Adjust CP4 Clock Frequency


Craig Young
 

Rolando mentioned in a recent posting (#70144) the tracking rate is dependent on the controller clock frequency.  Over the years of using my AP1600 with encoders I could never get APCC to provide the same tracking accuracy as other users, even with a 200 point model, so it had to be a problem with this particular mount.  Since this has never be resolved after several hours working with AP I wrote my own tracking program ATRACK which does provide very accurate tracking over several hours.

With ATRACK I know exactly what the RA and DEC tracking rates are because the target drifts no more than 1 or 2 pixels over 4 hours, without the use of an autoguider.  So comparing the 200 point model tracking rate with ATRACK I noticed a consistent error, in both direction and magnitude, which led me to the conclusion there is some base tracking error in the mount that APCC does not correct for.  One possibility would be the crystal frequency used for timing.  This is a known problem in radios which is why they revert to TCXO devices (Temperature Compensated Crystal Oscillators).  Also, on one of my other mounts, there is a setting to adjust the master clock frequency.  I know the basic tracking of the mount is excellent because ATRACK reports no sudden jumps (e.g., cable twists or mirror shift, etc).  And, if there is a clock frequency error then the encoders will provide excellent pointing (which they do) but accurate tracking would not work.  A 200 sample pointing model can provide excellent pointing and tracking but the tracking makes an assumption the sidereal tracking rate of the mount is perfect.  If the sidereal rate is too fast, like on my mount, then the model will not know this and accurate tracking is impossible no matter how big my model sample count or polar alignment.

To test this I ask the guys at Astro-Physics if there is a way to adjust the base clock frequency of the CP4 using APCC.  I made some quick measurements comparing ATRACK rates with APCC rates and found a discrepancy of 0.1% (0.001).  Since APCC is ALWAYS running fast I would then reduce the clock rate by 0.1% and see if it solves the problem.  The adjustment should reduce the tracking rate to match ATRACK.

Craig


Roland Christen
 


One possibility would be the crystal frequency used for timing.
You are heading down a blind alley on this one. Ray Gralack has been doing some development with APCC that may answer your issue.

If you really want to check your RA driving rate you can set the rate to Stop, place a piece of tape across the stationary and moving part of the RA axis. Cut the tape at the seam between the axis. Get an accurate timer, set it to the sidereal day (23hrs,56min, 4.09sec). Push the timer and set the rate to Sidereal at the same instance and let the axis track until the next day. Check when the tape lines up exactly with the timer. Or you can set it to Solar rate and set the timer to 24 hrs. We have done this a number of times with mounts that people swear are running fast or slow and every time the results are within a small fraction of a second.

Roland Christen
Astro-Physics Inc.


-----Original Message-----
From: Craig Young <craig.young.m8@...>
To: main@ap-gto.groups.io
Sent: Tue, May 5, 2020 4:23 pm
Subject: [ap-gto] Adjust CP4 Clock Frequency

Rolando mentioned in a recent posting (#70144) the tracking rate is dependent on the controller clock frequency.  Over the years of using my AP1600 with encoders I could never get APCC to provide the same tracking accuracy as other users, even with a 200 point model, so it had to be a problem with this particular mount.  Since this has never be resolved after several hours working with AP I wrote my own tracking program ATRACK which does provide very accurate tracking over several hours.

With ATRACK I know exactly what the RA and DEC tracking rates are because the target drifts no more than 1 or 2 pixels over 4 hours, without the use of an autoguider.  So comparing the 200 point model tracking rate with ATRACK I noticed a consistent error, in both direction and magnitude, which led me to the conclusion there is some base tracking error in the mount that APCC does not correct for.  One possibility would be the crystal frequency used for timing.  This is a known problem in radios which is why they revert to TCXO devices (Temperature Compensated Crystal Oscillators).  Also, on one of my other mounts, there is a setting to adjust the master clock frequency.  I know the basic tracking of the mount is excellent because ATRACK reports no sudden jumps (e.g., cable twists or mirror shift, etc).  And, if there is a clock frequency error then the encoders will provide excellent pointing (which they do) but accurate tracking would not work.  A 200 sample pointing model can provide excellent pointing and tracking but the tracking makes an assumption the sidereal tracking rate of the mount is perfect.  If the sidereal rate is too fast, like on my mount, then the model will not know this and accurate tracking is impossible no matter how big my model sample count or polar alignment.

