Linuxcnc - parallel port vs mesa card
08 Jul 2016 06:58 #77169
by jrsphoto
Linuxcnc - parallel port vs mesa card was created by jrsphoto
Hello all,
I'm currently using using linuxcnc 2.7 with a standard parallel port on a 2.8ghz intel core2quad processor, 16gb ram and a 4-axis cnc router, roughly 60" x 30" in size. I have two general question with regards to linuxcnc with mesa cards.
1) Should/Would one expect any noticeable difference in the performance or operation of linuxcnc when using something like the Mesa 7I76-6i25 combo, compared to a normal parallel port config. Just trying to get a feel for what I could expect.
2) Are there any benefit in using the Mesa 7I76e (ethernet) over the 6I25 - PCIE bus card, or just preference?
My understanding is that the mesa card would give you a LOT of additional configuration options that you don't have with the parallel port. It also seems that it would be possible to get more performance from the mesa cards because they can generate faster step pulses. But any performance gains would still be limited by the motors (impedance, voltage, etc) drivers I'm using, and the design limits of my router itself... Is that thinking correct?
Thanks
I'm currently using using linuxcnc 2.7 with a standard parallel port on a 2.8ghz intel core2quad processor, 16gb ram and a 4-axis cnc router, roughly 60" x 30" in size. I have two general question with regards to linuxcnc with mesa cards.
1) Should/Would one expect any noticeable difference in the performance or operation of linuxcnc when using something like the Mesa 7I76-6i25 combo, compared to a normal parallel port config. Just trying to get a feel for what I could expect.
2) Are there any benefit in using the Mesa 7I76e (ethernet) over the 6I25 - PCIE bus card, or just preference?
My understanding is that the mesa card would give you a LOT of additional configuration options that you don't have with the parallel port. It also seems that it would be possible to get more performance from the mesa cards because they can generate faster step pulses. But any performance gains would still be limited by the motors (impedance, voltage, etc) drivers I'm using, and the design limits of my router itself... Is that thinking correct?
Thanks
Please Log in or Create an account to join the conversation.
08 Jul 2016 12:50 #77175
by andypugh
Pretty much correct, yes. A Mesa card will give a smoother step output, and one with less granularity, but you still might not see any performance increase. Have you any feel for how your max-velocity at the moment correlates to your theoretical max step rate?
Replied by andypugh on topic Linuxcnc - parallel port vs mesa card
It also seems that it would be possible to get more performance from the mesa cards because they can generate faster step pulses. But any performance gains would still be limited by the motors (impedance, voltage, etc) drivers I'm using, and the design limits of my router itself... Is that thinking correct?
Pretty much correct, yes. A Mesa card will give a smoother step output, and one with less granularity, but you still might not see any performance increase. Have you any feel for how your max-velocity at the moment correlates to your theoretical max step rate?
Please Log in or Create an account to join the conversation.
- Todd Zuercher
- Offline
- Platinum Member
Less
More
- Posts: 5008
- Thank you received: 1441
08 Jul 2016 14:01 #77179
by Todd Zuercher
Replied by Todd Zuercher on topic Linuxcnc - parallel port vs mesa card
I upgraded a step/dir servo driven machine from software stepping to a 5i25/7i85S combo. The main advantages for me were that I could use the encoder feedback for tuning the servo drives, I no longer had a speed limit(I chose 600ipm rather than the previous 300), and I could set step pulse and encoder count ratio to be the same (rather than using the servo drives software gearing). By far the biggest advantage for me was having the ability to use halscope when trying to tune the servo drives. (That was worth the cost of the upgrade by it's self.) But the overall effect on machine performance has been minimal, mostly just higher G0 positioning speeds.
Please Log in or Create an account to join the conversation.
08 Jul 2016 15:11 #77183
by PCW
Replied by PCW on topic Linuxcnc - parallel port vs mesa card
One of the nice design decisions in LinuxCNC is that all the complicated motion operations are done by the host CPU
so all of LinuxCNC's features are available regardless of the external hardware chosen, from Parallel port up to a fancy Ethercat motor drive.
As Andy and Todd have said the main advantage of Mesa (or Pico or other) hardware is that they can generate faster and cleaner pulse streams and count encoder pulses faster as well. This may not bring any advantages if you are already limited by drive/motor speed limitations and don't use high resolution encoders.
We have had customers that have seen improvements in maximum step rate without stalling even if software stepping
can generate the required step rate. I believe this is due the cleaner pulse stream that lacks the beats between the base thread frequency and the requested step rate. These beats cause unwanted phase modulation that can excite the natural
resonance of step motors
so all of LinuxCNC's features are available regardless of the external hardware chosen, from Parallel port up to a fancy Ethercat motor drive.
As Andy and Todd have said the main advantage of Mesa (or Pico or other) hardware is that they can generate faster and cleaner pulse streams and count encoder pulses faster as well. This may not bring any advantages if you are already limited by drive/motor speed limitations and don't use high resolution encoders.
