In reality, only the dec belt needs to be tensioned as ra motor never reverses in use during guiding.
And since the dec belt inside is the most accessible to adjust I am unsure why ZWO does not put a pdf online!
Very large backlash in new AM5
a month later
- Edited
Got my new AM5 today. I downloaded its PE graph and inserted it into SolidWorks as a sketch picture. SolidWorks shows length and delta-x / delta-y of sketched lines, so it was a simple matter to fit lines and determine approximate slope and calculate its approximate worst-case PE as .17 arcsec/sec. Its slope is steep (~ 8.25), but its max/min PE are relatively low at 8.2/3.4 arcsec. The "partial zoom" isn't too radical, so I'm hopeful it'll guide well. More later...
- Edited
Interesting, it's actually closer to .13 arcsec/sec according to ChatGPT. This is possibly because I measured the slope of a shorter segment at ~ 8.5 and "cheated" to 8.25 for my initial calculation, whereas the slope I traced here is a bit under 8 (note that this is a newer "short period" AM5).
Here's a link to the ChatGPT analysis. Once you scale you units (whatever they me be; mm in my case), it's easy to plug in your own values and let the AI do all the work. Regenerating the AI's response sometimes yields nonsense, as it's evaluated different ways with each run. It's easy to pick out the wheat from the chaff, though:
https://chat.openai.com/share/754caa66-1c56-4823-a572-d7ade63aaaa5
tbhausen you will still need to test to make sure your slope is not greater at an area of less PE which is not shown in the partial. Mine is a bit higher PE, 432s period as opposed to your 288s period length. (Yes, some are not 432, most newer are 288s even thou the spec says 432s) and mine is almost as good a slope in the partial but elsewhere I have a steeper slope in a lower PE area that makes me have to increase my pulse duration from 22ms to 350ms to take care of it.
Unfortunately a partial only shows the worst PE, not the worst possible slope. Cheers and let us know how it guides! Even the bad ones are getting under 0.8rms but with a few blips. Your PE looks fairly sinusoidal in shape so it should be very promising!
- Edited
Kevin_A makes me have to increase my pulse duration from 22ms to 350ms to take care of it.
Hi Kevin... you may want to pay some attention when using 350ms for both declination and RApulxes, and guiding at a frame rate of 2 FPS.
The reason is this -- the guide exposure is 0.5 seconds. If that particular guide frame causes both RA and declination to use the full max pulse of 350 ms, depending on the guide program, this can cause the total time of the correction pulses to exceed 0.5 seconds, and if the camera is operating in synchronous mode (streaming), will make the latency to get the next usable exposure frame an extra 0.5 second longer. This could cause some large error for a frame or two.
Now, the mount protocol (the command language") that ZWO uses does allow RA and declination pulses to overlap. I.e., you can send a single command that says "move RA by 305 ms and declination by 270 ms" and the combination should finish executing in 305 ms if the mount firware does the right thing, and therefore not exceed the 0.5 ms exposure time.
However, I have not seen PHD2 make use of such a command. It usually sends an RA correction followed by the declination correction command sequetially, i.e., taks a little bit more than 305 + 270 ms to execute both corrections, and thus would add to no less than 575 ms and exceed the 0.5ms exposure time.
You may want to carefully monitor the guide graph's pulses to see if such a thing happens when you allow the two max pulse to go above 240ms or so when using 0.5 second exposure time.
The good news is that if the two axes are independent (i.e., no mechanical problem that will couple RA error to declination error, and vice versa), the probablility of both RA and declination gears acting up at the same time (and therefore need the full max pulses at the same time) is small. However, a wind gust can cause both to go nuts and guidng to go wild for a little while. It should be something that you can get a "feel" for by watching the pulses in the guide graph.
This is the reason why some mounts cannot use slow, calm 0.25x sidereal guiding, if the 0.25x sidereal guide pulses need to be twice as long as 0.5x sidereal guiding, and it is easy for the total pulse time to exceed 0.5 secs.
