Easiest Interface to Use With a Windows PC.

[lisa2]

Posted on November 13, 2012 at 14:37

Hi,

I am using the MCBSTM32E evaluation board to control some motors. I want to use my PC (Windows XP) to drive/record test data. Does anyone have an opinion on which interface is easiest to set up for communicating with a Windows PC? (e.g. USART, CAN, USB, etc.)

Thanks for any info!

#weakest-link #use-a-com-port

[Tesla DeLorean]

Posted on November 13, 2012 at 16:29

Serial is pretty easy to access from a user space application. USB can be, unless you need to create a custom drivers, ie pick a method that has built in support. For example you could record data to a flash card, then access it via USB MSC.

CAN? You have an existing board/driver supporting this?

[Andrew Neil]

Posted on November 13, 2012 at 16:42

I would definitely second the use of a COM port - whether a genuine RS232 connection to the PC, or via an RS232-to-USB converter.

This keeps you clear of any involvement at all with USB at either the target or the host end.

[John F.]

Posted on November 13, 2012 at 17:27

If asynchronous serial is fast enough (115200 baud works well usually) or you can buffer the data use the serial port. For a simple terminal and data logging to file I'd recommend Tera Term Version 4.75 is current.

[Andrew Neil]

Posted on November 13, 2012 at 17:45

''If asynchronous serial is fast enough (115200 baud works well usually)''

 

 

I have successfully used FTDI USB-to-Serial adaptors for years at speeds up to 921600 baud.

[jj2]

Posted on November 13, 2012 at 17:54

And are you recommending that - as a regular, suggested practice?  Had you run proper/thorough BER (bit error rate) test - over prolonged period?

Remainder of devices w/in that connection may be less robust - especially over time, temperature, and lot-run.   Keep in mind - top speed is always limited to the ''slowest'' device w/in the connection scheme/connection. 

Suggest that 115K should be adequate - and far more practical/achievable...

[Tesla DeLorean]

Posted on November 13, 2012 at 18:06

I think it was offered as an achievable rate, given it doesn't use our Grandma's 8250 UART card with a 1.8432 MHz crystal.

Async serial, at all rates, is prone to undetectable errors or framing issues, you'd typically layer a protocol on top to address that, like X/Y-Modem and variations which are certainly viable at 460.8K or 921.6K

[Andrew Neil]

Posted on November 13, 2012 at 20:20

''I think it was offered as an achievable rate''

 

Correct.

''you'd typically layer a protocol on top to address that, like X/Y-Modem and variations which are certainly viable at 460.8K or 921.6K''

Yes, XMODEM was used.

And, of course, it does require careful design of the whole chain; eg, not all RS232 transceivers are suitable for that speed, and you wouldn't want to be using long and/or cheap cables...

[jj2]

Posted on November 13, 2012 at 22:07

Poster stated she seeks to drive/control, ''some motors'' and record key test data.  So along with the valid points you two have made wrt appropriate line drivers and cables - we must add likely noise spikes to our mix.

Thus - imho - less than, ''Full Throttle'' seems wisest choice.  (i.e. past stated 115K or slower)   

Real world has nasty habit of ''intruding'' upon our best hopes/theories...

[Andrew Neil]

Posted on November 14, 2012 at 00:01

Interesting thought:

Using USB-to-Serial, the async connection could be of minimal length - so what is USB's noise immunity like...?

[russdx]

Posted on November 14, 2012 at 01:13

may as well use the built in usb stack, using the cdc example floating around with libusb you could have data running back and forth in no time :)