Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
PEC training
|
I followed the recommended PemPro PEC programming method, and so far have found it to make things worse. I'm assuming that it's operator error, me being the operator. I expect that it will improve things if done right, but I haven't seen that yet. The problem is that I'd rather spend time imaging when conditions are good enough for that, so PEC training doesn't get much of my time.?
I'm wondering why it did go wrong for me though. For one thing, I only recorded 1/2 hour for the PEC training instead of the recommended 1 hour. For another I'm not sure if the seeing conditions were good enough for PEC training. Does bad seeing have a big enough of an effect on PEC training to make it worth waiting for a good night to do it? One more thing is that I'm just not sure if the PEC training that I did, got loaded into the mount. I'm not sure how to verify that PemPro loaded it.? We're supposed to get some rain here soon, so maybe there's something I can do with PemPro connected to the mount indoors to verify everything is loaded, and the data is good.? Jamie |
|
Hi Jamie,
By "worse", what do mean?I followed the recommended PemPro PEC programming method, and so far have found it to make things worse. 1. Autoguiding appeared worse with PEC enabled? 2. Or the measured periodic error with PEC enabled was greater than disabled? 3. Something else? -Ray Gralak Author of PEMPro Author of APCC (Astro-Physics Command Center): Author of Astro-Physics V2 ASCOM Driver: -----Original Message----- |
|
Autoguiding was worse.?
There are so many variables with autoguiding that it's hard to tell what is going wrong, at least to me. So last night I was primarily interested in capturing SII data to complete a set of three targets before the rains come in. My secondary goal was to further an evaluation of PHD's new "multi-star" feature. Finally I wanted to test out the PEC training that I had loaded.? 1: So I got an awesome set of SII images. Primary goal achieved! Here's the first target processed:? 2: multi-star was doing great. I was getting one of the best sessions of guiding ever, yay! 3: enabled PEC in the mount to see if there was improvement. This one, not so good.? What I saw after setting PEC to "replay" in the GeminiII web interface was large jumps in RA and DEC.? I am using the PEC algorithm in PHD and I'm thinking that the PHD2's PEC training that I did without enabling PEC in the mount may have caused the bad results. I'm not sure how to clear PHD's PEC training, so I just set PEC aggression to zero, and reactive aggression to 100 for over 10 min after enabling PEC in the mount. I'm not sure if this re-trained PHD.? Like I said, PEC wasn't the highest priority item on my list for the night. If it was, I would have cleared out PHD data, and created a new profile etc from scratch with PEC enabled in the mount.? Jamie |
|
Jamie,
What I saw after setting PEC to "replay" in the GeminiII web interface was large jumps in RA and DEC.PEC only affects RA, so that you saw large jumps in Dec also indicates that the auto-settings were not optimal. That said, after programming PEC, you should always measure PE again with PEC enabled in PEMPro. This is necessary to confirm that the PEC curve is working. Also, note any changes to the amount of drift with PEC enabled. If drift increases substantially, that can have a negative effect on auto-guiding. Also, decide on a guide rate and stick with that rate (I recommend 0.5x). Changing the guide rate after having programmed PEC can introduce RA drift. -Ray Gralak Author of PEMPro Author of APCC (Astro-Physics Command Center): Author of Astro-Physics V2 ASCOM Driver: -----Original Message----- |
|
Thanks Ray,?
