Thanks for the detailed analysis. It seems you're coming down pretty strong on the side of an Intel 486DX2-66 needs a heatsink and fan, or if no fan, a ginormous heatsink. I had only one system with a stock 486DX2-66 and I barely remember it, but I believe it had a small heatsink and no fan.

I added active cooling to a 486 DX2/66 (heatsink and small fan), and I have a large one on a DX33. Those chips get quite warm, and while they can operate, I think for longevity it is best not to allow them to heat up significantly, to avoid stress of thermal expansion.

There is no harm to adding a heatsink, or even a heatsink and fan, as long as it is properly secured.

I have a small heatsink on my main Intel 486DX2/66. Never experienced any stability problems with that.

I really don't think it matters.

Longevity is unlikely to be affected much either way vs the standard ravages of time, since most 486DX2/66 computers are not actually putting on that many running hours.

BoraxMan wrote on 2020-07-12, 02:17:

I added active cooling to a 486 DX2/66 (heatsink and small fan), and I have a large one on a DX33. Those chips get quite warm, and while they can operate, I think for longevity it is best not to allow them to heat up significantly, to avoid stress of thermal expansion.

There is no harm to adding a heatsink, or even a heatsink and fan, as long as it is properly secured.

Don't use a big one in a tower though, then your introduceing mechanical stress on the CPU directly, a low profile heatsink is ideal

BinaryDemon wrote on 2020-07-11, 18:53:

hwh wrote on 2020-07-04, 08:29:

Anything above 25Mhz needs a heatsink.
Never hurts to have one, though I'm not sure what the adhesive they used for the early chips was.

I always heard it was 50mhz. I know I ran my 486sx-33 naked back in the day.

Well it's my opinion based on, essentially, OEM practice I have observed. For instance, this thing I have:

Now, they didn't make that for fun. They made it because the chip got too hot. It's clear enough that some 66s were equipped bare, but best practices indicate that's a limit, not a virtue. I suggest a contemporary heatsink.

The Serpent Rider wrote on 2020-07-04, 12:33:

Anything above 25Mhz needs a heatsink.

Not really. Yes, even if taken in context.

I'm not really interested in your opinion of my opinion. Shocking, I'm sure.

hwh wrote on 2020-07-12, 03:41:

The Serpent Rider wrote on 2020-07-04, 12:33:

Anything above 25Mhz needs a heatsink.

Not really. Yes, even if taken in context.

I'm not really interested in your opinion of my opinion. Shocking, I'm sure.

Hopefully you also liked johnnycontrario's analysis. I liked his scientific approach. It seems what Intel is telling us in their datasheet is that to safely use a 486DX2-66 with no heatsink, there either needs to be a constant 2.03 m/s (7.3 km/h or 4.5 MPH) wind blowing past it, or the temperature needs to be below freezing. So to me that sure settles it in favor of one. Whereas AMD's 3.3 volt 486DX2-66 can bask in 49.4 Celsius ambient heat with no heat sink or fan.

jakethompson1 wrote on 2020-07-12, 03:58:

Hopefully you also liked johnnycontrario's analysis. I liked his scientific approach.

Me? No. You could somehow meter the airflow you expect in your case, or incorporate ducting, or do some kind of correction for room temperature, because it's a given that the higher the ambient temperature, the less effective the cooling provided by the airflow is, and go on and on to some kind of formula that tells you according to the specifications if you absolutely need a heat sink, or if, after all these calculations you consider the effort (or whatever it is) to place the heatsink a "waste;" you can get away without it.

As far as a fan is concerned, that raises the bar, IMO beyond a 486 because in my own experience observing OEM practice some P233s didn't have fans; they they just had very tall heatsinks and a little case airflow. The majority probably did have fans, and some 486s had tiny fans to complement their tiny heatsinks.

If you aren't taking measurements you're just guessing anyway. I prefer to rely on examples provided to me rather than try to establish a scientific basis for what I was going to do anyway.

Interesting ! I put a fan on an Intel ODPRDX4-100 because it got nearly too hot to touch in a very ventilated case and I have enough experience to know that is not good. Also swapped an AMD 286-16 for a Harris 286 because the AMD ran hot and has a 3w+ TDP where the Harris has half that, now runs very cool. Did not need some case airflow tests or other analysis to tell me that. Not saying that taking measurements should be ignored but there are also other things to consider beyond case design docs, Intels thermal guidelines, manufactures design specs and taking measurements. Nothing wrong with adding a heatsink or fan if you think it helps to keep parts cool... just my opinion.

Based on Intel's datasheet, I would guess their reference cooler for a DX2-66 was probably the heatsink with a 40mm fan. That would easily achieve between 200 and 400 LFM, resulting in a max ambient temperature between 40 - 56C. OTOH, it's conceivable that a PC manufacturer could have taken the PSU fan into account and decided that it provided enough airflow to raise the maximum ambient temperature enough to use only a heatsink.