To test this I ask the guys at Astro-Physics if there is a way to adjust the base clock frequency of the CP4 using APCC.  I made some quick measurements comparing ATRACK rates with APCC rates and found a discrepancy of 0.1% (0.001).  Since APCC is ALWAYS running fast I would then reduce the clock rate by 0.1% and see if it solves the problem.  The adjustment should reduce the tracking rate to match ATRACK.

Craig


Craig Young
 

How does the CP4 determine Sidreal Rate when using encoders?  For example, does the CP4 count encoder pulses over a given period of time?  Assuming the pulses represent an arc of the mount and knowing precisely the arc change per given time period it would seem this could be used to check the tracking rate.  Which would then imply a precise calibration of pulses per arc length.  How is that done?

Craig


Roland Christen
 

You are asking for confidential information.
The sidereal drive rate is the same for you when you are doing the drift measurements as well as for APCC when it is doing the sky modeling. Any error present is present for both in the same amount.

Do the test I outlined if you are unsure about the accuracy of the mount's sidereal rate.

Roland



-----Original Message-----
From: Craig Young <craig.young.m8@...>
To: main@ap-gto.groups.io
Sent: Tue, May 5, 2020 4:44 pm
Subject: Re: [ap-gto] Adjust CP4 Clock Frequency

How does the CP4 determine Sidreal Rate when using encoders?  For example, does the CP4 count encoder pulses over a given period of time?  Assuming the pulses represent an arc of the mount and knowing precisely the arc change per given time period it would seem this could be used to check the tracking rate.  Which would then imply a precise calibration of pulses per arc length.  How is that done?

Craig


Craig Young
 

I'm not sure the test will be accurate enough.  If the diameter of the circle the tape is on is say 150mm then the circumference is about 471 mm (c = pi * diam).  If the clock error is 0.001 then the error after 24 hours would be 471mm * 0.001 = 0.47mm .. not sure if I can accurately see that.  On the other hand, the number of encoder pulses for a period of time should be known very accurately .. without revealing the confidential information, how about a test in APCC where you set the time and the software reports the number of pulses recorded.  Seems like that is a basic test and could be added to the AE tab to check calibration, tracking rate, etc.  That would be far more accurate than cutting a piece of tape.

Craig


Roland Christen
 

1100 mount diameter is 175mm, yielding an error of more than 1/2 mm (.022inch) which you will definitely see with the tape lined up.


-----Original Message-----
From: Craig Young <craig.young.m8@...>
To: main@ap-gto.groups.io
Sent: Tue, May 5, 2020 5:31 pm
Subject: Re: [ap-gto] Adjust CP4 Clock Frequency

I'm not sure the test will be accurate enough.  If the diameter of the circle the tape is on is say 150mm then the circumference is about 471 mm (c = pi * diam).  If the clock error is 0.001 then the error after 24 hours would be 471mm * 0.001 = 0.47mm .. not sure if I can accurately see that.  On the other hand, the number of encoder pulses for a period of time should be known very accurately .. without revealing the confidential information, how about a test in APCC where you set the time and the software reports the number of pulses recorded.  Seems like that is a basic test and could be added to the AE tab to check calibration, tracking rate, etc.  That would be far more accurate than cutting a piece of tape.

Craig


Craig Young
 

If the Sidereal rate is 15 arcsec/second then a 60 second image will have a movement of 15 x 60 = 900 arcseconds.  If the error is 0.001 then the drift will be 900 x 0.001 = 0.9 arc seconds which is about 2 pixels on my plate scale.  A 5 minute image will then be 0.9 x 5 = 4.5 arc seconds .. or about 10 pixels .. this is what I am seeing.  But what if the drift is lower than this .. say 0.0001 then the tape measurement must be accurate to 0.05mm.  I think when a mount of this calibre has encoders (at over $6000 cost) and can easily determine pulses per period of time .. why don't we have this?  This is not confidential information, just a simple metric to make sure everything is calibrated properly.  I think using a piece of tape is not the method a mount of this calibre deserves.

Craig


Roland Christen
 

If you are seeing a 0.1% error in your basic Sidereal tracking rate (assuming that you have polar alignment for min drift pointing straight up) then something is definitely wrong and you should make arrangements to send the RA axis and your controller in so we can check it out.

