wiretap wrote on 2021-03-05, 14:52:

That's awesome. 😁 I was looking for that specific article, but it wasn't on the Wayback Machine. But it looks like the UK site still has it. What needs to be done is to make a pinout table for mapping the AMD pin header to the TE 1734037-4 slot connector footprint in KiCAD. That should only take a few minutes to generate.

As far as I can tell A20 on the PCB edge is 1 on the AMP connector.
PCB edge---AMP connector
A20-------1
A19-------2
.
.
.
A2--------19
A1--------20
___
B20-------21
B19-------22
.
.
.
B2--------39
B1--------40
Somebody willing to counter check?
Edit: Pins according to AMP connector pinout in KiCad and CPU edge pinout in Tom's Hardware article.

wiretap wrote on 2021-03-05, 15:04:

The diode and 220-ohm note is on page 4 at the bottom of the UK article.

Oh, they're on page two in the German article. Anyway good thing I stumbled across that schematic and PARKE did that reverse search! 😁

Doornkaat wrote on 2021-03-05, 15:06:

Anyway good thing I stumbled across that schematic and PARKE did that reverse search! 😁

Tom's Hardware articles often seem disappeared via Google search on titles. But there is a russian site that has stored a ton of them with missing links for the graphics. A detour via copy/paste of a substantial piece of text into Google can lead to the original, as in this case.

I checked and the cache software is freeware and still avaliable on multiple sites (i.e. https://www.ocinside.de/links/ocsoft_links_d/)
WCPUA2 (by H.Oda! who exclusively holds all copyrights and wants this mentioned wherever this is shared!) to change Slot 1 cache multipliers:
WCPUL2 for 32bit Windows, WCPUL2D for DOS.) will let you choose Slot A CPU cache divider in 32bit Windows.
K7L2DOS (by Thomas Combeléran) will let you choose Slot A CPU cache divider in DOS.
L2_speed (by MORF based on Thomas Combeléran's info) will also let you choose Slot A CPU cache divider in DOS.
FixCache will set a fixed cache divider in an Asus K7V BIOS file. You have to flash the BIOS file after that. The program is based on the findings by H.Oda!.

I have never tested any of them, use at your own risk.

Also here are two more tools by H.Oda! (who exclusively holds all copyrights and wants this mentioned wherever this is shared!) to change Slot 1 cache latencies:
WCPUL2 for 32bit Windows, WCPUL2D for DOS.

I remember reading this article back in the day! It’s funny how you forget things and then you see it again and it’s suddenly remembered.

So the ninja clone follows the Toms Hardware schematic logic using the updated resistor values and diodes to generate the 3.3v.

I did start mapping the pins on the connector to the edge connector I don’t have my notes because I’m at work.

I was confused by the ninja compared to free speed and am now even more confused seeing the Toms article. The Ninja device replicates the whole schematic as shown in the yellow picture, the free speed is not.

So somehow the free speed is attempting to reduce the amount of logic in a way that I can’t seem to wrap my head around.

I think you mean the Northwind GFD.

Yea the smaller/simplified ones use the K7 processor resistors looping back through in same way. It doesn't really seem like a good design IMO, but rather a way to minimize size and reduce cost. I think they put additional strain on the CPU which requires an elevated vcore, resulting in more instability of the OC.

wiretap wrote on 2021-03-05, 15:56:

Yea the Freespeed uses the K7 processor resistors looping back through in same way. It doesn't really seem like a good design IMO, but rather a way to minimize size and reduce cost.

The Innovatek GFD follows the same principle. It's totally meant to reduce size and cost. Following the AMD design is probably the proper way.
Still I have bought a second hand 800MHz K7 with the device installed and set to 1GHz that has been running like this for over five years (according to the seller) and it shows no problems.

Doornkaat wrote on 2021-03-05, 15:06:

As far as I can tell A20 on the PCB edge is 1 on the AMP connector. PCB edge---AMP connector A20-------1 A19-------2 . . . A2-- […]
Show full quote

wiretap wrote on 2021-03-05, 14:52:

That's awesome. 😁 I was looking for that specific article, but it wasn't on the Wayback Machine. But it looks like the UK site still has it. What needs to be done is to make a pinout table for mapping the AMD pin header to the TE 1734037-4 slot connector footprint in KiCAD. That should only take a few minutes to generate.

As far as I can tell A20 on the PCB edge is 1 on the AMP connector.
PCB edge---AMP connector
A20-------1
A19-------2
.
.
.
A2--------19
A1--------20
___
B20-------21
B19-------22
.
.
.
B2--------39
B1--------40
Somebody willing to counter check?
Edit: Pins according to AMP connector pinout in KiCad and CPU edge pinout in Tom's Hardware article.

I just went through the original AMD schematic, and mapped it from how the Ninja is setup to the KiCAD footprint of the GFD interface. Please double check. Looks like we came to the same conclusion.

wiretap wrote on 2021-03-05, 17:41:

I just went through the original AMD schematic, and mapped it from how the Ninja is setup to the KiCAD footprint of the GFD inte […]
Show full quote

Doornkaat wrote on 2021-03-05, 15:06:

As far as I can tell A20 on the PCB edge is 1 on the AMP connector. PCB edge---AMP connector A20-------1 A19-------2 . . . A2-- […]
Show full quote

wiretap wrote on 2021-03-05, 14:52:

That's awesome. 😁 I was looking for that specific article, but it wasn't on the Wayback Machine. But it looks like the UK site still has it. What needs to be done is to make a pinout table for mapping the AMD pin header to the TE 1734037-4 slot connector footprint in KiCAD. That should only take a few minutes to generate.

