Mike wrote:
....a Elba Corp USB-TTL. converter.

I think I will ahe to go that route. I would rather the convenience of the electronics INSIDE a cable.

I will have to check around. I assume that my cable is a no-name chinese version re-packaged by whomever I got it from.

I will check around for something that uses the FT232R chip specifically, AND is conveniently packaged.

I will also try to run the USB serial port at something less than 115Kbaud. Unfortunately, this would require modifying the bootloader to use a specified lower speed, or would require an auto-speed testing/sensing section of code. Fortunately there is room for a rather large piece of code left in the currently reserved bootloader space.

-Tony

I bought a converter locally so I could return it if it didn't work. Luckily it did.

I have thought about adding a auto-baud rate detector to the Oak Micros bootloader but for simplicity decided not to.

Mike wrote:
I have thought about adding a auto-baud rate detector to the Oak Micros bootloader but for simplicity decided not to.

It is a bit of a pain in the neck to write, but while 115Kbaud seems pretty slow, it is rather fast in a noisy environment, and a slower speed, even 57.6K is considerably more reliable.

Last night I thought I had nuked the bootloader in the OM128 somehow with the failure of the USB downloads. I could not even get the serial port to download! Nothing was working. I was trying to get my ISP programmer (AVR Dragon)to recognize the OM128, and it was not working. I realized that 1) I need to throw the RS232 switch in the other direction, and 2)that I need to use pins 19 and 20, not 20 and 21, I was still not getting a connection to the OM128. But even then I was not getting communication. It turned out that I had a bad connection between the OM128 and the 2 headers included with the OM128. I had not gotten the pins to seat fully in the headers. I removed the headers and just plugged the OM128 into the breadboard. Then the serial connection worked and I did not pursue the ISP programming.

Maybe I should write a custom version of the bootloader which uses some SLOW communication speeds, e.g. 14.4K. However, I will be on vacation (in Austin, TX of all places!) for the next week starting on Saturday, so I will not have much time to do more work on programs (other than writing them in a simulator) during that time.

I decided today to get a usb->serial converter working if it killed me.

I had gotten a new FTDI EVAL232R adapter. This is a fully implemented FT232RL device. If ANY converter is going to work, this one should.

As you may recall, I had a different FTDI-based converter cable which would not work properly, although it almost worked.

I tried my EVAL232R and it totally failed to work!!! I pulled it out and tossed it onto my shelf in disgust.

I went to the FTDI website to see what I was doing wrong. The answer is either nothing or everything.

I learned about latency in one of their AppNotes (AN232B-04) which I found on this page: http://www.ftdichip.com/Documents/AppNotes.htm

I learned about changing the advanced settings in the serial port control in the System hardware manager. The default latency is 16 milliseconds. This is almost certainly too long for the OM128 bootloader. I changed it to 1mS and then tested the EVAL232R.

It STILL did not work. Not at all. But then I realized I needed to set the proper COM port. When I did that, it worked. Perfectly. Did I have the right COM port selected when it failed the first time I tired it? I am not sure. I went back and reset the port settings back to the 16mS default. The converter STILL worked fine.

I then installed the cable converter that did not really work properly, and tried it again. It again failed. I set the letency timer back to 1mS and it this one works too!

If this had not worked, or if I have glitches in the future, I will try to adjust the "USB Transfer Size" (also in the advanced section of the port settings). It defaults to a relatively large 4096K. I would set it down to the minimum of 64 bytes and see what happens there.


I am not sure if NON-FTDI converters have the same settings, but if they do, then try it.

-Tony

I tested both my no-name clone and the FT232R-based Elba adapter and both work fine. I checked my settings and the latency is 16ms so I'm not sure why you have this problem.

BTW I also used the FTDI MPROG utility and discovered that the no-name adapter was not recognized. Not a surprise of course.

The FT232R-based adapter could read the serial output from the OM128 no problem, but it could not get the timing down well enough with the 16ms latency to have a successful download. I am not sure why.

But when I changed the latency it worked. I also reduced the buffer sizes from 4096 to 64.

The other FTDI-based adapter cable NEEDS these changes, but then works well too.

I checked the device info with one of the FTDI utilities. I got bcdDeviceID as 0x400 for the cable. The FT232R has an ID of 0x600, as does my rally computer's FT232RL. Interestingly, this was the FIRST time I ever plugged-in the usb in the rally computer!

So, it appears that the adapter cable in an FTDI chip, but not the FT232R.

-Tony

EDIT:
addendum:

I tried MPROG and it found and read the EVAL232R, but it could not read the FTDI cable. I think it uses an older chip. I can not tell for sure which one. I think it does not support the CDM software, but requires the older D2XX software. I removed that software today when I updated to the newest stuff.

Anyway, the bottom line is that I got both of the devices to work with some simple changes.

-Tony