Cooling a 68040 CPU with a Big Copper Heatsink

[@ymkdevices4052]

Excellent video! The acrylic visual was great and it looks like you got the amount perfectly dialed in. On the visual artifacts, wishus knight makes a good point. You mentioned you got a higher stable speed without the IIe card. Could the card be pulling down your 5V rail? That and the OS-CONs suggest rail voltage might be a factor. I'm curious about what your 5V rail measures with and without the IIe card. You may have the headroom for safely increasing it. With overclocking, a few mV can make all the difference. I do enjoy your music selection. It adds a nice touch without stealing the spotlight. What is annoying about some other creators' videos, are tracks with strong vocals (often rap) which they attempt to talk over. Fortunately, this is not an issue with your videos. Thanks for the MacSD mention. I wish you and your family the best.

[@JDW-]

Thank you for the extensive feedback. And truly, without your help, this video would not have been possible. I also appreciate your thoughts on the music. I deliberately avoid tracks that have vocals because they annoy me too when listening to somebody talk on a KZbin video. I also try to keep the music volume lower than background tracks on other channels where possible. That's tough because it might be too hard to hear on some speakers, and then people will say it needs to be louder, but folks with am amazing audio system would say the opposite. I'm sorry I didn't have time to get everything read to demonstrate the temperature sensing of MacSD, as that would have added something fantastic to the mix. I will need to put a scope on the 5V rail to see fluctuations over time better. I can then update everyone on the result of that test. Thank you for not only making MacSD but for improving it. I do indeed plan to do a video on the version 2.x firmware at some point. It's pretty amazing.

[@JDW-]

I just posted detailed 5V rail measurements for you over at TinkerDifferent, including photos of my scope and Fluke benchtop meter: tinkerdifferent.com/threads/macsd-multi-device-scsi-adapter-with-audio.1362/page-3#post-18389

[@drywinddotnet]

Another great vid JDW! The fastest 68K Compact Mac! Now @ActionRetro needs to take the Cursed SE/30 up a notch!

[@michaelmullett7327]

JDW Fantastic Video, it shows how far we can take these old Mac’s!

[@Stryder_The_Nite_Owl]

Great video as always! Thanks for the incredible in-depth content. If only other channels did all of the thorough testing that you do...

[@JDW-]

Thank you for the very kind words, Stryder!

[@AppliedCryogenics]

I appreciate how methodical your process was. Doing test presses with the watch glass was brilliant! BTW, have you lent your voice to a text-to-speech synthesizer? I've noticed several "viral videos" here on KZbin, across multiple channels, which use the same voice, so I assumed it was some new TTS... which sounds like you! (No offense intended, you are clearly a real person.)

[@JDW-]

Thank you for your kind words. I've never lent my voice to a text-to-speech synthesizer, so I guess the similarity in voice must simply be a coincidence. 🙂

[@AppliedCryogenics]

@@JDW- sorry to leave it hanging like that. I just spent the last 34 minutes trying to locate one of these alleged KZbin shorts which have this voice.. and dangit, I can't find one! Murphy's Law strikes again. I'm going to keep trying for a little longer. I'm on a mission now... ;)

[@JDW-]

@@AppliedCryogenics By "shorts" you mean "short circuit"?

[@DrDavesDiversions]

Hi JDW - nice experiment, and work with the apparatus, and visualization with the acrylic. 23:45 Great to capture the ambient temp as you did, since the chip tests are perhaps best considered an offset from that. What do you think of a mechanical clip combined with paste? Point of comparison: my 68040 Amiga's CPU card (from early '91, predating Macs w/ '040) has thermal paste and a clip to the PGA socket. Aside, for casual temperature testing, I like the ThermoWorks Thermapen MK4 (usually used for cooking, smoking meats). It has a very fine tip that can fit in between most heat sink fins. For those that don't have a meter w/temp measurement, using cooking equipment is not a bad option, some with K-type probes, etc.

