Have you tried with the lowest possible front side bus (50 MHZ)? Sometimes chips of unknown quality fail at nominal frequencies.

nice i hope you get it working i have this board but the Cache is BAD ill have to pic up some Chips and see if i can fix it.

majestyk wrote on 2024-05-30, 06:00:

Have you tried with the lowest possible front side bus (50 MHZ)? Sometimes chips of unknown quality fail at nominal frequencies.

if i install all 4 32k32's for 512k what would the jumpers be?
right now i have 256k but the cache is bad so i removed them and ordered 4 32k32.

@ontrca

Worse case you just run a K6-II+ / III+ @233Mhz

Topaz wrote on 2024-05-30, 23:31:

if i install all 4 32k32's for 512k what would the jumpers be?
right now i have 256k but the cache is bad so i removed them and ordered 4 32k32.

In this case at the TAG chip R161 and R164 must be 0 ohm, R162 and R163 are not populated.

At the C585VP R200 (10K) remains untouched, R147 must be 0 ohm, R148 and R150 are not populated.

As for the K6-III+ question - The lowest core voltage the PA-2005 can provide is 2.7 / 2.9V. This would harm any K6-III+ since these CPUS need 1.6-2.0V.

majestyk wrote on 2024-05-31, 06:08:

As for the K6-III+ question - The lowest core voltage the PA-2005 can provide is 2.7 / 2.9V. This would harm any K6-III+ since these CPUS need 1.6-2.0V.

I wish these interposers were more readily available in 2024:

https://www.ardent-tool.com/CPU/Pentium_interposer.html

majestyk wrote on 2024-05-31, 06:08:

In this case at the TAG chip R161 and R164 must be 0 ohm, R162 and R163 are not populated. […]
Show full quote

Topaz wrote on 2024-05-30, 23:31:

if i install all 4 32k32's for 512k what would the jumpers be?
right now i have 256k but the cache is bad so i removed them and ordered 4 32k32.

In this case at the TAG chip R161 and R164 must be 0 ohm, R162 and R163 are not populated.

At the C585VP R200 (10K) remains untouched, R147 must be 0 ohm, R148 and R150 are not populated.

As for the K6-III+ question - The lowest core voltage the PA-2005 can provide is 2.7 / 2.9V. This would harm any K6-III+ since these CPUS need 1.6-2.0V.

i removed all the jumpers on VR1 and it will drop the Vcore down too 1.27V in the passed i have just removed all jumpers and used a 10 turn pot on a jumper wire to set V core too.

majestyk wrote on 2024-05-24, 08:40:

One single un-/ loose-connected pin at one of the chips will prevent the whole L2 cache from getting initialized by BIOS.
Check all the solderings carefully for shorts and bad contacts!

It's not pretty but all of my solder points are good according to my multimeter.

majestyk wrote on 2024-05-30, 06:00:

Have you tried with the lowest possible front side bus (50 MHZ)? Sometimes chips of unknown quality fail at nominal frequencies.

I'll try that 😀

I’d flux those up and with a fine tip melt each one again, they don’t look very uniform. Those pins are pretty small. It’s not easy work! Massive amounts of patience and the right tools are critical. I’ve handed stuff like this to the better equipped before, if you know someone with better tools, skills. Don’t be afraid to ask for assistance 😀

douglar wrote on 2024-05-31, 11:54:

majestyk wrote on 2024-05-31, 06:08:

As for the K6-III+ question - The lowest core voltage the PA-2005 can provide is 2.7 / 2.9V. This would harm any K6-III+ since these CPUS need 1.6-2.0V.

I wish these interposers were more readily available in 2024:

https://www.ardent-tool.com/CPU/Pentium_interposer.html

You can build one 😀 I use the ones I made when tinkering all the time!

Topaz wrote on 2024-05-31, 11:56:

i removed all the jumpers on VR1 and it will drop the Vcore down too 1.27V in the passed i have just removed all jumpers and used a 10 turn pot on a jumper wire to set V core too.

