overclock

[evildragon]

Well, not really an overclock, but installed as if it is.

My Model 25 PS/2 has a 10MHz V30 installed, but it only seems to be running at 8MHz. This would make sense as the 8086 that was installed before it was 8MHz, and the motherboard only seems to have a crystal for that.

What I want to do: Lift the CLK pin on the V30, solder a 10MHz crystal to it, and then put it in the socket (with the CLK NOT touching the socket)..

This should work, right? I know 2MHz really won't be much, but for a specific task I want it to do, it will..

 

[JoJo_ReloadeD]

Well, not really an overclock, but installed as if it is.

My Model 25 PS/2 has a 10MHz V30 installed, but it only seems to be running at 8MHz. This would make sense as the 8086 that was installed before it was 8MHz, and the motherboard only seems to have a crystal for that.

What I want to do: Lift the CLK pin on the V30, solder a 10MHz crystal to it, and then put it in the socket (with the CLK NOT touching the socket)..

This should work, right? I know 2MHz really won't be much, but for a specific task I want it to do, it will..

I did that one week ago on my 8088 8mhz board (now a v20 12mhz board), and done this way:

The board had a 14.3..mhz cristal (14.3/3=4.77 -> the isa clock speed) and a 24mhz cristal (24/3=8mhz -> the cpu clock speed). I dunno why is that divider there, but it works.

Then I replaced the 24mhz cristal with a 36mhz one and that's it! A 12mhz v20 with a 8087 rocking togheter!

The 8087 goes pretty hot, by the way, I had to attach a heatsink to it

I tried a 48mhz cristal too (to get to 16mhz, but that was too high and didn't work)

Hope that helps

 

[evildragon]

I did that one week ago on my 8088 8mhz board (now a v20 12mhz board), and done this way:

The board had a 14.3..mhz cristal (14.3/3=4.77 -> the isa clock speed) and a 24mhz cristal (24/3=8mhz -> the cpu clock speed). I dunno why is that divider there, but it works.

Then I replaced the 24mhz cristal with a 36mhz one and that's it! A 12mhz v20 with a 8087 rocking togheter!

The 8087 goes pretty hot, by the way, I had to attach a heatsink to it

I tried a 48mhz cristal too (to get to 16mhz, but that was too high and didn't work)

Hope that helps

won't this speed up the real time clock though? that's what I hear, and I want to try to keep all the original oscilators onboard, but introduce a new one for the soul purpose for only the CPU... (i don't know if that'll upset expansion cards though)

 

[mbbrutman]

If you don't have a wiring diagram of your machine and know how to read it, I wouldn't dare to just start changing crystals.

First, computer chips are spec'ed for a particular speed. Anything above that speed is not guaranteed. Excessive frequency differences can cause damage.

Second, you might change timings of dependent components on the system by changing the crystal. You are not just overclocking the CPU, you might be overclocking much more. That can lead to flakey operation, data corruption, etc.

Do some research on your specific machine before attempting this. Don't just post here with a half-baked idea .. I'd hate to see something get ruined for lack of better preparation.

Finally, everybody should try to make their posts as readable as possible. I'm not an English teacher, but things like not using the shift key on your keyboard, not knowing the difference between 'soul' and 'sole', mispelling 'crystal', etc. are just annoying.

 

[evildragon]

If you don't have a wiring diagram of your machine and know how to read it, I wouldn't dare to just start changing crystals.

First, computer chips are spec'ed for a particular speed. Anything above that speed is not guaranteed. Excessive frequency differences can cause damage.

Second, you might change timings of dependent components on the system by changing the crystal. You are not just overclocking the CPU, you might be overclocking much more. That can lead to flakey operation, data corruption, etc.

Do some research on your specific machine before attempting this. Don't just post here with a half-baked idea .. I'd hate to see something get ruined for lack of better preparation.

Finally, everybody should try to make their posts as readable as possible. I'm not an English teacher, but things like not using the shift key on your keyboard, not knowing the difference between 'soul' and 'sole', mispelling 'crystal', etc. are just annoying.

Read my original post, my CPU is a 10MHz processor. The computer itself at this moment is DOWNCLOCKING it to 8MHz. I need to get it up to the speed in which it was designed for. I made that clear when I said "has a 10MHz V30 installed, but it only seems to be running at 8MHz"...

I'm good with electronics, and don't appreciate you calling it a half-baked idea. I've been overclocking Motorolla 68000's for 5 years now, so I'm not a newb at this... I just want to know if the same processes can be applied to the 8086 line of processors, that is all..

I knew that replacing the crystal would be a bad idea, that's why I said something, and asked if the real time clock would be faster (as it overclocked other things too).. That's why I proposed the idea I had, by just connecting an oscillator directly to a lifted CLK pin. But I don't know if the 8086 can accept that, like the M68K's can...

