I can certainly test the CLK2 pin with the CPU removed. I can probably also test it with the CPU installed, but I do not have thin enough jumper clips to fit between the pins, so I will have to wrap up the CLK2 pin with a 30 AWG solid core wire and hope it doesn't slip off when I insert the CPU. Alternately, I can solder a wire to the bottom of the PGA168 and hook the oscilliscope test lead onto that wire.

What I meant by "packin'" was if you have the PGA168 or PGA132 interface board on hand for your logic analyser? If you have everything on hand and know what to look for (and are willing), I could ship the prototype to you.

The lowest cost on ebay.ca for the HP 1670G is over $600.

feipoa wrote:

The lowest cost on ebay.ca for the HP 1670G is over $600.

thats cheap. Look at the price of the new ones, even the entry level models.

edit: look online for your local hackerspace. ( https://en.wikipedia.org/wiki/Hackerspace ) larger ones should definitely have a logic analyzer you could use.

Yeah, but not worth it for the current goal at hand.

600? i got mine for like $150.

this one is $299 with cables, make an offer

http://www.ebay.com/itm/152401840048

not worth a one time use tho!

That is a listing for shipment to the USA only. They do not ship to Canada. The majority of the world are not as fortunate cost-wise as Americans. Even for $300 with a PGA168 interface, I am financially strapped right now. Wife is on maternity leave (where did that third child come from!?) and some concrete worker stole a good amount of money from me.

A lot more financially viable would be if you are interested in plugging my prototype into your fancy equipment.

feipoa wrote:

A lot more financially viable would be if you are interested in plugging my prototype into your fancy equipment.

Maybe that's where that third kid came from, eh?

🤣

haha, I suppose you could call it a prototype if it was the world's first circumcision.

BloodyCactus wrote:

test clock pin, is the CLK signal coming through the interposer?
is the cpu sending out its mem fetch signal on the bus? test address line 0, it should pretty well oscillate as the lowest address for fetching data..

Attached are the oscilloscope waveforms for CLK2 and A2. CLK2 is CH1 and A2 is CH2. I currently have a 66.6 MHz oscillator installed in the motherboard, so the frequency displayed is correct. CH2 (A2) indicates a logic zero value, though it is oscillating among 0.5V and 0 with some phase shift. When I hit the motherboard's reset button, the A2 waveform jumps up to 3.6 V, then back down to 0. Can we conclude anything from this?

You mentioned to check Address 0, but address 0 is derived from the Byte Enable output pins and some boolean algebra, whereby A0 = (BE0# · BE2#) + (BE0# · BE1#) and A1 = BE0# · BE1#, so I checked the next lowest order address, which is A2.

I used the pin and wire-wrap method to probe the pins as shown here,

however, it is beginning to tear apart the socket getting the CPU out - it is way too tight in there. For any other pins to test, I would need to solder the test lead to the bottom of the PGA168.

I'm not quite sure how to proceed with testing from here. Perhaps someone with experience in this arena can walk me through it?
Here is a symbol list for this CPU.

luckybob wrote:

TA-DA!

http://imgur.com/wP4IsUf
http://imgur.com/TDuiZKT

I also have this very same module, but it is currently in store and have not yet had time to test it in any customised systems.

Could you confirm if this works on any standard 386 motherboard?

Thanks

Believe it or not, i have absolutely ZERO 386 boards other than my model 80. I used to have a really nice 386 board with VLB that could take the ti 486 chip (and it had one) but I sold it.

I might have one somewhere, I will look around. That said I would not be surprised to find out it is IBM only. Judging by the jumper settings here: http://ohlandl.ipv7.net/CPU/Kingston_486-Now.html

That 386 board with VLB... I still have and is one of my favourites, thank you! 😁

I will try the interposer module on an old 386 board I have. My guess is, if it is going to work, it will do so under the Model 80 settings @25MHz and above. I have put in many 386 CPU's in the Model 80 and I know the CPU orientation.

The reason for the jumpers I suspect is because this Kingston module is it is an after market part, as in, made long after the Model 70 and other IBM systems were released. These older IBM systems (possibly) use a different pin out which required certain logic signals which enable the CPU to be inserted 90 degrees out, either this or the PCB was obstructed!

In theory, if it can work in a Model 80 in place of any other 386 CPU, it should work with the same settings elsewhere....

... famous last words! 🤐

feipoa wrote:

I planned to start with a proof of concept prototype, then possibly create a proper PCB if there was sufficient interest.

Here is the prototype:

SXL2_PGA168_to_PGA132.jpg

Having posted here previously, I have only just realised what this thread is about! I got confused thinking this thread was the other: -
"386 upgrade kits and the Transcomputer 486HPi - Am5x86-160 anyone?" thread.

What can I say, after reading the initial post this is an amazing project and would be severely impressed if you got it to work!

What I don't understand though is, would it not be easier to fabricate another Transcomputer 486HPi and design the SXL2-66 around this interposer?

If you need any help, i.e. traces going from certain pins of the QFP-144 to the PAL's/GAL's, let me know, I'd be happy to connect the dots.

luckybob wrote:

Believe it or not, i have absolutely ZERO 386 boards other than my model 80. I used to have a really nice 386 board with VLB that could take the ti 486 chip (and it had one) but I sold it.

I might have one somewhere, I will look around. That said I would not be surprised to find out it is IBM only. Judging by the jumper settings here: http://ohlandl.ipv7.net/CPU/Kingston_486-Now.html

FYI,

Good news, it works on all other 386 motherboards!

386_junkie wrote:

What I don't understand though is, would it not be easier to fabricate another Transcomputer 486HPi and design the SXL2-66 around this interposer?