[@JDW-]

Dave, thank you for sharing that and for your kind words. The reason I decided to measure on the top of the heat sink, deep down inside the fins was because my Fluke meter’s temperature probe wire was too fat to fit through the slot in the socket to measure the temperature of the chip at the bottom middle area of the chip. It would be best to use MacSD for that, but I didn’t have time and all the parts to do that.

[@DrDavesDiversions]

@@JDW- Oh, I totally agree with your methods; just mentioning it for others and more casual measurements in (smaller) heatsinks. Great vid as usual!

[@alexandermcalpine]

Great Videos!

[@JDW-]

Thank you!

[@DatBlueHusky]

i would try using another cpu, its most likely that cpu is not stable at higher clock speeds regardless of cooling. I mess with amigas with 68060 cpus and some can overclock to 75mhz max while others can do 100mhz all day with minimal cooling. It just depends if you lucked out on a good cpu or not. I also have a 040 thats 25mhz, overclocked to 40mhz and it runs very cool and has no issues at all. The other thing i noticed is the masking on your cpu is coming off, it could mean its a sign its a fake/slower chip as I never seen that before with legit chips

[@JDW-]

Actually, I have tried another 68040 and after testing the two for many days, I determined that the one I show in the video could be overclocked higher than the other. More specifically, the other CPU I tested could only be overclocked to about 48.4MHz without the strange pixel artifacts. And yes, I tested multiple pairs of 60ns VRAM during all that testing. The pixel artifacts also varied by the 60ns VRAM too. One pair of VRAM sticks allowed me to overlock my "better" 68040 CPU to about 49MHz without any pixel artifacts (prior to my adding the IIe Card), while the other pair of 60ns VRAM allowed me to overclock to 50MHz stable. And it was after all that extensive testing that I later added the IIe Card and found that I was then getting pixel artifacts at 50MHz with my best 60ns VRAM sticks and best 040 CPU, which led me to believe something had changed. And indeed something did -- the addition of the IIe Card. So it was then I explored better cooling and shot a video of that work for everyone to see. It seems to have helped a little bit, but it didn't quite eliminate all the pixel artifacts at a 50MHz overclock. And I'm also not sure why MacBench 3.0 quits almost exactly 1 hour into "All Tests" when the clock speed is set higher than about 44MHz.

[@DatBlueHusky]

@@JDW- is the analog board been recapped at this point? They are known to leak mostly where the 2 bigger ones by the front right corner of the mac.

[@JDW-]

@@DatBlueHusky Yes it is.

[@another3997]

If the CPU was struggling with the clock speed, it would have crashed the machine under heavy loads, especially intensive CPU games like Marathon. But it clearly didn't crash. It could well be the vram or some intermediate component not liking the overclocked bus, or even the 2E card. There are so many components involved it's hard to tell.

[@computerhobbyshop]

Thanks and I agree about the VRAM causing artifacts in video displays. Anyway, you had a great idea on showing the spread of the thermal paste with the acrylic plastic demo, just brilliant! Sadly, I don't have a Mystic setup, but hopefully, I will find one someday and I would want to cool it also. Plus, here is an idea others never consider. Everyone wants to cool the top of CPUs but people don't think about the bottom. Anything you can do to get hot air off the back of the motherboard near the area CPU heats up will definitely lower the overall temps as well. I cooled the back of an AMD FX Bulldozer chip the other day and got a 20-degree Fahrenheit drop! I know it is not a Mac but it's an idea just offered FYI by @computerhobbyshop

[@JDW-]

Cooling the "top" side of a CPU using a heatsink and/or fan is the normal way to do it, but how did you go about cooling the "pin side"?

[@computerhobbyshop]

@@JDW- An under case super thin fan can blow air on "the back of the motherboard". It points at the CPU area, on "the back of the motherboard" and the temps I got from my example were 20°f lower than before. Of course, there has to be a case mod for the cold fan air to get through. For example, imagine an LC II case fan. It has holes in the middle of the front part on the case bottom. A 2nd thinner fan under the case near the CPU area holes would look about the same but the fan is thinner and blows in, not out. Of course, an LC II does not need it but is just an example we can imagine.

