Bug Bytes July 1994
by John Steiner
Written in April 1994

Amiga Modem Connection Problem
Greg Suire sent EMail describing a problem in connecting a Zoom 14.4 KB Fax/Modem to his Amiga 1000. He is using the same cable that previously worked just fine with his 1200 baud modem. When he turns on the CPU, the MR light on the modem begins to blink. He wonders if pin 21 on the Amiga side might have some incompatibility with the modem. RTS/CTS was set in preferences and in software, and the proper initialization string is being sent. Any suggestions from those who have run into a similar problem? Drop a line, and well pass them along.

Ramworks and Workbench 2.0 Revisited
Brett McDonald sent CompuServe EMail regarding Clete Bakers question from the April 1994 Bug Bytes. He writes, <i>  I am successfully running the Applied Engineering RAMworks 2000 RAM Card under Kickstart/Workbench 2.1 (and formerly 2.0).  I have had experience with the problem that Clete Baker of Omaha, NE is describing.  The problem isn't exactly a conflict between the RAMworks card and Kickstart/Workbench 2.1.  More precisely, it seems the problem is flaky DRAM chips on the RAMworks that the auto-memory configuration of Workbench 2.0+ has problems with.  The card and Kickstart, in trying to autoconfigure, start to get conflicting indications of how much memory is on the Card - the Memory Configuration LED's start to flash in alternating sequences - and the system attempts to restart after failing part of its POST routine.  The solution is to move the memory chips around on the card (or remove a bank and shift the rest), re-seating them.  Also check and see if there are mixed speeds of DRAM (i.e.. some 80 ns, some 70 ns, etc.) installed.  If so, they must be installed with the slowest speed chips installed first at the front end of the card.  Any 100 ns or 120 ns DRAM chips in the card may have worked fine under Workbench 1.3, but Workbench 2.1 seems to have problems with them. I tried re-installing the 2 MB of 100 ns chips that I removed previously, and now the card works fine when I have 80 ns and 70 ns DRAM installed (256Kx4's), but if I install the 100 ns chips, very shortly after boot-up the system crashes (if it boots at all) and the RAM Cards LEDs start flashing.

  The 100 ns DRAM is the memory that my card came configured with from Applied Engineering, so if Clete bought the card fully loaded (8 MB) and they are all 100 ns chips, it could be an expensive proposition getting the card working again.  But then again, its one of the few cards that you can expand in 512K chunks (last week I was running mine with 1.5 MB on the card with no problem, but now back up to 3 MB installed on the card (yep, not 2 MB, not 4 MB, but honest to goodness 3 MB, autoconfigured!).<i>

Kickstart 2.05 ROMs and Early A500s
Brett McDonald also sent along a note that might be of value to readers with older A500s who are considering upgrading to Kickstart 2.05. Brett writes, <i>Kickstart 2.05 ROM's often don't work in older Amiga Machines.  There was a technical note from Commodore which indicates that Kickstart 2.05 ROM's are incompatible with early model A500s (particularly revision 5 and earlier motherboards).  Hence people have to watch just which Kickstart ROM they are getting.  The 2.04 ROM is still available (I know, having just received a 2.04 ROM and a 2.1 upgrade kit a month ago after Commodore Canada had been out of stock for the previous 6 months, and yet getting a 2.05 ROM out of a 2.0 upgrade kit that another supplier just happened to have in stock).  I ended up with a used A500, revision 5, and I COULD NOT keep it running consistently with the 2.05 ROM, yet it runs fine under 1.3, and that 2.05 ROM worked great in my A2000.<i>

