Roger Meier's Forums

I am seeing the following problem which I do not think is related to the 6809 SBC. The SBC is a dead simple machine, a 1MHz 6809, with a serial port controlled by an ACIA (6850) which is crystal locked running at 9600,8,N,1. No interrupts, no DMA, 52k of memory available running the very old and well known monitor PTMon (running from ROM) which is patched for the I/O address of the ACIA. Communication between my PC (Win11/Pro 24H2 build 26100.7840) and the SBC is solid using a USB to serial converter. There is no flow control of any kind, and I have added some delay at the end of the line to allow the processing at the end of a line to catch up. This is reliable except when it comes to file transfers from the PC to the SBC which crashes if too many characters are sent from a file. I can send the CoolTerm config file if that is useful. And sample S19 files as well if useful.

The symptom is this:
Send a raw file via menu -> connection, file is received correctly. Try another file and the transfer crashes, and Coolterm must be closed and opened to restore the send capability. Several smaller files can be sent, but eventually the limit is reached and the transfer crashes. Disconnect/Reconnection and Clear the data do not restore operation. CoolTerm must be closed and restarted.

Any assistance gratefully received.

Regards,

Hi,

I recommend also adding a delay after each character (i.e. "Use transmit character delay") in addition to the transmit line delay. The input buffer running on your 6809 may not be deep enough (or fast enough) to keep up with characters arriving in rapid succession. It might be fast enough when typing, but not necessarily when sending a raw file with will send the text as fast as possible unless delays are added. The ACIA 6850 can only buffer 1 character at a time, and it needs to be fetched by software before the next one arrives. If something slows the software down for any reason, then you might start losing characters.

As an example, I'm using a character delay of 30ms and a line delay of 100ms when I'm sending text to MBASIC running on my Altair simulator. Anything less than that results in characters getting lost.

Another thing to check are the line terminations in the file you are sending. Does your 6808 expect CR+LF? or just CR? or just LF? The line terminations in the file you are sending need to conform with what your target expects. I have has issues with files where the line terminations was incorrect.

I hope this helps.

Roger

Thanks for your reply and my apologies for not getting back to this sooner. Life intervenes and I needed time for further investigation to refine the problem definition.

I have since investigated this considerably further, including running Coolterm on my Ubuntu MATE 24.04LTS, also fully up to date, Linux system connected to the same Vintage 6809 Hardware.

To be clear i have always had delays inserted in the file stream using the delays in the Transmit section of the connection options Window, and without these the sending of files is unreliable consistently. I entirely expected this phenomena, directly related to not having flow control, and character/line delays is not related to the problem I am seeing. The vintage machine was intended to be connected to an ASR33 paper tape reader, not exactly known for their speed so compensation was expected for modern higher speed usage.

I will describe a scenario which covers the failure mode:

1) I open a Coolterm session and load the config file which sets 9600/1/N and a delays of 1ms and 40ms for end of character and end of line respectively. The vintage machine is connected to the PC via an Arduino style USB/Serial interface based on the FT232RL. (These devices have some reputation for having driver problems, my Windows system is Win 11, 26H2 fully patched, and the these devices work fine with my Arduino boards e.g Mightycore 1284P, and the vintage 6809 hardware, with the only exception below.)

Powering the vintage machine or resetting it results in it communicating instantly with Coolterm and the splash text appears and full communication is established.

2) The next step is to put the vintage machine into Load mode by typing L. The monitor program on the vintage machine is now expecting a string of S19 records, terminated by CR/LF, although the termination is unimportant if you look at what the load code is doing. When an S19 record completes with the checksum, the loader starts looking for an S, being the next record start character. If CR/LF arrives it serves only to format the output if echo is on. A CR/LF is not required for a successful load.

3) Next in Coolterm I navigate to Connection -> Send Files and choose an S19 formatted file and send it. And it all stops. The file is never sent. If I look at the data in HEX mode, there is nothing, no control characters, no S19 records, nothing. The vintage machine will wait forever for S19 records so I usually just reset it and again I see the splash text and communications are as expected.

I can repeat steps 2 and 3 forever with the same result. The file never sends. It is not as if it is sent and fails, it never sends.

If I now close Coolterm, re-open it, reload the config file and try steps 2 and 3 again the file downloads, is processed by the loader code in the vintage machine and the object program is stored as expected in the vintage machines memory. Once the process is working it is just fine, but the first run of the day on my Windows machine, fails to work every time. End Coolterm, and re-run it, and no problem. File size is no problem, sending an equivalent of an 8K byte binary works perfectly, once it is running.

With the same vintage computer, the same serial/USB device and the same configuration I have not yet seen this problem on the Linux computer. It just works first time every time.

The end of character/end of line time delays I am using may seem short but when it works it is totally reliable. If delays were the problem I would expect intermittent failures, and that is not what I see. The problem is the failure to send anything at the first start of Coolterm. This is a mystery, why should it do that. I realize that this problem could be hard to reproduce. If you need further information I am happy to help.

Regards,

Thanks for posting back. This is weird indeed. Does the TX LED in the bottom right corner of the CoolTerm flash when you're transmitting?

Instead of using the Connection > Send Files menu item and then navigating to your S19 file, can you try to drag and drop the S19 file into the CoolTerm window? This will also cause the file to be sent, but bypasses the file dialog. This would rule out any issues with the file dialog.

Another thing you can try is to create a loopback at your FT232RL adapter instead of connecting to your vintage hardware to see if the sent data is echoed back.

Lastly, do you have another USB-UART adapter you can try?

Meanwhile, please send me your settings and some S19 data to roger.meier@pobox.com.