You can do this test and report the results:
1) point the scope straight up with counterweights in the West. Check the drift rate of a star for 60 seconds, or better yet 5 minutes, in RA.
2) repeat the test with the counterweights in the East.
Note the drift rate as + or -.
3) send the scope to a star toward the eastern horizon and note the drift rate in Dec.

Roland Christen



-----Original Message-----
From: Craig Young <craig.young.m8@...>
To: main@ap-gto.groups.io
Sent: Tue, May 5, 2020 5:50 pm
Subject: Re: [ap-gto] Adjust CP4 Clock Frequency

If the Sidereal rate is 15 arcsec/second then a 60 second image will have a movement of 15 x 60 = 900 arcseconds.  If the error is 0.001 then the drift will be 900 x 0.001 = 0.9 arc seconds which is about 2 pixels on my plate scale.  A 5 minute image will then be 0.9 x 5 = 4.5 arc seconds .. or about 10 pixels .. this is what I am seeing.  But what if the drift is lower than this .. say 0.0001 then the tape measurement must be accurate to 0.05mm.  I think when a mount of this calibre has encoders (at over $6000 cost) and can easily determine pulses per period of time .. why don't we have this?  This is not confidential information, just a simple metric to make sure everything is calibrated properly.  I think using a piece of tape is not the method a mount of this calibre deserves.

Craig


Craig Young
 

By chance we will have a clear night tonight (and near full moon) so I should be able to run the tests.  To make sure, I assume you want APCC to not use the pointing model or any tracking corrections during these tests?  I will do a plate solve at the start and end of each test (5 minutes).  This will give me a precise measurement of the drift rates in both RA and DEC.
The third test is a problem though.  The building the mount is located in does not permit observations below about 45 degrees.  So the best I can do is point at DEC 0 and HA -3 .. is that okay?

Craig


Roland Christen
 


So the best I can do is point at DEC 0 and HA -3
It will have to do.

I was of course assuming that your drift rate of 0.1% was using the raw sidereal rate and not a custom rate in APCC or any other model rate. Or are you saying that the drift rate of 0.1% is caused by your APCC model?

Roland Christen


-----Original Message-----
From: Craig Young <craig.young.m8@...>
To: main@ap-gto.groups.io
Sent: Tue, May 5, 2020 6:48 pm
Subject: Re: [ap-gto] Adjust CP4 Clock Frequency

By chance we will have a clear night tonight (and near full moon) so I should be able to run the tests.  To make sure, I assume you want APCC to not use the pointing model or any tracking corrections during these tests?  I will do a plate solve at the start and end of each test (5 minutes).  This will give me a precise measurement of the drift rates in both RA and DEC.
The third test is a problem though.  The building the mount is located in does not permit observations below about 45 degrees.  So the best I can do is point at DEC 0 and HA -3 .. is that okay?

Craig


Craig Young
 

ATrack is used to determine the actual tracking rate.  I know this is correct because the drift (RA,DEC) is +/- 0.57 arcsec or less over 60 seconds, typically about +/- 0.2 arcsec.  If I then turn off ATrack correction and turn on APCC tracking correction (w/refraction) I then get a significant drift.  When comparing the two rates I get: ATrack Rate / APCC rate = approximately 0.001, or 0.1%.  This is consistent no matter which hemisphere or DEC,HA I have the telescope pointed at.  In other words, both ATrack and APCC change the RA and DEC tracking rates to accomodate the error in the mount pointing and refraction.  They both change in the same direction and about the same amount.  The problem I think is the Sidereal Rate is running too fast.  APCC assumes sidereal rate is accurate and adds the pointing model correction to it.  Which means the resultant corrected tracking rate is still running too fast.  So the base Sidereal Rate has to be slowed down by about 0.1%.  On the other hand, if the sidereal rate is corrected using the encoders and the encoder calibration is off then the sidereal rate will be off and APCC tracking correction will also be off.

Craig


Ray Gralak
 

With ATRACK I know exactly what the RA and DEC tracking rates are because the target drifts no more than 1 or
2 pixels over 4 hours, without the use of an autoguider. So comparing the 200 point model tracking rate with
ATRACK I noticed a consistent error, in both direction and magnitude, which led me to the conclusion there is
some base tracking error in the mount that APCC does not correct for.
Or... the more likely reason, is your scope has unmodeled complex flexure. This can easily cause apparent tracking errors. Changes in temperature of just a few degrees C can change the Young modulus of materials in the scope enough to cause apparent tracking rate changes at the pixel level.

