G540 Timings (defaults too slow)

More
12 Apr 2024 17:31 - 12 Apr 2024 17:35 #298088 by jtrantow
After years of using the default stepgen timings I've noticed they are much larger than needs be. Using PnCConf Wizard and selecting G540 generates:

DIRSETUP   = 5000
DIRHOLD    = 5000
STEPLEN    = 10000
STEPSPACE  = 10000

www.geckodrive.com/support/g540-manual-rev8
Step Pulse “0” Time2 uSStep Pulse “1” Time1 uSDirection Setup (Before step rising edge)200 nSDirection Setup (Hold after pulse rising edge)200 nS

PNCConf G540 timing will require 20usec per step when it only requires 3usec. I believe the PnCConf Wizard should be generating:

DIRSETUP   = 200
DIRHOLD    = 200
STEPLEN    = 1000
STEPSPACE  = 2000

I realize these timings are not the limitation in most cases. I think stepgen has similarly loose timing but I'm no longer using that. I have not tested mesact to see if this problem is across the board.

 
Last edit: 12 Apr 2024 17:35 by jtrantow. Reason: formating

Please Log in or Create an account to join the conversation.

More
12 Apr 2024 18:08 #298097 by Aciera
In my opinion the PnCConf Wizard should produce a working config for a beginner.
While the default timings do seem rather conservative it would certainly not be wise to set them to the minimum values stated in the manual.
The following user(s) said Thank You: zz912

Please Log in or Create an account to join the conversation.

More
16 Apr 2024 18:18 #298396 by zz912
I agree with Aciera. I have bad experience with short timing. If the hardware is not ideal, a step may be lost. Longer timing has already solved my hardware problem several times.
The following user(s) said Thank You: tommylight

Please Log in or Create an account to join the conversation.

More
17 Apr 2024 14:52 #298450 by jtrantow
Were these bad experiences with short timing on systems with a Mesa fpga board and G540?

I too have helped friends debug systems with PP and chinesium drivers where increasing the timing solved stepping problems.

I'm not asking for changing the timing for all boards. I do think specifically selecting the G540 driver should use values in the right ballpark. 25X DIR, 7X STEP aren't even in the same zipcode.
I've had good experience with Gecko drivers and tend to believe their specifications. I'd have less confidence in parallel port stepping. It might be reasonable to apply a 10%, 20%, or even 100% safety factor.

I emailed Gecko support to get their take on this matter and will report back when I hear more.

Please Log in or Create an account to join the conversation.

More
17 Apr 2024 15:14 #298455 by PCW
I would probably double minimum timings for step time and step space
for adequate margins.

The setup and hold times have basically 0 effect on performance so
it's not harmful to give them larger margins.

Please Log in or Create an account to join the conversation.

More
25 Apr 2024 01:05 #299036 by jtrantow
I received the following reply from Marcus Freimanis. I believe he's the designer. Based on this information I believe the following 110% values should be the g540 defaults.

DIRSETUP   = 220
DIRHOLD    = 220
STEPLEN    = 1100
STEPSPACE  = 2200

Thank you for contacting Geckodrive technical support. I apologize for missing your email when it first came in.                 There is zero advantage to a much longer signal being sent; the G540 will only pay attention to the minimum timing (e.g.: the first 2uS of a 10mS signal, essentially wasting 9.998mS). If it is six-in-one and half-dozen in the other it is always wise to err on the side of a longer signal as there is no explicit downside as long as you can output the frequency you need with that pulse timing.                 The only reason you would NEED a very specific and very short signal would be if you were driving the motor to run at a high step pulse frequency, which would result in what would likely be a speed beyond the stall speed of the motor. A safety margin of 10% will take into account component tolerances across multiple devices.                 The timing on all revisions is identical and they all use the same circuit, with an HCPL2531 optoisolator.                 Again, my apologies for missing your email! Marcus Freimanis Geckodrive, Inc.1540 S Grand AveSanta Ana, CA 92705 P: (714) 832-8874F: (714) 832-8082  www.geckodrive.com

Please Log in or Create an account to join the conversation.

Moderators: cmorley
Time to create page: 0.273 seconds
Powered by Kunena Forum