 

[mbbrutman]

It is a half baked idea if you think you can lift one pin on a chip, connect it to a different clock crystal, and get a faster system.

It's really half baked when you consider that the chip will now have a different bus timing cycle than the rest of the system that it is connected too.

Get the wiring diagram. Read it. Good luck.

 

[evildragon]

It is a half baked idea if you think you can lift one pin on a chip, connect it to a different clock crystal, and get a faster system.

It's really half baked when you consider that the chip will now have a different bus timing cycle than the rest of the system that it is connected too.

Good luck.

You can easily lift one pin.. the chip isn't even soldered! it's not that hard at all! like i said, I do this on Motorolla 68000's all the time, and they work great at the higher speed...

THIS IS WHY I FREAKIN ASKED..

EDIT 2: Removed my own anger.

EDIT: This is pin lifting.. And this is a schematic of what I want to do...

 

[JoJo_ReloadeD]

won't this speed up the real time clock though? that's what I hear, and I want to try to keep all the original oscilators onboard, but introduce a new one for the soul purpose for only the CPU... (i don't know if that'll upset expansion cards though)

Nope, RTC it's working ok... I had luck that my board uses a crystal just for the cpu and other for all the rest, so changing it only changed cpu frecuency

P.S.

Just to get this straight, the crystals in my board were like this:

That is, a quartz crystal alone. The oscilator circuit (an oscilator is NOT only a crystal) was left like it was.

This is an oscilator :

Any tries i've done to overclock systems with oscillators (not crystals) were all failures (I had to resolder the original oscillator and ready to go.. only that..)

So, if your system uses an oscillator I think that it will be hard to overclock it...

 

[evildragon]

Nope, RTC it's working ok... I had luck that my board uses a oscilator just for the cpu and other for all the rest, so changing it only changed cpu frecuency

my computer has a 25MHz crystal on it.. not a 24MHz one.. That would infer 3.34MHz, if it was /3, but I don't think it is. I think it's just a different division slightly. My system also lacks the clock driver that's common in the XTs, it probably has a different one.

I just need to know if the above schematic I came up, will work or not.. I'm not going to replace crystals.. I have a spare Model 25 motherboard if I had to "test" this, but I would rather not.

When I find a 10MHz crystal, I'm going to test my theory first (test motherboard, once again)... If it works, I'm going to run another benchmark on it, specifically, I'm going to have a couple opcodes that are NECV30 specific, run over and over again for a certain amount of time, and post the results... I could write up a simple program to do this. If not, i'll just write a QBASIC program, and publish it too for users to benchmark their own XT.. (i have already written a 68000 benchmark for game consoles)

EDIT: ARGH! I once again forgot that oscillators were called oscillators! I keep confusing terms. Mine has 3 oscillators, and NO crystals..

I'll have to check later tonight, with an oscilloscope, if the waveform generated by the IBMs normal motherboard, is the same as an oscillator.. Glad I do electronics for a living, otherwise getting an oscope would be freakin hard, at a cheap price that is..

 

[JoJo_ReloadeD]

my computer has a 25MHz crystal on it.. not a 24MHz one.. That would infer 3.34MHz, if it was /3, but I don't think it is. I think it's just a different division slightly. My system also lacks the clock driver that's common in the XTs, it probably has a different one.

I just need to know if the above schematic I came up, will work or not.. I'm not going to replace crystals.. I have a spare Model 25 motherboard if I had to "test" this, but I would rather not.

When I find a 10MHz crystal, I'm going to test my theory first (test motherboard, once again)... If it works, I'm going to run another benchmark on it, specifically, I'm going to have a couple opcodes that are NECV30 specific, run over and over again for a certain amount of time, and post the results... I could write up a simple program to do this. If not, i'll just write a QBASIC program, and publish it too for users to benchmark their own XT.. (i have already written a 68000 benchmark for game consoles)

EDIT: ARGH! I once again forgot that oscillators were called oscillators! I keep confusing terms. Mine has 3 oscillators, and NO crystals..

I'll have to check later tonight, with an oscilloscope, if the waveform generated by the IBMs normal motherboard, is the same as an oscillator.. Glad I do electronics for a living, otherwise getting an oscope would be freakin hard, at a cheap price that is..

Good luck then, and like I said before, all times overclocking did not work on certain system, I only had to resolder old oscillator and all went ok... They are not 1000+ mhz athlons, so overheating is not a problem.. at least for little time

 

[evildragon]

for the cool factor though, i want a nice heatsync with fan

maybe some blue LEDs too! just kidding..

 

[mbbrutman]

This isn't about how you lift pins or solder components. This is about trying to speed up your CPU without any knowledge of what it means to the rest of the system.

