Grey code encoder & Differential input driver

[DWIM]

Hello
The KFlop board looks impressive. i would be happy to have some starting help. I'm planning a closed loop setup.

My Encoders have a grey code output (10Bits, Open Collector). So I need for each stepper 10Bits/Pins. The advantage is, the position is clear at startup.
My Motor driver has differential STEP/DIR/HALF STEP/WINDINGS OFF/CURRENT CUTBACK inputs.

So, what boards do I need for that? Do I need an analog board for the drivers differential inputs (KFLOP outputs)? From reading the documents one can mirror pins and some have different function when board "X" is installed. I think about the analog board which has a large connector and might block some IO's. Does it add inputs or does it limit the digital IO's?
Before going too much into C-Code I would like to ask how many "grey-code/Diff-Input" stepper axis can I run with a KFLOP board?

If somebody can give a rough estimate I would really appreciate that.
Thanks
Ben

[TomKerekes]

Hi Ben,

Do you have detailed specifications on the encoders and Drivers?

Do you know what the maximum encoder rate will be?

How many axes?

Can the encoders be wire-or'ed together and enabled separately?

[DWIM]

Thanks for the reply,

>Do you have detailed specifications on the encoders and Drivers?
I attached some images. The encoder is a KOYO TRD-3A1024-2610. So it has 1024 positions per 360°.
I have no manual,I assume the outputs are Open Collector type.

The driver is from Vexta (CSD5807N-P)
http://site.motadistribution.com/manual ... manual.pdf

>How many axes?
The goal is 5 at the moment.

>Can the encoders be wire-or'ed together and enabled separately?
The encoders have only 24V VCC and 10Bit outputs (total 12 wires, each axis)

>Do you know what the maximum encoder rate will be?
To be honest, no. I expect stellar speed thanks to KFLOP In reality, the driver will probably be the limit.

[DWIM]

Ok, I just studied more...
It seems not to be a clever idea to combine a 1024 pulse encoder with a 5-phase motor (500 steps per rev) because there is not a single point where they match. It is 0.35156° per step for the encoder and 0.72° for the stepper. Am I completely wrong? They sell them as industrial surplus on ebay. It seems someone paid once big money for these..?
The control loop would probably constantly/desperatly try to reach its target....
I probably ditch the 1024 encoder and go with another one with a decimal number of steps.

[TomKerekes]

Hi Ben,

Ok, I just studied more...
It seems not to be a clever idea to combine a 1024 pulse encoder with a 5-phase motor (500 steps per rev) because there is not a single point where they match. It is 0.35156° per step for the encoder and 0.72° for the stepper. Am I completely wrong? They sell them as industrial surplus on ebay. It seems someone paid once big money for these..?
The control loop would probably constantly/desperatly try to reach its target....
I probably ditch the 1024 encoder and go with another one with a decimal number of steps.

KFLOP can scale the encoder position so the resolutions don't need to match.

5-phase steppers have mostly been replaced over the years with 2-phase steppers and microstepping. For example with a 200 full step stepper and KStep's 16X microstepping there are 3200 microsteps per rev.

You may want to consider whether you really need absolute encoders. With 5 of those encoders there will be 50 input bits. One approach might be to connect them all in parallel to 10 inputs with pull up resistors. Then switch on power to them one at a time to read them one at a time using the same 10 inputs. I'm not sure how fast that could be performed. Maybe only do that on power up then treat them as incremental thereafter. The bottom 2 bits of an absolute grey scale encoder is the same as AB quadrature of an incremental encoder.

[DWIM]

Hi Tom

I do not understand why it is irrelevant to have matching drivers. There is no microstepping on these drivers. So be it, your surely know.

Thanks for the tip with the lower 4 bits! This makes it easier on the pin count.
So, the 5 axis are easy with KFLOP.

Last question: The drivers have differential inputs. What card combination do I need to have differential outputs on KFLOP?

Thanks for your help.
Alesch

[TomKerekes]

Hi Alesch,

I do not understand why it is irrelevant to have matching drivers

I don't understand what you mean. But the encoder can be scaled by any factor. So whatever the resolution is it can be scaled so that if for example you move the motor 1000 steps the encoder can read 1000.

Thanks for the tip with the lower 4 bits! This makes it easier on the pin count.

I think you meant to say 2 bits?

The drivers have differential inputs. What card combination do I need to have differential outputs on KFLOP?

KFLOP (or any of our option boards) do not have differential outputs. The drive's inputs are basically an LED so you probably don't need differential drivers. KFLOP's 5V open collector mode would drive them.

But unfortunately the Drives Specify CW/CCW pulses of 5us minimum. KFLOP's pulses are a maximum of 3.78us. So they are incompatible and probably won't work reliably. The only way to make them compatible would be to add an external pulse stretching circuit.

You might read this.

[DWIM]

Hi Tom,

Bits:
Of course, I meant 2 bits. Yes.

Accuracy, a sample:
Imagine an encoder with only three(!) points. At 0°, 120° and 240°.
Further we take a stepper with only 4(!) steps. At 0°, 90°,180°,270°
So, this seem for me not accurate. The encoders result and the steppers postion never match!
It helps a bit with 200 Steps and 1024 points, because the differences get smaller, but it is not accurate.

Regarding the timing:
This is a bummer. I can't believe that there is no solution.
I just need an encoder to Step/Dir converter. There are thousand of Step/Dir configurations out there which one could "update" with an encoder. Maybe it is not available as a product because it is too simple?