Topics

TDM Question


r_hoskin@...
 

Jerry,

You mentioned TDM in another post...  I'd seen that come up in other posts over at CN and never quite understood how it worked.  So I looked up some posts, and also discovered a product listing for the unit on the ES website. 

That listing filled in some blanks:
"...compares the incoming signals from the high precision rotary encoder, which is mounted onto the RA shaft, with the time signal of its quartz oscillator. Any deviations of the angular velocity of the RA shaft from the prescribed drive rate are compensated..."

We have encoders on our drive motors, and if I recall correctly, the PMC-8 has its own oscillator... 


With the understanding that the TDM module probably has much better encoders than we do...could the PMC-8 be enhanced to provide a degree of TDM-like improvement? 

Thanks!

- Bob



hubbell_jerry@...
 

Hi Bob,

Yes, the TDM uses a very-high resolution incremental encoder system (interpolated sin/cos signal) which provides a precision of about 0.125 arc-seconds. This is made up of 12 bits for the cycles per revolution, and 11 bits of interpolated precision, for a total of about 23 bits. The rate accuracy for this system is about twice this at 0.25 arc-seconds/second. I wrote a paper back in 2012 when I first purchased mine:

 
I have worked with the company in Hungary that produces this system for several years and developed some software that we use to record data. The program also has a PID (proportional, integral, derivative) controller also that reads the encoder signal from the TDM and uses Pulse-Guiding to correct the drive rate like a camera would. 

I have always planned on integrating a very-high resolution incremental encoder system in the PMC-Eight for this very purpose. I have a good idea on the design of the interface circuit but have not had time to go to the next level with the PMC-Eight. I expect to get to that over the next year or so. One of the main goals for the PMC-Eight was to provide a very precise measurement of the RA and DEC axes position without having to resort to a high resolution encoder, or an encoder of any kind, that is why I used the stepper motors. The encoders on the servo motors used on other mounts are not near as precise as stepper motors when integrating the counts, they are used to maintain the proper tracking rate on the RA axis, and on both axes when configured as Alt/Az mounts.

In our factory testing, the PMC-Eight on the G11 and on the EXOS 2 has a repeatability of better than 1 arc-min. That is over a slew of 360 degrees in both directions. That is the uncorrected pointing error. For most users this is adequate to place the object on the CCD chip and in the Eyepiece. There are of course other factors involved with actually pointing the scope accurately such as polar alignment, cone error, and orthogonality of the axes, but  those are solved other ways. 

There are a lot of other features that would be nice to have, but we have to balance the cost with what people are willing to pay. Also, it depends on the size of the mount. We have actually fitted the TDM to an EXOS 2 mount, but with the mount costing $899, adding a $1500-$1800 drive correction system to it is probably not the most practical. I have to say though,

The TDM for me is the biggest time saver there is when you are observing. You basically turn it on and you get near perfect tracking. Good enough to do unguided imaging for 10 minutes. My time is worth a lot so that is why I purchased the TDM originally for my Skywatcher EQ6 Pro mount back in 2010. I love it. Just turn it on and forget it. It just works. My goal is to make all our equipment like that. It should disappear.

Jerry Hubbell
Director Electrical Engineering
Explore Scientific, LLC.



Robert Hoskin <r_hoskin@...>
 

TDM Question

Jerry,

Thanks!  So the current steppers do not have encoders?  My bad, I thought they had.  I was wondering if there was more that could be done with the hardware you already had in place, but you did that already :-)

Yes, introducing an additional $1500-1800 cost probably wouldn't work.  You can see folks working through that on CN with the TDM add-on and deciding to pass.  

Do the encoders you'd need drop off enough in cost vs capability that you could leave "Best" to the existing (expensive) TDM unit and deliver a less expensive "Better" with wider appeal?  

