DMM DYN4 Test findings

Figured I'd make a new post about this but this is basically in response to what was discussed in this thread:
forum.linuxcnc.org/27-driver-boards/4122...de-instead-of-analog

The gist of that discussion is that there is a 5ms delay on the emulated encoder output of the drive. I didn't quite agree with that based on the halscope images posted but there did appear to be somewhere around 3ms of delay. Long story short I found that the drive's encoder output has much less delay. It's almost not even worth mentioning. Analog (velocity) control has what appears to be about a 400us delay vs the real encoder but it is only the first pulse. By about the 3rd pulse it settles to more like 200us as it's sampling speed better. Over serial (position) there is almost no delay at all, it's more like a 0-200us deviation rather than delay as the pulses actually tend to be slightly ahead of the real encoder.

That said there actually CAN BE a decent sized delay when using analog control, it's in the drive's response to the analog input (ie actually making the motor turn) and it's greatly affected by the settings used on the drive. It's hard to say how much of it is some issue with the drive and how much of it just boils down to things like acceleration settings that will affect the encoder reading on any drive. I wasn't able to get super methodical about trying to figure out what parameters had the biggest effect but it seems like it's related to how the deadzone is calculated as mentioned on P78 of the manual.

  
Test rig is a 86M-DHT-A6MK1 750w motor and a DYN4-H01A2 drive. The motor is coupled to a 1000ppr Omron encoder and the drive itself is set to output 1000ppr as well (Line Number).

My scope is only 2 channel. So the first test I did was to check the DYN4 encoder output vs the Omron encoder over serial:
 
Yellow ch1 is the drive's encoder output, red is the real encoder. There's no delay here at all, the drive's output is actually slightly ahead of the real encoder.

This is the same test with Analog control. It's off slightly more on the first pulse but plays catch up by the 3rd. 1st pulse on the drive is about 400us late then gets to about 200us.
 

Here's where the actual issue occurs, the analog input. I don't know how other servo setups behave with analog control. I realized I can't really capture an analog slope very well so I pulsed the drive's analog input to 5v. Yellow ch1 is the analog input and ch2 is the drive's encoder output:
 
This actually does show about 5ms delay. DMM had sent me a scope shot of a 1ms delay between the analog input pulse and the encoder output. Using these settings:
MAX_ACCELERATION = 127
MAX_SPEED = 24
OnPosRange = 127
Torque Filter = 48
GEAR_NUM = 4096
LINE_NUM = 2500

I used LINE_NUM = 1000 to match the real encoder PPR and I used MAX_SPEED = 127. Reducing MAX_SPEED actually does reduce the delay a bit but it definately was not 1ms and I didn't mess with it enough to really figure out the correlation. I'll have to play with it some more but for now I'll say that it certainly can have a 5ms delay on the analog input itself, the encoder output is fine in everything I've seen. I'd try this with step/direction but I don't have anything setup at the moment to do it.
 

Just an update....

DMM sent me a video of a a pulse generator or something hitting the analog input of my same drive 6 or 7 times like 3 seconds apart and it was showing 1ms delay on the analog input again each time.

I repeated my test and kept getting about 4ms each time. The conclusion is it's probably the firmware as I suspected. They're going to send me a new drive and I'll test that when I get it.