friedl@mtndew.UUCP (Steve Friedl) (02/15/90)
Hi folks, I am having a very rough time finding a decent I/O card for my 386, and weeks of research have been really discouraging. I am very picky about what I need the card to support, and it seems that everybody is lacking in this or that area (hell, I have enough trouble getting people to understand what I want in the first place). This is what I need -- the board must... * not lock up all the time. I had a Bell Tech/Intel ACE card in my machine, and it would lock up several times a day. This is unacceptable. * support hardware flow-control plus full modem control at the same time. Many boards support TD + RD plus a single control line in and out, but I need CTS+RTS+DCD+DTR. * do full-time hardware flow control. This means that RTS/CTS must be used without me asking it with some kind of stty operation (and I don't really care if it is under software control at all -- hardwired is OK). * support *concurrent* hardware and software flow control. It is beyond me why any vendor would make this mutually exclusive, but AT&T did it this way for their 3B2 EPORTS card. I hate them for it. * be supported in AT&T UNIX System V Release 3.2. * have the bidirectional modem port interlocking ("uugetty in the kernel"). When the outbound port is closed, DTR must drop for a brief time (so my TrailBlazer will reset) and then go high if there is a getty waiting on the inbound port (letting the modem answer again). It looks like Bell Tech really got this right. * it would be nice if the outbound modem port supported full modem control, but this is not a requirement. I think that most of the outbound ports are non-modem-control. * have responsive tech support. I am tired of having to give RS-232 lessons to tech support personnel. Having the primary tech support in this country would be a big win (the UK is too durn many time zones east of me). I am not really concerned about VPIX or DOS, and performance and cost are not that important either (within reason). I have looked at the following: Bell Tech/Intel ACE - I had the board in my machine for about six months and had to get rid of it. It locked up all the the time -- it was maddening -- has only part-time hardware (but concurrent with softawre) flow control, supports full modem control, has *great* bidirectional port locking, not so hot performance on interactive sessions. Tech support is friendly but of limited help. Intel is getting harder and harder to deal with (especially for small-fry like me). The latest driver (a week or two old) is on its way to me, so I might try this again. I understand that their current two-layer board sometimes has noise problems, so they are going to a four-layer board with power planes and such. Specialix - in my machine now on ten-day evaluation. *Very* good performance, full-time and concurrent hardware flow control (!), full modem control (!), but bidirectional port locking that doesn't work right at all. The USA representative for tech support has been friendly and responsive, but the UK folks who do the real work seem to be slow to respond and uninterested in the finer points. It looks like this board is going back, which is too bad because Specialix is who I wanted in the first place. Equinox - I am told that they have extremely high performance cards, but there is no fulltime HFC, and they have limited modem-control lines. Oh well. Digiboard - no fulltime hardware flow control, the ports lock up often in a couple of customer machines. Corollary - fulltime hardware flow control, supposed to be really nice performance, but probably doesn't have the bidirectional port locking. I spoke at length with the guy who writes the driver, and while he seems interested in making things work right, nothing is likely to be forthcoming in the short term. Anyway, this has been very frustrating. I spend at least two hours a day on the phone with Specialix tech support, with other vendors looking for somebody to answer my questions, or trying something new on a particular card to see if it will work. This really shouldn't be that hard. If anybody has any suggestions, I would love to hear them. Vendors who think they can do this are very much encouraged to call. I really want to find a board I can recommend to my customers. Steve -- Stephen J. Friedl, KA8CMY / Software Consultant / Tustin, CA / 3B2-kind-of-guy +1 714 544 6561 voice / friedl@vsi.com / {uunet,attmail}!mtndew!friedl "Winning the Balridge Quality Award is as easy as falling off a horse." - me
bill@twwells.com (T. William Wells) (02/19/90)
I'm using APT Technology's HSS08 (8 16550As) with Jim Murray's free serial driver and a Telebit. No fuss, no muss, no system lockups, and most everything you might want as well. The card was very inexpensive too, though rumor says that they've raised the price. I can't say anything about customer support because I never needed any. --- Bill { uunet | novavax | ankh } !twwells!bill bill@twwells.com
eric@egsner.cirr.com (Eric Schnoebelen) (02/21/90)
In article <1990Feb19.134803.6464@twwells.com> bill@twwells.com (T. William Wells) writes: - I'm using APT Technology's HSS08 (8 16550As) with Jim Murray's - free serial driver and a Telebit. No fuss, no muss, no system - lockups, and most everything you might want as well. The card was - very inexpensive too, though rumor says that they've raised the - price. I too am using the HSS08 from APT Technology. I am rather pleased with it. I am currently using APT's drivers for release 3.2 based 386 Unixen, with fairly good success. - I can't say anything about customer support because I never - needed any. I have been in contact with APT's customer support, and I can say they are very good, and very responsive. Case in point, I sent a bit of e-mail about a problem I had while installing the driver, and the following afternoon I received a phone call requesting more info. After providing them with all the needed info, they were able to provide several ideas to try. But before I even had a chance to try them, I received another call saying that they had located the problem, and told me how to fix it. This second call occurred after normal business hours. That's what I would call service! Just a satisified customer. -- Eric Schnoebelen eric@egsner.cirr.com schnoebe@convex.com "Operator?, I'd like to make a syscall, please."
