Signals Output by LinuxCNC to Breakout boards

LinuxCNC doesn't do buffering. Even with the FPGA cards that generate steps independently of the main control loop they only ever make steps at a constant rate for the 1mS servo-period, then make the steps at a different commanded rate for the next period.

If you have a realtime requirement and the budget you could look at doing everything with LabVIEW RealTime (if that still exists, it's been about 15 years since I was involved in a project using that).

Depending on what you want to log in tandem with the stepper motion you might be able to use LinuxCNC for logging. It all rather depends on what time-stamp resolution you need for the data.

You can do 7500 steps per second without any special hardware at all, that is within the capability of a parallel port.

Things like the Mesa cards (5i25 for example) can output steps at 10 MHz, but can only _change_ step frequency every 1mS.

I think a more important question is what you want to log at the same time as making the step pulses.

jsimeroth wrote:

You say I can do 7500 steps per second. Is that statement based on the assumption that I am using Linux, or is that true of Windows too?

That was assuming that you were using Linux _and_ that that version of Linux was running a real-time patched kernel.

It is somewhat dependent on the PC hardware but 20uS timing jitter is about average with the RTAI Linux kernel and some machines get down to 3uS.

Mach3 runs under windows and manages to be functional for many people without recourse to a buffering breakout board, and so it seems fair to assume that it is possible to do it in Windows too. (Mach uses some rather special tricks to make this all work)

I have to say that if I wanted to send stepper pulses and read RS232 at the same time, and be sure of the time-stamp alignment, I would probably be using an Arduino Mega (you need one with multiple UARTs, one to send data back to a host PC on the serial-to-USB connection and one to read the real-time data.

The type of "breakout" that you seem to be thinking of (Probably the Smoothstepper?) isn't a simple breakout. It is basically a complete motion controller that is smart enough to buffer a few tens of mS of instructions. I don't think this will help you at all, because the Smoothstepper wouldn't be able to read the RS232 for you to get the timestamp alignment.

jsimeroth wrote:

So, if I was using a non-buffering breakout board, what would its benefit be? It still offloads some of the processing to the dedicated hardware, right?

Very little actual processing. The advantage is that the controlling PC isn't being interrupted every 25uS to make a step pulse (if one is needed at that timestamp) and that the step pulses are made faster and more regularly than can be managed by software. The step-time jitter with software stepping is typically 5uS or so whereas an FPGA card can time steps to a few nS.

The question I am still confused about and cannot seem to find an answer to, is: what signals does a simple $30 breakout board expect? .

I don't know which $30 breakout you are talking about, so can't say. Most of the ones on eBay at that price don't do any processing at all, they just pass-through the signals from a parallel port. Things like the Smoothstepper cost $150 or so.