MESA cards
21 Feb 2024 01:02 #293874
by ajc771
MESA cards was created by ajc771
Hello,
First let me say I am a newbie and want to learn as much as I can from the wiser ones in this group. Further, I appreciate any and all help as I have been at this for over a year, on and off, trying to get my system working. I purchased a 7i76E and 7i85 board from MESA and feel that there documentation (although very thorough but hard to digest) is, in parts, over my head . MESA's support seems to be lacking and that may be because I am not a programmer or a wizard at computers. At this point in time, I would have not made the purchase knowing all the stress and time spent spinning my wheels.
I have asked a question, recently, about MUXed encoders and grayed out areas of the pncconfig program with no reply. Again its probably my fault because I asked the wrong question. Now I ran into another problem (one after another) concerning the "closedloop stepper signals" section of the HAL file (which I do not know how it got in the HAL file, since I do not have a closed loop steppper). In any event, seems the pncconfig program wants to net the x-pos-fb signal to stepgen.00.position-fb. This causes an error because I want the x-pos-fb signal to be connected to the x-axis encoder. In my logic (which is probably flawed) the program should know that the x-axis encoder info should reflect the steppers pulse count and at the same time, verifying the ball screw movement. Very confusing on how this program logically looks at real world data.
Any help in clarifying what is meant by MUXed (MUX x 4 or MUX x 16) encoder vs a TTL or differential encoder would be appreciated. I have TTL encoders and hope they will work with the above MESA cards. Also, I have a pendant control that has a MPG on it with x1, x10, x100 switch on it and will need help adding it to the HAL files. The encoder on the Pendant seems to be differential as it has A+, A-, B+, B- wires.
In regards to the problem with x-pos-fb; I just commented out the lines where they were netted to stepgen position feedback . Please, comment on this being the correct strategy.
Thanks,
ajc
First let me say I am a newbie and want to learn as much as I can from the wiser ones in this group. Further, I appreciate any and all help as I have been at this for over a year, on and off, trying to get my system working. I purchased a 7i76E and 7i85 board from MESA and feel that there documentation (although very thorough but hard to digest) is, in parts, over my head . MESA's support seems to be lacking and that may be because I am not a programmer or a wizard at computers. At this point in time, I would have not made the purchase knowing all the stress and time spent spinning my wheels.
I have asked a question, recently, about MUXed encoders and grayed out areas of the pncconfig program with no reply. Again its probably my fault because I asked the wrong question. Now I ran into another problem (one after another) concerning the "closedloop stepper signals" section of the HAL file (which I do not know how it got in the HAL file, since I do not have a closed loop steppper). In any event, seems the pncconfig program wants to net the x-pos-fb signal to stepgen.00.position-fb. This causes an error because I want the x-pos-fb signal to be connected to the x-axis encoder. In my logic (which is probably flawed) the program should know that the x-axis encoder info should reflect the steppers pulse count and at the same time, verifying the ball screw movement. Very confusing on how this program logically looks at real world data.
Any help in clarifying what is meant by MUXed (MUX x 4 or MUX x 16) encoder vs a TTL or differential encoder would be appreciated. I have TTL encoders and hope they will work with the above MESA cards. Also, I have a pendant control that has a MPG on it with x1, x10, x100 switch on it and will need help adding it to the HAL files. The encoder on the Pendant seems to be differential as it has A+, A-, B+, B- wires.
In regards to the problem with x-pos-fb; I just commented out the lines where they were netted to stepgen position feedback . Please, comment on this being the correct strategy.
Thanks,
ajc
Please Log in or Create an account to join the conversation.
- tommylight
-
- Away
- Moderator
-
Less
More
- Posts: 18563
- Thank you received: 6219
21 Feb 2024 01:37 #293878
by tommylight
Replied by tommylight on topic MESA cards
I recall your question so if you got no answer:
-i was very busy
-it was on the wrong section of the forum
Mesa related questions ( or Pico Sytems) should be on the "driver boards" section.
Feel free to blame that one on me as i always move stuff around to proper sections, but must have missed it.
Mesa support is absolutely the best, their owner is here helping always.
I have nothing with Mesa, i just buy their stuff like there is no tomorrow as they have proven extremely reliable.
Muxed encoders is something you should not bother with as it is not required knowledge for making machines work, at all.