- Bob





From: "hubbell_jerry@... [ESPMC-Eight]"
To: ESPMC-Eight@...
Sent: Wednesday, March 21, 2018 4:11 PM
Subject: [ESPMC-Eight] Re: TDM Question

 
Hi Bob,

Yes, the TDM uses a very-high resolution incremental encoder system (interpolated sin/cos signal) which provides a precision of about 0.125 arc-seconds. This is made up of 12 bits for the cycles per revolution, and 11 bits of interpolated precision, for a total of about 23 bits. The rate accuracy for this system is about twice this at 0.25 arc-seconds/second. I wrote a paper back in 2012 when I first purchased mine:

 
I have worked with the company in Hungary that produces this system for several years and developed some software that we use to record data. The program also has a PID (proportional, integral, derivative) controller also that reads the encoder signal from the TDM and uses Pulse-Guiding to correct the drive rate like a camera would. 

I have always planned on integrating a very-high resolution incremental encoder system in the PMC-Eight for this very purpose. I have a good idea on the design of the interface circuit but have not had time to go to the next level with the PMC-Eight. I expect to get to that over the next year or so. One of the main goals for the PMC-Eight was to provide a very precise measurement of the RA and DEC axes position without having to resort to a high resolution encoder, or an encoder of any kind, that is why I used the stepper motors. The encoders on the servo motors used on other mounts are not near as precise as stepper motors when integrating the counts, they are used to maintain the proper tracking rate on the RA axis, and on both axes when configured as Alt/Az mounts.

In our factory testing, the PMC-Eight on the G11 and on the EXOS 2 has a repeatability of better than 1 arc-min. That is over a slew of 360 degrees in both directions. That is the uncorrected pointing error. For most users this is adequate to place the object on the CCD chip and in the Eyepiece. There are of course other factors involved with actually pointing the scope accurately such as polar alignment, cone error, and orthogonality of the axes, but  those are solved other ways. 

There are a lot of other features that would be nice to have, but we have to balance the cost with what people are willing to pay. Also, it depends on the size of the mount. We have actually fitted the TDM to an EXOS 2 mount, but with the mount costing $899, adding a $1500-$1800 drive correction system to it is probably not the most practical. I have to say though,

The TDM for me is the biggest time saver there is when you are observing. You basically turn it on and you get near perfect tracking. Good enough to do unguided imaging for 10 minutes. My time is worth a lot so that is why I purchased the TDM originally for my Skywatcher EQ6 Pro mount back in 2010. I love it. Just turn it on and forget it. It just works. My goal is to make all our equipment like that. It should disappear.

Jerry Hubbell
Director Electrical Engineering
Explore Scientific, LLC.





hubbell_jerry@...
 

Hi Bob,

Yes, your are correct again. 3-4 years ago, I speculated the same thing that we could get away from using the very-high resolution incremental encoder and go with a less expensive lower resolution one. Something like a 1000 cycle Sin/Cos encoder with interpolated at 10-bits. This would result in an effective 20 bit resolution or about 0.5 arc-seconds per pulse.

What we are actually dealing with here is the rate change. For the 0.125 arc-second resolution encoder system, the pulse rate out of the encoder at sidereal rate is equal to 15/0.125 pulses/second or 120 pps. This gives you a maximum theoretical update rate of about 60 Hz. The TDM averages 12 samples to get an effective update rate of 5 samples/second. That averaging really nocks down the noise by sqrt(12) or by a factor of about 3.5.

If we wanted the same update rate on the lower cost incremental encoder with a 0.5 arc-sec resolution, that would be 4 times less pps, or 30 pps. The max update rate for that would be half that or 15 Hz. We could only average 3 samples to get the same 5 samples/second. I would be curious to see how much jitter there would be in the instantaneous rate independent of the gear motion. I probably could code a simulator that would let me investigate it further. Here is a link to a screenshot I posted in the G11 album


I was planning on modifying the TDM TE Recorder software that we use in the observatory to take the raw pulses out of the TDM and lowering the resolution in the program to see what effect it would have on the guiding. This way I could investigate different resolutions and determine the sweet spot for maintaining acceptable guiding performance at the lowest possible cost.

I would want to develop some specific software to perform this study with the TDM. It can be done. Just a simple matter of finding the time. I need to add that into my project list. It keeps getting pushed back, but it is time I start looking at it again.

Thanks

Jerry Hubbell
Director Electrical Engineering
Explore Scientific, LLC.