Do I need to do PEC of guiding?

[kirkster501]

^^^

What do you think - do I need to bother with PEC if I am going to guide anyway?

Thanks, Steve

[andrew s]

I don't have any experience with your mount but with mine, Paramount ME, I use PEC as it minimises the work the auto quider has to do. In controll enginering terms it is better to remove error before you use a a feedback loop that requires an error to have developed before you can correct it. Regards Andrew

[ncjunk]

Some say yes and some say no. So try it and see if it makes a difference.

If i could ensure the pec correction was aligned with the error correctly so as to remove it i would apply it. But i am too lazy

[ollypenrice]

I've never tried PEC with guiding on the grounds that I don't want to befuddle the system with conflicting signals. I must say that the best mounts I've used, Takahashi and Mesu, don't have PEC anyway. (Or I don't think the Mesu has and, if it did, I wouldn't bother!) As Neil says, experimenting is always the best way forward.

Most of the problems with guiding come from backlash.

Olly

[ian_bird]

Olly - "experimenting is always the best way forward."

Not when you live in England! Clear skies are so rare that you don't really have that luxury! You don't know how lucky you are!

Cheers

Ian

[ian_bird]

Actually - thinking abut it - it all depends on your focal length. If you are less than 600mm then piggy back guiding seems good enough with a good polar align. Over that - in my very limited experience - you need a good drift alignment - a really good mount and an OAG guider. But I am a bit of a newbie - so the experienced guys are who you should listen to.

[Shibby]

I've never tried PEC... You don't really need it if you're guiding. However, I suppose a preventative approach might be better on some occasions - when the seeing (or wind) is particularly bad, for example, and your guiding becomes less responsive.

I know what you mean about experimenting, Ian. I also find it hard to justify when I only manage an average of about 2 nights a month under the stars!

[Gina]

I've never had anything to do with PEC either. My guiding "just works"

[ollypenrice]

I'm not sure that PEC is really 'preventitive' with a guided system. It involves sending a guiding command based on an earlier error. The autoguider sends commands based on measured real-time error. Both follow their respective errors so PEC is preventitive in a somewhat limited sense, I think. The real issue is that the two sets of commands don't communicate with each other or, if they do, the conversation might go like this.

PEC I've just sent a pulse East. Leave it to me.

ST4 D'oh, I just sent one the same. Let me deal with it.

PEC STOP INTERFERING! We are two pulses out now and I can't do anything about it so you do it. Then leave it to me.

ST4 But you keep getting it wrong.

PEC No I don't!

ST4 Yes you do. If you didn't, why did they buy me? Ha! Got you. AH, one pulse west.

PEC I've got that one coming up in half a second, you idiot! Now you'll have to send one the other way.

Etc...

lly

[andrew s]

Not sure it is like that Olly. I was a control engineer and we used both feed-forward (e.g. PEC) and feedback (Guiding) all the time and a formal analysis will show they don't compete as you describe.

Feed forward is normally model based and corrects for errors the system "knows" will occur and so if no non modelled errors occur following would be perfect. Feedback requires an error to develop before a correction can be sent. This means that if you have periodic error it will have to cause some error before guiding alone can correct it.

However, both the model can be imperfect, repeatable errors may not be modelled and random errors will always occur so guiding is often required for long exposures.

Both PEC and guiding can make matter worse. If the model - PEC curve- is wrong, and it is not easy to get a good estimate of the PE, it will generate its own error. Similarly, if the gain (aggressiveness) of the guide corrections is too large it can lead to oscillations. To be effective both need to be set-up with care and require a good control system with little or no hysteresis.

On my Paramount ME I use two feed-forward systems PEC and PROTRACK . PEC corrects for the periodic error and PROTRACK corrects for telescope & mount flexure, misalignment of mount axis etc. and refraction. I also guide!

I know many just guide and that's fine by me whatever works for you. I find with long focal lengths 1.5 to 2m PEC & PROTRACK help. As a bonus with the TPOINT model that supports PROTRAC I get all sky pointing of about 13 acr secs.

Regards Andrew

[andrew s]

Sorry not sure what happened here it is again-

Not sure it is like that Olly. I was a control engineer and we used both feed-forward (e.g. PEC) and feedback (Guiding) all the time and a formal analysis will show they don't compete as you describe.

Feed forward is normally model based and corrects for errors the system "knows" will occur and so if no non modelled errors occur following would be perfect. Feedback requires and error to develop before a correction can be sent. This means that if you have periodic error it will have to cause and error before guiding alone can correct it.

However, both the model can be in error, repeatable errors may not be modelled and random errors will always occur so guiding is often required for long exposures.

Both PEC and guiding can make matter worse. If the model - PEC curve- is wrong, and it is not easy to get a good estimate of the PE, it will generate its own error. Similarly, if the gain (aggressiveness) of the guide corrections is too large it can lead to oscillations. To be effective both need to be set-up with care and require a good control system with little or no hysteresis.