That's great info. Actually now that I think about it, I'm not sure if DEC was doing anything different. I was only concerned with RA as I know that is where PEC works.? Next time I'm working on PEC, after the rains pass, I'll do a 1 hour training session, and run PEMPro on the trained mount to confirm how it's performing. I had never thought about drift, so I'll make sure to look at that also.? Jamie |
|
Hi Paul i agree DEC looks pretty?good. again the amount of backlash will change through the night, primarily due to changes in your scope orientation (i.e., altitude) RA seems to me to show issues related to a bad PEC. you have a lot of back-and-forth looking at your raw RA, i can't even make sense of the calculated raw periodic error with is another sign of bad PEC my suggestion is next run please do a guidelog that includes a calibration run, and do not use PEC at all. when you calibrate, make sure it's near the intersection of the meridian and celestial equator On Fri, Dec 11, 2020 at 12:07 AM Paul Goelz <pgoelz@...> wrote: On 12/10/2020 10:55 PM, Brian Valente wrote: --
Brian? Brian Valente portfolio |
|
On 12/11/2020 11:46 PM, Brian Valente wrote:
RA seems to me to show issues related to a bad PEC. you have a lot of back-and-forthYour included image looks perfectly understandable to me. It shows RA errors at three important frequencies.... worm, worm X2 and worm bearing. It also matches the frequency domain plot I get in PECPrep. See my comments below. my suggestion is next run please do a guidelog that includes a calibration run, and do not use PEC at all.Please note that only a small part of that 100 minute run was with PEC on. For the bulk, PEC was off because I have always found PEC makes things marginally worse. The run began with PEC off. Somewhere near the middle of that run I trained the PEC, which leaves PEC on after the training ends. Including the training, it was on for a total of maybe 5-10 minutes at most and when it looked on the guiding graph like RA was the same or worse, I turned it off again. If you look at the entire run in the time domain, you will note that the RA looks different for a stretch in the middle of that run. I had a look myself at that log in PECPrep and I see significant peaks in the frequency domain at the worm fundamental and second harmonic. The highest peak by far, though, is at 76.2 which corresponds to "worm bearing (ball passes outer race)". The bearing signal at 76.2 is double the worm fundamental signal signal at 239.4 and triple the worm 2nd harmonic signal at 118.7. See attached screen grab. The worm fundamental (239.4) I understand. The second harmonic (118.7) would seem to indicate an possible alignment issue with the Oldham coupler? The worm bearing signal at 76.2 could indicate the worm block might be crooked and in need of alignment? Note that for now, the RA axis on this mount is 100% as-received. I have not touched it but I am willing to do the same adjustments that helped the DEC axis if needed. When I was working through the DEC backlash issue I checked the RA on a terrestrial target and found the backlash was around 2000mS. Since backlash is not an issue in RA I just left it alone. But it would be easy to loosen the outer worm block cover screw and make sure the block was well aligned if that might be responsible for the large signal at 75s. Today I am going to have a look at all the old PHD2 logs that are stored from past sessions and see if there is anything consistent in them. when you calibrate, make sure it's near the intersection of the meridian and celestial equatorYes, that is always how I calibrate. I rotated the guide camera yesterday to match the orientation of the main camera so the next time out I will re-do the calibration. And unless it is a really good night I'll also get about 30 minutes unguided to see what that looks like. I also discovered that the RA axis was not perfectly balanced. It was VERY close (only noticeable by how long the axis coasted after being pushed) but not perfect. The slight unbalance biased the axis to the east on the east side of the pier and to the west on the west side. slightly. I have since corrected that imbalance. Paul -- Paul Goelz Rochester Hills, MI USA pgoelz@... www.pgoelz.com |
|
On 12/12/2020 6:31 AM, Paul Goelz wrote:
Today I am going to have a look at all the old PHD2 logs that are stored from past sessions and see if there is anything consistent in them.Replying to my self..... I had a look at the other stored PHD2 logs and while I find the same worm, worm x2 and bearing signals, they are not as significant and guiding was more precise. That leaves me puzzled as to why the session we have been talking about was noticeably worse. The only thing that is different is that this was the first guided session using my new Polemaster. The Polemaster worked perfectly so the PA for this troublesome session was possibly more precise than previous (casually drift aligned) sessions. ???? Paul -- Paul Goelz Rochester Hills, MI USA pgoelz@... www.pgoelz.com |
|
Paul,
I reviewed your PHD2 log and here are my findings. At the beginning and end of the log, when you stated PEC is off, the main frequency components are 1x, 2x and 3.15x. In the middle of the log, when PEC is on, the 1x and 2x components are well corrected and there is an added 3x component that was not there before. It indicates you probably included the 3x frequency in your PEC curve in an attempt to correct the 3.15x component, but this will not work. This will cause a beat frequency between the 3x and 3.15x components (they will go in and out of phase, causing worse PE at its peak). Therefore, I would rebuild the PEC curve using the 1x and 2x components only. The 3.15x component can only be resolved by mechanical alignment, not PEC. Eric |
|
Sonny Edmonds
I love sitting in the back of the class room with discussions like this.
