A DX4 will work, yes.

PD2JK wrote on 2024-06-04, 04:43:

A DX4 will work, yes.

Thank you !

The thing is... a 'normal' SX/SX2/DX/DX2 requires 5V.

DX4 = 3.3V

The manual doesn't mention any voltage selection or jumper. You should get a DX4 OverDrive which has the voltage interposer.
Check the cooling as well. They can get hot with zero air flow.

It's pretty weird that the manual mentions a DX-33, a DX/2-66 and a DX/4-100 equipped system, and it more or less implies that they all use the same motherboard. Yet, no voltage jumper to be found anywhere in the documentation....it even lists the jumper for setting cache in writeback/write through mode. I refuse to believe that NEC sold systems with Overdrive CPUs...never heard of such a thing.
Automatic voltage detection is not out of the question. I own a board like this...however, I wouldn't risk it. I'd have a look through old usenet postings if possible. There's probably an undocumented jumper somewhere.

A voltage regulator near the CPU socket might also indicates the board supports 3.3V CPUs.
Pics! 😉

Well, there's support and support.
My IBM Value point has a 5v only board, but support DX4. There's no voltage regulator.
My Compaq XE supports DX4 on the 5v board as well.
Both work fine with the interposer and CPU is detected correctly.

So yes, it's perfectly possible the board has "support" without the correct voltage.

Could it be that the motherboard simply runs at 3.3v all the time?

I don't know about the 486DX33, but I vaguely remember the 486DX2-66 had been installed in both 5v and 3.3v systems.
- Ok, probably because in a later version the 486DX2-66 had been re-released as 3.3v model (3.3 to 4v)..

But in practice, did this really matter so much?
A 3.3v version didn't immediately die if being operated at 5v and in reverse,
many 5v parts could still work if being undervoltaged, too.
(Static CMOS designs were quite flexible, all in all.)

The threshold for 5v TTL signals is 3v.
That's the lower point at which RS-232 starts to have trouble determining high/low states.
(3.3v is slightly above, but longer serial cables can still drop the voltage on other end below 3v).

Edit: Never mind, my reply was late.

Jo22 wrote on 2024-06-05, 06:28:

Could it be that the motherboard simply runs at 3.3v all the time? […]
Show full quote

Could it be that the motherboard simply runs at 3.3v all the time?

I don't know about the 486DX33, but I vaguely remember the 486DX2-66 had been installed in both 5v and 3.3v systems.
- Ok, probably because in a later version the 486DX2-66 had been re-released as 3.3v model (3.3 to 4v)..

But in practice, did this really matter so much?
A 3.3v version didn't immediately die if being operated at 5v and in reverse,
many 5v parts could still work if being undervoltaged, too.
(Static CMOS designs were quite flexible, all in all.)

The threshold for 5v TTL signals is 3v.
That's the lower point at which RS-232 starts to have trouble determining high/low states.
(3.3v is slightly above, but longer serial cables can still drop the voltage on other end below 3v).

Edit: Never mind, my reply was late.

Yes, the later Intel DX2 were both 5v and 3.3v capable. Many earlier ones do run reliable in 3.3v too. Some AMD and Cyrix DX2 are indeed rated 3.3v only.
The name Powermate 433d suggests the system was originally fitted with a dx33 and that one certainly does not work reliable at 3.3v. So this board definitely supplies 5v to the DX cpu.
A '3.3v DX4' wil probably handle 5v fine for a little while, but running wat too hot even with upgraded cooling. It'll wear out much faster and leave you with a useless ceramic-gold display object soon enough.
(Correct me if I'm wrong, but I don't think there's voltage auto sense in the socket 486/2/3 scene)

waterbeesje wrote on 2024-06-05, 07:31:

(Correct me if I'm wrong, but I don't think there's voltage auto sense in the socket 486/2/3 scene)

The M-tech R418 claims to support voltage auto-detect. Not really sure how that works. From what I gather, it can only sense if a 5V CPU is installed.
https://theretroweb.com/motherboard/manual/r4 … 76017214273.pdf

You could be right about the board being 5V only but supporting the DX4 in the firmware. I've seen Dell branded interposers for DX4s that imply they did something like that.

Some DX2 do work stably at 3.3V, I have one that does and one that doesn't but there are some other aspects such as IO actually not meant to receive voltages higher than the supply voltage, which will be the happening in a typical 486 board.

3.3V CPUs have a separate IO power pin that still stays at 5V. This prevents ESD diodes from activating and trying to burn away voltages above 3.3V that enter the chip via the IOs since the top will be connected to that 5V supply line.
However, 5V only CPUs lack that and will eventually kill the motherboard when ran at 3.3V like this since the chipset outputs get excess loading on them from the 1.6V difference being burned away in the CPU ESD diodes. Parts of CPU itself will heat up more than normal due to this too and may also get damaged...

EDIT: As far as auto detection goes, 3.3V CPUs have a "VOLTDET" pin which the board can use to know if the CPU is 5V or not, and change CPU supply voltage accordingly. No clue how many things actually use this feature.