here we go again...

luckybob wrote on 2024-03-11, 16:55:

here we go again...

yay. It's a crime to run a PPro all by itself.

WORKING Shuttle HOT-613 (Single to Dual - Pentium Pro)

Here is a similar thread.

Also, my sig has a socket 7 conversion. Idk if they are much help.

Guess we will just have to wait for a dual owner to show up and take some readings/pics. For best results the in question components should probably be measured removed. Except those with visible markings that identify them.

In meantime, order the socket and apic.

The Tyan S1662D is one of my favorite dual Socket 8 boards. Gave mine to a Vogons member years ago.

Re: What retro activity did you get up to today?

I will start with some measuring to find out the connections between the caps and CPU pins. Those connected to V I/O and Vcc are probably all the same capacity like 100nF, I could remove one relating cap inside the 1st CPU socket and mesure it´s capacity.
There are also some components I have to order in the meantime (Intel APIC etc.).

If an owner of a dual socket version shows up, all that´s left to do is reading the printed values of the resistors with a good magnifying glass.

majestyk wrote on 2024-03-12, 06:40:

If an owner of a dual socket version shows up, all that´s left to do is reading the printed values of the resistors with a good magnifying glass.

A good idea, but most of them are not marked. I will try.

I'll have to unbox it when i get to my garage today.

That´s good news!
I´m sure the resistors are all marked. Most of them have a greenish/sage color and the white printing has minimal contrast. Good lighting and magnifying will do the trick...at least I hope so 😉

In the meantime I found the values of the components in the first picture around the VRM-socket:

I also found that R51 is a 10K pull-up resistor at the "Power Good" pin of the VRM socket (pin 9A).

Now it´s time to have a look at the missing caps inside the 2nd CPU socket.
Generally all positions marked "CBxyz" are between Vss and either VccP (core voltage), VccS (cache voltage 3.3V) or VccT (Termination voltage 1.5V).
Capacitors mamed "Cxyz" are between CPU pins.
Intel recommends 40 capacitors 1µF around ind inside the socket for filtering / stabilizing the core voltage VccP. I assume all the caps connected to VccP belong to this group.
The ceramic capacitors that are already present are for VccS and VccT.

R254 -> 220R
R253 -> 330R
R260 -> 10K
(Edit:) R257 -> 10K
What I still need to know are
R8

Between the CPU sockets there are just 5 caps for decoupling / filtering of VccP

For the ones in the socket, could you not basically extrapolate based on cpu socket 1?

The layout inside CPU 1 is completely different. For determining the capacitors it was easier the way I did it.
But when it comes to the resistors, measuring at the respective resistors of CPU1 will be all that´s left - for the time being I hope luckybob can detect the printings on his board.

There are 6 resistors left that I cannot identify myself:
================================
= R8 (inside the 2nd CPU socket at the top) =
================================

================================
= R193, R205 (at the clock generator chip) =
= R267, R249, R250 (at the APIC chip) =
================================

Today I soldered most of the birdseed, as far as I had it on stock. The 2nd termination-voltage regulator got fitted in also.
Now waiting for the APIC chip to arrive. 2nd CPU socket and 2nd VRM socket will be added last.

Sorry, got distracted and completely forgot to snap some pics.

https://imgur.com/a/8FpSwmP

its amazing what pictures we can get with consumer grade cell phones.

Great - Thanx a lot!!

As soon as I get the APIC, the story continues...

If not hosting images here, at least make a guide listing values and positions please 😀 here in 5-10 years those images are gonna be dead links

I mean, imgur has been around since like 2010. I'm not exactly worried. Also, I'm keeping the motherboard around, so if the image host goes down and someone else wants to do this, I'll be here to take new pics.

As opposed to what everybody believed in the beginning, today there´s amnesia on the net everywhere. Websites, FTP servers and tons of Microsoft knowledge base articles are disappearing as we speak.

To be on the safe side: R8 inside CPU-socket 2 has been idientified as 330R.
And here´s the completed APIC / clock generator area:

The only component still missing is the Intel APIC chip, that seems to be stuck at customs.
Meanwhile I added the 2nd socket, VRM connector and the missing caps and resistors.

Should be able to fire it up with second cpu and do a smoke test at this point. Ohm out power rails make sure their resistance didn’t change much

A current limiter would be a good idea,

Sphere478 wrote on 2024-03-26, 18:02:

Should be able to fire it up with second cpu and do a smoke test at this point. Ohm out power rails make sure their resistance didn’t change much
A current limiter would be a good idea,

I tested the system a couple of times after adding components, checked for shorts etc. also. Last test was before I added the second CPU socket.

Yesterday the APIC chip arrived from a nice seller from Ukraine, so I could proceed.

After soldering the chip and doing some cleaning the system wouldn´t finish POST correctly in about 70% of the cases. Bending the PCP or pressing against the RAM sticks had an effect on the outcome.
So I seached for bad soldering joints and found about 10 of them at the legs of the chipset . Due to all that handling, bending and heating the board the solder junctions must have deteriorated. After finding all of them and resoldering the system started 100% reliable again.

Unfortunately the second CPU was not detected, it was detected occasionally at 2 or 3 starts out of 50.
I suspected flux remnants inside the new soldered CPU socket and cleaned it very carefully with IPA - to no avail.

While reviewing everything I had done, I noticed I skipped one resistor inside CPU socket #2: R257. I had even marked it in one of the first pictures, but then forgot it. It pulls pin "AA45" (PREQ#) up tu 3.3V. It´s a 10K resistor. Once it was added, everything´s working perfectly.
The board came with a BIOS from 1998, but reset doesn´t work with it (neither KB-shortcut nor switch), so I went back to V3.03 from Oct. 1996.