I'm beginning to confirm things I thought I was seeing correctly, but wasn't sure. I hope to be more observant of my RMS error [px], I know my RA Osc is very low, Often 0.0X, and never higher than 0.1X. I don't recall my total RMS error, off hand. OK, I'll shut up and listen more.... -- SonnyE (I suggest viewed in full screen) |
|
On 12/12/2020 2:26 PM, Cyclone wrote:
Therefore, I would rebuild the PEC curve using the 1x and 2x components only. The 3.15x component can only be resolved by mechanical alignment, not PEC.Thanks Eric, it had not occurred to me that PEC could not remove anything that was not exactly in phase with the worm rotation period. But of course that makes sense now that I think of it since the phase of the bearing signal relative the worm period drifts from one worm revolution to the next. I find the cost of PEMPro to be a bit prohibitive at $149 so I may end up either using the demo period to create a PEC curve or just leave PEC off and keep an eye on things. And of course it cannot address the major source of error.... the worm bearing. Actually, most of the time my guiding has been quite satisfactory. This last session was worse than usual and since the major culprit seems to be the worm bearing, that might explain why it was not present in prior sessions. It also might benefit from some adjustment. If I go by the Losmandy video, the rotation angle of the far end bearing block is not well controlled during adjustment and could easily be slightly cocked since the only thing that controls its rotation around the vertical axis is the bearing itself as you press the blocks together. Note that at present, the mount is only a couple months old and the RA is so far as-received. Since it usually works fine, I am not chomping at the bit to tear into it. At least not until I have a goodly string of starry nights ;) As received, I also had a DEC backlash issue that was easy to work on and solve because I could validate the adjustments on a terrestrial target during the day. But the RA axis needs stars to evaluate tracking and they are in VERY short supply right now here in SE Michigan :( Has anyone tried to evaluate tracking during the day using PHD2 on an artificial star while tracking at the sidereal rate? One worm period is what.... one degree? That means I could (barely) keep the star in my guide scope FOV for one worm period. If there was a way to load the guide log (guiding disabled) and subtract the steady state drift, it should reveal any inconsistencies in movement. ?? HEY.... the sun just came out! I think it is taunting me..... Paul -- Paul Goelz Rochester Hills, MI USA pgoelz@... www.pgoelz.com |
|
On 12/12/2020 2:26 PM, Cyclone wrote:
At the beginning and end of the log, when you stated PEC is off, the main frequency components are 1x, 2x and 3.15x.I had another think on this and I see what you are getting at. At the beginning and ends of the session in question, PEC was off. In the middle it was on and recently trained. However, when I load the log into PECPrep, it loads the entire log and the amplitudes of the three signals in question are the average for the whole session so I can't tell how they compare relative each other. Because it looks like it will be cloudy for the next decade here so I can't create another log, I'll try cutting the existing log into pieces and see if I can isolate sections with PEC on and with PEC off for analysis. Unless there is a hidden tool to do that in PECPrep? Paul -- Paul Goelz Rochester Hills, MI USA pgoelz@... www.pgoelz.com |
|
you can select log portions in PEMPro Log Viewer On Sat, Dec 12, 2020 at 12:16 PM Paul Goelz <pgoelz@...> wrote: On 12/12/2020 2:26 PM, Cyclone wrote: --
Brian? Brian Valente portfolio |
|
no one ever seems to think seeing conditions are ever a variable ;) On Sat, Dec 12, 2020 at 4:24 AM Paul Goelz <pgoelz@...> wrote: On 12/12/2020 6:31 AM, Paul Goelz wrote: --
Brian? Brian Valente portfolio |
|
>>>
