possible.
What is the mass of the copper you're going to use? I've run two of Brian's boards on a piece of brass that is about 25mm x 50mm x 80mm. I ran the chips at 35v, and while the block did get a bit uncomfortable to the touch, it wasn't blistering hot or anything.
If you do go with the fan, though, just be sure that you don't run it too fast. Otherwise, the noise might get annoying. A little breeze is all it takes.
If you do go with the fan, though, just be sure that you don't run it too fast. Otherwise, the noise might get annoying. A little breeze is all it takes.
Selexus said:
Thanks for the suggestions. The silentX fans look good. If i need some, those seem the ones to have. Maybe i can put one about halfway down each tube and have it draw air across one board and blow it out across the other...
I ripped apart an old 12v 1000ma wall wart i had laying about. The board in that looks perfect, 4 diodes and a cap. If i wind some more wire onto one of the trannys and use this board do you think it will do the job ? I could house it right in the rear tube, as far away from the chips as possible.
I'd say the wallwart would be fine (Safety warning, since you're using a conductive frame in sections, be careful! Your opened wall wart may have exposed conductors. Don't mean to be patronizing or anything, I've pulled spectularily stupid things because I found the 'perfect solution' and was in too much of a hurry to see if it works) I'd be concerned about the quality of the power coming from that little rectifier, the quality of the PSU's for many of these chip amps blow my mind (Sometimes I think people spend more time on the PSU than they do with the amp)
Papst has some radial fans that could be mounted inside tubes, then you wouldn't have the split. The Silenx could be remove from their frames.
Thanks for the pictures, that looks awesome. More info (and maybe a close up) of the knobs please.
ive just built a single channel bass guitar amp based on national's BPA-200 design from the relevant application note. a pair of bridged LM4780 in parallel, +/-38v supplies (according to a dmm) into 8ohm, right about the edge of the specs.
these chips run extremely hot and are quite durable, LM38xx and LM4780 all have 125w listed for case heat dissapation. The LM4780 has a temperature rating of 85C, but the 'spike' thermal protection doesnt engage until 150C. with both BrainGT's LM3875 kit and my personal dual LM4780 design, ive used thin finned all copper athlon xp heatsinks, thermaltake volcano 7+ or similar purchased from newegg.com, WITH FANS REMOVED. For coupling of the chip and sink i used arctic silver 3. i DO NOT suggest using this compound (explained later).
they heat up quickly, within minutes, under little load. under full loads, they become too hot to touch. however, the chips do sound as advertised even under such loads, minimal distortion likely attributed to the four series/parallel 7" 8ohm speakers at this point. MY CHIPS HAVE NEVER RUN HOT ENOUGH TO ENGAGE THE THERMAL PROTECTION. i dont know what it sounds like but from the images of what it does to a sine wave in the datasheets, theres no way i wouldnt notice as the amps sound dreamish thru the crappy enclosure i built for it.
i left the amp running towards full for 12 hours+ several days in a row, never did the amp sound stressed or in thermal protection mode, with the PCBs and heatsinks enclosed in a small airtight box (no cooling ventilation/ports).
i dont thinks fans are necessary for LM38xx or LM4780 chips if designed and run within spec with an eduquate heatsink. copper athlon xp sinks are neither expensive nor cumbersome, being small and easy to drill mounting holes due to the socket A mounting clip design. $15 or so for a good sink from newegg to insure your chips are properly cooled doesnt seem like to big an expense to me, considering what some pay for caps and xformers.
As for the arctic silver, while it does seem to work extremely well, it is made of silver and thus extremely conductive. after a few weeks of regular use, the amp experienced an oscillation situation for approx a minute followed quickly by catostrophic failure at 4am. dying overtures would make great alarm clocks. using BrainGTs board as an influence for my GNDing scheme, the amp was completely free of hum. i often would forget to turn it off, and it was on for days, sinks warm. the heatsink compound had actually dripped down the back of the chip and shorted out about 7 pins in two spots, mostly +/-V and GND pins. those pins no longer exist, other than black stumps on the PCB. however, as further testimonial to the durability of these chips, the case is perfectly is fine. regardless, being extremely careful about how much thermal compound you use or using mica washers is advised.
thank you for taking the time to read the post, know i talk a bit too much most the time.
rmr
these chips run extremely hot and are quite durable, LM38xx and LM4780 all have 125w listed for case heat dissapation. The LM4780 has a temperature rating of 85C, but the 'spike' thermal protection doesnt engage until 150C. with both BrainGT's LM3875 kit and my personal dual LM4780 design, ive used thin finned all copper athlon xp heatsinks, thermaltake volcano 7+ or similar purchased from newegg.com, WITH FANS REMOVED. For coupling of the chip and sink i used arctic silver 3. i DO NOT suggest using this compound (explained later).
