Quadra 700 overclocked at 37.5 MHz

[timtiger]

I was wondering, if you are interested to see my overclocked Q700? I replaced the 50MHz oscillator/quarz by a socket and put different clocks into it: 64MHz (to achieve 32MHz speed) and 75MHz - which achieved 37.5MHz and I kept in place.

Though the processor gets really warm - so I decided to add a cooling fan to the processors cooling fin.

The Quadra booted nice and smooth! Different benchmarks reflecting the increase of speed. I‘m now long running some benchmarks over the night to see, if the Quadra is still stable and reliable.

It maybe not a great deal - for me it is Never thought, I‘d be able to do something like this!

 

[Phipli]

Thats absolutely awesome

The things to check are if the floppy drive and serial ports still work as they tend to be the first to fall over when overclocking. You might want to heatsink the video chip too if it is anything like the Q630.

Fast VRAM might also help, but it seem like it is managing with what you have. Is it Apple VRAM?

 

[timtiger]

The things to check are if the floppy drive and serial ports still work as they tend to be the first to fall over when overclocking. You might want to heatsink the video chip too if it is anything like the Q630.

Ah good to know, I‘ll check that later.

Is it Apple VRAM?

Yes I believe something around 70ns? Is there a way to verify this?

 

[Phipli]

Ah good to know, I‘ll check that later.

Yes I believe something around 70ns? Is there a way to verify this?

Its written on the chips usually. The part mumber will end in something like -70 or -7.

 

[timtiger]

I have good news and bad news…

Good is: Even at 37.5MHz the floppy works fine. Read, write, copy, move and initialize works as expected.

Bad is: It seems that I have scrambled the serial ports:
- After trying to print using different printers I can say: The Quadra don’t want to print anymore
- Switching back to 25MHz don’t brought any change
- Resetting PRAM also doesn’t change anything
- Thought some kind of data seems to go through to the devices, but it is not enough to print. (Line Feed command for the LabelWriter works most of the time)
- I switched to clean system to avoid misconfiguration and side-effects of extensions - no changes
- Snooper reports handshake and different baud rates are ok for both ports (using a loopback adapter)
- I tried also an ImageWriter II to see what it is printing: The first print was okay until the very last character, since then it is printing line feeds most of the time

@Phipli do you have any idea what I destroyed or what to look for? Thaaaanks

 

[Phipli]

I have good news and bad news…

Good is: Even at 37.5MHz the floppy works fine. Read, write, copy, move and initialize works as expected.

Bad is: It seems that I have scrambled the serial ports:
- After trying to print using different printers I can say: The Quadra don’t want to print anymore
- Switching back to 25MHz don’t brought any change
- Resetting PRAM also doesn’t change anything
- Thought some kind of data seems to go through to the devices, but it is not enough to print. (Line Feed command for the LabelWriter works most of the time)
- I switched to clean system to avoid misconfiguration and side-effects of extensions - no changes
- Snooper reports handshake and different baud rates are ok for both ports (using a loopback adapter)
- I tried also an ImageWriter II to see what it is printing: The first print was okay until the very last character, since then it is printing line feeds most of the time

@Phipli do you have any idea what I destroyed or what to look for? Thaaaanks

Strange - I wouldn't have expected such a fault to stick so I think simething else is happening.

Have you tried another cable?

 

[timtiger]

Yes, I tried another printer cable (Though LabelWriter and StyleWriter are using different cables anyway)

 

[timtiger]

Could it be a soldering problem at the socket? It’s a multilayered board right?

I don’t have any schematics of the logic board to check all traces…

 

[Phipli]

Could it be a soldering problem at the socket? It’s a multilayered board right?

I don’t have any schematics of the logic board to check all traces…

Have you tried the printer from both ports?
Do you have another computer to try the ImageWriter with?

Don't worry, the computer (or printers, or both (one because of the other) is most likely just muddled and not actually damaged. Absolute worst case, the serial circuit isn't complex and isn't many chips.

But, given the loop back works, the serial on your Quadra is almost certainly fine.

A really daft thing to try that shouldn't work is to just unplug everything and take the PRAM battery out and come back to it in a couple of days. I once had a ln 8600 that would not boot no matter what and I thought I'd destroyed it - I was climing the wall as it was of sentimental value to me and my dad. I spent hours tryingvto get it running, changed the CPU, zapped PRAM, took the battery out, pressed the power button withvthe cord and battery out, reseated the RAM, removed cards... nothing worked.... a week later I poked the power button as I walked past and it started right up first time and hasn't caused any trouble since.

