Neat interposer!

Iirc, 386 also lived in PGA form factor. A socket version of this would be handy kit!

386DX is a 32-bit chip. 286 motherboards are 16-bit. It doesn't really make sense to strap a 386DX to a 286 board. Though despite that, Buffalo did it anyway (I own two).
386SX was available in PGA for military applications, but it's incredibly rare.
That leaves the option of the PQFP100 test socket, which is also very rare and expensive.

While you wont get 32bit wide databus, you DO get 32bit wide registers, v86 mode, the ability to return to realmode from protected mode, and larger on-die cache.

I was meaning a normal pga socket for a 386dx, a chips & technologies super 386, or an AMD AM386 on a pga sled. IIRC.

One might need a ribbon to the 287 socket to drive a 386dx chip though, not sure.

Address bus you could just truncate at 16bits, but databus would need some kind of bus transciever to cascade two 16bit reads together / break a 32bit write into 2 16bit ones, with forced waitstates. This would be true with the SX as well, no? (no, sx chips apparently have 16bit databus externally.)

386SX ist 16/32bit hybride (inside is 32bit, outside 16bit ) same as 8088 (16/8bit )

schematic reference is VTI databook from 1989
https://archive.org/details/bitsavers_vtidata … e/n415/mode/2up
page 417

Biggest problem is PLCC "plug"

I have using random PLCC-68 ICs and cutting their silicon out.
Then i have bend their leg and grinded it for safe contact into most PLCC socket.

Madao wrote on 2024-07-20, 09:06:

386SX ist 16/32bit hybride (inside is 32bit, outside 16bit ) same as 8088 (8/16bit ) […]
Show full quote

386SX ist 16/32bit hybride (inside is 32bit, outside 16bit ) same as 8088 (8/16bit )

schematic reference is VTI databook from 1989
https://archive.org/details/bitsavers_vtidata … e/n415/mode/2up
page 417

Biggest problem is PLCC "plug"

I have using random PLCC-68 ICs and cutting their silicon out.
Then i have bend their leg and grinded it for save contact into most PLCC socket.

Cool!
I think you can find some plugs but they cost in the hundreds of dollars.

There's a guy in China selling incomplete Make it 486 upgrades for 286 systems that contain this type of connector. I think he wants around $12 each. Looks like he has 1000 of them.

Caluser2000 wrote on 2021-07-12, 03:00:

maxtherabbit wrote on 2021-07-12, 02:55:

Caluser2000 wrote on 2021-07-12, 02:07:

There a a lot factors involved and it is well known that when 386sx cpus process 16-bit instructions they use approx twice the cpu cycles as equivalent speed 286 cpu.

That's simply untrue, and my testing supports this statement.

I share your enthusiasm for the 286, it's my favorite PC platform in fact, but I can't abide blatant misinformation

That's fine. I've in ssue with your view point at all. It would apeair some in the industry back then supported my view point..😉

The only time I saw a very large speed difference between a 386sx and a 286 of the same speeds was when the 386sx had a waitstate and the 286 did not.

I had a few sx16 and sx20 machines pass through my fingers that had a waitstate and very slow ram. God only knows why anyone manufactured that combination still with dipps.
The sx20 was a proprietary “full AT” motherboard with a slow MFM harddrive that made it seem worse than it really was.

The late lunchpail 1992+ era Harris 286 systems (20/25mhz) that had of all things 72pin simms were both configured (right or wrong) with 0ws and seemed to test abnormally fast in comparison to the 386sx systems. Sad they became ewaste instead of being preserved after they quit posting.

The same friend that got the very fast 25mhz 286 from work had an unexplainably slow 386sx40. I peeled some stickers off and it was an overclocked sx16, it was configured the same as the 286-25 but the vga and hd (I think) were very slow. The motherboard was some type of PCCHIPS and the bios was very generic without any real way to configure timing of anything.

Back in 1996 when I encountered that system I was too stupid to know how to properly test it, sort of curious what was wrong with the sx40 that it seemed slower than the 286-25 running the software they had and Windows stuff

Anonymous Coward wrote on 2024-07-20, 15:24:

There's a guy in China selling incomplete Make it 486 upgrades for 286 systems that contain this type of connector. I think he wants around $12 each. Looks like he has 1000 of them.

Nice! Since I 1st watche Cpugalaxy's videos I wanted to get one but it was too high.
Also, LCC to PGA adapters cost a lot when I looked up for them.

Could you pvt me the link of this guy?

Madao wrote on 2024-07-20, 03:36:

My try with homemade 386SX upgrade with reference from VTI /VLSI Databook 1989

It show lifesign on IBM 5170 board. Check on other motherboard come soon.

Nice, would be interested to know how it compares to a 8mhz 286 like mine which does not have some clock adjusting abilities except in bios between 6/8mhz.

386SX is a bit slower than 286 on same platform and same clock (IBM AT Typ 3 , ~ 5% slower )

But, you can executing 32bit-software 😀