[ezkcdude]

I'm going to build my DAC prototype this weekend, but I don't have any heatsinks. First question: Can I get some heatsinks for LM317's at RadioShack? Second question, has anyone here used the paper clip method for heatsinking? If so, how much current/voltage drop does this work for?

 

[psilosome]

RadioShack lists this TO-220 heatsink as well as this one. But the stores seem to be cutting down on their stock of these sorts of things so you might be out of luck.

The regulator won't overheat immediately, giving you time to yoink the power if you think it's getting too hot. It's probably worth checking Tagent's power supply calculator to estimate how much heat the thing will shed.

Good luck with the DAC. I've been following this project as a possible outboard DAC for my Roku Soundbridge.

 

[Xakepa]

Do you mind elaborating on the paper clip method

Depending on the power consumption and drop-out you can safely get away without. My LM7824 was merely hainging on the wires while dropping 2-3V and delivering about 40mA. Once I upgraded to fully Class A biased Diamond buffers (2 pairs of thansistors each buffer, biased at 20mA), my PIMETA is sqeezing ~120mV/24V and even with a heatsink the chip is pretty hot, 50-60C I guess.

 

[ezkcdude]

I'll be feeding the DAC +/-15VDC using my DIY LM317/337-based PSU (which also doesn't have heatsinks). The input will be stepped down to 3.3V (for CS8416, AD1896, the XO, and digital side of PCM1794), 5V (for analog side of PCM1794), and +/-12V for the op amps (AD8610, AD8065, or AD8033, most likely). It doesn't seem like that much current would be consumed, but I suppose the relatively large step down in voltage could make a difference here.

Oh, as for the paper clip, I think you just attach a paper clip directly to the regulator. I'll look for those TO-220 heat sinks first.

 

[Garbz]

Paperclips aren't very good heatsinks

If you have a sheet of aluminium at home grab some tinsnips and cut a strip, fold the strip in half and clip it on the heatsink. It's nasty but it works well while you buy proper heatsinks.

 

[Voodoochile]

If it's not dissipating much, you could get away with just the reg in free air. I have a little regulator that's been running that way for several years. But if it running very warm at all, almost anything would be better than a paperclip. It's not even making good contact, let alone being a poor conductor.

Epoxy a penny to the tab if you can't find a clip-on sink... or fold one from some aluminum flashing or a beer can. Extreme ghetto heatsinking, but still an improvement over a paperclip.

 

[tangent]

Quote:

Originally Posted by Voodoochile Epoxy a penny to the tab if you can't find a clip-on sink...

I was (and am) dubious about this.

A heat sink only helps to the extent that it improves heat transfer from the heat source to another medium belonging to someone else, thereby making it Not Our Problem. The room's air is a popular choice among engineers, because the engineer doesn't pay the A/C bill.

For passive heat sinks, the only way to improve heat transfer is to add surface area. So the question is, how much does a penny change the surface area of a TO-220 part? Considering just the metal parts (the plastic doesn't conduct heat well enough to matter), very roughly speaking, it doubles the surface area. Now, you get full benefit of that only if there is no thermal resistance between the penny and the IC. That isn't going to happen, so I suspect most of the benefit is wiped out, especially if you use something with poor thermal conductivity like epoxy.

You know, if you have a hot enough iron, you could literally solder a penny to a TO-220's tab. That would probably save this idea.

Needless to say, a penny's worse than useless for any bigger package.

 

[nikongod]

people used to use window-weld to glue computer chips to their heatsinks. glue is great.

gluing a peny on would be cool if it was just to prevent the part from burning up from a MOMENTARY short (like unplugging) but i agree a peny would be too small to use as a legetimate solution.

tehre are also purpose built clip on heatsinks that are much smaller.

 

[ezkcdude]

Here's another enigma. How does one go about determining if the regs or other components are getting too hot? I certainly don't think the "finger touching" method is a good idea.

 

[cerbie]

I think the first linked Radio Shack HS is a nice one. Based on the Digikey listeing of the same item, it's 17.4C/W. If Radio Shack doesn't have one on hand, try stocking up on that and other stuff over at Futurlec (limited but useful selection, very good prices, fast and cheap shipping, and decent packing). Once your order hits $5 sans shipping, you'll have beaten Radio Shack pricing.

