VOGONS


First post, by seleryba

User metadata
Rank Newbie
Rank
Newbie

Hi Vogons

I have bunch of Socket 7 motherboards. On most of them, heatsinks of the voltage regulators are super hot.
I mean, I can touch it only for 0,5s. I measured the temperature using IR sensor and it's around 80 * C.

On some boards this heatsink is just warm.

My question is: is that OK? Or it's sign of ageing of some kind?

For the tests, I'm using Pentium 166 MMX.

Reply 1 of 20, by dan86

User metadata
Rank Member
Rank
Member

Make sure the voltage jumpers area set right.
That being sad, the VRMs of a lot of socket 5 and older socket 7 bards were build in a way that makes alot of heat.
If the heat is a problem look into getting a PowerLeap PL-K6-III.

Reply 3 of 20, by ATauenis

User metadata
Rank Member
Rank
Member

It is a normal condition. These VRMs have very low efficiency and produce more heat than power. Transistors and PWM ICs used in them are high-temperature and in most cases working without any problems. But hiding them under a huge bunch of wires may fasten overheat of VRM.

dan86 wrote:

a lot of socket 5 and older socket 7 bards

More precisely, not "older socket-7", but simply "a lot of socket 7 except super-socket 7" boards. 😀

2×Soviet ZX-Speccy, 1×MacIIsi, 1×086, 1×286, 2×386DX, 1×386SX, 2×486, 1×P54C, 7×P55C, 6×Slot1, 4×S370, 1×SlotA, 2×S462, ∞×Modern.

Reply 4 of 20, by seleryba

User metadata
Rank Newbie
Rank
Newbie

Voltage jumpers are set OK: 2.8V and 3.3V in this 166MMX case.

Thanks for the replies - now I feel a bit more safe. I thought that's an issue of the old regulator, or some caps issue... Thank you.

Reply 5 of 20, by mpe

User metadata
Rank Oldbie
Rank
Oldbie

I'd say this is normal. These parts often have operating junction temperature up to 150 or even 200 °C. Regulators get hot during operation. Just make sure they are separated from other parts which might not handle such a high temperature.

On the other hand the dissipation is proportional to input/output voltage difference (and power) so if your power supply gives higher voltage as due to aging caps or something regulator could generate more heat than when they were new. However, chances are other things will break before regulator if your 5V is way off...

Blog||486DX-50|NexGen 586|S4

Reply 6 of 20, by dan86

User metadata
Rank Member
Rank
Member
ATauenis wrote:
dan86 wrote:

a lot of socket 5 and older socket 7 bards

More precisely, not "older socket-7", but simply "a lot of socket 7 except super-socket 7" boards. 😀

I have to disagree. Wile a lot of socket 7 boards use older VRM types that heat up, its mostly early ones. By the time the MMX CPU's were out most boards moved onto new VMRs and the heat problem was gone.

Reply 7 of 20, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie

I just checked my Fairchild LM7805 datasheet and it's rated to 125C. Yikes! I used to have a Cyrix 5x86 board that would instantly shrivel up heat-shrink tubing when I held it against the VRM. That's too hot for my liking, but those boards were usually designed with puny heatsinks, and no expectation of forced airflow over the sink, and yet they survive.

Reply 8 of 20, by Windows9566

User metadata
Rank Member
Rank
Member

the regulator heatsink on my Intel TC430HX and AN430TX gets hot too, it doesn't run as hot when i use a regular Pentium on them. at least intel added large heatsinks to them unlike the puny heatsinks that i've seen on many of the baby-at socket 7 boards.

Reply 10 of 20, by Vynix

User metadata
Rank Member
Rank
Member

I am having a very similar issue on my Shuttle HOT-555A motherboard (just your garden variety 430VX motherboard, nothing fancy or "out of the ordinary"), whenever I set the motherboard to use a Pentium MMX, one of the regulators would get scorching hot, so hot that this board became somehow unstable (such as random data corruption bugs when installing Windows 98, locks up systematically ⅓ of the time during a cold boot... And sometimes it would straight up reset by itself 😦). And weirdly enough if I put a regular (Non-MMX) Pentium, everything goes back to normal, with the normal Pentium, I noticed that the offending VRM only gets slightly warm, so I guess I'm onto something there.