tron1@tronsbox.UUCP (HIM) (02/22/90)
>Item: 817 by *Masked* at mtndew.UUCP >Author: [Steve Friedl] > Subj: ranting on serial I/O cards >Hi folks, > > I am having a very rough time finding a decent I/O card for >my 386, and weeks of research have been really discouraging. I am Well, I'll tell you . I had the same search --- and picked a nice simple solution. >DUMB< serial cards with 16550A's running the PD FASY driver for SYS 5 3.2 Does everything you want , never locks up . is REAL CHEAP. **************************************************************************** Everything I say is Copr. 1990, except the stuff I stole from someone else and the stuff I don't want responsibility for. Kenneth J. Jamieson: Xanadu Enterprises Inc. "Professional Amiga Software" UUCP: tron1@tronsbox.UUCP BEST PATH ---> uunet!tronsbox!tron1 Sysop, Romantic Encounters BBS - (201)759-8450 / (201)759-8568 ****************************************************************************
larry@nstar.UUCP (Larry Snyder) (02/23/90)
In article <25e2f06a:817.1comp.unix.i386;1@tronsbox.UUCP>, tron1@tronsbox.UUCP (HIM) writes: > Well, I'll tell you . I had the same search --- and picked a nice simple > solution. > > >DUMB< serial cards with 16550A's running the PD FASY driver for SYS 5 3.2 > > Does everything you want , never locks up . is REAL CHEAP. But dumb ports (multple ones) supporting high speed modems put a massive load on the primary CPU (don't they?) -- Larry Snyder, Northern Star Communications, Notre Dame, IN USA uucp: larry@nstar -or- ...!iuvax!ndmath!nstar!larry 4 inbound dialup high speed line public access system
sl@van-bc.UUCP (Stuart Lynne) (02/23/90)
In article <511202@nstar.UUCP> larry@nstar.UUCP (Larry Snyder) writes: }In article <25e2f06a:817.1comp.unix.i386;1@tronsbox.UUCP>, tron1@tronsbox.UUCP (HIM) writes: }> Well, I'll tell you . I had the same search --- and picked a nice simple }> solution. }> }> >DUMB< serial cards with 16550A's running the PD FASY driver for SYS 5 3.2 }> }> Does everything you want , never locks up . is REAL CHEAP. } }But dumb ports (multple ones) supporting high speed modems put a massive }load on the primary CPU (don't they?) } On Xenix even using 16450's will work for one or two Trailblazers quite nicely. On my system (25Mhz 386) I didn't really notice a tremendous load. With 16550's I have run three Trailblazers and three 2400 Hayes with little loading and no character loss problems. -- Stuart.Lynne@wimsey.bc.ca ubc-cs!van-bc!sl 604-937-7532(voice) 604-939-4768(fax)
larry@nstar.UUCP (Larry Snyder) (02/23/90)
In article <196@van-bc.UUCP>, sl@van-bc.UUCP (Stuart Lynne) writes: > > On Xenix even using 16450's will work for one or two Trailblazers quite > nicely. On my system (25Mhz 386) I didn't really notice a tremendous load. > > With 16550's I have run three Trailblazers and three 2400 Hayes with little > loading and no character loss problems. > Ahh - but with Xenix. Unix would have problems with 3 blazers and 3 2400 baud modems all going full bang on dumb ports. I wonder if 16550AFNs and a replacement driver (FASY, Murray) would work in the above installation (under 386/ix of course) supporting 6 modems? -- Larry Snyder, Northern Star Communications, Notre Dame, IN USA uucp: larry@nstar -or- ...!iuvax!ndmath!nstar!larry 4 inbound dialup high speed line public access system
sl@van-bc.UUCP (Stuart Lynne) (02/24/90)