 Brett also writes regarding another 2.05 ROM experience.  <i>I was setting up for a null modem link on my A3000-25/100 with a Hewlett-Packard 9000/360 workstation (I have successfully accomplished null modem hookups many times before with many different machines). When I plugged the null modem cable into the A3000, I heard a pop and later found out that my entire system was fried (except for the hard disk and the high density floppy). The repair man said that one of the pins carrying power must have shorted with something. I am a little skeptical about the diagnosis because I would have really had to have been off to make that happen (I wasn't). <i> You wouldnt be the first one  this happened to, however, the problem may not be as sever as you describe. The A3000 has a fuse connected to one of the pins on the serial port. (Its very small and soldered in, so dont expect to find it or replace it easily. They meant the fuse to protect the motherboard should the metal ground shield on an expansion cable come in contact with the power pin on the Amiga serial port.) The cost of replacement isnt inexpensive, either, which underscores the fact that one should always shut off the Amiga before plugging anything into its expansion ports. Mr. Johnson continued, <i>That incident left me with the only other Amiga in the house (A2000, 1 MB of RAM, 2 floppies). I was able to get a 2091 SCSI card for my hard disk, but could not use it until I upgraded the 1.2 Kickstart to 2.05.
 The 2.05 ROM allowed the 2091 to autoboot my hard disk. Right now the 2.04 OS from the dead A3000 is running on it.
 The guru problems began after the insertion of the 2.05 ROM. They are random and hard to predict. At first I though it had to do with my 1 MB of RAM (the OS would take up all but 380k), but I knew there was enough RAM for the requested task when the computer would bomb out. Here is a list of all the gurus I have kept track of:
Word Perfect 4.1.9: 8000 0004 Task: 00C0670A
Disk Master 2.0: 8000 0025 Task: 00C6B090
Quarterback 4.3: 8000 0003 Task: 00C0BA38
Quarterback 4.3: 8000 0003 Task: 00C0D1F4
Amiga Monitor (PD): 8000 0003 Task: 00C62AF8
Disk Master 2.0: 8000 0004 Task: 00C0670A
JR Comm 1.02a: 8000 0008 Task: 00C0670A 
 I added 2 MB of fast RAM to the 2091 card. After that I told Quarterback Tools 1.52 to do a routine scan my Work: partition (80 MB). After a few minutes it came up with a new guru: 
Quarterback Tools 1.52: 8000 000B Task: 0020B750
Another unrelated one happened not too long after:
Rebooting: 8000 0003 Task: 00206422 I have no idea why, but it seems like a lot of numbers/variables/listings tend to lock my Amiga.
 I hope this information will help you track down this major annoyance. If you find out any answers, please let me know. <i> Any comments or suggestions?

Recolor Fixes Icon Problems
Maxx Daymon sends EMail with a response to Jim Shaffers Icon question in the March 1994 Bug Bytes. He writes, 
<i>The reason Mr. Shaffer is having problems is because some of those older  icons actually HAVE data in the higher bitplanes (thus, a "FixIcon"  program won't fix it - It doesn't see anything wrong!) Some of the older  icon editors apparently wrote garbage to these areas which the computer  (able to open higher bitplane Workbench screens) thinks is part of the  icon image.

 What he needs is "Recolor" from Fred Fish disk #490. Recolor was intended to swap the colors of icons to/from 1.3 & 2.0. The Recolor ALSO truncates bitplanes. 

 Get Recolor. Click on the icon and select "Information...". Now, click on "New" and type in "TRUNCATE=2" (will truncate all but the first two bitplanes). Make sure you don't use this on 8 or 16 color icons!<i>

 Maxx also wrote with several comments regarding various items listed in the April 1994 Bug Bytes.

 Workbench 2.1 Gurus 
<i>There are a few possibilities regarding Mr. Myers Software Failures.  Since he has had the crash on bootup - 2.1 probably isn't the culprit,  not directly anyway. 2.04/2.05 are the only possible ROMs in this  situation, since 2.1 is a disk based update and has no ROM counterpart.  There are a few things that might be happening:

 There could be a virus (as yet undetected by virus checkers) that causes a  software failure, rather than triggering because of OS differences. (This  would explain the failure on boot-up)

 The 2.x versions of Kickstart access some of the chips on the motherboard  differently, and faults will show up under 2.x more readily than under  1.3. The read/write errors could be due to a faulty CIA - 2.x accesses  the CIA chips differently, and by default speeds up the step rate of  drives. Older drives cannot handle this increased step rate and fail.

 The errors: 
8000 0003 (Address Error: Word access on odd byte boundary)             
8000 0004 (CPU Error: Illegal instruction)

 The 2.1 Setpatch also has the drive fix (Patch for 1010 drives) You can  type "Setpatch" at any time to get a list of installed patches. A3000  users might try replacing U202 on the motherboard. The 2.0 setpatch  should not be used in an OS 2.1 environment.