Now I've noticed that there are three distinct regulators on my board, two by the CPU socket (these two never get hot or warm), and another one which is not too far the memory slots, this is the one that get very hot whenever I set the board to use a MMX Pentium.

And to top things off, this board has no heatsinks for the VRMs whatsoever... I don't know how should I go about that, try to unbolt the regulator, then fit a heatsink on it and bolt it back on or set up a fan blowing some air over the offending module?

This really is grinding my gears not because I paid anything for this board (got it for free, only paid 10 bucks or so for the Pmmx) but because I bought a CPU to max it out only to find that my motherboard isn't able to cope with it even though it's listed as compatible with my board.

So I'm going to follow this topic, because I'd like to see if there is anything that can be done about these toasty regulators.

Proud owner of a Shuttle HOT-555A 430VX motherboard and two wonderful retro laptops, namely a Compaq Armada 1700 [nonfunctional] and a HP Omnibook XE3-GC [fully working :p]

Reply 11 of 20, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie

There are a gazillion options for heatsinks assuming your VRM is a normal TO-220 package (if you don't know packages, just Google that number -- does it look like that?) Bolt-on is easiest. Keep in mind that the metal back of most TO-220 ICs is actually connected electrically to one of the pins. Keep it isolated from anything else conductive and/or use a thermally-conductive (but electrically isolating) pad, and a non-conductive through-hole washer for the bolt.

Reply 12 of 20, by seleryba

User metadata
Rank Newbie
Rank
Newbie
Vynix wrote:

I am having a very similar issue on my Shuttle HOT-555A motherboard (just your garden variety 430VX motherboard, nothing fancy or "out of the ordinary"), whenever I set the motherboard to use a Pentium MMX, one of the regulators would get scorching hot, so hot that this board became somehow unstable (such as random data corruption bugs when installing Windows 98, locks up systematically ⅓ of the time during a cold boot...

I have EXACTLY this same case, including the cold boot one. When the regulator is on its maximum of celsius, when I turn off the PC and turn on again I have big chances that it won't boot the board. I need to wait 10-15 seconds, then it works.

Reply 13 of 20, by Vynix

User metadata
Rank Member
Rank
Member
SirNickity wrote:

There are a gazillion options for heatsinks assuming your VRM is a normal TO-220 package (if you don't know packages, just Google that number -- does it look like that?) Bolt-on is easiest. Keep in mind that the metal back of most TO-220 ICs is actually connected electrically to one of the pins. Keep it isolated from anything else conductive and/or use a thermally-conductive (but electrically isolating) pad, and a non-conductive through-hole washer for the bolt.

Yup that's a TO-220 packaged regulator, and it seems that the metal back is connected to ground, as currently it's placed on a thick ground trace, it's hard to describe by text, but I'll see if I can snap a picture of the whole thing. Truly it's one of the most bizarre designs I've ever seen.

Proud owner of a Shuttle HOT-555A 430VX motherboard and two wonderful retro laptops, namely a Compaq Armada 1700 [nonfunctional] and a HP Omnibook XE3-GC [fully working :p]

Reply 14 of 20, by Deksor

User metadata
Rank l33t
Rank
l33t

I have a HOT 557. Not exactly the same board, but exactly the same behaviour with K6-2s.

I managed to solve the problem though ... by screwing a fan in the transistor's heatsink 😁
Then the cooler was super cool 😀

Trying to identify old hardware ? Visit Ultimate Hardware 2019

Reply 15 of 20, by Repo Man11

User metadata
Rank Member
Rank
Member

Vogons member Feipoa recommends putting a small fan near the VRM on the Biostar MB8500TTD to prevent overheating when using a 400+ MHz CPU such as a K6-2 or K6-2+. Even with an Intel 233, it gets too hot to touch.