MUX = multiplex meaning 2 or more encoders are "mixed" together and sent to the FPGA (controller) where they are demultiplexed and values sent to corresponding axis/joints.
I hope i did not mess the explanation.
Normal TTL A/B/Z or A/B/I encoders work always with Mesa boards, just make sure they work with 5V supply, for any higher voltage encoders "voltage dividers" must be used, easy to make with 2 resistors for each encoder signal.
One thing to check is if the encoders have differential or single ended outputs and set the Mesa boards acordingly. Differential ones have signals like A+ and A-, B+ and B- etc, single ended have only one wire per signal.
And yes, you did correctly omit the stepgen feedback, if you are using encoders.
BUT, and this is a big but (
) you will have to omit the encoder feedback first, make the machine work in open loop with feedback from stepgens, set scaling and speeds/accelerations, etc, only then go back to encoder feedback and do the tuning.
Keep in mind that stepper with encoder feedback and loop closed in LinuxCNC behave like servo systems, so a bit of tuning is required, usually by lowering the P value quite a bit from the default 1000 and maybe a bit of FF1 testing of values from 0 to 1.
-i was very busy
-it was on the wrong section of the forum
Mesa related questions ( or Pico Sytems) should be on the "driver boards" section.
Feel free to blame that one on me as i always move stuff around to proper sections, but must have missed it.
Mesa support is absolutely the best, their owner is here helping always.
I have nothing with Mesa, i just buy their stuff like there is no tomorrow as they have proven extremely reliable.
Muxed encoders is something you should not bother with as it is not required knowledge for making machines work, at all.
MUX = multiplex meaning 2 or more encoders are "mixed" together and sent to the FPGA (controller) where they are demultiplexed and values sent to corresponding axis/joints.
I hope i did not mess the explanation.
Normal TTL A/B/Z or A/B/I encoders work always with Mesa boards, just make sure they work with 5V supply, for any higher voltage encoders "voltage dividers" must be used, easy to make with 2 resistors for each encoder signal.
One thing to check is if the encoders have differential or single ended outputs and set the Mesa boards acordingly. Differential ones have signals like A+ and A-, B+ and B- etc, single ended have only one wire per signal.
And yes, you did correctly omit the stepgen feedback, if you are using encoders.
BUT, and this is a big but (
![:) :)](/media/kunena/emoticons/smile.png)
Keep in mind that stepper with encoder feedback and loop closed in LinuxCNC behave like servo systems, so a bit of tuning is required, usually by lowering the P value quite a bit from the default 1000 and maybe a bit of FF1 testing of values from 0 to 1.
Please Log in or Create an account to join the conversation.
- tommylight
-
- Away
- Moderator
-
Less
More
- Posts: 18563
- Thank you received: 6219
21 Feb 2024 01:38 #293879
by tommylight
Replied by tommylight on topic MESA cards
Moved to "driver boards".
Please Log in or Create an account to join the conversation.
21 Feb 2024 03:16 #293884
by PCW
Replied by PCW on topic MESA cards
A couple things:
I missed your post about encoders
Whether encoders are muxed or not has no real effect
on configuration file setup or hardware setup.
Almost all mesa encoders (other than MPG encoders)
can be jumpered to select TTL or differential inputs.
This is true of the 7I76E and 7I85. Typically MPG inputs
are single ended, this means if you have a differential
output MPG, you only use the two A,B lines and ignore
/A and /B
Typically 7I76E firmware with 7I85 or 7I85X puts the
first 4 encoders on the 7I85 or 7I85S
so that the sensible encoder/stepgen 0,1,2 = X,Y,Z
mapping can be used.
Also note that stepgen systems with encoder
feedback are rather complex and uncommon.
In fact, if you are setting up such a system,
its better to get a pure open loop stepgen
system configured and working before you
change over to encoder feedback
With pncconf if you setup encoder feedback
the hal file will have the proper connections
(feedback comes from the encoder)
If setup without encoder feedback, the local
stepgen position feedback is used.
This may be pncconf version dependent
but current (2.9.2) does the right thing.
Note that tuning will be quite different
with stepgen feedback or encoder feedback
and with encoder feedback you must get
the stepgen scale and encoder scale and directions
correct.
I missed your post about encoders
Whether encoders are muxed or not has no real effect
on configuration file setup or hardware setup.