 It may be software (like a commodity) that is running in the background  that is causing these problems. Also, the errors are somewhat general and  can be caused by different things. 2.x uses a different numbering system  for errors which leads to some confusion.<i>

 Vertical Pin Striping Problem on A4000/030 
<i>The problem comes from crosstalk between two of the traces leading to the  video connector. It generally only shows up in 31 kHz modes. Open up your  silver monitor adapter (DB-23 to DBH-15) and solder a 100-ohm resistor  between pin 15 (of the 23-pin connector) and ground. This should  eliminate the banding.<i>

 Temperature problems with A500
 <i>Many of the chips in the A500 are socketed. Certain temperatures can  cause these sockets to lose or have bad contacts with the chips - while  warmer temperatures can cause the sockets to be more forgiving.  Re-seating his chips, or replacing the sockets with a higher quality  style might fix his problems for good. Condensation is also a concern in  some geographic areas.<i> Our local Amiga service center technician also commented that its quite common with the A500 to develop problems that are fixed by simply taking the cover off, and pressing down firmly on all socketed chips.

 Toolmanager 2.0 and Quarterback  
<i>PROGDIR: is a temporary assignment made by AmigaDOS when you double click on a program icon. PROGDIR: is the path where the program was launched.  In my experience working with Quarterback, you must insert a high density disk BEFORE Quarterback locks the drive. Quarterback locks the drive in whatever mode it happens to be in (determined by the disk currently in the drive, in most cases) and does not accept another density until the drive is unlocked. Mr. Hyatt should make sure he is using the WorkbenchStarter with  Toolmanager to insure that Quarterback's PROGDIR: is correctly assigned.<i>  Thanks to Maxx for all of his comments.

 GVPs I/O Extender Fix Corwin Wong of Toronto, Ontario sent an EMail letter regarding Barlow Sopers GVP I/O Extender switch question mentioned in the March 1994 Bug Bytes.

 He writes, <i>Users who have GVP's I/O Extender no longer have to cry about not being able to use GVP's card for their printer. I would also like to say that the user is now able to use their printer and ParNet at the same time. Something that everyone has been looking forward to, especially us GVP I/O Extender owners. 

         How I accomplished this was with the use of a program called ParPrefsV1.0 off a local BBS. The author is Benoit Mortier. What this program does is allow the user to output their printouts to any device, other than the Amiga's internal parallel or serial port.

         The main feature that I like ParPrefs is due to the fact that GVP's redirector program does not work properly. With ParPrefs Im also able to use ParNet at the same time my printer is spilling out my homework. ParPrefs requires you to have at least one MB of RAM and Workbench2.x to run. <i>

 Quarterback 6.0.1 Supports Supra Turbo 28 Terry M. (No last name provided) sent EMail regarding the release of a minor upgrade to Quarterback that fixed a compatibility problem with Quarterback 6.0 and the Supra Turbo 28 accelerator. He writes, <i>I have an A2000 which I recently upgraded with a Supra Turbo 28. When Quarterback 6.0 came out, I immediately upgraded because of the promised faster compression times.  To my dismay, the use of the compression feature sent me to the GURU!  I wrote to Central Coast Software detailing my problem and system configuration.  I received a disk with Quarterback 6.0.1.  The shipping notice stated it was to correct problems ,like mine, with the Supra Turbo and compression.  To my delight, it does work and compression is much faster than 5.0.<i>

 Vortex GG486SLC and CrossDOS
 Doug N. sent EMail with a question for Bug Bytes readers. He writes, <i> I use the Vortex GG486SLC. I have an MS-DOS partition set up for the board. I would like to directly read this partition from the Amiga side using CrossDOS. I have posed this question to both Vortex and Consultron. I got suggestions, but never a working answer. Has anyone been able to accomplish this? There is a Mountlist-like file used by the GG, but if I mount a CrossDOS partition using its information, CrossDOS still cannot correctly read the partition. Before owning the GG, I used the software PC emulator CrossPC. I was able to read an MS-DOS formatted partition with no problem. Incidentally, I'm using version 5.06 of CrossDOS.<i> How about it readers, any suggestions? 

X-CAD Support 
Bob Miller-Rhees of Bainbridge Island, WA  sent a large packet of copies of correspondence with Digital Multimedia Services in England. Bob writes a newsletter he publishes for CAD on the Amiga, and notes that he and a couple of his newsletter readers are dissatisfied with the service provided by that programs distributor. Basically, he notes that they agreed to provide registered X-CAD3D users with a copy of X-CAD3000 for an upgrade fee, which was never done. The distributor then agreed to compensate him by providing copies of Digitals Symbol Libraries, which to this date has also not been done. Space here doesnt allow me to reprint the materials he provided, but in addition to his problems with Digital Multimedia, he wanted to make readers aware that Grafx Computing of Panama, NY does a super job of supporting X-CAD. He complimented technical support persons Juan Wilson and Amy Swan on their capable support. In a follow-up letter he received from Richard Nolan, representing the developers of X-CAD 3000, they noted that the authors of the program are a completely separate entity, and are not connected with the distribution company, Digital Multimedia Services.

