"The 56k Sweet Dream" (is made of this) -A Technology Thriller, based on Real life.- :) Chapter 1: the Hayes Accura --------------------------- First, I purchased a Hayes Accura 56k/voice. BZZT WRONG! :) After connecting it, I dialed my ISP only to find it ignored the "special" 56k beep of the k56flex handshake and proceeded to connect at 28.8. LESSON #1: The early Accuras (and the ones from Motorola and Diamond) have not only NON-UPGRADEABLE firmware, but also a BAD, totally BRAIN-DAMAGED DATA-PUMP, this makes it IMPOSSIBLE to establish a connection in K56flex rates. The only way to fix these modems is to return them to the manufacturer. How to detect it: If you want to check whether yours is upgradeable, you must issue either an ATI6 or ATI7 command to your modem and look for a response that starts with RC. The text following that first word either looks like: L8570 Rev 13.2/13.2 or 47BA. The L8570-type modems CAN BE UPGRADED (the data pump is DSP-based and can be upgraded by software), while the 47BA modems can NOT. Chapter 2: the external ------------------------- After about a month waiting, my hardware provider finally phoned me to tell me that he had received a new shipment of Hayes Accuras, this time external, and that he would exchange it at no extra cost. Great!. The first thing I did after connecting it was typing "ATI6". Result: "47BA". LESSON #2: Both the INTERNAL *and* external Accuras have this data-pump bug. Hayes claim they fixed it and that non-upgradeable modems are no longer shipped. However, the market is flooded with these modems (at least locally, maybe they ship them to south america knowing they won't come back home :). So... that one went inmediatelly back to the box. Chapter 3: the Motorola VoiceSURFR ---------------------------------- After a dozen or more visits, my usual hardware provider didn't have to talk, he just took the Accura and gave me a brand new Motorola VoiceSURFR 56k/voice. I dialed my ISP... CONNECT 33600. GRRRR.... so I headed to dejanews... "Motorola VoiceSurfr k56flex". Oh boy... it turns out that ALL the "K56flex" modems, no matter which brand, have *ANOTHER* problem: the level of k56flex code installed. So ***you can have an UPGRADEABLE, "good datapump" k56flex, yet not get speeds higher than 33600***. LESSON #3: The level of "K56flex" firmware is VERY important. This Motorola VoiceSURFR, with a manufacture date of "06/97" contained "v0.54" level of K56flex. You can check the level of K56flex code in your modem by typing the command "ati3". An ati3 revision of 0.5xx means that the protocol is really Rockwell's K56plus (before the agreement with Lucent) and not K56flex (co-developed with Lucent and Motorola). This code level, for example, CANNOT CONNECT AT ALL WITH LIVINGSTON ROUTERS of a certain revision level. (and by 'not connect' I mean NOTHING, NADA). I typed "ati3". The modem's answer: 0.54. By this time I was thinking of using my brand-new modem to start the BBQ fire on sunday. Yet I decided to continue "playing" (remember, all I wanted to do is lower my phone bill (there's no such thing as free, unlimited local calling down here) and *USE* my brand new modem, however the current state of technology prohibits me from doing with my modem what it is intended for). So... I headed to Motorola's web site looking for flash upgrade code. LESSON #4: The Motorola's 56k flash upgdates are *NOT* available on the www.mot.com site. You have to either call long distance to motorola's US or UK tech support BBS, or download the code from a few third-party sites (ISPs who had set up "k56flex help" pages, go figure). Chapter 4: If it's upgradeable, it doesn't mean it's easy. --------------------------------------------------------- I finally had on my HD the magic file: 9-20-97 6:44a 772,233 42 a--- 56K1009.EXE (I wonder why Phil Katz and others have developed ZIP and other multi-platform CRC-aware archive formats if developers decide to ship sensitive data files as flash updates on a plain DOS EXE ;). Maybe this is a conspiracy to make Mac and unix users see the DOS prompt to unpack a file?. ...this is indexed everywhere as "k56flex update to v1.009 for Motorola ModemSURFR". Mine is a *VOICE*surfr so I was not sure if this update was for my modem. The TXT inside however indicated that the archive contains multiple files for each model, including the external and internal VOICEsurfers. Now... if