Almost all mesa encoders (other than MPG encoders)
can be jumpered to select TTL or differential inputs.
This is true of the 7I76E and 7I85. Typically MPG inputs
are single ended, this means if you have a differential
output MPG, you only use the two A,B lines and ignore
/A and /B
Typically 7I76E firmware with 7I85 or 7I85X puts the
first 4 encoders on the 7I85 or 7I85S
so that the sensible encoder/stepgen 0,1,2 = X,Y,Z
mapping can be used.
Also note that stepgen systems with encoder
feedback are rather complex and uncommon.
In fact, if you are setting up such a system,
its better to get a pure open loop stepgen
system configured and working before you
change over to encoder feedback
With pncconf if you setup encoder feedback
the hal file will have the proper connections
(feedback comes from the encoder)
If setup without encoder feedback, the local
stepgen position feedback is used.
This may be pncconf version dependent
but current (2.9.2) does the right thing.
Note that tuning will be quite different
with stepgen feedback or encoder feedback
and with encoder feedback you must get
the stepgen scale and encoder scale and directions
correct.
Please Log in or Create an account to join the conversation.
21 Feb 2024 03:33 #293887
by ajc771
Replied by ajc771 on topic MESA cards
Tommy,
Thanks for getting back to me so quickly. Concerning my MPG; it has 4 wires, A+, A-, B+ and B-. It is 100 PPR but the 7i76e terminals only take A+ and B+ and of course the 5 volts. On my oscilloscope I get a pulse from A+ and B+ by connecting to the probe grounds to negative.. A and B signals are 90 degrees out of phase. Maybe I should just leave the A- and B- unconnected?
In reference to your suggestion to run a PID tuning; can that be done without running PNCongfig? If so how?
Agian, thanks for your help.
ajc
Thanks for getting back to me so quickly. Concerning my MPG; it has 4 wires, A+, A-, B+ and B-. It is 100 PPR but the 7i76e terminals only take A+ and B+ and of course the 5 volts. On my oscilloscope I get a pulse from A+ and B+ by connecting to the probe grounds to negative.. A and B signals are 90 degrees out of phase. Maybe I should just leave the A- and B- unconnected?
In reference to your suggestion to run a PID tuning; can that be done without running PNCongfig? If so how?
Agian, thanks for your help.
ajc
Please Log in or Create an account to join the conversation.
21 Feb 2024 03:47 #293888
by ajc771
Replied by ajc771 on topic MESA cards
Hi Peter,
Thanks for your quick reply. I am confused about what stepgen feedback is? I do not have a closed loop stepper. What I did is mount an Omron encoder at the opposite end of the axis's. They of course rotate in the opposite direction of the steppers. I put this info in the pncconfig file along with all the other info and calculated the no of pulses of the encoder per the steps of steppers. I did not run a tuning sequence though and will leave that for later.
Is that correct? Also, how can I run a tuning sequence (PID) without running the pncconfig program again and losing all my manual edits to my HAL file?
Thanks,
ajc
Thanks for your quick reply. I am confused about what stepgen feedback is? I do not have a closed loop stepper. What I did is mount an Omron encoder at the opposite end of the axis's. They of course rotate in the opposite direction of the steppers. I put this info in the pncconfig file along with all the other info and calculated the no of pulses of the encoder per the steps of steppers. I did not run a tuning sequence though and will leave that for later.
Is that correct? Also, how can I run a tuning sequence (PID) without running the pncconfig program again and losing all my manual edits to my HAL file?
Thanks,
ajc
Please Log in or Create an account to join the conversation.
21 Feb 2024 05:03 #293892
by PCW
Replied by PCW on topic MESA cards
Standard hardware stepgen setup for Mesa (and other)
systems uses local feedback (from the stepgen position)
to have position control, so there is always a local feedback
loop in the hal file.
PID tuning is typically done live with the "calibrate"
menu which allows live adjustments of PID parameters
and has the ability to save these changes back to the INI file.
systems uses local feedback (from the stepgen position)
to have position control, so there is always a local feedback
loop in the hal file.
PID tuning is typically done live with the "calibrate"
menu which allows live adjustments of PID parameters
and has the ability to save these changes back to the INI file.
The following user(s) said Thank You: ajc771
Please Log in or Create an account to join the conversation.
Time to create page: 0.345 seconds