bruce@sonyd1.Broadcast.Sony.COM (Bruce Lilly) (01/08/91)
In article <133@limbic.ssdl.com> gil@limbic.ssdl.com (Gil Kloepfer Jr.) writes: > >Would you please explain to me what evidence you have that the Telebit >modem is doing something to violate the async serial timing constraints? >I have forwarded the problem to someone very knowledgable at Telebit >(it was Cerafin, by the way, who is now no longer with Telebit), and >he hadn't heard of any such bug. Have you reported this to Telebit, >and have THEY admitted any timing problem? Hmm... See below (perhaps he forgot?) > [ ... ] >If you do have some evidence as to how and why this is happening, please >follow-up to the net and explain in-depth how and why this is happening. >Otherwise, look to another source of the problem. Glad to. I've x-posted to c.s.att and comp.dcom.modems, follow-up remains in unix-pc.general (unless/until somebody changes that). This is a summary of the correspondence which is still in existence. Sorry this is so long, but the question was asked... It started when I had a problem communicating with another system. I was using the unix-pc OBM, the other system (sonyusa) was using a Telebit (T100 I believe, but don't quote me). I posted (over 7 months ago): From sonyd1.uucp!bruce Wed May 23 15:09:04 EDT 1990 Article 11 of unix-pc.bugs: Xref: sonyd1 comp.sys.att:249 unix-pc.general:260 unix-pc.uucp:3 unix-pc.bugs:11 comp.dcom.modems:1 Newsgroups: comp.sys.att,unix-pc.general,unix-pc.uucp,unix-pc.bugs,comp.dcom.modems Path: sonyd1!sonyd1.uucp!bruce >From: bruce@sonyd1.uucp (Bruce Lilly) Subject: AT&T 3B1/Telebit modem compatibility problem Date: Wed, 23 May 90 18:37:35 GMT Distribution: usa Organization: Sony Communications Products Company, Teaneck, NJ Sender: bruce@sonyd1.uucp (Bruce Lilly) Message-ID: <1990May23.183735.9239@sonyd1.uucp> It seems that the built-in modem in the AT&T 3B1 (a.k.a. 7300) has some problems when communicating with Telebit modems at 1200 baud. The problem occurs when the OBM calls out to a Telebit which is set to provide the PEP answer tones first. The symptom is a sequence of correctly received characters (about 20) followed by gibberish. My theory is that the OBM is responding to the V.22 bis tones sent by the Telebit (before it gets to 1200 212A), which results in synchronization errors due to the slight speed difference between V.22 bis and 1200 baud Bell 212A. I've tried the stock cu and the HDB version with both 3.51 and 3.51m kernels with the same results. A program written to use the dial(3) routine also has the same problem. At this point, I suspect either a hardware problem, or a problem with the driver for the OBM (assuming it hasn't changed between 3.51 and 3.51m). Has anybody else experienced this problem, and if so, are there any workarounds (short of using 300 baud, which does work) or bug fixes? One final note: the characters sent by the 3B1 seem to be received OK at the Telebit end, but the symptoms described above affect the flow of data from the Telebit to the 3B1. Also, the 3B1 is set up not to use the guard tones, etc. which would be used for V.22 bis compatibility. Changing this also does not appear to improve things. OK, as it turned out my theory was wrong. Note that the symtoms are the same as described in <1991Jan7.083645.7349@yenta.alb.nm.us> by David B. Thomas (recently posted to unix-pc.general -- refer to it for more details). I received a few responses to my posting -- some were dead ends; the following were not: From vmp!uunet!attcan!ncrcan!becker!hybrid!mdapoz Fri Jun 1 10:35:12 EDT 1990 Article 12 of unix-pc.bugs: Xref: sonyd1 comp.sys.att:290 unix-pc.general:277 unix-pc.uucp:5 unix-pc.bugs:12 comp.dcom.modems:2 Path: sonyd1!vmp!uunet!attcan!ncrcan!becker!hybrid!mdapoz >From: mdapoz@hybrid.UUCP (Mark Dapoz) Newsgroups: comp.sys.att,unix-pc.general,unix-pc.uucp,unix-pc.bugs,comp.dcom.modems Subject: Re: AT&T 3B1/Telebit modem compatibility problem Message-ID: <1990May28.012449.3507@hybrid.UUCP> Date: 28 May 90 01:24:49 GMT References: <1990May23.183735.9239@sonyd1.uucp> Distribution: usa Organization: The Home for Unemployed Basselopes, Toronto, Ontario, Canada Lines: 30 In article <1990May23.183735.9239@sonyd1.uucp> bruce@sonyd1.uucp (Bruce Lilly) writes: >It seems that the built-in modem in the AT&T 3B1 (a.k.a. 7300) has some >problems when communicating with Telebit modems at 1200 baud. The >problem occurs when the OBM calls out to a Telebit which is set to >provide the PEP answer tones first. The symptom is a sequence of >correctly received characters (about 20) followed by gibberish. This problem only occurs when talking to a telebit which has locked the interface speed (ie. S51 = [0-5]). I