Minimum timing between encoder pulse edges for acorn?

[DannyB]

What is the minimum edge separation time (in μs) the encoder input on the acorn can handle?

I'm looking at RLS's ROLIN encoders, and they require knowing the minimum edge separation time.
(presumably because they build lowpass filter into the thing).

In particular, i only care to know: Is it <= 0.07us, <=0.13us, <=0.5us.

Reading the beaglebone green specs, it looks like it actually could do <=0.07us depending on how it was programmed (IE with a PRU wired GPIO), but i have no idea how the acorn is set up.

[ScotY]

How much do those encoders cost?

[DannyB]

Only a couple hundred bucks total (for the accuracy and options you get, that's a good price).

The rings are ~100, the RoLIN encoders also ~100. The straight linear encoders are closer to 200 (but work on both rings and magnetic scales)
You have to understand the technical/electrical enough to know what to get, but good value for the money IMHO.

If you have a properly magnetized ring (they happily give you the specs), etc, you can save half. I've never bothered.

I've used them on spindles before, and i've used their linear encoders on saw fences. The magnetic straight track is about 100 bucks for 4 feet at +-10um accuracy, and half that for +-40um accuracy. 8 foot is exactly double the price (so pricing seems reasonable), i've never priced more than that.

[tblough]

Minimum timing for the encoders depends on the maximum Velocity. If you are using these for spindle encoders, they have a PDF file that gives max RPM for the different bit widths and ring pole counts.

https://www.rls.si/eng/fileuploader/dow ... les_01.pdf

The datasheets for the different read heads have max velocity tables when used as linear encoders.

Either way, you will be limited by the encoder rating. For Acorn, Oak, and ALLIN1DC, the encoder input bandwidth is far greater than the max encoder output bandwidth.

[ScotY]

Thanks for the info, Danny. I think I'm going to pick up a Tormach 440 and will probably want to convert it to run off an Acorn but the problem is the spindle encoder. It's a 10k rpm spindle so one of those ring type encoders might be easier to implement, assuming it can process the info fast enough? I'm clueless so if you figure out what you need to know, it would be great if you could post some more info on how you end up using one of these encoders.

[DannyB]

Hey Tom, i understand the minimum encoding timing i can use depends on the pulse count/velocity i want.

But it also depends on what acorn's encoder input can handle, and that's the variable i don't know.
Plenty of high speed input counters can't handle better than 0.5us pulse separation.
That's a 2-3mhz input full bore. It could also be a 100khz input with really fast pulses and a lot of idle time.

(that's why they define it in edge separation time for something like this - the spindle may not be moving at max velocity, but the pulse separation will still be the same whether it's going 1rpm or 10000rpm)

0.07us is a 15mhz signal. I don't believe the GPIO on the beaglebone green can handle that (or at least, having used them, nothing suggests they can).
The best i've ever seen them handle is 12.5mhz with the PRU handling the code.
(the FPGA's on the allinonedc and oak are a different beast, i could believe they could handle faster signals, they are often built to handle full clock GPIO, and so may be able to handle a 400mhz signal no issue)

Thus, without more, i don't believe your statement "Either way, you will be limited by the encoder rating. For Acorn, Oak, and ALLIN1DC, the encoder input bandwidth is far greater than the max encoder output bandwidth."

This isn't also about bandwidth but latency and pulse timing . As above, the edge separation timing does not change with speed. That only changes the number of pulses output. I totally believe it can handle most PPR i can reasonably throw at it, but it doesn't matter if those pulses are 0.07us and it sees those as noise.

I'd rather like confirmation it can handle very fast pulse timing before i spend the money on the encoder.

[eng199]

The ACORN can take an extremely high encoder frequency. However, there may be filtering added in the future to reduce the frequency to 1 - 5MHz. To maintain future compatibility, it would be best to consider the maximum frequency 1MHz.

The best bet is to follow tblough's advice. Frequencies should be kept as low as possible for best noise immunity and lowest emission. The multiple choices are probably made available so that slower slew rate drivers can be used where higher frequency is not required.

It could also be a 100khz input with really fast pulses and a lot of idle time.

(that's why they define it in edge separation time for something like this - the spindle may not be moving at max velocity, but the pulse separation will still be the same whether it's going 1rpm or 10000rpm)

This is incorrect. Quadrature frequency or edge separation is a function of encoder speed. If your A or B channel is not 50% duty cycle at fixed speed, the encoder is bad.
Here is a quick overview of quadrature encoder signals:
https://www.dynapar.com/technology/enco ... e_encoder/

[tblough]

I getting ready to use an RLS encoder on a Bridgeport conversion I'm doing. Because of where I'm fitting it, I'll be using the RLC2IC bare board with a MR050C 72 pole axial ring.

Max speed on my spindle will be 5000rpm and I want the widest pulse I can get at that speed with at least 1000ppr/4000cpr. I ended up going with 6-bit interpolation (64 interpolated steps * 72 poles = 4608cpr/1152ppr) and a min pulse width of 1us/1MHz max frequency. That combination is good for up to 6073 rpm.

[ScotY]

Hi Tom,
Can you post some pics of your encoder and how it will be installed? I have a Tormach 440 on order and would like to somehow be able to run it on an Acorn with an encoder. I'm looking for ideas on if/how it can be done. The spindle is rated at 10k rpm.
Thanks, Scot

[tblough]

Scot,

I don't want to hijack Danny's thread so check here: viewtopic.php?f=20&t=1944
The encoders should work on your Tormach, but the method I'm using certainly wont.