I tried this recently, Vcore dropped to 2.5V on my revision 1.31. With a 390R resistor across pins 7-8 I got 2V and did some tests. A K6-2 450 works without Write Allocation being enabled by BIOS (latest one). The system started with none of the K6-plus CPUs. Then I tried an interposer - same result.
I´m sure the BIOS would need a massive makeover to run plus-CPUs, even with a modded VRM.

majestyk wrote on 2024-05-30, 06:00:

Have you tried with the lowest possible front side bus (50 MHZ)? Sometimes chips of unknown quality fail at nominal frequencies.

I tried 50mhz and it didn't work...

I've been putting it off because it's so tedious, but I guess I should try reflowing the solder points now

Make sure you use some really good flux this time - like the "MG Chemicals 8341".

i have that MG Chemicals 8341 its like magic

ontrca you have inspired me to fix my PA-2005 motherboard so i ordered some Cache but there only W25P010AF so if i put all 4 on ill have 512 but ill be happy to just get the Cache just working i hope you get it all figured out.

i got mine off just lost 1 pad 🙁 and i was using hot air and low melt solder.

Topaz, did your PA-2005 come with all Tantalums?

Another nice alternative on this board is a Tillamook:

Sadly, L2 cache is always wtite through with this BIOS...

majestyk wrote on 2024-05-31, 06:08:

Topaz wrote on 2024-05-30, 23:31:

if i install all 4 32k32's for 512k what would the jumpers be?
right now i have 256k but the cache is bad so i removed them and ordered 4 32k32.

In this case at the TAG chip R161 and R164 must be 0 ohm, R162 and R163 are not populated.

At the C585VP R200 (10K) remains untouched, R147 must be 0 ohm, R148 and R150 are not populated.

this is how i set up the jumpers it detected 512 but would give me a mem error loading up DOS HiMEM i reversed the jumper links and its back to 256 cache and will run Quake fine. going to look for some solder bridges but every thing looks real good reflowed the chips with a lot of flux.

as for the Tantalums i put them on some of the 10uf caps where damaged or ripped off so i just put them in as i have a box of 500 i for from a fellow Amateur radio operator.

BLUE is where a 0 ohm resistor is installed and RED is no resistor is this right for 512kb?

The changes at the TAG chip and at the 82C585VP are 100% correct.

A thorough cleaning of all the grease with lots of IPA and a small brush and good drying afterwards is recommended.

majestyk wrote on 2024-06-04, 05:25:

The changes at the TAG chip and at the 82C585VP are 100% correct.

A thorough cleaning of all the grease with lots of IPA and a small brush and good drying afterwards is recommended.

Ok well something is not happy with 512k so I'll just roll it back to 256k and live with it. At least the 256k works.
Thanks for the help

Hi, just playing with this MB and have found this cache mod.
My MB is originally equipped with 2 tmtech T33L3232A-6Q - 256kB
I found spare parts T35L3232A-6Q and soldered them on. I cannot find datasheet for T33L3232A but hope they are compatible.
After moving 2 described resistors I got detected 512kB in summary boot table but system fails to boot. Instead of veryfiing dmi pool data I got just "Failure" message and it hangs. I tried lower FSB at 50MHz - didn't help. Only disabling ext. cache in setup makes it boot. Did I missed something? What about missing R180, R181, R183 near cache chips? Anybody esle succeeded with the mod?

Other mod I tried was CLK3 jumper and it seems run 75MHz fine. Have someone checked also other hardwired jumpers MFG1-4 near the CPU socket?

I reflow my hand soldering by hot-air but still fails.
I found another 2 cache chips Winbond W25P010A-6 so I replaced that tmtech and it finally boots! So probably I had bad spare cache chips. I don't remmeber where I desoldered it from. So now I got 2FPS perf. gain in Quake (320x200). Did it worth? 😁 I didn't expect much of course. So those with bad chips don't worry much about this mod...