On my Paramount ME I use two feed-forward systems PEC and PROTRACK . PEC corrects for the periodic error and PROTRACK corrects for telescope & mount flexure, misalignment of mount axis etc. and refraction. I also guide!

I know many just guide and that's fine by me whatever works for you. i find with long focal lengths 1.5 to 2m PEC & PROTRACK help. As a bonus with the TPOINT model that supports PROTRAC I get all sky pointing of about 13 acr secs.

Regards Andrew

[ncjunk]

I'm currently a control, which means nothing!, but i agree with andrew.

I would only apply pec on a system that i could guarentee alignment of the pec curve with the actual error. So i would require absolute position information via encoders. I would want to be sure that at x i am correcting for x error and not x+1, for instance.

Hence i never performed pec correction on my eq6 as there was too much room for error, i mean i probably would have mucked it up.

[Jessun]

Doesn't PEC simply alter the rate of rotation, meaning that if the recorded PEC is on the money then it's no trickier to guide with PEC on than with the mount rotating in the first place?

[ollypenrice]

Interesting. There's no PEC on the Mesu but there the great advantage is that there's no backlash. And on the Tak mounts there's no PEC. I dare say the PEC on the Paramount is well ahead of that on the only mounts I have which offer it, the EQ6. I'm not using the EQ sixes at the moment but perhaps I'll try PEC plus guiding to see how it works out once I dig them out again.

Olly

[andrew s]

This could be a very long thread!

Doesn't PEC simply alter the rate of rotation, meaning that if the recorded PEC is on the money then it's no trickier to guide with PEC on than with the mount rotating in the first place?

Yes it speeds up and slows down the rate in exact opposition to the rate changes caused by the PE so the net rate stays constant on the sky.

Interesting. There's no PEC on the Mesu but there the great advantage is that there's no backlash. And on the Tak mounts there's no PEC. I dare say the PEC on the Paramount is well ahead of that on the only mounts I have which offer it, the EQ6. I'm not using the EQ sixes at the moment but perhaps I'll try PEC plus guiding to see how it works out once I dig them out again.

Olly

Olly, I think all mount have some backlash but I am willing to except it is too small to worry about on the Mesu and Paramount ME. For a belt drive mount when you reverse direction the tension in the belt and the deformation it causes on the wheels it is on reverse and these will cause some (possibly very small) backlash. The Mesu wont have the major PE component due to the worm of a worm drive.

My conclusions are: if you can only have one form of control is it is best to close the loop on all errors via guiding, bad PEC is worse than no PEC, good PEC and guiding is better than guiding alone.

Andrew

[ncjunk]

Interesting. There's no PEC on the Mesu but there the great advantage is that there's no backlash. And on the Tak mounts there's no PEC. I dare say the PEC on the Paramount is well ahead of that on the only mounts I have which offer it, the EQ6. I'm not using the EQ sixes at the moment but perhaps I'll try PEC plus guiding to see how it works out once I dig them out again.

Olly

Theres no need for pec on the mesu as the error is small and slow moving i.e. easy to guide out.

With the eq6 as an example with 20 - 30 peak to peak error oscillating away nicely then any improvement would be a welcome one.

I quite like the tdm method of sampling the speed of rotation of the axis with an encoder and perfoming corrections on the fly. The high rate of sampling keeps the mount tracking acurately.

I may try adding an encoder to my eq6 one day

[ncjunk]

This could be a very long thread!

Yes it speeds up and slows down the rate in exact opposition to the rate changes caused by the PE so the net rate stays constant on the sky.

Olly, I think all mount have some backlash but I am willing to except it is too small to worry about on the Mesu and Paramount ME.

Andrew

The way the mesu is transmitting the force from the motors to the axis it has practically no backlash in comparison to the eq6 say. But i would assume there must be some very very slight backlash it's just small enough not to be noticable at our tolerance levels.

[andrew s]

For the mega nerd I can recommend Telescope Control by Trueblood & Genet. They cover all these issues in some detail and while the implementation parts are way out of date (my copy is 1997) the basic anaysis still holds up. For example confiming Neil's point above it says "The best approach is to avoid gear errors... ...[by] using a Friction Drive. Such drives are still subject to periodic errors, but they are small and of low frequency, because the high frequency errors (e.g., small bumps on the roller or drive segment) tend to be ground down over the yers with use. Friction drives have been used successfully on telescopes up to 3.5 meters aperture, but may lack the stifness required in some application."

I am that nerd - Andrew

[melsky]

Thinking about it I guess it's very important if you're using PEC to turn if off before calibrating your guider other wise the guider won't have a constant reference.

Mel

[andrew s]

Thinking about it I guess it's very important if you're using PEC to turn if off before calibrating your guider other wise the guider won't have a constant reference.

Mel

Not at all leave it on otherwise the PE will add an error to the calibration.

Andrew