As far as I can tell A20 on the PCB edge is 1 on the AMP connector.
PCB edge---AMP connector
A20-------1
A19-------2
.
.
.
A2--------19
A1--------20
___
B20-------21
B19-------22
.
.
.
B2--------39
B1--------40
Somebody willing to counter check?
Edit: Pins according to AMP connector pinout in KiCad and CPU edge pinout in Tom's Hardware article.

I just went through the original AMD schematic, and mapped it from how the Ninja is setup to the KiCAD footprint of the GFD interface. Please double check. Looks like we came to the same conclusion.

Yup, we assigned the same pins/signals on the PCB edge the the AMP connector. Great! 👍 I have no objections. (Apart from changing PB_FID2 to BP_FID2 in the table you posted which is obviously just a harmless typo.)

oops, haha. Here's the Excel doc.

Awesome!😁👍

I did notice one difference between the CPU world schematic, and the AMD schematic. A18 and A19 seem to be backwards on one or the other. According to the Ninja schematic and how dip switches normally go from 0 through 3 for four positions, it makes sense that the yellow AMD schematic from Tom's Hardware has A18 and A19 backwards.. any thoughts on this?

Do we actually have a datasheet for the AMD CPU that identifies what these pins are? I didn't see it in the normal AMD K7 datasheets.

Updated based on the above information:

Your reasoning sounds plausible imho. I tried to verify this by looking at photos of Athlon PCBs and what resistors are installed where for what frequency but I'm too tired to understand anything right now.
Maybe Sunday I'll find some proof.

I got my solder stencil today from JLCPCB for the Ninjamicro Freespeed clone. PCB's should show up later this week if the tracking is correct. However, I still don't have a working K7 board to test with. I bought a SD-11 which was completely dead and wouldn't even post, so I had to return it.

wiretap wrote on 2021-03-09, 17:49:

I got my solder stencil today from JLCPCB for the Ninjamicro Freespeed clone. PCB's should show up later this week if the tracking is correct. However, I still don't have a working K7 board to test with. I bought a SD-11 which was completely dead and wouldn't even post, so I had to return it.

My PCBs shipped today, I had them do the SMD solder work, we’ll see how it turns out. Should be here in 2-3 days. My mouser order for dip switches has not shipped so I may not be able to start right away anyway.

gbeirn wrote on 2021-03-09, 18:23:

wiretap wrote on 2021-03-09, 17:49:

I got my solder stencil today from JLCPCB for the Ninjamicro Freespeed clone. PCB's should show up later this week if the tracking is correct. However, I still don't have a working K7 board to test with. I bought a SD-11 which was completely dead and wouldn't even post, so I had to return it.

My PCBs shipped today, I had them do the SMD solder work, we’ll see how it turns out. Should be here in 2-3 days. My mouser order for dip switches has not shipped so I may not be able to start right away anyway.

If the soldering works out well, could you share the BOM and placement file that you used? I didn't have much time for retro-related activities recently, so I didn't get around to preparing the files yet, so I haven't ordered the Ninjamicro clone PCBs (which I absolutely intend to do).

gbeirn wrote on 2021-03-09, 18:23:

wiretap wrote on 2021-03-09, 17:49:

I got my solder stencil today from JLCPCB for the Ninjamicro Freespeed clone. PCB's should show up later this week if the tracking is correct. However, I still don't have a working K7 board to test with. I bought a SD-11 which was completely dead and wouldn't even post, so I had to return it.

My PCBs shipped today, I had them do the SMD solder work, we’ll see how it turns out. Should be here in 2-3 days. My mouser order for dip switches has not shipped so I may not be able to start right away anyway.

Mouser has been having some significant delays for me. It's sad when I can upload gerbers and have China ship me completed PCBs to by door before Mouser even finishes picking my cart order..

edit: oh good, Mouser just upgraded me to 1-day delivery and my parts order will be here tomorrow morning.

gex85 wrote on 2021-03-09, 18:43:

gbeirn wrote on 2021-03-09, 18:23:

wiretap wrote on 2021-03-09, 17:49:

I got my solder stencil today from JLCPCB for the Ninjamicro Freespeed clone. PCB's should show up later this week if the tracking is correct. However, I still don't have a working K7 board to test with. I bought a SD-11 which was completely dead and wouldn't even post, so I had to return it.

My PCBs shipped today, I had them do the SMD solder work, we’ll see how it turns out. Should be here in 2-3 days. My mouser order for dip switches has not shipped so I may not be able to start right away anyway.

If the soldering works out well, could you share the BOM and placement file that you used? I didn't have much time for retro-related activities recently, so I didn't get around to preparing the files yet, so I haven't ordered the Ninjamicro clone PCBs (which I absolutely intend to do).

I absolutely will once I make sure what I receive in hand is actually correct. I’m assuming since the order completed without issue that’s it’s fine but I should know for sure in a few days.

Assembly. My solder paste is a little old, but it turned out ok.