On these older systems the CPU is tightly coupled with the rest of the system. You can't just change the timing of the CPU without considering what it does to the bus cycle of the rest of the system. Ever hear an orchestra where one of the musicians is not in synch with the others?

You can get as 'FREAKIN' angry as you want; just explain how your system running at 8Mhz with a bus cycle based on that is going to interface to a 10Mhz CPU. Then you'll have your answer on if this will work or not.

 

[evildragon]

This isn't about how you lift pins or solder components. This is about trying to speed up your CPU without any knowledge of what it means to the rest of the system.

On these older systems the CPU is tightly coupled with the rest of the system. You can't just change the timing of the CPU without considering what it does to the bus cycle of the rest of the system. Ever hear an orchestra where one of the musicians is not in synch with the others?

You can get as 'FREAKIN' angry as you want; just explain how your system running at 8Mhz with a bus cycle based on that is going to interface to a 10Mhz CPU. Then you'll have your answer on if this will work or not.

you once posted a link to PC Sprint stuff.. then explain how that works? you had no problem sharing that.. what does that do different then, without pissing the system off? I mean, wouldn't overclocking the computer be impossible then? But it is possible, we know that much..

 

[mbbrutman]

PC Sprint provides a separate timing IC for the CPU that is faster than the system bus one. But it preserves the synchronization of the bus cycle between the CPU and the rest of the system, which is the key point I've been trying to make to you.

http://www.umich.edu/~archive/msdos/un_indexed/

Look for pcsprint.arc


Other overclocking solutions speed up the entire system, not just one part. When you do that, things stay in synch. That is much easier, and was often used on the 6Mhz PC AT.

Your solution is more like the PCSprint, but you just can't go lifting legs of ICs, drop in a different crystal/oscillator/timing IC, and go .. it requires some thought. Hence my imploring you to do some homework first.

 

[evildragon]

well im gonna have to find out which the timing IC is in the model 25, because it doesn't have the 8284..

 

[Big Blues]

Guys, take it easy. Can't we all be friends?

Mike, I don't think he made any claims saying this would definitely work. It sounds like the lad likes to tinker. So as long as he's not going to hurt himself or anybody, let him go. Even if you're 100% sure that his project will fail, let him go. If he releases the magic blue smoke that makes the system run, well that's his loss.

Just my $0.02.

Joel

 

[evildragon]

Guys, take it easy. Can't we all be friends?

Mike, I don't think he made any claims saying this would definitely work. It sounds like the lad likes to tinker. So as long as he's not going to hurt himself or anybody, let him go. Even if you're 100% sure that his project will fail, let him go. If he releases the magic blue smoke that makes the system run, well that's his loss.

Just my $0.02.

Joel

i got a spare board i call it my "test 25" board.. if i want to modify my 25 in anyway, i test it on this board... (it's a pretty beat up board.. HD controller is dead, NPU socket is destroyed, etc)..

 

[JDT]

Guys, take it easy. Can't we all be friends?

Mike, I don't think he made any claims saying this would definitely work. It sounds like the lad likes to tinker. So as long as he's not going to hurt himself or anybody, let him go. Even if you're 100% sure that his project will fail, let him go. If he releases the magic blue smoke that makes the system run, well that's his loss.

Just my $0.02.

Joel

well put!

Although I hate it when the smoke escapes from my toys!

 

[mbbrutman]

Guys, take it easy. Can't we all be friends?

Mike, I don't think he made any claims saying this would definitely work. It sounds like the lad likes to tinker. So as long as he's not going to hurt himself or anybody, let him go. Even if you're 100% sure that his project will fail, let him go. If he releases the magic blue smoke that makes the system run, well that's his loss.

Just my $0.02.

Joel

Perfectly valid, but why bother posting here for advice or help if you are not going to listen?

I suggesting doing some homework first before trying, and knowing how a CPU interacts with it's bus is kind of critical before even thinking about this kind of thing. And yet the reaction to that was kind of bizarre and defensive ... what's the point in knowing how to use an oscilloscope if you don't know what the signals on the pins mean?

I think it's stupid to try letting the smoke out of something without doing a basic level of homework first. It is beyond me why somebody would whip out a soldering iron or an oscilloscope or pretend to explain to me out to lift a single leg from an IC when the person doesn't even have the schematic for what they are about to hack up.

 

[evildragon]

i was pissed when you used insulting terms, which are not very polite, especially from a moderator... i mod two forums, and never would treat a member the way you did..

second, i was defensive on it, because you still haven't explained why i can easily do the pin-lift on the 68000, and it works just fine.. that's why i was defensive, you didn't clearly seperate the differences, for me to believe you..

i mean, im not new to electronics.. i'm a certified technician in hardware, but you know, the books don't seperate the 68000 to the 8086 when it comes to the CLK pin.. that's why i asked you, but you failed to separate it..