smith@Nrl-Aic.ARPA (04/14/84)
From: Russ Smith <smith@Nrl-Aic.ARPA> A while back I posted a request for help concerning problems I was having using MDM7xx on a homebuilt z80 based machine with a Hayes 1200 Smartmodem and an h-19 terminal. The problem was that I could hook the Hayes directly to the terminal and talk both ways to another machine without any problem but if I hooked the Hayes and the h-19 to my machine I could only send stuff to the other machine, received stuff being totally garbaged. The problem seemed to go away after I just turned everything off and reseated boards...it didn't. Well, the recent net traffic about modem heating problems got my synapses firing... I took the front cover off the Hayes, placed a muffin fan blowing at it about 4" away, and all symptoms have "permanently" disappeared. Apparently my chicken-wire-and-spit homebuilt machine is close to the limit with the "receive data from modem" baudrate. Hence when the Hayes started drifting due to heat, the received characters started getting garbaged (framing error). This caused the modem-to-computer-to-terminal setup to go south. When I "reseated" the boards, I had to turn everything off, thus giving the Hayes time to cool down and causing the problem to go away (temporarily). This red herring behavior led me astray until I read the recent net stuff. When just hooked in as a modem-to-terminal setup everything worked okay, so I have to conclude that the h-19, using newer technology (software programmable UART), is more tolerant of out-of-spec received baudrate than my "old" hardware-programmable-UART-based Godbout Interfacer (sans II (inside "joke"...)) Now the problem is to find a way to sufficiently cool down the Hayes without the silly muffin fan... Keep them informative messages coming, they've helped more than once, Russ <Smith@nrl-aic>
phil@amd70.UUCP (Phil Ngai) (04/25/84)
Hmmm, we have two Hayes Smartmodem 1200s sitting in a Western Electric
modem rack with very poor air circulation (it's got a door on the
front and no fan) and NEVER had any trouble with them, heat related
or otherwise. We have passed many megabytes through them and logged
many hours of connection time.
I have seen some sleazy circuits which used baud rate clocks which were
a little bit off. As an example, one circuit was about 2% off. No problem,
right? The error adds up. With an average 10 bit character (1 start, 8
data, and 2 stop bits) the accumulated error at the last bit will be 20%.
If the other circuit were off 2% the other way, there'd be a system error
of 40%. It's acceptable, but not what I would call good engineering. It
does save the price of a crystal ($2).
I don't know what your interface board uses, but it seems more likely a
cause of the problem than the Hayes.
--
Phil Ngai (408) 988-7777 {ucbvax,decwrl,ihnp4,allegra,intelca}!amd70!phil