-Ray Gralak
Author of APCC (Astro-Physics Command Center): https://www.astro-physics.com/apcc-pro
Author of PEMPro V3: https://www.ccdware.com
Author of Astro-Physics V2 ASCOM Driver: https://www.siriusimaging.com/apdriver


-----Original Message-----
From: main@ap-gto.groups.io [mailto:main@ap-gto.groups.io] On Behalf Of Craig Young
Sent: Tuesday, May 5, 2020 2:24 PM
To: main@ap-gto.groups.io
Subject: [ap-gto] Adjust CP4 Clock Frequency

Rolando mentioned in a recent posting (#70144) the tracking rate is dependent on the controller clock frequency.
Over the years of using my AP1600 with encoders I could never get APCC to provide the same tracking accuracy
as other users, even with a 200 point model, so it had to be a problem with this particular mount. Since this has
never be resolved after several hours working with AP I wrote my own tracking program ATRACK which does
provide very accurate tracking over several hours.

With ATRACK I know exactly what the RA and DEC tracking rates are because the target drifts no more than 1 or
2 pixels over 4 hours, without the use of an autoguider. So comparing the 200 point model tracking rate with
ATRACK I noticed a consistent error, in both direction and magnitude, which led me to the conclusion there is
some base tracking error in the mount that APCC does not correct for. One possibility would be the crystal
frequency used for timing. This is a known problem in radios which is why they revert to TCXO devices
(Temperature Compensated Crystal Oscillators). Also, on one of my other mounts, there is a setting to adjust the
master clock frequency. I know the basic tracking of the mount is excellent because ATRACK reports no sudden
jumps (e.g., cable twists or mirror shift, etc). And, if there is a clock frequency error then the encoders will provide
excellent pointing (which they do) but accurate tracking would not work. A 200 sample pointing model can provide
excellent pointing and tracking but the tracking makes an assumption the sidereal tracking rate of the mount is
perfect. If the sidereal rate is too fast, like on my mount, then the model will not know this and accurate tracking is
impossible no matter how big my model sample count or polar alignment.

To test this I ask the guys at Astro-Physics if there is a way to adjust the base clock frequency of the CP4 using
APCC. I made some quick measurements comparing ATRACK rates with APCC rates and found a discrepancy
of 0.1% (0.001). Since APCC is ALWAYS running fast I would then reduce the clock rate by 0.1% and see if it
solves the problem. The adjustment should reduce the tracking rate to match ATRACK.

Craig


Craig Young
 

I disagree Ray that this is the more likely reason.  If it were the reason then I would see a divergence between ATrack and APCC.  I don't, they follow each other in parallel which means there is a constant factor that is causing the difference.  The two likely constants are: a) the sidereal rate is not sidereal, or b) the encoder, if used to determine sidereal rate, is not calibrated.

Craig


Ray Gralak
 

I disagree Ray that this is the more likely reason. If it were the reason then I would see a divergence between
ATrack and APCC. I don't, they follow each other in parallel which means there is a constant factor that is causing
the difference. The two likely constants are: a) the sidereal rate is not sidereal, or b) the encoder, if used to
determine sidereal rate, is not calibrated.
That doesn't prove anything. That coincidence could be accounted for by many things, including different refraction calculations.

I am certain that the version of APCC you are using doesn't account for changes in modeling parameters because of temperature.

-Ray Gralak
Author of APCC (Astro-Physics Command Center): https://www.astro-physics.com/apcc-pro
Author of PEMPro V3: https://www.ccdware.com
Author of Astro-Physics V2 ASCOM Driver: https://www.siriusimaging.com/apdriver


-----Original Message-----
From: main@ap-gto.groups.io [mailto:main@ap-gto.groups.io] On Behalf Of Craig Young
Sent: Tuesday, May 5, 2020 5:34 PM
To: main@ap-gto.groups.io
Subject: Re: [ap-gto] Adjust CP4 Clock Frequency

I disagree Ray that this is the more likely reason. If it were the reason then I would see a divergence between
ATrack and APCC. I don't, they follow each other in parallel which means there is a constant factor that is causing
the difference. The two likely constants are: a) the sidereal rate is not sidereal, or b) the encoder, if used to
determine sidereal rate, is not calibrated.