[@wishusknight3009]

i didn't recall if it was mentioned you were getting the artifacts before installing the ][e card? But this is probably a case where for myself i would run it 49 or whatever it was that clears up the artifacting.. i have never liked that kind of thing happening at all and usually associate it to some damage taking place.

[@JDW-]

Yes, I was getting artifacts prior to adding the IIe Card, depending on the clock speed used. I tested more than one pair of 60ns VRAM and two 040 CPUs and found one combination allowed me to overclock to 50MHz without the artifacts. I then added the IIe Card and started getting artifacts until I down-clocked to 49MHz. But without the IIe Card, I didn't need to down clock to 49MHz. The cooling seems to have helped reduce, but not eliminate, the artifacts at 50MHz. So I'm not entirely sure of the root cause of the artifacts.

[@wishusknight3009]

@@JDW- ug... its so close! I can probably only guess bus noise?

[@JDW-]

@@wishusknight3009 Hmmm. Bus noise... Well, I'm not sure how to best determine if it is that. But prior to reading your comment, one thing I had been pondering was the XC88916DW55 chip shown in my earlier overclocking video here: kzbin.info/www/bejne/gKPQq3yZmaeBhbs If the clock speed was too high for that chip, I would think it simply would stop working altogether. I don't see how it would create odd pixels on the screen. But it is a fact that chip is being driven beyond is specifications.

[@wishusknight3009]

@@JDW- I think the graphics display logic is also being driven higher. But I don't know if it existed on a separate clock domain. I would assume if it did, then it would not need faster vmem. And Although the ][E card is buffered, It is still directly connected to the local bus. it may be enough that the signals on the bus are not quite perfect enough. I don't know how the bus is terminated, if it is at all. Perhaps probing them with an oscilloscope would reveal some insight. But i am pretty much pulling a fart out of the wind on this one. It is just a wild guess that makes a bit of sense but also serves as one of them things that we can blame any mystery on... "its gremlins". I would also ask what is wrong with running 49mhz.. but I know the answer would be "because 49mhz is not 50mhz"... At least that is how i would answer lol.

[@JDW-]

The reason I want to find out what is limiting the clock speed in terms of generating those stray pixels is because Spicy O’clock can overclock to 52 MHz or so, plus, Kay Koba is about to release a new version that can overclock all the way to 100 MHz, which means it’s all the more important to find out what it is in the hardware that is limiting the overclocking.

[@tenminutetokyo2643]

You should get a NeXT Slab and refurb it next.

[@JDW-]

NeXT is one of those machines, like a LISA, that I never had a great interest in acquiring, regardless of cost. I relocated those items to the "if someone donates it to this channel someday, I'll do a video on it" category. 😇

[@Coffeeology]

Sir, another banger of a video. I am trying to track down that book about hacking the Color Classic from your friend. Do you know the name off the top of your head? Also is there an ISBN? どうも

[@JDW-]

Thank you for your kind words. I showcase "Doping Mac" in my VGA Mod video, and I put a Buyee link in the text description: kzbin.info/www/bejne/p5Cmc2qdhrmsaZo

[@SianaGearz]