they heat up quickly, within minutes, under little load. under full loads, they become too hot to touch. however, the chips do sound as advertised even under such loads, minimal distortion likely attributed to the four series/parallel 7" 8ohm speakers at this point. MY CHIPS HAVE NEVER RUN HOT ENOUGH TO ENGAGE THE THERMAL PROTECTION. i dont know what it sounds like but from the images of what it does to a sine wave in the datasheets, theres no way i wouldnt notice as the amps sound dreamish thru the crappy enclosure i built for it.
i left the amp running towards full for 12 hours+ several days in a row, never did the amp sound stressed or in thermal protection mode, with the PCBs and heatsinks enclosed in a small airtight box (no cooling ventilation/ports).
i dont thinks fans are necessary for LM38xx or LM4780 chips if designed and run within spec with an eduquate heatsink. copper athlon xp sinks are neither expensive nor cumbersome, being small and easy to drill mounting holes due to the socket A mounting clip design. $15 or so for a good sink from newegg to insure your chips are properly cooled doesnt seem like to big an expense to me, considering what some pay for caps and xformers.
As for the arctic silver, while it does seem to work extremely well, it is made of silver and thus extremely conductive. after a few weeks of regular use, the amp experienced an oscillation situation for approx a minute followed quickly by catostrophic failure at 4am. dying overtures would make great alarm clocks. using BrainGTs board as an influence for my GNDing scheme, the amp was completely free of hum. i often would forget to turn it off, and it was on for days, sinks warm. the heatsink compound had actually dripped down the back of the chip and shorted out about 7 pins in two spots, mostly +/-V and GND pins. those pins no longer exist, other than black stumps on the PCB. however, as further testimonial to the durability of these chips, the case is perfectly is fine. regardless, being extremely careful about how much thermal compound you use or using mica washers is advised.
thank you for taking the time to read the post, know i talk a bit too much most the time.
rmr
los_ren said:
I am using a heat sink with a C/W of ~1.6, with no fan the 1 kHz THD at 115 watts was definitely higher -- like 0.10%, vs 0.02% with a fan -- does this matter in the listening ? As I said in another post, the SPIKE seems to have some latency.
I also found, by accidentally leaving the darned thing on, that it would run at 115 W without blowing itself apart even without the fan. The heat sink temperature was probably hot enough to fry an egg, however. These are very durable devices.
I did fashion mill a bracket to pull the device tight to the heatsink.
Are you sure it was Artic Silver 3? Artic Silver 3 is not electrically conductive Only the original Arctic Silver (i.e. ver 1.0) was conductive.
Also, IMO, if you used enough Arctic Silver to drip, then you used way too much by an order of magnitude. For the size of the die that I'll be coupling with my LM3875 (Don't know if your chip is larger) I will be using a drop a quarter the size of a grain of barley (or grain; maybe 1/10 the size of a kernel of corn).
The best implementations are always with the least amount of thermal compound on a thoroughly cleaned and flat surface (i.e. lapped) with a high mounting force. My copper heat sinks get lapped to x6000 grit.
I have moved to Arctic Ceramique for my thermal compound because of it's lack of conductivity.
Please don't take this as harshness, I feel you used too much compound and that caused you problems. I know that using a good thermal compound (lapped, so flat heatsink and chip, so only two surfaces, only one can really have imperfections, which the thermal compound is designed to fill) will be better than a mica washer (washer has two sides, plus the heat sink, plus the chip... so four sides that could have improper contact).
Oborous said:
If they claimed only AS1 was conductive, then I wouldn't trust the claim that AS3 was not
See, ~2 years ago I modified a bigger heatsink for a Radeon 8500 videocard to cool it passively. I used AS2 in the process and a very small amount of the stuff got on the AGP connector of the card. Few days later the card started putting out garbage on the screen. Luckily the stuff didn't get on power pins and cleaning it cured the problem. Otherwise I agree with your post and I use Arctic Ceramique too, shouldn't conduct (
) and it's quite a bit stiffer than the silver compounds (easier to work with imo).
1200 grit and WD40 on a disc sander.AudioFreak said:
An externally hosted image should be here but it was not working when we last tested it.
well, theyre wrong dude. do you want pics? cuz im not about to debate wheather silver is conductive or not. also, spike thermal protection turns on at 150C, the chips rated at 85C, well beyond what a modern CPU can tolerate. just cuz it isnt conductive at 40C dosnt mean it wont be at 120C.
no duh. the bead that sqeezed out the bottom of the chip/sink junction after mounting was about the thickness of a piece of thread. clearance between the bead and the pins was ~1.5mm. the possibility of it dripping down and shorting pins was considered during contruction. just wanted to sere if itd happen, i have 10 extra LM4780 to play with. but regardless, if it isnt conductive why would it matter how much i used and what it dripped on?
the error was in construction not design (so i dont care that much cuz the achievment was in a PCB that worked well).
An externally hosted image should be here but it was not working when we last tested it.
looks conductive to me =]
the chip wasnt quite as messy with the silver goop after pulling it from the amp. i hate getting that stuff on me so i wiped most of whatever was left off, got all over the chip. anyway, the silver compound around where the pins shorted isnt goopy anymore, its rock hard, so like there likely is a point at which the silver goop changes chemically.