you had the luck of NOT buying a buggy Hayes, and the LUCK that your Motorola is of the UPGRADEABLE type... *AND* the luck of having managed to learn about the "v0.54" problems *AND* obtained the well-hidden v1.009 update... HOW THE HECK DO YOU INSTALL IT???. Motorola's answer: "BY USING THE INCLUDED WINDOWS 95 UPDATE PROGRAM". !!! Of course!!, everyone is running Windows 95, right? WRONG DAMNIT!. LESSON #5: The txt describes in a few words a procedure for updating "with other operating systems". It involves loading a terminal program, setting hardware handshake on both the modem and software, setting ascci upload to disable carriage returns and every 'tweaking' that might be available, setting terminal type to TTY, port speed to 115200, typing "AT**" and 'ascii upload' the new k56 code file to the modem. I wonder how long it would have taken for them to code a *DOS* version of this updater that can be run everywhere. I'm not even thinking about an OS/2 version. (which would have been easier). Oh well, I did the update dance. After about one minute of seeing dots on the screen (my fingers where crossed, remember I was the one who lost two mobos due to a bios update getting dead at 99%), the Moto displayed "update succesfull". It now shows: ati3 V1.009-K56_DLS OK Chapter 5: The never-ending story. --------------------------------- So, I have a good-datapump, flash upgradeable, just-updated-to-v1.009 Motorola VoiceSURFR. I should be able to connect to 56k right?? BZZT WRONG! CONNECT 33600. (begginning to hate Hayes, Motorola and the whole modem industry). The only difference is that with this new code, now the modem ATTEMPTS a handshake at 56k (the first beep from the ISP's router is replied with a series of fast, low sounding beeps from the modem, after that the k56flex handshake quits and a normal 33.600 carrier/handshake is heard). LESSON #6: Now the last tip I've found on the newsgroups suggest that v1.009 is not the real thing. It seems that "K56flex 1.1" has been released, but Motorola has not released flash updates to this specification yet. They are supposedly "working on it". But no word from Motorola (by looking at the Motorola or Hayes web pages, NOTHING about all this mess is even mentioned). Oh well... Chapter 6: it's not my fault, is it my telco's? ---------------------------------------------- A couple posts on usenet and the ascend site (http://www.ascend.com/737.html) suggest to check line quality by dropping to command (+++) mode once connected and use the Rockwell AT%L and AT%Q commands. AT%L reports "line signal level" in -dbm. AT%Q reports "line quality index". (?) I dialed my ISP, and got the following figures: AT%L 030 (-30 dbm, dbm=decibels/milliwatt) AT%Q 035 According to Ascend's info, "with anything higher than -24dbm, you will be seeing bad connects or possibly none at all". And "higher than 15 on the second (line quality index), and you've got real line problems on your circuit that your telco should be able to sort out". (ho ho ho) Well, I'm sorry to contradict Ascend, but with these figures I never get slower than 28,800 connects (and 33,600 with all the Flex), and I *never* got disconnected. I think my usage record was 12 straight hours @ 28.8, and I *never* got a disconnect, this on a no-name taiwanese 28.8. And this is *THE SAME LINE*, dialing to the *SAME AREA CODE* where I consistently get 50K connects using an USR with X2 protocol to the local compu$erve node (but at $15/hour access surcharge, I'd prefer to be using the internet @ 300bps ;) Here's the X2 logon to prove it: atdt311-0600 CONNECT 52000/ARQ/x2/LAPM/V42BIS <--- (!!!) User Access Verification Username: +++ OK ATI11 U.S. Robotics Sportster 56000 Voice Link Diagnostics... Modulation x2/V.34 Carrier Freq (Hz) None/1920 Symbol Rate 8000/3200 Trellis Code None/64S-4D Nonlinear Encoding None/ON Precoding None/ON Shaping ON/ON Preemphasis (-dB) 7/2 Recv/Xmit Level (-dBm) 25/10 Near Echo Loss (dB) 14 Far Echo Loss (dB) 0 Carrier Offset (Hz) NONE Round Trip Delay (msec) 6 Timing Offset (ppm) 916 SNR (dB) 61.74 Status : 0001,0011,0000,0000,0000,0000,0000,0000 ------------------------------------------------ I wish the Rockwell chipset had an easy reading and informative screen as the X2's. The ascend page continues: "if you've got noise and/or low line level, you need to contact your telco provider. Explain to them that you are