Sorry to butt into this discussion, but I just want to point out that my Shuttle HOT-433 v4 seems to have a damaged chipset due to the wrong voltage RAM being used by dumb-bunny-me. Attempted repairs on the BIOS didn't do anything noticable for stability so I'm almost certain that it's the chipset.
My Rev. 4 board seems to handle the AMD 5x86 CPUs quite well as well as the Cyrix based ones. Haven't tried my POD in there yet though.

That's very interesting - thanks a lot for your contribution. (I found your previous Vogons discussion here.) Also in that same discussion, feipoa says where to buy RTC sockets, which is also useful.

I wonder, does anyone have access to the shuttle hot 433 manual? It would surely have a warning about RAM and voltage in it. I would be very interested to read any official documentation about this mobo. Failing that, has anyone spotted any relevant posts on old usenet archive about this specific issue?

I'm also wondering - does anyone have any old mobo manual that says "warning: don't put 'x' voltage RAM in to this mobo (where 'x' is 3.3v)"?

It is just seems logical that you were to put in 3.3 V RAM into a 5.0 V RAM slot, then too much current will flow into the chipset, thereby burning out some component. This is not limited to 486 motherboards; I had a friend a few years ago order the wrong type of RAM for his laptop. The RAM fit in fine, but damaged the laptop. He couldn't boot into Windows anymore, even after putting his old RAM back.

As I mentioned, I've damaged an M919 using the wrong type of RAM. I don't recall anymore exactly what made the RAM wrong; it was IBM server RAM. I figure it must have been 3.3 V modules.

I was going to attach the HOT-433 manuals, but they are over 5 MB. Look out for an e-mail from me.

If you are going to try the RTC socket replacement yourself, I recommend testing this out on a know POS board you don't care about first. It is the desoldering that is tricky. I now have a 100% full-proof method and can do a whole conversion in 15-20 min, but I made many mistakes before getting to this point. Never use any force, even if you are frusturated. The trick is having an assortment of soldering tips and knowing exactly what temperature to use.

Thanks a lot for the info. Is the following a reasonable suggestion to make? Put 3.3V RAM in to a 5.0V RAM slot, and then measure the voltage flowing from the RAM in to the chipset. If that is a valid suggestion, would one require an oscilloscope [edit: sorry, it must be a multimeter, not an oscilloscope. end edit] to perform this measurement? If so, one could also use the correct voltage RAM as a secondary test, and do the measurement again, and if the voltage was significantly different, then this would double-check and verfiy this issue.

feipoa wrote:

It is just seems logical that you were to put in 3.3 V RAM into a 5.0 V RAM slot, then too much current will flow into the chipset, thereby burning out some component. This is not limited to 486 motherboards; I had a friend a few years ago order the wrong type of RAM for his laptop. The RAM fit in fine, but damaged the laptop. He couldn't boot into Windows anymore, even after putting his old RAM back.

As I mentioned, I've damaged an M919 using the wrong type of RAM. I don't recall anymore exactly what made the RAM wrong; it was IBM server RAM. I figure it must have been 3.3 V modules.

There are 3.3v SIMM's??

Tetrium wrote:

There are 3.3v SIMM's??

http://en.wikipedia.org/wiki/SIMM

Under the heading 72-pin SIMMs, there's a footnote which reads,
# These lines are only defined on 3.3V modules.

This seems to imply that there are. Be careful of those.

feipoa wrote:

http://en.wikipedia.org/wiki/SIMM […]
Show full quote

Tetrium wrote:

There are 3.3v SIMM's??

http://en.wikipedia.org/wiki/SIMM

Under the heading 72-pin SIMMs, there's a footnote which reads,
# These lines are only defined on 3.3V modules.

This seems to imply that there are. Be careful of those.

Very interesting. In the table on that wikipedia webpage, I wonder what they mean by "3v3"?

Also, @nemesis is there any chance you could please take a photo of your 3.3V RAM? If possible, both sides please. Thanks a lot if possible.

I'm pretty sure 3v3 is short-hand for 3.3 V

feipoa wrote:

I'm pretty sure 3v3 is short-hand for 3.3 V

OK thanks a lot. I didn't want to make any assumptions, because I wanted to be sure - just in case. I wonder why there's no warning message in the Shuttle manuals about this issue? An oversight on someone's part, I guess.

