From: Groepaz (groepaz_at_gmx.net)
Date: 2004-01-15 16:24:40
On Thursday 15 January 2004 15:21, Spiro Trikaliotis wrote: > On Thu, Jan 15, 2004 at 01:05:56PM +0100, Groepaz wrote: > > but a missing rts/cts connection is pretty much ridiculous - this wont > > ever work right except when both end are faster than the actual > > transfer is. > > There is also a software-handshake method (XON/XOFF, for example). uhmz yes. you can also send snail mail or morse code :=P seriously, software handshake is no option these days. it just plainly sux, you dont want it. never. > As I told, I'm not the expert. Anyway, from my understanding, this > problem arises because the power lines on both sides can get different > potential ("Potenzial" in german) from the PSU, so there is some current > running through, trying to put balance this difference. yes, but the diodes will avoid that (they are a "one way" for current) ever seen one of these cheap centronics interfaces for c64 that have a lot of diodes at the userport connector? guess what they are good for :) > From what I remember, this happened on the serial lines (RS232C) in the > company I told, and there were some Ampere running (but not damaging > anything seriously). Flipping the power cord in the outlet solved the > problem before they used the permanent solution: Using optocoupler to > galvanically isolate both sides. very common mistake, and done in 99% of all installations that use rs232 1) you are ***not*** supposed to connect protective and signal ground, this is a violation of the rs232 standard (only protective ground is directly connected to PE, signal ground -just like VCC- _is_ already galvanically isolated by the transformer in the PSU!) [note: the optocopplers would still make sence for protecting the interface chips themselves by other means, but thats not the point here] 2) you are ***not*** supposed to connect protective ground on both ends of the rs232 cable, thats again a violation of the standard. (this is what makes those high unwanted currents possible) now guess what, those two pretty effectivly eliminate the problem you are describing :) its btw also the main reason for any failing rs232 transfers i have seen so far (besides broken/wrong cables) btw, using rs232 cables >10m are also a violation of the standard (and if i had the time to waste i could probably prove by calculation that within this range the resistance of the wire isnt high enough to provoke any kind of harmful voltage differences) solution: use ethernet damnit :=P > Have a look here (sorry, german only): > > http://www.wut.de/e-8wwww-13-apde-000.php3 > > According to this text, it makes a difference. it makes a (subtle) difference when using that kind of bulk equipment you get at walmart, yes indeed. (same reason for starcommander working here and not there). ***decent*** rs232 solutions do not have this problem at all. (that includes not violating the standard by any means, including removing those common 1000m cables :)) oh btw that page has a slight mistake....according to VDE *any* kind of "new" installation (that includes installing a new power outlet or so) requires you to install proper seperated pe/n/l atleast within the next main "circuit". this is rarely done (because its not quite cheap) but it would solve the problem nevertheless (atleast make it less bad) :) (and if you ever see such a "wrong" installation and want to be mean - a simple call to VDI or simelar institution will do the trick) gpz Message was sent through the cbm-hackers mailing list
Archive generated by hypermail pre-2.1.8.