using a modem on your line, you are getting poor cct quality figures from your modem, and that the line level is low. You are entitled to have this problem corrected. (...) Ask to have them turn the AGC (automatic gain control) turned OFF and your line setting at the exchange set to position "5". This should give you a good cct and level and cure your connect problems at a stroke". I cannot imagine the faces of the local wire-pullers if I call for repair and start talking about dbm or "flex" version levels. (Local phone service is only guaranteed for VOICE, and the phone co. has no obligation to provide data-quality lines, AND they have a monopoly so basically I'm f*cked if I expect them to "data grade" my line). So Rockwell, Motorola and Co. better make K56flex handshake on this line quality in the same way X2 *works* now. In fact, I even took my Flex modem to a friend's house downtown, and it failed to connected at more than 33600, with a %L of -027 dbm, yet the X2 connected at 50k. Unless the whole country's phone network is rewired this won't change imho. I can't believe all major ISPs in the USA have selected Flex with the current BUGGY state of this technology. Chapter 7: Maybe it's my ISP's fault? ------------------------------------- I've read a lot by now about all versions of "flex". How "0.5anything" is not flex really but rockwell's pre-flex technology. But what about my ISP's routers???. Maybe they need a firmware update to their flex specs? Here's the output of their routers: ** Bienvenido al Servicio Internet Impsat ** Username: os2user Password: Impsat_E2_56K>show revision ASCEND IMPSAT system revision: ek.m40 5.0Ap16 Impsat_E2_56K>ppp Chapter 8: The Phone Network, unanswered questions: -------------------------------------------------- Are there any fundamental differences between x2 and K56flex to make one work and the other to fail on the same line?. Is there any way I can improve the dbm figures from my end? (If there are any gizmos to correct the line quality that I could purchase or build, I can try). "turn off Automatic Gain Control"... how do I translate this for a level-1 tech support people to understand?. Are there any added services (conference calling, call waiting, etc) that automatically disable AGC?. Chapter 9: USR vs K56flex: the big question ------------------------------------------- Why suddenly I must be a rocket scientist to get two modems to connect, while X2 works right out of the box?. I know, I know x2 and k56flex are not modulation techniques but rather complicate approachs to use the phone network, but x2 works and flex doesn't. Someone reported on the newsgroups: |> we could not get our Motorola ModemSURFR or our |> Supra Express modems to connect at anything above 33.6K speeds, |> even after applying the 1.X upgrades (visit www.56k.com). |> |> Then we simply changed the init string to &FS202=32 and now we |> consistantly get 52K and sometimes 54K connections. I have yet |> to see a connection rate of less than 52K, and the worst it has |> ever renegotiated to is 50K. Downloads are very smooth and |> very quick. I am even seeing renegotiations back UP to 52 |> or 54. I've tried this &FS202=32 and nothing changed. (I don't know what is it supposed to do since the fu**ing thing doesn't even include a reference card with the modem's commands and registers, in fact you have to ORDER the printed documentation if you want it. The included docs in the cd (adobe acrobat format) does not list "&FS" commands, only s-registers). Chapter 10: Moral of the story? ------------------------------ There should be a class action suit against all the K56flex consortium for screwing up so many people and selling non-functional beta products that should be back to the drawing board because they simply not work yet in the real world. ALL THESE MODEMS (including the useless, brain-damaged Hayes Accuras with crappy place-holder dummy datapump) HAD NO INDICATION AT ALL OF THE CRAP INSIDE OR POSSIBLE SURPRISES, but bright stickers with "56k!" and glamorous promises. I dunno why this reminds me of women with the 'wonder bra'... ;) Chapter 11: Here's where I'm gonna be soon with companies like Hayes, Motorola, et. al. ----------- THY END...? ------------------------------------------------------------------ Fernando Cassia Author of the 'OS/2 Warp and (HP) CD Writers F.A.Q.' http://os2warp-cdr.home.ml.org