PC Archeology: One of first Compaq Deskpro 386 made... but how early?

[dhamilton]

So I recently acquired a Deskpro 386/16, dated from September 1986, the month they began shipping.

Some clever person has already explained either here on VOGONS, that the beginning of the old Compaq serial numbers give you the year and week a machine was assembled.

This machine's serial number is 4637AJ4B0052. So this machine was build week 37, 1986. This dates it between September 8th and 12th 1986. Amazing thing is, the Deskpro 386 was announced to world and began sales on September 9th 1986.

So it's pretty certain this was one of the first machines to roll off the line... but how close? The serial number ends with "0052." Does that mean this could be "the 52nd machine" off the line? I don't know how to decode AJ4B. Not sure if anyone does, but here's to hoping.

Does anyone have theories about what AJ4B might mean?

 

[Timo W.]

Parts of AJ4B most likely refer to the factory it was made in. 0052 is not an absolute number, as I'm sure they would not limit the max. units possible to 9999 from the outset. Very likely to be the 52nd system built in week 37 of 1986.

Note, however, that it is not that early as you might think. Production starts months before something hit the market, as logistics require this. You can't produce a computer and have that very unit in a store next week. When your Deskpro was made, production was running for 2 or 3 months already.

 

[dhamilton]

Parts of AJ4B most likely refer to the factory it was made in. 0052 is not an absolute number, as I'm sure they would not limit the max. units possible to 9999 from the outset. Very likely to be the 52nd system built in week 37 of 1986.

Note, however, that it is not that early as you might think. Production starts months before something hit the market, as logistics require this. You can't produce a computer and have that very unit in a store next week. When your Deskpro was made, production was running for 2 or 3 months already.

That's a very good point. Compaq worked with dealers and didn't do mail order like Dell or Gateway, so they certainly would have needed to build up inventory.

The motherboard is Rev A, but the BIOS is already on Rev E. I've not seen any public references to any earlier revisions. So I guess in this case it's possible there's a Rev D. version out there, or perhaps they'd iterated already to E by the time they made them available.

Upon further inspection the 386 processor in this unit is the B1 stepping that wasn't as buggy. It would have been neat if it was one of the earlier A steppings, more bleeding edge I guess. I've read somewhere (probably pcjs.org) that Compaq had to work around bunch of bugs in the 386 when they were working on it.

It's still neat the machine is from the week of when they were announced.

 

[oldpcguy]

That's a very good point. Compaq worked with dealers and didn't do mail order like Dell or Gateway, so they certainly would have needed to build up inventory.

The motherboard is Rev A, but the BIOS is already on Rev E. I've not seen any public references to any earlier revisions. So I guess in this case it's possible there's a Rev D. version out there, or perhaps they'd iterated already to E by the time they made them available.

Upon further inspection the 386 processor in this unit is the B1 stepping that wasn't as buggy. It would have been neat if it was one of the earlier A steppings, more bleeding edge I guess. I've read somewhere (probably pcjs.org) that Compaq had to work around bunch of bugs in the 386 when they were working on it.

It's still neat the machine is from the week of when they were announced.

Does the processor have the ∑∑ on it?

 

[rmay635703]

Does the processor have the ∑∑ on it?

If it’s early it won’t be marked either way.

Sometimes the first batch were 32bit clean.

I owned one where the person purchased a “bad” 386 beige box full tower on purpose to save money.

Really sucked because it actually couldn’t run as much software as a 286.

 

[wrljet]

Mine wasn't marked.

 

[resman]

How does that 287 work? IIRC, it was clocked at 8 MHz (I think yours is 5 MHz).

 

[wrljet]

I don't know. But it did work. It was in the system when I received it from the original owner's son.

(I no longer have that machine, so can't try anything)

 

[resman]

I imagine it was pretty common to run slower 287s on that motherboard. But yours was definitely a slower one for 1986.

 

[oldpcguy]

Mine wasn't marked.

View attachment 1275774

That's awesome as it means Intel didn't test it. You have one of the earliest 80386 processors.

 

[dhamilton]

Does the processor have the ∑∑ on it?

Mine also does not have the marking. Mine is just like the one pictured above (40344). Same S-SPEC number. These are known as the "B1" stepping, according to information compiled over on pcjs.org:

Early 80386 CPUs

PCjs offers a variety of online machine emulators written in JavaScript. Run DOS, Windows, OS/2 and other vintage PC applications in a web browser on your desktop computer, iPhone, or iPad. An assortment of microcomputers, minicomputers, terminals, programmable calculators, and arcade machines...

www.pcjs.org

 

[dhamilton]

Here are some more photos of my system. S-SPEC S40344, 386-16. Also had a 287-8. The BIOS on this board is Rev. E. I dumped it and passed it along to Jeff over at pcjs.org.

 

[dhamilton]

I don't know. But it did work. It was in the system when I received it from the original owner's son.

(I no longer have that machine, so can't try anything)

Are you still in contact with the person who has the system today? It'd be interesting to compare notes...

 

[wrljet]

No, sorry.