Those chips should be super flat. The flatness of those alu heatsinks though is usually absolutely atrocious. Since the extrusion process makes it come out of the die a little bit warped, and then they just cut them down and dress them up a little but they don't machine the bottom flat; if they did, it would have been pretty good, even if they did a super rough job. Fundamentally nothing speaks against sanding alu heatsinks flat after they've been manufactured and eloxated, i've done this a number of times, on a mirror with sandpaper glued to it, though it takes a while. The two contact lines are probably along the extrusion direction, but sometimes they warp while being cut to length as well. Honestly the thermal impedance (°C/W) of the IC packaging with all that thick ceramic is so high that it's a little pointless to fret over how good the thermal compound is just as long as it has proper coverage unlike happened with the tape on alu heatsink, so something that appears adequate to me - after assuring that surfaces are of good enough flatness, is to actually use thin thermal tape. It's not the best material, way short of paste's performance, but it shouldn't add much to the overall impedance chain. Back before processors had a thermal diode on the die and when temperature control already became important, it was common to place a thermal diode into the socket under the processor, that worked fairly nicely. If the board had a spot there where you can just drill through without damaging internal traces and place a t-couple on the gold plated bottom lid of the processor package, bet it would be a superb way to measure the temperature. Of course there's a massive *if* in there, i can't think of many boards i'd dare drill into just like that, i can't fathom drilling into this one. A copper heatsink of the same exact external dimensions (envelope) as the aluminium one, given both are adequately designed, will usually transmit almost twice as much heat (correspondingly lower thermal impedance). So if you had marginal cooling that works but you find somewhat unconvincing, once you go copper, there's rarely a need to go up the size as well. If you want fairly nice and compact active cooling, there are old aftermarket VGA heatsinks, pure copper a couple millimetres thick with a fan mounted in the middle, very low profile, quite feisty really. The fins are arranged on the perimeter. There are also all sorts of stock heatsinks from the early mid 2000s era which have a copper base and copper accordion soldered to it, with a side blower fan, very low profile but for cases when you can go wide. You can also just buy a blower fan size 4010, it has a wide opening, and somehow fashion a side attachment to any given heatsink you happen to have. Just deliberating of course, you've done well, it works well.

[@fattymcboomboom9254]

Cant we control the cpu clock without overclocking the bus? Maybe by cutting the clock line to the cpu and soldering an external clock gen only for the cpu?

[@JDW-]

Theoretically possible. @Kay Koba has considered that, but I don't believe he has had actually success yet.

[@mikek1187]

60 degrees C on a ceramic die with thermal tape barely making contact? Ouch! There's a huge thermal delta between the chip and the ceramic surface where the heatsink actually sits, so if you were reading 60 degrees, the poor CPU was likely being roasted, hence the persistent black pixels showing up in your menu bar. Just my opinion, not based on any facts, except observation that ceramics is a poor insulator. Maybe try this again with a fresh 68040 CPU that hasn't been overclocked, with the new MX4 thermal paste and large copper heatsink and see if the black pixels persist? But, also, I wonder if with the new large heatsink and MX4 paste if you even NEED a fan? Excellent video, as always. Thank you!!!

[@JDW-]

Actually, I tested more than one CPU prior to adding the IIe Card and found that one overclocked to a maximum of 48.4MHz while the other "better one" (shown in my video) could overclock to 50MHz with the right pair of 60ns VRAM sticks -- I tested more than one pair of 60ns VRAM, and there are variations between VRAM pairs, probably due to variations in the RAM chips themselves. As to the chip being roasted, it didn't get much beyond 60°C even with no heatsink at all. It was after I did extensive testing without a heatsink that I later added that aluminum heatsink with thermal tape, and it was only during the recent making of this new video that I removed the gold heatsink and discovered it had not been making contact in the center of the chip. If you want to see more of my overclocking and testing adventure, prior to adding the IIe Card, I have two videos on that topic here: kzbin.info/www/bejne/gKPQq3yZmaeBhbs kzbin.info/www/bejne/e5C9eWuZp9mMi5I

[@mikek1187]

@@JDW- Copy. These were data points I was not aware of when I made my comments. Also, I'm curious if perhaps you've been able to monitor any voltage drops at the CPU, especially when the black dots appear on screen? Would an oversized capacitor possibly assist in this situation? Edit: I did watch both of your linked videos in the past. In fact, I have a Mystic Classic and now, thanks to you, a couple Spicy O'Clocks, with one waiting to be installed. ;)

[@JDW-]

I haven’t made any voltage measurements yet, but it is a good idea. I do know that you cannot overclock much past 42 MHz unless you recap the motherboard with extremely low ESR OS-CON capacitors. So yes, capacitance and ESR does play a role when overclocking on this particular motherboard.