But it's funny, back in the day, I had an Intel DX33 with no heatsink and until recently, I thought that was normal. After looking at the datasheet, I'm kind of horrified. Anecdotal Evidence & Wild Speculation Ahead: Between the Am5x86 and DX33 chips that I currently have, the 5x86 part feels way hotter than the DX33. Both datasheets quote maximum ambient temperatures for a case temperature of 85C. I would hazard a guess that PC manufacturers maybe assumed that because PCs would live their lives in home and office environments, the ambient temperature may not get that hot. Or maybe the CPUs can safely dissipate more heat than the datasheet claims. End of Anectdotal Evidence and Wild Speculation

All that said. I don't think CPU cooling should be this controversial or difficult; find the datasheet and look at the reference cooling solution. As long as your heatsink is greater than or equal to the size of the reference heatsink, and that you have a CPU fan if the datasheet calls for airflow, you should be fine; you don't really need to do all the calculations I was talking about. I was simply helping jakethompson1 understand what the datasheet was saying.

Datasheets are what manufacturers use to design the PCs, cases, and cooling solutions that go on the shelves. It's not guessing if you know how to use component specs to calculate how a configuration will perform. The manufacturers of the fans, heatsinks, and CPUs have done the measurements for us; that's how the datasheet was created. Now if you want to talk about specs on retail packaging, then that's another thing. Marketers love to lie to consumers. I'm talking about the engineering datasheets.

I probably already said this, but seeing as nobody makes these chips anymore, it's probably better to err on the side of more cooling. Heat does wear chips down and can lead to eventual failure.

Whereas AMD's 3.3 volt 486DX2-66 can bask in 49.4 Celsius ambient heat with no heat sink or fan.

And they did. Some embedded stuff had these DX2 without heatsink.

I think I'd an OEM 486 33Mhz maybe from IBM I don't know that had a factory glued heatsink on it, low profile.

Really surprised alot of OEMS for the longest time stood by passive heatsinks. I just got a P2 450 taken from a Compaq and even that only had a chunky heatsink on it.

Makes me wonder how heat hurt these chips coming from OEM machines got. Cooking all that time without a fan, slowly draining life.

Not gonna lie. Modern computing really got me crazy over how cool my CPUs are 😜

SodaSuccubus wrote on 2020-07-12, 22:04:

Really surprised alot of OEMS for the longest time stood by passive heatsinks. I just got a P2 450 taken from a Compaq and even that only had a chunky heatsink on it.

I have a Compaq P2-266 with a Ginormous heatsink but no fan, the case it came from had a front fan that sorta blew on it. I wondered if it really kept it cool enough in a hot environment. added: I still have the Mobo too, it is a semi standard ATX 440BX (not EX or LX). Not sure why they had a 66Mhz P2 on it, would have expected something with 100Mhz bus like your cpu...

OEM's have the benefit of knowing which case the CPU will live in so can run control tests to see what cooling is really required.

My slot 1 based Compaq server for example the stock cooling was rated upto Katmai 500Mhz with passive heatsinks, any faster you needed a Thermal Upgrade kit which included a fan with higher CFM and ducting.
I've no doubt if they couldn't get temps to something reasonable they just wouldn't list it on the hardware compatibility list.

hwh wrote on 2020-07-12, 03:41:

Well it's my opinion based on, essentially, OEM practice I have observed. For instance, this thing I have: […]
Show full quote

BinaryDemon wrote on 2020-07-11, 18:53:

hwh wrote on 2020-07-04, 08:29:

Anything above 25Mhz needs a heatsink.
Never hurts to have one, though I'm not sure what the adhesive they used for the early chips was.

I always heard it was 50mhz. I know I ran my 486sx-33 naked back in the day.

Well it's my opinion based on, essentially, OEM practice I have observed. For instance, this thing I have:

Now, they didn't make that for fun. They made it because the chip got too hot. It's clear enough that some 66s were equipped bare, but best practices indicate that's a limit, not a virtue. I suggest a contemporary heatsink.

The Serpent Rider wrote on 2020-07-04, 12:33:

Anything above 25Mhz needs a heatsink.

Not really. Yes, even if taken in context.

I'm not really interested in your opinion of my opinion. Shocking, I'm sure.

I'm looking for that type of heatsink with clip that snaps onto CPU's package.
Where can I get one like this?

pentiumspeed wrote on 2020-07-13, 14:33:

I'm looking for that type of heatsink with clip that snaps onto CPU's package.
Where can I get one like this?

I have no idea, sorry. I've had it about 15 years...and I don't remember what computer (which I may or may not still have) I got it out of.

eBay has them

I always use active cooling on any CPUs it never hurts to have more cooling than necessary. I also try to make sure my cases have as much air flow as possible because I'm crazy like that.

candle_86 wrote on 2020-07-14, 12:54:

eBay has them

I have not found like one with metal clip that grips the ceramic package on ebay. What was your search term for ebay?

Cheers,

All my DX2/66 systems came with heatsinks. Most were upgraded from SX or DX25s.