In article <90.25E55172@csource.oz.au> david@csource.oz.au (david nugent) writes: >In a message of <Feb 23 23:18>, Stuart Lynne writes: > > >On Xenix even using 16450's will work for one or two Trailblazers quite > >nicely. On my system (25Mhz 386) I didn't really notice a tremendous load. > > > >With 16550's I have run three Trailblazers and three 2400 Hayes with little > >loading and no character loss problems. > > >Unless the serial driver is written specifically to handle and will enable the 16550's FIDO queues, then the CPU overhead is the same as for 16450, so you shouldn't notice any difference at all. Running a 16550AN without it's FIFOs "disabled" (well, the FIFO's themselves are always enabled, except the IRQ trigger levels are set at 1) will however aleviate some character loss in itself, since the FIFO is still present, and receive overruns become more rare. It was. To amplifify my above remarks: I have run two Trailblazers under Xenix with 16450's using SCO's std serial driver .... I have run three Trailblazers and three 2400 Hayes under Xenix using my own serial driver with a HUB6 card with 6 16550's ... I have run two Trailblazers under Xenix with 16450's using my own serial driver .... -- Stuart.Lynne@wimsey.bc.ca ubc-cs!van-bc!sl 604-937-7532(voice) 604-939-4768(fax)
david@csource.oz.au (david nugent) (02/24/90)
In a message of <Feb 23 23:18>, Stuart Lynne writes: >On Xenix even using 16450's will work for one or two Trailblazers quite >nicely. On my system (25Mhz 386) I didn't really notice a tremendous load. > >With 16550's I have run three Trailblazers and three 2400 Hayes with little >loading and no character loss problems. Unless the serial driver is written specifically to handle and will enable the 16550's FIDO queues, then the CPU overhead is the same as for 16450, so you shouldn't notice any difference at all. Running a 16550AN without it's FIFOs "disabled" (well, the FIFO's themselves are always enabled, except the IRQ trigger levels are set at 1) will however aleviate some character loss in itself, since the FIFO is still present, and receive overruns become more rare. david -- UUCP: ...!munnari!csource!david <Fido/ACSNET Gate> Internet: david@csource.oz.au FidoNet: 3:632/348
tron1@tronsbox.UUCP (HIM) (02/24/90)
>Author: [Larry Snyder] > Date: Fri Feb 23 1990 06:53 >> >DUMB< serial cards with 16550A's running the PD FASY driver for SYS 5 3.2 >But dumb ports (multple ones) supporting high speed modems put a massive >load on the primary CPU (don't they?) Well, with the FIFO's on the 16550A's enabled I havenet noticed MUCH .. of cource, it is more load that a dedicated CPU card .. but a good solution in the interim. **************************************************************************** Everything I say is Copr. 1990, except the stuff I stole from someone else and the stuff I don't want responsibility for. Kenneth J. Jamieson: Xanadu Enterprises Inc. "Professional Amiga Software" UUCP: tron1@tronsbox.UUCP BEST PATH ---> uunet!tronsbox!tron1 Sysop, Romantic Encounters BBS - (201)759-8450 / (201)759-8568 ****************************************************************************
jeg7e@watt.acc.Virginia.EDU (John E. Gefaell) (02/26/90)
But that limits you to 2 serial ports, right? Under what conditions and with what equipment can you use more than 2 serial ports, given that you are using the other IRQ's on the first controller, like 2 for the mouse ??? Are there dumb serial port cards that can use the second set of IRQ's? or can SysV/386 somehow manage sharing interupts? Please lemme know, I've got a card with two ports, on a 16450 and on a 16550. . . I'd love to get four ports working with CTS/RTS flow control, etc. . .