Video

Nah this is for sure a real copper heatsink. You can tell just when you hold them that the density of metal is much higher. If it was any other metal it couldn't have possibly performed nearly this well. You expect to achieve approximately as good a performance in half the total volume envelope when using copper as opposed to aluminium, for ideally designed heatsinks, and in this regard, it comes remarkably close, being as tiny as it is and half as tall as the aluminium one plus a little smaller in footprint. Both heatsinks have some design compromises, the aluminium has its base and fins maybe a little too thin, while copper one has its base too thick and radiation area is too small.

@@SianaGearz The Air velocity and temperature are limiting performance here. If the "copper" heatsink were steel or zinc with copper plate you are unable to tell the difference by picking it up The dimensions of the "copper" heatsink make no sense for a material with such good thermal conductivity.

@@2OO_OK If it was a steel or zinc heatsink, it would not come anywhere near close to the performance that it has shown, it would fare much much worse. Don't expect magic from copper heatsinks just because it's copper. Of course it can be demonstrated that it's same material throughout and not just surface plated by cutting into it. I had cut into heatsink looking just like this a long long time ago and it was genuine, though of course it doesn't mean that new production is. But i also pay about 6 times more for copper heatsinks than for aluminium ones, they don't cost the same. I had as such not run into fake copper heatsinks yet, but it would indeed be easy to fake by plating, so buyer best beware.

My rule is below 60C.

Every 10°C increase roughly doubles the speed of degradation. But there's no hard rule to how long they have to survive at a given temperature, many semiconductors last for 20 years continuous 80°C and more. Indeed if you can touch it bare hand and not feel being burned alive, so on the order of 55-60°C maximum, that is definitely well on the safe side. As long as you don't get condensation, run classic chips as cold as you can manage to. Deep cooling under the ambient temperature is potentially a little unwise for this reason, but there's no harm in adding airflow and indeed this will increase heatsink efficiency drastically.

@@SianaGearz A cool project would be mount an effective fan that sucks, rework the chassis/body so it is sealed the right places, you can add heat sinks that you connect with the RF shield, you redo the RF shield in aluminium, no edges, just smoothly radiused bends where needed, add a correctly sized bellmouth where the air should get in. Oh, that would be a nice flowing cooling system. You could do a ligth version though, with only cooling where needed, but a custom Aluminium RF shield with louvers and fins would probably help if done correct.

Water cool your C64 because it'd be funny

Fans, ie air pumps and openings.

The plastic insulation on power cords increases the heat transfer from them - critical radius of insulation for PVC is a bit under 1", so up to 1.5" diameter increasing the insulation thickness means they run cooler. Plastic is widely used to dissipate heat, despite being widely dismissed for that role. The fins on that sink should help a lot under natural convection, as the vertical surface is aligned with the flow of hot air. Aluminium sheet is a cheap way of buying heatsinks, until you run into the sinks for other components on the board!

@@maxhammick948 interesting. I know that painted bus bars have higher current carrying capability than unpainted ones, but I never heard about that on wires. Got any data or referrence on that?

@@wernerviehhauser94 The extra surface area helps more than the added thermal resistance, because one increases linearly with radius and the other with ln(radius). It all boils down to r(crit) = thermal conductivity / convective heat transfer coefficient; youtube doesn't like links in comments but the derivation is widely available online if you google "critical radius of insulation"

@@wernerviehhauser94 CZcams seems to have eaten my first reply, so second attempt: - adding more thickness adds surface area faster than thermal resistance, for small diameters (and small depends on the thermal resistance of that material and the convective heat transfer coefficient) - for a full derivation, google "critical radius of insulation" - it's in many places online

​@@wernerviehhauser94CZcams keeps eating my comments, so hopefully this gets through. Below a certain size the extra surface area helps more than the extra thermal resistance - the derivation for critical radius of insulation is replicated in many places online if you want to see the maths

Also being black only helps it. At least if the coating is done properly of course, too thick and it will become an insulator instead of helping with radiation.

yeah, the penny copper things are sad compared to ones I've seen eons ago that had much taller fins and less thermal mass in the base. I think it was around 20 years ago they stopped making good motherboard VRM heatsinks when motherboards started having fashion accessories over the VRMs. it's a bit sad as them blocks are a good size for cooling TO-220/TO-263 ICs. BTW, the copper is only good for looks, the tape and air contact area make the metal used a bit moot for small low watt devices, lol.

You still need airflow and heat air exchange, otherwise the internal airspace of a piece of equipment just gets to X number degrees above ambient depending how well the enclosure exchanges heat with the ambient air. A small fan pushing air into the computer will dramatically lower the internal temps.

NOPE!!!! it is about getting the heact out of the chip... and you can have the biggest assed heat-sink in the world and it will not make the slightest difference, if you are at the transfer rate of the interface material. Which is usually the resin and you can find that data on the chip specsheet. the heat goes from the silicon to the resin & leadframe. , and THIS is why socketed chips are a BAD idea., you are actually increasing the internal temp of the silicon die. , but you will get some smart ass who will measure the case temp and say nope!!! but they forget that the temp they are measuring is the resin case, which has a maximum thermal interface temp., it's NOT a perfect heat transfer system. if you look at the chip ends.. usually you will see 2 dots of metal at each end (old style chips), this is the lead frame the silicon sits on. the pins sit inside the resin & are connected by the bond wires. they remove the lead frame carrier on most modern chops.

A bit of Dremel work to make the grooves deeper would probably help a bit.