No, because I think the Transcomputer module needs the PAL logic chips to function properly. Each of those PAL's are custom programmed and there are a lot of them. I do not have the code and it would take more time than I have - or am interested to put in - to reverse engineer it. My vision was that the SXL2-66 in the PGA168 format should not need any of these glue logic chips, which would greatly simplify the design. The other point, at least in my mind, is that the Transcomputer module is too much of a 486, whereas the 486SXL feels more 386. As a hobby, I'm not really looking for what is the most practical, but what fits my mentality.

I noted in the original text of this tread that I tried the SXL2-66 on the Transcomputer, but could not get it working in its native state. Appendix D in the 486SXL reference guide goes into detail about what differences are needed to get the 486SXL2 working vs. that of a 486SX, but this would require modifying my Transcomputer, which I am not interested in doing due to rarity. Also, and again for my mentality, even if the 486SXL2-66 worked in the Transcomputer module, I would not want to use this particular configuration because the Transcomputer can work with an Am5x86-133/160, meaning that for a particular component on the motherboard (the Transcomputer) it is not "maxed out".

386_junkie wrote:

If you need any help, i.e. traces going from certain pins of the QFP-144 to the PAL's/GAL's, let me know, I'd be happy to connect the dots.

The CPU collector in my province did not return my inquiry about borrowing his QFP144 SXL2-66 upgrade module (the same one you have), so it looks like I won't be able to computer what is done on this upgrade myself. If you would be willing to confirm a few things, it would be greatly appreciated. Attached here is the QFP144 pin outs and the PGA-132's pinouts. Could you determine which pins are one-for-one, e.g. which pins have A13, BS16#, D4, etc on PGA132 going to A13, BS16, D4, etc on QFP144 and which pins go to the glue logic SMD chips?

feipoa wrote:

Could you determine which pins are one-for-one, e.g. which pins have A13, BS16#, D4, etc on PGA132 going to A13, BS16, D4, etc on QFP144 and which pins go to the glue logic SMD chips?

Sure, I'll map the pins listed above and list the part #'s they meet.

Nothing like Trans-Atlantic teamwork... a virtual-inter-continental high-five! 🤣

Great! How's the progress?

386_junkie: Any update?

Yo!

Yea this is not going to be as easy as first thought.

I’ve spent quite a bit of time on this already and I’ve not even mapped half of the IC’s pins! So far… I’ve identified the IC’s, both with the relative data sheets are below… along with the pins that I’ve mapped already. Some of the pins are real difficult to reach as they go right underneath the socket… for these one’s I used the tip of a wire cutting. There may be some pins which are not used maybe or there could be more connections I haven’t found yet. All mapped pins below are with reference to the PGA.

AMD PALCE16V8: -

http://pdf1.alldatasheet.com/datasheet-pdf/vi … /PALCE16V8.html

Pinout: -

1, BEO#..20, Vcc
2, ??? … 19, ???
3, ???… 18, ???
4, R/W#..17, ???
5, ???… 16, ???
6, RDY..15, A31
7, ???… 14, A9
8, ???… 13, A8
9, ???… 12, A2
10, Vss… 11, ???

Triple 3-input NOR Gate: -

https://assets.nexperia.com/documents/data-sheet/74F27.pdf

Pinout: -

1, A10… 14, Vcc
2, A11… 13, A12
3, A13… 12, ???
4, A14… 11, ???
5, A15… 10, ???
6, ???…. 9, ???
7, Vss… 8, ???

On initial analysis, there are not enough pins on the IC’s to cover all address lines. I couldn’t find any data lines going to the IC’s, so it may be they are used only for addressing and control signals.

I hope this gives you a better idea of what’s going on.

Thanks for looking into this. I am having trouble following what connection points you are refering to. Could you add some clarificadtion?

What is the following in words?
1, BEO#..20, Vcc

e.g. "Pin 1 on PALCE16V8 connects to BEO# on QFP144..." Then what is 20? What is Vcc written there for?

Ideally, I was thinking something like

x,y,z pins on QFP144 connect directly to their corresponding pin assignments on the PGA132, that is, A27 goes to A27, D28 to D28, etc.
with the exception of pins a,b,c on QFP144 (which do not go directly to PGA132), but instead go to pins 1,2,3... on PAL/Gate/VRM. Maybe this is what you've written, but I am having trouble following.

I wrote it in socket form... i.e. as if you were putting the IC into the socket made of syntax on your screen.

e.g. Pin 1 on PALCE16V8 connects to BEO# on PGA132, ... then Pin 20 on PALCE16V8 connects to Vcc
e.g. Pin 6 on PALCE16V8 connects to READY on PGA132, ... then Pin 15 on PALCE16V8 connects to A15 on PGA132

All mapped pins are referenced with the PGA. Trying to operate on the PCB inside the socket whilst maintaining a probe on the side with the QFP-144 is no easy task... had a look around for where BS16# of the QFP-144 goes... but couldn't find it, it goes somewhere to the opposite end of the PGA socket from where the PALCE is (i.e. to where the transistor & Phillips IC is) but couldn't find the beep.

There could be other connections I haven't found yet... but finding the one's I already have took some time. I'm getting to know this little CPU better than I thought I would! 🤣

If it helps put you in my shoes...

The pins on the top row of this picture are mostly address lines. The pins at the bottom row are mostly control lines / data lines.

The 3-input NOR gate top left i've mainly mapped to the address lines, pins up top row of this picture.

The PALCE has address lines on the top half of the IC, but control functions on the bottom half of the IC... they go to pins on the bottom row of the picture.