 

[timtiger]

Have you tried the printer from both ports?

Yep

Do you have another computer to try the ImageWriter with?

On the SE/30 it’s printing as it should

A really daft thing to try that shouldn't work is to just unplug everything and take the PRAM battery out and come back to it in a couple of days.

I‘ll check the solder connections of the socket (I’m sure that will not make a difference) and will finally do as you recommended.

Thanks for your kind words, the Quadra is my most favorite machine…

 

[Phipli]

I‘ll check the solder connections of the socket (I’m sure that will not make a difference) and will finally do as you recommended.

It would have to be impacting both sockets. If it is something on the board it will be closer to the... Zilog seral chip... the 8530. After that I think the signals travel independently, through two transmitter / receiver chip pairs (on my 630 anyhow, sadly I'm not lucky enough to have a Q700).

Good luck. Let us know how it goes. Even if there is a fault, remember anything is fixable with enough time and communal brain power

 

[Phipli]

I don't know the board and I can't quite read the chip labels in the photo, but I feel that the highlighted bit is the interesting bit. The first and second pairs of chips look like the transmitter and receivers, and the one labelled SCC is the Zilog Z8530 serial chip thingybob. As it is a Q700, at least we know it hasn't suffered from cap damage, and the battery is miles from this part of the board.

Edited to add two more chips and I need to remind myself what the third pair of chips are :

 

[timtiger]

the one labelled SCC is the Zilog Z8530

Thank you for supporting me so much. Tomorrow I shoot a closeup of this area.

Sadly I cannot find any schematics of the Q board. Such were very helpful understanding and bugfixing the SE/30 with @Bolle last year.

Is the Zilog somehow connected to the quarz oscillator?

I‘m wondering if I missed to fill out the sockets leg holes completely with solder. (But it looks good for me) Hm…

Anyway, the board takes a cleansing bath and dries over night… (I never had to clean the board - I was lucky to get this nice Quadra for 100€ in last January. Though it took several years to find a Q700 which was not overpriced too much)

 

[Phipli]

Is the Zilog somehow connected to the quarz oscillator?

I‘m wondering if I missed to fill out the sockets leg holes completely with solder. (But it looks good for me) Hm…

It is, but I think the clock also feeds other things, so you'd know about it I think.

There are schematics available for similar machines. A similar aged machine would be informative enough I'd probably start with the IIci or IIsi. There are Bomarc schematics online for those machines.

 

[Phipli]

The IIci and IIsi have basically the same components. It looks like the chips that were confusing me are resistor arrays.

https://recapamac.com.au/wp-content/uploads/2019/07/mac_iici_logic.jpg

https://recapamac.com.au/wp-content/uploads/2019/08/IMG_5981.jpg

 

[timtiger]

BTW: VRAM is 80ns

 

[jeremywork]

My thought is the data on the hard drive may have become corrupt and something in the printing system is not loading correctly. NewerTech's manual for the Variable Speed Overdrive mentions an idiosyncrasy with the Quadra 700 specifically, that the SCSI controller uses the same clock as the CPU, suggesting that I/O errors might occur if the speed is not reduced during SCSI access. I'd reinstall the OS/restore from a backup while at 25MHz and see if that fixes anything.

 

[timtiger]

I'd reinstall the OS/restore from a backup while at 25MHz

Interesting. The Q700 SCSI implementation is weird, isn’t it?

I already switched at 25MHz to an old installation to do some prints: Same effect…

 

[jeremywork]

Good to know. In that case I'd agree with @Phipli that since the loopback seems to be working it's still more likely a fault in some code layer than a hardware fault caused by overclocking. Not impossible, however. The 'leave without power for a while with no battery' trick has some weight to it, in weird cases like this.

Just another thing to confirm: the printer was tested working with this Quadra before the overclock?
If not, it may be a preexisting hardware issue or something in ROM has been corrupt for a while.

 

[jeremywork]

The Q700 SCSI implementation is weird, isn’t it?

It might be this note doesn't matter much, as the other times I recall bringing this up it hasn't turned out to be the cause of any overclock symptoms being discussed (or isn't really true? There's a 100MHz oscillator directly next to the SCSI chip... At least as recently as 2015 it doesn't seem like a schematic is available for the 700.)