My amp uses 4 OPA541 parralleled and bridged. I use two 60mm fans running at 7.5v, and even running the amp at full power for 6+ hours, the temperature does not increase at all. The heatsink does get warm (not hot) under the same conditions with the fans off. They arnt really needed but I like going OTT. 
I would say look at my website, but I havent taken any pictures of the finished amplifier.
A note about noise. At 7.5v, my 60mm fans are barely audiable. At 5v they are silent. I would avoid 40mm fans like the plauge. Even run at low voltages, they are usually noisy. I have some Akasa 80mm fans (second one on this page: http://www.overclockers.co.uk/acatalog/Online_Catalogue_Akasa_Fans_97.html) for another project. Even at 12v, they are barely audiable.
I'll stop typing now.
I would say look at my website, but I havent taken any pictures of the finished amplifier.A note about noise. At 7.5v, my 60mm fans are barely audiable. At 5v they are silent. I would avoid 40mm fans like the plauge. Even run at low voltages, they are usually noisy. I have some Akasa 80mm fans (second one on this page: http://www.overclockers.co.uk/acatalog/Online_Catalogue_Akasa_Fans_97.html) for another project. Even at 12v, they are barely audiable.
I'll stop typing now.
Does anyone know what the cracking pressure of the LM3875 chip is? Many of the CPU heatsinks have a clamping force of ~40 pounds of pressure. I'm thinking of doing a clamping strap like Peter Daniel does for his mono-block amps (Sorry, Peter D is just too prolific for my Search-fu to find the picture, he show cased it with that beautiful lexan TVC Pre-amp, the amp has cork sidings).. Anyways...
I'm thinking heat conductors on both front and back should make things a bit cooler, as I really don't want to have to implement a fan solution on my GC. I'm hoping five kilos of copper bar stock is nough to keep it cool.
Has anyone considered a Peltier device as a heat pump to keep the chip cool no matter what the heatsink temp is?www.peltier-info.com
Heat Pipes still seem an excellent option for the design, chips in the tube, coupled to the heat pipes, put a 90º bend at the end and heat sink the 'back' of the amp.
One possibly extreme/messing possibility is Mineral Oil submersion. None of the components of the gainclone 'amp' board would be harmed by mineral oil submersion, then you'd get stronger thermal coupling in the tubes where they're held. Problem would be that you'd spread the heat pretty evenly, and that could force some weird drift in specs of the resistors...
I don't think going beyond these exotics would be beneficial. Hirsch/Vortex tubes might be interested, but would be an outboard solution (Demonic cooling (Maxwell Demon) would fit with ultra-high end audio... which cable manufacturer uses a product called pixie dust in their power cords?)
I'm sorry, I can't see the picture. I would like to see a worst case scenario on what happens. Not that I doubted what happened, or why... AS3 is capacitive... and I found evidence that with enough pressure the imersive effect (Colloidal particles) of the silver breaks down and AS3 does become conductive.
Ceramique for me!
I'm thinking heat conductors on both front and back should make things a bit cooler, as I really don't want to have to implement a fan solution on my GC. I'm hoping five kilos of copper bar stock is nough to keep it cool.
Has anyone considered a Peltier device as a heat pump to keep the chip cool no matter what the heatsink temp is?www.peltier-info.com
Heat Pipes still seem an excellent option for the design, chips in the tube, coupled to the heat pipes, put a 90º bend at the end and heat sink the 'back' of the amp.
One possibly extreme/messing possibility is Mineral Oil submersion. None of the components of the gainclone 'amp' board would be harmed by mineral oil submersion, then you'd get stronger thermal coupling in the tubes where they're held. Problem would be that you'd spread the heat pretty evenly, and that could force some weird drift in specs of the resistors...
I don't think going beyond these exotics would be beneficial. Hirsch/Vortex tubes might be interested, but would be an outboard solution (Demonic cooling (Maxwell Demon) would fit with ultra-high end audio... which cable manufacturer uses a product called pixie dust in their power cords?)
los_ren said:
I'm sorry, I can't see the picture. I would like to see a worst case scenario on what happens. Not that I doubted what happened, or why... AS3 is capacitive... and I found evidence that with enough pressure the imersive effect (Colloidal particles) of the silver breaks down and AS3 does become conductive.
Ceramique for me!
autoexec said:
Definitely, but I have a problem in knowing when to stop, and I use to lap things with my grandfather, he is Extreme for quality/fit/finish. I never could find any specific data on what level of progressive grit was good enough to consider it flat for heat transfer. I wish there was a process like Fire-lapping for firearm barrels... 30 bangs and you're pretty much lapped.
Vikash said:
That was just WD40 and a disc sander?? Would you mind telling me a bit about what technique you did, any difficulties you encountered?
I lap by hand, that looks way better than the other 'mechanically aided' lapping that I've seen. That would save so much time.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.