Your included image looks perfectly understandable to me.? It shows RA errors at three important frequencies.... worm, worm X2 and worm bearing.? It also matches the frequency domain plot I get in PECPrep. Sorry i didn't mean that.? i meant they are extremely noisey - normally i see just a simple distribution curve and a smooth hump. Yours has many spikes in and around those periods. that suggests to me there is mis-alignment of your PEC which i would expect in a low quality PE correction, such as using PHD and recording it (the guiding adjustments happen after the actual event) I realize PEMPro isn't free, but if you are really focused on getting the most from your mount, it is a good investment. you just aren't going to get a high quality PE correction any other way (at least right now). On Sat, Dec 12, 2020 at 3:31 AM Paul Goelz <pgoelz@...> wrote: On 12/11/2020 11:46 PM, Brian Valente wrote: --
Brian? Brian Valente portfolio |
|
On 12/12/2020 3:26 PM, Brian Valente wrote:
no one ever seems to think seeing conditions are ever a variable ;)Not me. To me, seeing is a HUGE variable. When I went back and loaded older logs, I found similar signals but the overall frequency domain was a lot more noisy. That makes sense because the troublesome session from the 10th was probably the best seeing I have had in a month. The older session had a lot of random noise due to seeing. The end result is that the session on the 10th LOOKED worse because the signals were much farther out of the noise floor. In truth, they may be the same amplitude.... just more buried in the noise so they LOOKED lower. If I seem to be contradicting myself, it is only because I am learning minute by minute from this process. I know a lot more about what I am dealing with now than I did this morning ;) Paul -- Paul Goelz Rochester Hills, MI USA pgoelz@... www.pgoelz.com |
|
>>>
?I know a lot more about what I am dealing with now than I did this morning ;) that's how it should work :) On Sat, Dec 12, 2020 at 12:35 PM Paul Goelz <pgoelz@...> wrote: On 12/12/2020 3:26 PM, Brian Valente wrote: --
Brian? Brian Valente portfolio |
|
PS i hope that didn't sound snarky i meant it as a compliment. usually it takes me days haha It's a constant learning curve for everyone.? On Sat, Dec 12, 2020 at 12:37 PM Brian Valente via <bvalente=[email protected]> wrote:
--
Brian? Brian Valente portfolio |
|
On 12/12/2020 3:38 PM, Brian Valente wrote:
PS i hope that didn't sound snarkyNot at all. It got me thinking again. And it raised a question.... unless someone has a ready answer, I may end up posting this on the PHD2 list. My recollection from the session on the 10th is that peak guide error was occasionally as much as +/- 2" or maybe even a bit more. However, PECPrep shows it as something less than +/- 1". Is it possible that when using multi star guiding that the displayed guiding error (the graph) does not accurately show the actual guiding error? If the corrections are based on the average centroid of all the guide stars being used, I'm wondering if the graph is only based on the one guide star in the crosshairs same as always. But the guide star's instantaneous position is NOT the same as the average centroid of all guide stars, causing the displayed error to be worse than the actual error????? See what happens when it is cloudy with no prospects of imaging for as far as the long range forecast can see ;) Paul -- Paul Goelz Rochester Hills, MI USA pgoelz@... www.pgoelz.com |
|
On 12/12/2020 3:34 PM, Brian Valente wrote:
Sorry i didn't mean that.I suspect it is noisy because that was a log of an imaging session with guiding active. So there was a lot going on in addition to just the mount's contribution. That doesn't change the fact that the signals are there, but it probably does explain the noise. If I EVER see stars again, I'll do a run with guiding disabled to get a better look at things. But am I chasing a phantom here? PECPrep seems to say my guiding was excellent at less than 1" P-P. At the time I didn't think it was that good. Paul -- Paul Goelz Rochester Hills, MI USA pgoelz@... www.pgoelz.com |
Jamie Amendolagine
to navigate to use esc to dismiss