Encoder math check - usable on a 7i77?

Hi, all;

If I'm using a Mesa 5i25/7i77 in mode 3, according to the manual I can get a usable encoder count rate that's something around 6mhz . I'm thinking that's per encoder, but I wanted to double check since encoders are multiplexed on these cards.

I have acquired some surplus encoders for a set of AC servo drives I'm contemplating using on one of my machines along with a 5i25/7i77 kit (I have one on my first machine and like it). They have commutation tracks, which I want/need, but they are relatively high resolution - 5000 cpr. So if my math is right, that's 20,000 pulses per revolution, and on a servo running at up to 5000 rpm (not often, but possible) that would be (20,000*5000)/60 or 1,666,667 pulses/second.

I will have up to four axes on this machine. So can the 7i77 keep up counting pulses if I have the possibility of 6.66 million per second (I know that is an edge case, but possible) coming from four encoders total plus maybe a few more simultaneous pulses from an MPG?

If it can do this (I'm 90% sure it can because it seems to me the Mesa hardware is very well designed for this sort of thing) would it still make sense to reduce the amount of incoming pulses? I won't need the resolution, and my servo drives will be processing the encoder input so I think they can scale the pulse train.

If the 7i77 can handle it without any expected issues I will probably use full resolution just because I can.

Thanks,
Erik

Andy -

Good to know. I'm pretty confident the 7i77 can handle it at this point, so I'm going to keep planning as if it can.

I got the servos surplus off of some kind of industrial robot system made in the UK... it was designed to run baskets of parts through successive ultrasonic baths without human intervention. So, I got some nice rack gears and pinions off of it too. The electronics were custom, with a funny (odd voltage) hall setup in the back cap of the servo (SEM HD92C4-64T, I think) and some limit switches that provided all the necessary feedback.

I have the electronics too.. I was hoping I'd open the boxes and find off the shelf servo drives, but nope.. they have some kind of custom IGBT based system integrated with their microprocessor electronics in a custom case.

So, to use these on a mill I'll be adding an encoder to the unused mounting location in the back cap... I might be able to use the weird hall sensor, but I'm not counting on it, hence the encoders with commutation tracks.

Erik

Yes, encoder count rate limits are per channel (all channels are independent)

for 1.66 MHz and some margin for quadrature distortion you will probably have to raise the default mux frequency from 4.16 MHz to say 8.33 MHz and probably add the muxed encoder skew setting (30 ns + 3 nS/foot of cable). This is easily done in master, not so easily in 2.5.4

Ok, great. I'm running master now.

However... hypothetically I could run 2x 7i77 cards on one 5i25 and use stock settings, with two encoders running this speed on each channel?

Erik

Encoder count rate is independent of the number of encoders so it
doesn't matter if you have 3, 6 or 36 encoder channels.
It only depends on the filter settings (3 or 15 counts settable per channel)
and the filter frequency (which is global)

Global with respect to the FPGA card?

What I'm understanding is that how fast the card counts depends on how much filtering of the signal is needed... a signal on a short cable with little noise can use less filtering and get faster counting. I suppose the ideal would be using a back to back connector and balancing the 7i77 on top of the PC aligned with the 5i25, but that seems a bit excessive (although someone here did it, I don't have a link handy).

The filter rate is per FPGA card

The maximum count rate of multiplexed encoders depends on how fast you can mutiplex the signals from the encoder inputs, This depends on signal integrity issues which get worse in an exponential fashion with the cable length. Because of this exponential relationship, a direct connection is not significantly different than a 1-2 foot cable.

I have tested 12 MHZ multiplex rates with a 15 foot cable, These allow up
to 6 MHZ count rates and require skew compensation.
Shorter cables will allow more margin.

A separate issue is how how good your external signals are. You may find that have a high encoder input bandwidth will result in noise issues causing count errors.

Okay, the exponential increase makes sense. I would probably use the standard cable, then.

I'll have some testing to look forward to, it sounds like. I haven't worked much with signals problems before, so it should be an education