think the problem was that one of the modems (the telebit I believe) is cutting the stop bits a little bit short of what the other expects, and as a result you get garbage. For some reason the problem disappears when you get the telebit to autobaud the interface. >At this point, I suspect either a hardware problem, or a problem with the >driver for the OBM (assuming it hasn't changed between 3.51 and 3.51m). As you've found out, the OBM is very picky with what it's communicating with. Do yourself a favour and pick up a cheap 2400 baud modem and hook it up to the serial port. >Has anybody else experienced this problem, and if so, are there any >workarounds (short of using 300 baud, which does work) or bug fixes? See above, I got it to work that way but I don't guaranatee that it'll still work since I don't use the OBM and my telebit now has V5 roms. -- Managing a software development team | Mark Dapoz is a lot like being on the psychiatric | mdapoz%hybrid@cs.toronto.edu ward. -Mitch Kapor, San Jose Mercury | ...uunet!mnetor!hybrid!mdapoz Oops, I seem to have lost John B. Milton's email on this subject. Thanks anyway to John. Next step was the following email to Telebit: From bruce Fri Jun 1 11:30:28 1990 Subject: Stop bit problems? To: modems@telebit.COM Date: Fri, 1 Jun 90 11:30:28 EDT Cc: ryan@sonyusa, mdapoz@hybrid, celebr!jbm X-Mailer: ELM [version 2.2 PL16] Status: RO I'm having a problem communicating at 1200 baud with some Telebit modems. The problem seems to occur when the Telebit is set to use a fixed bit rate to the CPU, independent of the modem-to-modem communication rate. I have been informed by mdapoz@hybrid (Mark Dapoz) and by celebr!jbm (John B. Milton) that the problem is caused by improper timing of the stop bit duration put out by the Telebit modem under these conditions. Are you aware of this problem, and is there a fix to correct the stop bit timing (if that is indeed the problem)? For various reasons, it is not possible to use an automatic baud rate selection for the CPU-modem connection in this installation. Thank you in advance for your help. Cerafin E. Castillo of Telebit sent the following professional, helpful reply: From bruce Sat Jun 2 09:30 EDT 1990 >From uunet!america!telebit!modems@vmp Sat Jun 2 09:30:25 1990 Received: from vmp.UUCP by sonyd1.Broadcast.Sony.COM (5.61 5/25/90) id AA05896; Sat, 2 Jun 90 09:30:25 -0400 Received: by vmp.com (smail2.5) id AA09663; 1 Jun 90 20:21:13 EDT (Fri) Received: from telebit.UUCP by uunet.uu.net (5.61/1.14) with UUCP id AA21976; Fri, 1 Jun 90 19:47:58 -0400 Received: from america.TELEBIT.COM by telebit.TELEBIT.COM (4.0/TELEBITGENERIC.3) id AA19806 to sonyd1!bruce; Fri, 1 Jun 90 12:58:22 PDT Received: by america.TELEBIT.COM (4.0/TELEBITGENERIC.3) id AA08995 to uunet!sonyd1!bruce; Fri, 1 Jun 90 12:57:10 PDT Date: Fri, 1 Jun 90 12:57:10 PDT From: america!telebit!modems@uunet (Telebit Support acct) Message-Id: <9006011957.AA08995@america.TELEBIT.COM> To: bruce@sonyd1 Subject: Re: Stop bit problems? Status: RO Bruce, We are aware of this 'bit-shaving' problem in the T1000 modems when S66=1 and S51=1-5 is set. Apparently, S66=0 with S51=254 for auto baud has a bug as well in certain releases. I recommend that you call 1-800-TELEBIT and talk to Jim in Tech Support. He is well aware of this problem and may have some info for you. =============================================================================== Cerafin E. Castillo |\\\\\\\| UNIX Communications Specialist |\\\\\\\| Telebit Corporation |\\\\\\\| 1315 Chesapeake Terrace T E L E B I T Sunnyvale, CA 94089 (408) 734-4333 UUCP: {sun,pyramid,uunet,ames}!telebit!cec (408) 745-3104 INTERNET: cec@telebit.com "...A kinder modem for a gentler UNIX community..." ================================================================================ A little over a week from posting to a good, solid lead -- not bad. At that point, my contact at the system with the Telebit modem contacted Telebit and discussed the problem further. I don't know the details of the discussion. The solution at that time was the addition of a Hayes 1200 baud modem to sonyusa for calling sonyd1 to avoid the problem. Since then, sonyd1 has moved and is now using an external quasi-modem which appears to work well with the Telebit up to 2400 baud (the limit of the quasi-modem) (that is, it works when I can get through the phone system and modem bank :-(). I hope this answers the question... -- Bruce Lilly, Product Manager, | bruce@Broadcast.Sony.COM Digital Television Tape Recording, | uunet!{sonyusa,vmp}!sonyd1!bruce Sony, 3 Paragon Drive, Montvale, | lilb@sony.compuserve.com (slow) NJ 07645-1735 | Telephone: 1(201)358-4161 | FAX: 1(201)358-4089