Jaggies In WYSIWYG Screen Displays
Robert Bennett of Jericho, NY is very happy with Final Writers feature set, but he uses an Amiga 1084S and an older Amiga (he does not say which model). He is unable to get the program to display text with good enough resolution to be satisfactorily readable. He notes<i> I realize that more and more people are buying the A1200 or A4000. Both of these have the new chip set and allow greater screen resolutions. However, software producers should be aware that there are still a great many 500s and 2000s out there. It is all right to write a program which utilizes the functions of the AGA chipset but if such a program needs that chipset and the higher capabilities those chips offer the public should be made aware of that requirement. it is the same as telling people that a program needs a hard drive or needs  DOS 2.0 in order to operate properly.<i>

OpalVision Problem With New Motherboards
Robert Iacullo writes with a question about the OpalVision board. He notes, <i>It worked fine when I first got it, but the computer has had two new motherboards since then. The Opalvision board does not work in 24 bit mode with the two new motherboards. Does anyone have a clue about this?<i>

Amiga and Ethernet
Everett Greene of Ridgecrest, CA writes <i>I am interested in seeing what responses you receive to the Ethernet item in your February Amazing Amiga column. I was recently investigating the Ethernet subject with the goal of obtaining products to allow the US Navy, China Lake to connect several Amigas to the Base-wide network. Among the non-definitive results of that investigation are: Commodore has disavowed all knowledge of Ethernet and has dropped its A2065 board and associated products. The TCP/IP update supposedly in the mill will probably never happen. Additional copies of the last release are no more.<i> I suspect that other Amiga users would like to connect their Amigas to company Ethernet LANs. Comments from those who have already done so, especially without the above discontinued products, are welcome, and will be noted here.

A3000 SCSI Problem Revisited
Mr. Greene also commented, <i>The version of the 33C93 chip may only be part of the problem. I suspected the WD chip as a possible cause of the problem in my personal A3000 so I replaced the 33C983A with a 33C93B obtained from Western Digital, no help.

Our experience at China Lake with the 33C93 is that all versions are faulty in one way or another. We had to alter the design of our in-house boxes to circumvent the defects when WD couldnt seem to get the defects corrected through several revisions and versions of the chip. Lest this seem overly critical of WD, our experience with TI and NCR chips has been that those devices also have their less than optimal design and/or operational features as well.

Expansion Systems Dataflyer and the A1000
Mr. Greene also included a bug report and workaround for the Expansion Systems Dataflyer. He writes, <i>the hardware works fine, however, the programs which worked correctly on an A3000 got spectacular flameouts on the A1000. The failures were deduced to be an apparent bug in the SCSI driver supplied by Expansion Systems. The driver seems to completely mishandle the auto-sense option of SCSI Direct calls. The bug can be circumvented by not using the auto-sense option and placing the burden of handling the check condition on the application program. <i>

Problem with Magtape command
On a similar issue, Mr. Greene reported, <i>Ive encountered a near bug with AmigaDOS 2.Xs Magtape command. Units which report a check condition for the first command after a hard reset (per the ANSI standard) are reported to be not ready by magtape. This is no big deal if youre familiar with the standard and realize that the purported error really isnt. However, it can be quite disconcerting to the uninitiated.

My complaint about magtapes misleading error report is that I have been unable to find a way to kill the report when invoking it from ARexx programs. Ive developed some data reduction programs for the Navy which are designed to be canned processes for use by technicians and unsophisticated computer users. In these processes, magtape is invoked to ensure the tape having been rewound from any previous use prior to retrieving data from it. The only thing I can seemingly do is verbally tell the users to ignore the error reported by magtape and continue operation. There has to be a better way.<i>

That's all for this month.  If you have any workarounds or bugs to report, or if you know of any upgrades to commercial software, you may notify me by writing to:

  John Steiner
  c/o Amazing Computing
  Box 2140
  Fall River, MA 02722

...or leave EMail to
  John Steiner on Portal
  73075,1735 on CompuServe
  Internet mail can be sent to
  John_Steiner@cup.portal.com
  FAX John Steiner at (701)280-0764
  (8:30 A.M. to 5:30 P.M. Central time, Monday-Friday)