Just to satisfy my curiosity, I would like to test this. I am looking at a UMC chipset-based 486 mobo (not a Shuttle Hot 433 because I don't have one yet), and where the northbridge chipset is located, on the underside of the mobo, I can see lots of tiny "circles" on the PCB. Do I put a multimeter probe directly on top of these "circles", to see what the voltage reading is?

I could also perform this test on another UMC chipset-based 486 mobo, to see if there are any differences in multimeter readings. In order to probe the underside of the mobo, it would have to be powered on, with the mobo positioned upside down. I would have to balance it carefully, so that the video component card does not fall out. I'm sure that this can be done, using a bit of imagination.

I remember there was a "trick" to identify non-5v SIMM modules by reading the part numbers. I once tried to identify all of my SIMMs and most manufacturers used a similar way to mark their memory chips.
And iirc those non-5v memory modules were very uncommon.
I'll have a look this evening or so as I'm low on time right now.

And Nemesis, please do take a picture (or pics) of your memory module!

Without doing any testing, I guess sending 5v in to a SIMM that only needs 3.3v would make it heat up too much. But would it make extra voltage "leak out" from the SIMM, and go in to other areas of the mobo, such as the chipset? I am not an engineer, and so I have no knowledge about these things, and therefore I cannot answer that question, but what is the likelihood of this happening? Is it very obvious that this event would occur, or is there an element of doubt?

I'm just brainstorming here: I wonder if it could be something else. I wonder if the little metal connectors on the SIMM slot on the mobo get heated up, as the RAM SIMM gets too hot, and this in turn could damage some copper trace connections somewhere. I wonder if someone could power on a Shuttle Hot 433 mobo with 3.3V RAM in it, and touch the RAM to see just how hot it gets.

@Nemesis, have you retested your board recently, with a completely "fresh" set of components - such as a couple of never overclocked CPUs (one 5V, the other ~3V), with completely different RAM that has never been used inside this board, and a simple configuration that just consists of just one IDE HDD, using a controller that hasn't been used inside this board before?

Here are the pictures that were requested. I hope they help in some way.

Uploaded with ImageShack.us

@Nemesis, have you retested your board recently, with a completely "fresh" set of components - such as a couple of never overclocked CPUs (one 5V, the other ~3V), with completely different RAM that has never been used inside this board, and a simple configuration that just consists of just one IDE HDD, using a controller that hasn't been used inside this board before?

I haven't done all of that, and I plan to. I'll post the results in the original post though so I don't hijack this one. Time to start digging through my messes again 🤣 .

Oh wow! Those SIMMs really do look different to your "common or garden" SIMMs. And they're clearly marked 3.3V, which is really handy. Thanks a lot for posting those pics, they were very interesting to look at.

Seems clear as daylight that using 3.3 V SIMMs in a 486 leads to some broken components. We have 2 cases in support of this now. The question that is still somewhat fuzzy is if that broken component is the Northbridge chipset, or is some other motherboard component.

Someone will have to do their own research here in exactly how the Northbridge, cache, and memory talk to each other. Maybe the problem is with a fried address decoder or with a fried surface mount resistor. My gut tells me its the Northbridge, but I haven't investigated this matter with any lab equpiment yet.

retro games 100 wrote:

Oh wow! Those SIMMs really do look different to your "common or garden" SIMMs. And they're clearly marked 3.3V, which is really handy. Thanks a lot for posting those pics, they were very interesting to look at.

That is weird. I have a couple such SIMMs (with just 2 of those really flat chips) and those were 5v iirc.
But the sticker clearly mentions those SIMM modules are 3.3v.
Me ponders what they were used in. Perhaps only in a printer or something?

Me ponders what they were used in. Perhaps only in a printer or something?

That is exactly what I have been thinking lately. Either that or server, but the server chips usually look very different. Strange still is that these are supposedly 50ns.

What do server SIMMs look like?

The IBM server SIMMs that crippled my M919 looked similar to regular SIMMs, except I think the width of the IC's were larger. It has been too long really. That was 8 years ago now; I luckily got the seller to take the RAM back.

retro games 100 wrote:

What do server SIMMs look like?

I got a bunch of weird looking SIMM's. Perhaps will take a pic tomorrow.

Tetrium wrote:

I got a bunch of weird looking SIMM's. Perhaps will take a pic tomorrow.

Several 'tomorrows' have now passed. Any luck with the photos?