Newser.device update information -------------------------------- V2.10 --- June 27, 1991 This version is very similar to V2.00. It adds a routine for controlling the RTS and DTR control lines directly. The method is compatible with that of ASDG's Dual Serial Board. See the source code for more information. A couple of potential problems were fixed as well. This will likely be the final "official" version of newser, except perhaps for minor bug fixes. HAVE FUN! Documentation for V2.00: Please read this completely, since it contains information you'll probably need to know sometime... The time has come for another update to the serial device driver for the I/O expansion board. Previous versions basically worked, but had various shortcomings. This, the third release, tries to finally do everything right...(and seems to succeed!!) What's new? ----------- * Better performance. The last version increased read performance to the max; this version addresses write performance as well. Context-switching overhead has been reduced to a minimum to make writing much faster than the previous version, and faster than the regular serial.device too. The overall performance of the driver is now extremely satisfying! * CTS handshaking now works. This is a major fix. Previous versions would mysteriously hang if CTS was ever deasserted. (The serial chip documentation was inadequate...). It is now safe (and recommended) to turn on CTS handshaking. Actually there is no way to turn off CTS handshaking (a limitation of the serial chip) but if you specify no CTS handshaking, the driver will check the status of CTS before starting a write operation: if CTS is not asserted, the write is aborted. On the other hand, if CTS handshaking is turned on, and CTS is not asserted (e.g. the modem is turned off), the driver will wait forever for CTS to be asserted; expect this. Terminal software is supposed to provide a time-out to avoid "hanging" in this case, but I doubt the authors' of these programs bothered to do it. * The "random lock-up" bug reported by some users is fixed. * IOPatch received a small update: It now calls CacheClearU after duplicating itself, when running under kickstart 2.0. Also the gadget text was changed to be slightly more intuitive. One bug report remains an unsolved mystery. A user reported total failure when using the I/O board in a system with a GVP Series-II controller; when the controller was removed, the board functioned normally. I have checked the driver very carefully and don't believe that it could be responsible for such a drastic failure. Instead it's probably a hardware problem. First suggestion: add a pull-up resister on the reset line, as specified in the assembly manual. Tips for using newser.device ---------------------------- * Turn off XON/XOFF handshaking unless you really really really need it (you probably don't). If XON/XOFF is on, the driver can't use its "RAD_BOOGIE" mode to speed up reads; instead of doing a fast bulk copy, the driver is forced to check each byte for XON/XOFF...this is extremely slow!!!! * If you are communicating with a modem at a speed higher than the connect rate, CTS handshaking MUST be turned on for ZModem transfers to work properly. * Higher numbered serial "units" have higher interrupt priorities. If, for example, you are using two ports, one at 19,200bps and the other at 2400bps, then a good arrangement is to connect the 19,200bps line to unit 1, and the other to unit 0. The other way around will give poor results. Compatibility ------------- This version of the serial driver was tested with the following programs: * JRComm 1.02 * Handshake 2.12a/2.20c * Ncomm 1.9 * C-Net 1.95t (BBS software) Special thanks go to John Hoyt for his beta-testing and encouragement. Call his excellent C-Net BBS at 213-633-5002. Important notice to all I/O board users! ---------------------------------------- Due to the longer cable length that the moved CIA must drive, many of you are experiencing erratic floppy operation, even after grounding the I/O board as recommended. Fortunately I seem to have run into the cure! It turns out that this problem is highly dependent on the CIA revision. The CIAs that came with my A2000, labeled 8520A-1, worked absolutely horribly, BUT when I replaced it with an 8520R2 from my old A1000, the erratic floppy problem COMPLETELY disappeared even when the board had no extra ground/power attachments. If you are having this problem, trade CIAs around! Note: New Amigas are being shipped with a CIA revision that promises much better drive characteristics. However, this new version is not a perfect replacement, requiring some additional pull-up resisters. The bottom line is this. If you have the latest CIA revision, you are probably not having any problems. If you have an old CIA revision, you can't substitute the latest revision without some hacking; instead, try swapping with an old A1000 like I did. (The 1000 won't know the difference). The Future ---------- I'm still interested in a Zorro-II version of the I/O board, maybe even with a 6502 on board...(or am I dreaming?). I talked to one person who expressed interest in designing such a board, but I haven't contacted him in several months. Don't be shy; give me a call and get started!! ---------------------------------------------------------------- Since I put so much time into this upgrade (unpaid of course), I would appreciate receiving a few disks in the mail...I especially like programmers tools/utilities, megademos, and anything fun. Thank you for your support!!! My address... P.O. Box 1532 Southgate, MI 48195 USA Phone: (313)-479-1056 Internet: dxb132@psuvm.psu.edu Bix: danbabcock