We have had customers that have seen improvements in maximum step rate without stalling even if software stepping
can generate the required step rate. I believe this is due the cleaner pulse stream that lacks the beats between the base thread frequency and the requested step rate. These beats cause unwanted phase modulation that can excite the natural
resonance of step motors
Please Log in or Create an account to join the conversation.
16 Feb 2017 05:24 #88037
by Karri
Replied by Karri on topic Linuxcnc - parallel port vs mesa card
Hello
How mesa card can affect to read faster encoder pulse than parallel port?
I mean if you'r pc latency allows for example 5000 step pulse, how it can read faster encoder pulse's.
I have understood it that way that on same cycle you make pulses and read encoders, buttons and so on
I'm asking cause i found quadrature encoder with index pulse. But it was 2000pulse / rev - way too fast for my pc.
Kind regards karri
How mesa card can affect to read faster encoder pulse than parallel port?
I mean if you'r pc latency allows for example 5000 step pulse, how it can read faster encoder pulse's.
I have understood it that way that on same cycle you make pulses and read encoders, buttons and so on
I'm asking cause i found quadrature encoder with index pulse. But it was 2000pulse / rev - way too fast for my pc.
... This may not bring any advantages if you are already limited by drive/motor speed limitations and don't use high resolution encoders
Kind regards karri
Please Log in or Create an account to join the conversation.
- tommylight
- Online
- Moderator
Less
More
- Posts: 19277
- Thank you received: 6452
16 Feb 2017 05:56 #88038
by tommylight
Replied by tommylight on topic Linuxcnc - parallel port vs mesa card
If i remember correctly, mesa cards can read up to 10MHz on encoder inputs, that is 10.000.000 pulses per second. It reports that back to Linuxcnc at 1000 times per second, if the servo thread is 1000000.
That is plenty for any encoder, even for some extreemly high resolution ones.
On a machine i retrofited lately i have 4096 cpr absolute encoders attached to a mesa 7i74 and from that to a 5i25. Id does 5000 rpm on the spindle easily.
I doubt you will come acros servo drives that do more than 5000 rpm, usualy they top at 3000 rpm, during work they rarely go that high.
Bottom line, they are way to fast for any combination you can think of.
Regards,
Tom
That is plenty for any encoder, even for some extreemly high resolution ones.
On a machine i retrofited lately i have 4096 cpr absolute encoders attached to a mesa 7i74 and from that to a 5i25. Id does 5000 rpm on the spindle easily.
I doubt you will come acros servo drives that do more than 5000 rpm, usualy they top at 3000 rpm, during work they rarely go that high.
Bottom line, they are way to fast for any combination you can think of.
Regards,
Tom
The following user(s) said Thank You: Karri
Please Log in or Create an account to join the conversation.
17 Feb 2017 07:47 #88102
by Karri
Replied by Karri on topic Linuxcnc - parallel port vs mesa card
Thanks,
So Mesa card will calculates the pulses itself and the reports value of position (or speed? ) to Linuxcnc - not the individual pulses?
Karri
So Mesa card will calculates the pulses itself and the reports value of position (or speed? ) to Linuxcnc - not the individual pulses?
Karri
Please Log in or Create an account to join the conversation.
- tommylight
- Online
- Moderator
Less
More
- Posts: 19277
- Thank you received: 6452
17 Feb 2017 12:22 #88107
by tommylight
Replied by tommylight on topic Linuxcnc - parallel port vs mesa card
Yes
Please Log in or Create an account to join the conversation.
17 Feb 2017 13:57 - 18 Feb 2017 11:58 #88111
by BigJohnT
Replied by BigJohnT on topic Linuxcnc - parallel port vs mesa card
I asked Peter when I was adding an encoder to my BP knee mill and he said the encoder input can count up to 1MHz or 1,000,000 cycles per second. So if you have a 2500 ppr encoder you can spin it at 12,000 RPM. Pretty impressive I'd say.
Edit: this was for my 7i77 card.
JT
Edit: this was for my 7i77 card.
JT
Last edit: 18 Feb 2017 11:58 by BigJohnT.
Please Log in or Create an account to join the conversation.
18 Feb 2017 00:58 #88147
by PCW
Replied by PCW on topic Linuxcnc - parallel port vs mesa card
Yeah the maximum encoder count rate for the Mesa cards depends on a number of variables
1. ClockLow rate (33 MHz to 100 MHz depending on card)
2. Muxed on Non-muxed, muxed encoders top out at a couple MHz Max, non muxed up to about 10 MHz
3. Digital filtering settings (settable to 3 or 15 sample clocks, sample clock can be up to 8 MHz on mulltiplexed
encoders and up to ClockLow on non Muxed counters.)
1. ClockLow rate (33 MHz to 100 MHz depending on card)
2. Muxed on Non-muxed, muxed encoders top out at a couple MHz Max, non muxed up to about 10 MHz
3. Digital filtering settings (settable to 3 or 15 sample clocks, sample clock can be up to 8 MHz on mulltiplexed
encoders and up to ClockLow on non Muxed counters.)
Please Log in or Create an account to join the conversation.
Time to create page: 0.079 seconds