[@ymkdevices4052]

Because of the fin density, these heatsinks aren't great for passive / convection cooling. They need some airflow, but a little goes a long way.

[@mikek1187]

@@JDW- Oh, and a little tip: If you're not satisfied with using alcohol as a cleaning solution on copper, try muriatic acid for 1 minute and a generous water rinse afterwards. I do electroplating as a hobby and use muriatic acid to clean metal before plating. For your situation, the thermal benefits may be negligible, but I do it with liquid metal cooling and I do see a 1-2 degree C difference between oxidized copper and deoxidized. Just a sidebar :)

[@jimjimx5418]

please STOP the background music

[@JDW-]

Understood. I appreciate the input. I tend to put music at the beginning and end, but for this video I added more so as to make it perhaps less boring for people with shorter attention spans. I've come to find that younger people tend to like background music, so I was keeping them in mind when I made the decision to add more music this time. Anyway, such that I can better understand your feelings on music (which may or may not apply to other viewers as well), is it because you (1) dislike all background music in general, or (2) dislike the particular selections I chose for this video, or (3) dislike the volume of the music being too loud? With regard to volume, I did not choose that at random. I've actually consulted with people on past videos who said the music was either too loud or not loud enough. And based on that feedback, I made the volume decision for this new video while listening to the entire video multiple times with both headphones and with the internal speaker of late 2015 5K iMac and with the internal speakers of a new model MacBook Pro. The end result was a happy medium where I got the volume level loud enough to be heard on headphones and the Macs, yet not so loud it drowned out my voice. With that said, some people hate all music in the background, and it could be you are in that group. So that is why I asked you the questions in the first paragraph. Based on your kind answer, I can better consider what to do for future videos.

[@mikek1187]

@@JDW- I personally didn't mind the music. For me personally, please keep the music, but you can't please everyone ;)

[@JDW-]

@@mikek1187 Thank you for telling me that. In all the years I've been on KZbin, I only received 2 complaints about music. I'm serious. The first was by private email from a friend this summer when I sent him a preview link to one of my recent videos at that time. He said the music was fine but a tad too loud for his nice speaker system. So because that was before I made the video public, I reduced the volume of the audio, let him listen, and with his blessing I re-uploaded the video and later made it Public. Nobody said anything about the music at all. The second complaint was the one today by jim jimx. So 2 complaints in total, and one of the 2 people was satisfied with my volume change because he kindly told me it was the volume and nothing else. He also is a friend and he said a lot of good things about my video too, which is of course why I went out of my way to ensure the audio didn't hurt his ears. I have a rather detailed weighting system when it comes to complaints. First, I appreciate all comments, even complaints. However, I cannot weight all feedback equally. Logic and reason must dictate how I ponder each bit of feedback. For example, jim jimx only lodged a complaint and made no mention of the primary content of my video. Logic dictates that I give less weight to those comments. Also, if jim jimx never answers the questions I asked him, I will further need to reduce the weight of his criticism. And then of course, I need to give ear to those people like yourself who say that all is well when it comes to the selection of music and the volume. I must admit, I don't really like adding music at all. It eats 2 days of my editing time just trying to figure out what FREE music to use and from where. I prefer no attribution music where possible. I don't pay for a music subscription because I don't get enough donations to cover that. Despite the pain involved, I realize my videos would be even more boring without music, especially for younger viewers, so I spend a lot of time trying to find the best compromise on music and volume for them. I know I can't please everyone, but I spend a huge amount of time trying. It's my hope some people will see those "little things" and appreciate it. That's also why I keep Ads switched off on my videos. It wouldn't bring in much money anyway, and I know it would annoy people greatly. I want people who watch my videos to be happy, not sad.

[@johnsnook2358]

@@JDW- I like the music.

[@JDW-]

@@johnsnook2358 Thank you for making time to tell me that, John. I really appreciate your feedback!