VOGONS


First post, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

I finally got this build far enough along that I'm wanting to try to do some more overclocking to it

I want to try to get the fsb to 83 or more mhz but right now I'm stuck at 75mhz

motherboard: p55xb2
(random pic from internet, I'm using a k6 III+ cpu)

The attachment s-l1600 (3).jpg is no longer available

clock gen: pll52c62-01xc

The attachment DA126C71-1A21-457A-9519-DEF82A6C4FC0.jpeg is no longer available

here is a screen grab from the datasheet:

The attachment 0C01C79B-EC9A-4D2B-B1FF-5D62024E147B.png is no longer available

this next picture shows the jumper setting that gives 2x multiplier and 75mhz fsb
It has a weird way of setting the processor ratio and bus speed all on one jumper group:

The attachment C4F4F368-DA43-42ED-A4EE-61AAE63D800E.jpeg is no longer available

so far based on what speeds I can get by setting one or more jumper position combinations it seems that they are programming registers f0 and f1
I tried using a 10k resistor to ground on pin f2 but it had no effect on a setting that appeared to be giving me 55mhz bus so I am assuming that pin f2 is permanently pulled down to ground so my theory is if I heat it up and lift it I may be able to set 83mhz

but these pins are very small and I'm not confident my soldering equipment is up to this task and I really don't want to mess it up.

so I had another idea.. what if I buy a socket for the quartz and buy an assortment of crystals? has anyone ever done that?
will the computer still think it's going at the same speed or can it tell by comparing to the rtc that the bus is going faster when the quartz is changed?

and what speed quarts crystals should I get?

The attachment image.jpg is no longer available

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 2 of 18, by Tiido

User metadata
Rank l33t
Rank
l33t

When you change the 14MHz crystal you will cause some essential things such as system timer to speed up, which will break a lot of software. The very least you must make sure that all the 14MHz outputs remain what they are. Other things wouldn't be a problem as long as the PLL itself will be able to handle the higher clock input which it might not. Datasheet will tell what range there is that is guaranteed to work.

T-04YBSC, a new YMF71x based sound card & Official VOGONS thread about it
Newly made 4MB 60ns 30pin SIMMs ~
mida sa loed ? nagunii aru ei saa 😜

Reply 3 of 18, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
Tiido wrote on 2021-03-05, 17:12:

When you change the 14MHz crystal you will cause some essential things such as system timer to speed up, which will break a lot of software. The very least you must make sure that all the 14MHz outputs remain what they are. Other things wouldn't be a problem as long as the PLL itself will be able to handle the higher clock input which it might not. Datasheet will tell what range there is that is guaranteed to work.

So the quartz idea is a bad one?

So messing with f2 is my only option?

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 4 of 18, by gnif

User metadata
Rank Newbie
Rank
Newbie

According to the datasheet, this will screw up a ton of things. Your ISA bus clock, FDD clock, and if the motherboard has USB (looks like it doesn't) it will break USB as it requires a 48MHz clock.

In short, changing the crystal is a bad idea.

Reply 5 of 18, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
gnif wrote on 2021-03-06, 05:37:

According to the datasheet, this will screw up a ton of things. Your ISA bus clock, FDD clock, and if the motherboard has USB (looks like it doesn't) it will break USB as it requires a 48MHz clock.

In short, changing the crystal is a bad idea.

it has usb.

Okay. What about a software option for overclocking to 83mhz?

Is messing with that f2 pin my only hope?

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 6 of 18, by gnif

User metadata
Rank Newbie
Rank
Newbie
Sphere478 wrote on 2021-03-06, 05:46:

Okay. What about a software option for overclocking to 83mhz?

None, machines of this era never had such ability.
Datasheet clearly shows your clock is programmed by the jumper settings and not via any communication bus.

Reply 7 of 18, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
gnif wrote on 2021-03-06, 05:57:
Sphere478 wrote on 2021-03-06, 05:46:

Okay. What about a software option for overclocking to 83mhz?

None, machines of this era never had such ability.
Datasheet clearly shows your clock is programmed by the jumper settings and not via any communication bus.

Do you agree that it appears that f2 pin needs to be ungrounded to open up the 83mhz option? Via the already present jumpers?

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 8 of 18, by gnif

User metadata
Rank Newbie
Rank
Newbie
Sphere478 wrote on 2021-03-06, 13:57:
gnif wrote on 2021-03-06, 05:57:
Sphere478 wrote on 2021-03-06, 05:46:

Okay. What about a software option for overclocking to 83mhz?

None, machines of this era never had such ability.
Datasheet clearly shows your clock is programmed by the jumper settings and not via any communication bus.

Do you agree that it appears that f2 pin needs to be ungrounded to open up the 83mhz option? Via the already present jumpers?

The photo of the PLL shows that F2 (pin 20) appears to be left disconnected, hard to tell without pulling the chip to be sure. To get access to 83MHz you need to ground this pin, the datasheet shows that F1, F2 and F3 are all internally pulled high when disconnected, so to unlock 83MHz you need to pull it to the ground (ie, 0V). You would need to very carefully solder some fine wire on to do this, I would suggest using Kynar Wire (aka, wirewrap) as it's perfectly suited for this kind of mod.

