Telescope Reviews: Guiding Issues PHD and the CG-5 in DEC only

Ok...I think I know what is going on with your mount, because it does the same thing with mine. The problem is backlash and I have yet to eliminate the problems....only to try to minimize it.

I'll try to explain, but it may be confusing and long winded. First off, I will assume you know all about backlash and you have tried to eliminate most of it by adjusting the worm gear mesh....but even with the adjustments you cannot completely get rid of the backlash.

Note: Although cdndob says not to have any backlash settings setup, I do on DEC (but not in RA). I adjust my backlash settings to the point where there is no jump in my PHD graph when it switches from north to south guiding and vice versa.

First off...As for what you see during PHD guiding calibration....that is normal. PHD guiding winds the gears to clear the backlash when it first starts North calibration. It will send X number of pulses to reach the required # of pixel movement for the selected star. When it finishes the calibration, it IMMEDIATELY starts moving back in the south direction WITHOUT clearing the backlash and sends the same number of pulses in the south direction that it did in the north direction for calibration. When it finishes the X pulse it starts guiding. The difference in position is basically a measure of the residual backlash in the gears.

Now for the really confusing part.

You may have noticed looking through an eyepiece or on the display from your guide camera....if you switch directions from south to north or north to south, that the star may not immediately move, that is partly due to the backlash. When the star stars to move, it may move slowly at first, and then it starts to move faster. I think this is because when the backlash is cleared and the gears touch, the gears are not really fully meshed and loaded, it takes a bit more movement of the motor and gears for everything to get going.

For example (these are some madeup numbers): Lets say you have a star on your guide camera You had just moved it in the south direction. The star is at point Y = 0. You now send say a 50ms pulse in the north direction. The star doesnt move. Send a second 1000ms north pulse and the star moves 1 pixel (backlash is now eliminated). Third 1000ms pulse it moves 2 pixels....4th pulse = 4 pixels (it is now loading the gears)....you continue to send pulses and say at the 5th pulse, and every pulse afterwards, the star moves 6 pixels. The gears are now fully loaded.

I'm not sure just how Craig Stark programmed PHD guiding, but I dont think his calibration routine takes into account this "loading of the gears". It seems to measure the amount of pulses needed to reach X pixels of star movement when calbrating (with my DSI-1 X= 24 pixels).

Now for a hypothetical calibration using the numbers above. Each pulse is for 1000ms. Calibration starts when the star starts moving:
Pulse 1 = 1 pixel
pulse 2 = 2 pixels (3 total)
pulse 3 = 4 pixels (7 total)
pulse 4 = 6 pixels (13 total)
pulse 5 = 6 pixels (19 total)
pulse 6 = 6 pixels (25 total)

North calibration ends and then the program sends six 1000ms pulses in south direction and the program starts guiding.
Again....im not sure how Craig does the calibration math, but lets just say it calculates the numbers as 6 x 1000ms pulses = 25 pixels ... or ... 240ms per pixel

How about some hypothetical guiding with the calibration done above. The graph in DEC starts drifting south, the drift reaches the threshold for a guidepulse to be sent....say 1 pixel (again this is theoretical and not real world numbers...these numbers are exaggerated). Program sends a 240ms guide pulse to correct....but the gear isn't loaded! In the example above, it took an unloaded gear 1000ms to move the star one pixel! It may take 3 or more guide pulses to get that star to move one pixel. During that time the guidestar may have drifted another pixel because the gears were still being loaded.

You may see what is coming next...

Eventually, PHD will send enough pulses so the gear is fully loaded. But guess what? When the gear is fully loaded, it should only take a 167ms pulse to move the star one pixel! PHD calibrated things to the tune of 240ms per pixel! Because the gear was not loaded previously, it wasn't correcting fully, and now the star has drifted 5 pixels off axis. Gear is fully loaded now, PHD sends a 1440ms pulse, but because the gear is loaded, it will move 7 pixels, causing an overshoot to the other side of the axis

This is what I see happening when my guiding gets kinda screwy. Things will drift in one direction with PHD guiding not being able to keep up, and then it overshoots badly when the gears are fully loaded. I have never really been able to get rid of this completely. But luckily, the drifting and overshoots in my guiding has been under 1 pixels and doesn't ruin my subs.

I guess the only real way to get rid of the problem would be to try to get rid of as much backlash as you can. The other thing I can say to try is before you calibrate, make sure that your mount gears are fully loaded in the North direction. This should help get rid of the overshooting problem (theoretically), but wouldn't get rid of the insufficient corrections when the gear is loading.

Sorry for the long post, but I wanted to relay to you what I think has been going on with my guiding (and my interpretation on what is going on). If anyone knows if the stuff I said about the PHD guiding is wrong, please correct.

--------------------
Check out my Astrophotos on Flickr
C6-N 150mm f/5 Newtonian
CG-5 ASGT mount
Canon 300D self modded + IDAS LPS-P2 FF
Meade 70AZ + Meade DSI for autoguiding