[ESjaavik]

The reason I want a step and dir output is that I want the contraption to be my handwheel when the PC is not there and I want to use the mill in manual mode. I just want to feed the steppers with pulses from my "handwheels" to make it a non-CNC mill in "fly-by-wire" mode. I don't have handwheels anymore. If I had, the axes would creep when milling because of the high efficiency of the ballscrews, it must be kept in place by active motors.

I also want the AVR to send continous pulses on command (button) to regain the auto-feed I had to remove because it was incompatible with the ballscrews.

Finally I would like to be able to send "cheater pulses" to reposition after toolchange. I've seen in the forums that this could be a bit inconvenient using some control programs.

Greg: If I run a resistor from the output back to the +input, that would give hysteresis, not reduced gain? I think that's what Mariss say in #16. If I run it back to the -input, it will reduce gain. Am I right?

If your code is in C, it would be very interesting. If its PIC assembly, I'm afraid it will be about as clear to me that this thread is to HuFlungDung. :-) PIC assembly is one reason I use AVR.

[Mariss Freimanis]

You can build an X4 quadrature to step/dir decoder using a single quad XNOR gate package, 2 resistors and 2 capacitors. It will set you back less than $0.50 USD. If someone tells me where and how, I'll post a .pdf of the schematic.

Mariss

[Greg Fill]

Yes I messed up it does provide hysterisis. In my defense it was late at night :-), but I feel hysterisis is not necessary with the slow opamp. What we need to do is to ensure is when we turn the motor it is turned hard enough so that it produces a good voltage, and when a good pulse is not made it is ignored and this is where our micros come in to play.

As for my code it is "C" code so not to worry, and it is readable. I don't have any Atmel equipment and I personally have everything for PIC's so I always use PIC's. I also probably have every flavour of the 16XXX family of PIC there is in my parts drawers so if one part doesn't do the job I have something else that will. I like all your ideas for the handwheel and it seems to outgrown my PIC12C508. If I wanted to implement everything you are going to do with the AVR I would need to use a different part now.

[IJ.]

Originally posted by HuFlungDung
I understand some of the words they're using in this thread

Hu: LOL so far your post is the ONLY one I understand !

[Greg Fill]

Mariss instead of writing your reply in the post reply box use the post reply button on the last message to respond and attach the file through the enhanced menu. A cross coupled SET - RESET FLIP FLOP by any chance? I wonder how it will respond to the poor pulses.

I.J Just wait until I start asking all my questions....ACME screw 1 start, 2 start , 4 start what is better and so on. I think I will be the one getting confused. I really have to stop designing the electronics for my machine and get on with the mechanics but electronics is fun as well.

[Mariss Freimanis]

OK, let's see if this works. My circuit is a little more bizzare than a flip-flop. It counts every edge (X4 decode).

Mariss

[Greg Fill]

Thanks Mariss, I finally got around to your diagram. Yes I see what you are doing. It will work with an optical sensor but not the stepper motor encoder. Unlike an optical sensor once you stop turning the motor the quadrature outputs will both go back high (or low depending on how you have wired the circuit). When you stop turning an optical sensor the quadrature outputs will stay high or low depending on the wheel position. The 4X decoding is not necessary for the motor encoder as I actually slow the decoding down depending on how fast you turn the motor. The best solution for the motor encoder is a simple micro. Noisy motor pulses in is resolved to STEP/DIR and JOG A and JOG B outputs.

Greg