Reply 9 of 18, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
gnif wrote on 2021-03-06, 14:22:
Sphere478 wrote on 2021-03-06, 13:57:
gnif wrote on 2021-03-06, 05:57:

None, machines of this era never had such ability.
Datasheet clearly shows your clock is programmed by the jumper settings and not via any communication bus.

Do you agree that it appears that f2 pin needs to be ungrounded to open up the 83mhz option? Via the already present jumpers?

The photo of the PLL shows that F2 (pin 20) appears to be left disconnected, hard to tell without pulling the chip to be sure. To get access to 83MHz you need to ground this pin, the datasheet shows that F1, F2 and F3 are all internally pulled high when disconnected, so to unlock 83MHz you need to pull it to the ground (ie, 0V). You would need to very carefully solder some fine wire on to do this, I would suggest using Kynar Wire (aka, wirewrap) as it's perfectly suited for this kind of mod.

I think it is connected to a wire below the chip.

What about connecting it to the f1 pin while it’s set to ground?

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 10 of 18, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

I’ve been doing tests on the jumpers.

The attachment 4C1732F1-3B9A-4D72-A07C-514B4A1AAF7A.jpeg is no longer available

I’ve been able to set these speeds

The attachment image.jpg is no longer available

Here is the table:

The attachment C0B20198-AEF1-40E6-979B-8B9D6CBA2102.jpeg is no longer available

Some other speed settings were from doing this:

The attachment image.jpg is no longer available

Does it look like any of those speeds may be based on 83mhz?

Last edited by Sphere478 on 2021-03-06, 16:00. Edited 1 time in total.

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 12 of 18, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

Okay, so I set it to 450 which should be 75x6 and shorted pins f1 to f0 and f2 and no post

So I tried pins f1 and f2 and still no post. No short it posts at 450..

I wasn’t sure that 0 meant grounded but seems it does. And it seems the async settings don’t work.. any idea why?

Based on that it seems I may need to replace the clock gen with a different unit?

Hasn’t anyone come up with a universal clock generator you can wire in and take over a motherboard with?

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 13 of 18, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

Proof of concept. Unfortunately.

I was able to achieve the as of yet unattainable 330 mhz setting by shorting f0 to f1 while jumper was set to c200 aka 450 75x6. The short creating the setting 55x6 330

This unfortunately confirms that my theories about how it’s set are correct and that the async settings including the 83mhz setting aren’t working for some reason.

Any ideas why async settings aren’t working on this clock gen?

Last edited by Sphere478 on 2021-03-07, 04:12. Edited 1 time in total.

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 14 of 18, by gnif

User metadata
Rank Newbie
Rank
Newbie
Sphere478 wrote on 2021-03-06, 16:28:
Proof of concept. Unfortunately. […]
Show full quote

Proof of concept. Unfortunately.

I was able to achieve the as of yet unattainable 330 mhz setting by shorting f0 to f1 while jumper was set to c200 aka 450 75x6. The short creating the setting 55x6 330

This unfortunately confirms that my theories about how it’s set are correct and that the async settings including the 83mhz setting aren’t working for some reason.

Any ideas why asuny settings aren’t working on this clock gen?

Sorry no, without having the motherboard on hand and a DSO to probe/verify there is no way to know. If you did have a DSO with a bandwidth of 100MHZ or better you could check the output of the PLL.

Reply 15 of 18, by rmay635703

User metadata
Rank Oldbie
Rank
Oldbie
Sphere478 wrote on 2021-03-06, 15:56:

And it seems the async settings don’t work.. any idea why?

Async never did work on anything before PC100, to be honest it should have never been included at all.

Most early 75/83mhz FSB Socket 7 boards would run async with only Cyrix branded CPUs and even then I had strange stability issues.

Put in an Intel or AMD chip set async no post.

Just how it was, my manual even stated async for Cyrix only.

To run at 75 or 83 you need to leave the pci divisor at /2 not async

Reply 16 of 18, by ghost.dog@mail.ru

User metadata
Rank Newbie
Rank
Newbie
rmay635703 wrote on 2021-03-07, 03:37:
Async never did work on anything before PC100, to be honest it should have never been included at all. […]
Show full quote
Sphere478 wrote on 2021-03-06, 15:56:

And it seems the async settings don’t work.. any idea why?

Async never did work on anything before PC100, to be honest it should have never been included at all.

Most early 75/83mhz FSB Socket 7 boards would run async with only Cyrix branded CPUs and even then I had strange stability issues.

Put in an Intel or AMD chip set async no post.

Just how it was, my manual even stated async for Cyrix only.

To run at 75 or 83 you need to leave the pci divisor at /2 not async

Hello, do you remember the model of motherboard and cyrix CPU that worked in async pci bus mode ? I have Asus motherboard with Intel 430fx chipset which also doesn't post if I select async mode. I replaced clocker on it with one which can do 75/83 in both sync and async modes. I also have measured pci frequency in async mode and it definitely can do async.

Reply 17 of 18, by SSTV2

User metadata
Rank Oldbie
Rank
Oldbie

P5/P6 class chipsets from intel never supported asynchronous bus operation, this feature was meant for other chipset manufacturers, that did support it at the time.