Too hot: good question. I've gone with the non-grounded finger method, so far

. Generally, though, estimate the power needs, and get a HS that can handle that and more. If it doesn't have to be pocketable, a bigger heatsink generally won't hurt. Mainly because you don't want to find out by it shutting off.

If my math isn't too shabby:
deltaT/W - ICR - PasteR = Rating in C/W

Let's call the LM317 (ICR) 5 degC/W and grease 0.5 in degC/W (I think both are actually lower), and deltaT as 50 degC (keep it below boiling). So:

W = 50/(5.5+Rating)

Plug in and get a low estimate of max power consumption. Then, some heatsinks may be eliminated, or you may be able to use any one that fits. The first linked RS HS should get you to a max of 2.2W wasted on the regulator.

With separate ICs doing the job for each section, it likely won't be too bad.

 

[Xakepa]

-A penny (brass) and TO-220 (tinned copper if I'm not mistaken) have different coef of thermal conductance, deltaT per W. Well might be that tinned copper is a better thermal conductor than (an oxydized) brass, have to look in the book

-The surface could be twice as big, but also could be much more - all you have to do is to scratch the penny with a Dremel or a rough file. Surface could easily go up tenfold, so the thermal conductance. I love to scrub the paint and scratch the bottoms of the heatsinks in my computer

-From my expirience, if you don't use smth like Arctic Silver (which I do), the contact b/w the penny/heatsink and the chip will be a major bottleneck, no matter how large of a heatsink you're going to get

-I'll be curious to see degradation curves for some of those chips, but from my (modest) expiriece electrolytic capacitors are suffering from heat the most, while output transistors can run very hot (above 60C) for many years with no audiable degradation of the sound

Just my 2c

 

[blaken]

just go to your local compusa or something similar and buy a north bridge heatsink. One without a fan. Then cut it up with a dremel. Use a little super glue on the edges and your good to go. I would also put a little thermal grease between the heatsink and chip. You could also buy video ram sinks but they would probably be more or even an old socket 5/7 cpu heatsink if you could find one.

That penny thing is just a bad idea.

 

[Garbz]

Quote:

Originally Posted by ezkcdude Here's another enigma. How does one go about determining if the regs or other components are getting too hot? I certainly don't think the "finger touching" method is a good idea.

Why not, burn your finger, too hot

Although yeah it would be a problem if the tab of your regulator or transistor sits high than 50V, if you aren't careful and provide a path to ground that is. If you use silicon or a different isolator between the part and the heatsink like I often do than this isn't a problem.

I use the KProbe on my multimetre to take a temperature reading.

 

[tomb]

Quote:

Originally Posted by Xakepa ... snip ... -The surface could be twice as big, but also could be much more - all you have to do is to scratch the penny with a Dremel or a rough file. Surface could easily go up tenfold, so the thermal conductance. I love to scrub the paint and scratch the bottoms of the heatsinks in my computer. ... snip ...

I doubt this is the case. The Reynolds # confirms that the size of air molecules is fixed. For instance, increasing surface area by thinning fins has no improvement unless you make certain there is also a corresponding space for the air. Your scratches will still be seen as a single surface by the air molecules. In fact, heat transfer may be inhibited, because the optimum surface area exposed to the air is actually a smooth surface.

This is analogous to the use of thermal grease - the greatest contact is if the mating surfaces are smooth. Any roughenss will inhibit the thermal contact, and thus, the heat transfer. In the case of contact with air, air is simply another contact surface (although a moving one, hence - convection). Increasing the area of that contact surface has benefit, but not through simple surface roughness - those are not "fins."

 

[ezkcdude]

Real-time updates! I just started building. The CS8416 was easy peazy. Just before I started on SRC4192, I accidentally grabbed the tip of soldering iron with two fingers. Whoops! That fricken' hurt. I'll deal it with later. Back to work

.

Screwed up the first SRC4192...But luckily, I pulled it off the board, and was able to get the second one soldered. Now, I just need to get PCM1798 done and the reset chip, and everything else should be cake.