The flat HS is merely named as such, its junk. better tape will make no effective difference, the HS just sits there at the ICs temp and cannot radiate, a very high speed air nozzle etc etc, but the noise and the air pipes lol.

Undervolt them slightly. Say -0.15V

Hence why audio amplifier ICs have ground pins tabs coming out the side, or flush (to PCB) HS plates under them. See the TPA3110 TSSOP package.

@@simontay4851 TTL runs between 4.5V and 5.5V, ie rated at +/- 10% Replacing some TTL logic with 74HCL might be an option if the timings are not different and critical. Replacing any 78 or 79 series regulators with small SMPS types would removes some of the internal heat.

I mean, it's maybe an easy mistake to make if it's a "high temperature" tape?

Yes. The DF Robot was meant for a Raspberry Pi. I ordered these because of the price and width matching the DIP IC being cooled. It wouldn't surprise me if I missed something better and cheaper. 😄

@Retrommodore digitec sells a brand which are a dollar each and well worth it.

Industrial PCBs for potentially harsh environments get acrylic conformal coatings.

actually I'm not sure if sealing 40 yo chips today would be any good. About the surface area, usually the sides of the IC don't run very hot, since the chip itself stands in the middle. Of course, since the price of this heatsink is just cents, why not to put along the whole surface... but the hotter one will be always the one in the middle. When I open one of my Commodore computers (VIC, 64, etc.) I use to put heatsink in all the big ICs and as rule of thumb, when I open an old computer I observe what is the temperature of every chip; it runs hot, it gets an heatsink. In the C=64 I use more I also installed a little fan to blow air onto the VIC-II. It's a 12 volts fan with a resistor; it runs slower because of the resistor, so it's quite silent, but it's enough to keep the VIC-II totally cold (afaik it's the hottest chip in the C64).

it's a double edged sword, actually it is maintaining consistent temperature. during that era , the dynamics of surface separation were not well understood.

Wouldn’t two different metals touching cause a galvanic reaction which would lead to corrosion?

@@trippycat only if there's an electrolyte, which there should not be in this case.

@@gwong940 would a graphite thermal sheet further reduce the chance of galvanic corrosion due to humidity in the air?

That's just wrong. Copper is great at conducting temperature, both heat and cold. It doesn't magically easily gets hotter and don't release the heat outside, otherwise we would have a perpetual motion machine based on this. Also, this is against laws of thermodynamics. The problem with the copper heatsink in the video is the surface area, it's multiple times smaller than that of the aluminum heatsink.

​@@dirtyclanner2250 No idea if you are right or wrong, but I'm talking from experience. I tried a copper heatsink to cool a chip and it just gets really hot, really fast. You can put your hand close to it and not feel the heat in the air. Tried the same thing with an aluminum heatsink, even a little smaller than the copper one, and it stays at a lower temperature and you can feel the heat in the air. I also don't think it goes against thermodynamics. If you take more energy than what you lose or release, you get hotter until you meet equilibrium or it melts. Maybe I have some concept wrong.

You can see three of the copper heatsinks mounted on the VIC at 6:06 so I assume that's how they were tested, but I agree, these small heatsinks have too little thermal mass and too little surface area for what I'd consider long-term-solution temperatures. Still better than nothing, but yeah ¯\_(ツ)_/¯

One heat sink of each kind was used for the test. When it comes to VIC and VIC-II I put three on.

Thanks. I'll look into that. Assembly is a must!

Yeah and specifically using the /FLAG2 as I do in my recent video where you can trigger an interrupt that way you don’t waste cycles polling.

One more suggestion: turn off the video interrupt in the VIC chip as well. Any interrupts will introduce errors and with a CPU that slow it won’t be simple jitter, it’ll be not even in the ballpark.

I must say that I was surprised about how low a frequency that the basic program failed. The clock frequency of the 6510 is 1MHz, allow 100 clock cycles per measurement yields 10KHz. Another way to think about the maximum frequency achievable is to think about the maximum supported modem speed that a C64 could support. From memory I think this was 2400bps half duplex which suggests a maximum counting limit of about 1KHz. At a guess I think machine code would be between 10 and 100 times faster than basic. It would be interesting to see how much screen blanking would increase the maximum. (I think about 20%.) You could also try crunching your program by using colons between code lines compressing two or more lines into a single line, eliminating spaces in the lines and entering as tokens rather than English style text. All efforts to reduce the basic interpreter overhead. Cheers David

Thanks to both of you for the advice. I'm starting to look into using machine code which I'm still learning. It wouldn't surprise me if my basic program is far from optimal. I expected it to be very limited in what it could measure.

@@davidbaker4867 I'm not surprised at all. I've built a few retro 8-bit systems which run Microsoft BASIC; mostly of the MC6809 variety. It really helps running BASIC and a UART to do board-bring-up since you're testing your devices quickly from a sort-of command-line. Devices that need long delays between communicating with in machine code need no such thing in BASIC when using PEEK/POKE in the most efficient way possible. It's just a real dog to constantly re-interpret text over and over the way these BASICs are implemented. Higher frequencies could've been attained had BASIC compiled to byte-code and interpreted that way or into some sort of threaded code as in FORTH. But that wasn't the goal for these 8-bit BASICs running on systems with such limited memory.

Thanks. I've though about measuring the VIC-II when during idle vs. lots of graphics on the screen.

@@Retrommodore I would expect, there is no difference. The VIC-II is doing the same work, no matter of the content. That‘s my guess, prove me wrong 🙂