Many moon ago I had a Biostar MB8500TVX, and when I had a bad flash with my Abit KT7A, I ended up knocking together a system using that board to get me by until my new BIOS chip arrived. I had a K6-2 450 that I ran at 400 (if there are settings to run the FSB over 66 MHz on that board I didn't know them) at 2.6 volts as that was the lowest core voltage (again, if there's a lower setting, I didn't know it). It was working okay, and I was playing Jane's WW-2 fighters on it one hot late summer or early fall night, when I heard a loud pop, and it shut off, never to work again. I now suspect that it was the VRM that blew.

Reply 16 of 20, by Vynix

User metadata
Rank Member
Rank
Member

Well, well, well, I tried putting a fairly large fan blowing on the VRM that would overheat, aaaand.... Nothing. Windows boots to a "SCANNING REGISTRY" prompt and ends up rebooting, each boot I'm back to square one: this damn registry thingy again.

Reinstalling windows causes all sorts of bugs (SUWIN crash, data corruption on the hard drive.... Yes I even tried with a SCSI hard drive and its matching controller)

I'm practically about to give up about this motherboard, it's driving me insane, and I don't wanna pay the egregiously high prices for a replacement S7 mobo that some sellers want on FleaBay (looking on the local classifieds as I'm writing this...)

I'm not entierly sure what to do at this point... I don't have any other PSU, and this motherboard refuses to start if I use a ATX-to-AT adapter harness, I'm sure it's because my ATX PSU doesn't have a negative 5v rail, so I'll have to either get my AT PSU recapped, or find a ATX PSU that is old enough to have a negative 5V rail.

I don't know why, but I feel like old computers REALLY like to troll me somehow.

Proud owner of a Shuttle HOT-555A 430VX motherboard and two wonderful retro laptops, namely a Compaq Armada 1700 [nonfunctional] and a HP Omnibook XE3-GC [fully working :p]

Reply 17 of 20, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie

Haha... part of the creedo of being a retro computer enthusiast is knowing you have to have at least two of everything. 😉 This really could be caused by any number of things, and it's only possible to tell by swapping out parts until the problem goes away. I had a similar issue on a Pentium system with Win 95. I was playing Warcraft II on it, and I would get to a certain mission, the sound would start going weird, and it would eventually develop graphical glitches, crash, and then I would be left with an unbootable Windows partition with tons of corrupt file/dir entries. I reinstalled, and it happened again, coincidentally on the same mission. I was starting to think it was a bug in the game, but on the next reinstall it happened sooner. Turned out to be a bad hard drive. You just can't trust those early single-GB silver Western Digital drives. They only come in two flavors: Failed, and failing.

Reply 18 of 20, by Vynix

User metadata
Rank Member
Rank
Member

Eh, I already tried with a CF card adapter that I borrowed from a friend, same result, so I think we can rule out the hard drive.

Now if that's the PSU or the motherboard that are faulty... I'm not sure what to do, unless it's the CPU itself that is faulty (this would be a surprise given how CPUs are usually resilient to damage).

Don't have a spare AT PSU, same for the motherboard... I can smell the loads of headdesking trying to track down a faulty component causing a ruckuss. Curiously enough when I tried today, with the PMMX, the VRM didn't get hot... Much to my surprise and disappointment...

I'm sure the fault is something quite obvious, so yeah I got the TO-220 heatsinks coming up soon (I have a few dead/junk PCBs with TO-220 regulators and heatsinks, just need to find where did I put these boards in my junk heap)... So far the fan didn't do the trick (and it was a fairly chunky fan)

Proud owner of a Shuttle HOT-555A 430VX motherboard and two wonderful retro laptops, namely a Compaq Armada 1700 [nonfunctional] and a HP Omnibook XE3-GC [fully working :p]