Craig


Craig Young
 

I don't think it is a coincidence Ray.  But it is easier to make that statement than to try and find the real problem.  Notice that it could easily be determined if we had a feature in APCC AE tab to run some diagnostics on the encoder and tracking.  But that appears to be asking too much, so it is simpler to try and explain it away.  What is the harm in adding some diagnostics?  Seems like a professional mount like the 1600 would give the user some additional tools to fine tune it.  Or is this too complex for us amateurs and should only be left to the hands of experienced engineers?

Craig


Dale Ghent
 

On May 5, 2020, at 9:07 PM, Craig Young <craig.young.m8@gmail.com> wrote:

I don't think it is a coincidence Ray. But it is easier to make that statement than to try and find the real problem. Notice that it could easily be determined if we had a feature in APCC AE tab to run some diagnostics on the encoder and tracking. But that appears to be asking too much, so it is simpler to try and explain it away. What is the harm in adding some diagnostics? Seems like a professional mount like the 1600 would give the user some additional tools to fine tune it. Or is this too complex for us amateurs and should only be left to the hands of experienced engineers?
Yo, just a bystander watching an otherwise quite technical thread here, but check the tone before proceeding. You're either investigating a discrepancy and its cause, or you're trying to accuse people of things with a chip on your shoulder. Pick one.

Thanks
/dale


Craig Young
 

Dale,
Good point, I would like to continue investigating the discrepancy.  To me some diagnostics are needed and I am asking if AP would add them to the AE tab.

Craig


Ray Gralak
 

Craig,

Say there is a tracking rate error of 0.1%. What makes your software immune to this tracking rate error? That is do you have to adjust the calculated rate you get by 0.1%?

-Ray Gralak
Author of APCC (Astro-Physics Command Center): https://www.astro-physics.com/apcc-pro
Author of PEMPro V3: https://www.ccdware.com
Author of Astro-Physics V2 ASCOM Driver: https://www.siriusimaging.com/apdriver

-----Original Message-----
From: main@ap-gto.groups.io [mailto:main@ap-gto.groups.io] On Behalf Of Craig Young
Sent: Tuesday, May 5, 2020 6:14 PM
To: main@ap-gto.groups.io
Subject: Re: [ap-gto] Adjust CP4 Clock Frequency

Dale,
Good point, I would like to continue investigating the discrepancy. To me some diagnostics are needed and I am
asking if AP would add them to the AE tab.

Craig


Roland Christen
 

Or C you are not polar aligned or D you have differential flex which I have seen in the larger SCTs in fact of this magnitude and more.
It is virtually impossible for the encoder to have such large errors- your error is off by 100x. I just ran a test of a random encoder mount, tracking at sidereal, counting pulses from a Renishaw encoder over time. The tracking rate is constant over a 30 minute period of less than 0.2 arc seconds. I then changed the tracking rate by 0.1% and recorded the error - I can upload an image showing the result if you don't believe it.

Roland Christen


-----Original Message-----
From: Craig Young <craig.young.m8@...>
To: main@ap-gto.groups.io
Sent: Tue, May 5, 2020 7:33 pm
Subject: Re: [ap-gto] Adjust CP4 Clock Frequency

I disagree Ray that this is the more likely reason.  If it were the reason then I would see a divergence between ATrack and APCC.  I don't, they follow each other in parallel which means there is a constant factor that is causing the difference.  The two likely constants are: a) the sidereal rate is not sidereal, or b) the encoder, if used to determine sidereal rate, is not calibrated.

Craig


Roland Christen
 


But that appears to be asking too much,
The software for that is not presently available in APCC. We have it here for internal testing, but never will it be available for general use and pirating by our overseas competition.

Roland


-----Original Message-----
From: Craig Young <craig.young.m8@...>
To: main@ap-gto.groups.io
Sent: Tue, May 5, 2020 8:07 pm
Subject: Re: [ap-gto] Adjust CP4 Clock Frequency

I don't think it is a coincidence Ray.  But it is easier to make that statement than to try and find the real problem.  Notice that it could easily be determined if we had a feature in APCC AE tab to run some diagnostics on the encoder and tracking.  But that appears to be asking too much, so it is simpler to try and explain it away.  What is the harm in adding some diagnostics?  Seems like a professional mount like the 1600 would give the user some additional tools to fine tune it.  Or is this too complex for us amateurs and should only be left to the hands of experienced engineers?

Craig