[labnetwork] Plasmatherm ISA Cards

P. Scott Harris psharris at trilicium.ca
Wed Sep 30 22:06:58 EDT 2020


Hi Shawn,

Just for future reference and anybody else who runs into this kind of thing, 
since you found that you needed to adjust the I/O speed of the bus, I looked 
more closely at the ITE8888G chip documentation 
(http://www.ite.com.tw/uploads/product_download/IT8888G_V0.9_05162005.pdf) and 
found that there is a register that you could set that would slow down the bus 
accesses by extending the recovery time. It's on page 30, and called the 'IO 
recovery register' #51. By default it looks like the ISA bus runs at top speed 
(3.5 BCLK cycles where a BCLK cycle is 1/4 of the PCICLK which seems to be 
33MHz). So the ISA bus clock is running at 8.25 MHz and the recovery time looks 
to be about 425 nsec which might be too fast for your old boards. You could, 
with some simple DOS code, extend the recovery time to as much as 1.4 usec 
(8-bit I/O) or  900 nsec (16-bit I/O) which might possibly make your boards 
happy. Usually this sort of control is provided by the BIOS but that doesn't 
seem to be the case for this motherboard.

You would have to play a little bit with this timing register in order to 
confirm that this actually fixes your original problem. You have no real 
incentive to play since you now have a working system but the information might 
be useful for some other applications in the future.

On 9/30/2020 6:53 PM, Shawn Wright wrote:
> Hi Everyone,
>
> Thanks for the help and suggestions, we’re back online now.  It turns out the 
> old analog boards (ADA88’s from ROBOTROL) weren’t able to run at the normal 
> speed of an ISA bus and need more like 4.77MHz.  Luckily Plasmatherm sells 
> more modern replacements from Nova Research (ADS168) and those are fine at the 
> higher speed (8.25 MHz in this case since the PCI-ISA bridge chip divides the 
> PCI clock by 4).
>
> Shawn
>

-- 
Best regards,

P. Scott Harris, M.Eng., P.Eng.
H&L Associates
21 Parkmount Crescent, Ottawa, Ontario K2H 5T3
(613) 828-1462



More information about the labnetwork mailing list