Guiding a strain wave mount is like being stuck between the frying pan and the fire. If you increase the exposure to 1 second from 0.5 second, you can then also increase the maxpulse durations -- but then your sawtooth amplitude goes up, and that causes its own problem :-). I am lucky that my RST-135 allows me to use 0.25x sidereal guide rate and still keep the two max pulses small enough (150ms to 180ms-ish) to not sum over 0.5 seconds.
Some strain wave owners would probably think this type of guiding is nuts, but I get it when it is calm:
Was trying to capture the Milky Way a little south of Sadr with a 100mm lens and full frame 6200. I will be trying tonight for the same region, but with the Samyang XP-50 and the APS-C frame. Even the Canon 100L has too much star distortion at full frame. It will be pretty much "first light" for the Samyang since installing an EAF.
Chen
- Edited
w7ay i am guiding mine at 1s Chen… not 0.5s. If i were doing 0.5s, I would reduce it by half. I will be trying guiding at 0.5s as soon as zwo make asiair available with adjustable guide rates for the AM5.
I get fairly good guiding still, but not very consistant and always around 0.7. I am in the process of finding out if my mount is one with a bad belt profile supplied. My asiair sometimes needs 10-12 steps of backlash adjustment during calibration before it starts to guide.
w7ay thanks for the synchronous tip! I do try to give some headroom and if i were guiding at 0.5s I would drop my pulses down to around 175 each roughly…. and it I could change my guide rate to 0.25x, then back up again to 350ms. Just passing the time away now until zwo gets back to me. I like my Rokinon 135 as it does not care about great guiding and 2 hours only gets me decent results even without much post processing.
- Edited
Kevin_A i am guiding mine at 1s Chen… not 0.5s.
That's 0.25x sidereal guiding rate, right? (I.e., 3.75 arcsec/second.) Which the ASIAIR does not support for their own mounts.
With your guide camera, you should be able to do 0.5 second exposure durations to achieve 2 FPS. That is the time interval that the sum of the two max pulse durations should not exceed.
Chen
- Edited
Kevin_A My asiair sometimes needs 10-12 steps of backlash adjustment during calibration before it starts to guide.
That is somewhat unusual for strain wave gears, which have very little backlash (at least the ones from Harmonic Drive LLC). Either the backlash is poor with the Chinese strain wave gears, or ZWO has some bad machining for parts other than the gear itself.
If you align the camera angle so that it is one of the compass cardinal points, you may be able to tell better where the error is coming from. Easiest to do by switching the ASIAIR to plate solve the guide scope -- the camera angle is right in the results. That's assuming that there is a convenient way to adjust the camera angle. This way the x-y errors the PHD2 report will correspond directly to only the RA or only the declination error (i.e., RA would just be the x error or the y-error, with no off-diagonal terms in the rotation matrix (which PHD2 applies after it measures the actual camera angle). I regularly set my camera angle within a couple of degrees of the cardinal points, not because it is needed, but because it makes it easier to analyse.
I like my Rokinon 135 as it does not care about great guiding
Yeah, that is one of the reasons I pulled the trigger on the encoder version of the RST-135 -- unguided short focal length lenses. The other is to simplify planetary and solar work where the mount only has to keep the planet/Sun inside a small ROI, since the exposures are so short (I have a Hinode solar guider for daytime, but that is just one extra item I can leave out -- can't handle complex thinking in my old age :-).
Chen
w7ay my calibration orthagonality is perfectly 90 degrees and these many backlash steps increased largely after the last asiair update so not sure if it is software driven and actualy not a mechanical issue. I was wondering if it was a result of lowering pulse lengths to 350ms from their nominal 2000ms pulses.
- Edited
Kevin_A I would like the option to try reducing the sawtooth patterns.
Methinks you would like it. The sawtooths really calm down.
But it won't help the sporadic jumps that you are seeing, though. That has to some mechanical flaw if it were not caused by wind gusts.
Next, they need to allow more flexible guide rates than the four that ASIAIR provides. This way, if the mount has too much error to use 0.25x sidereal, you can choose 0.3x siderate rate or 0.4x sidereal rate, which would still be better than 0.5x sidereal. Or, at least provide 4 more useful rates, by dropping the 0.75 and 0.9x rates.
Funniest thing is that I have seen people push an RST-135 to 0.9x sidereal rate, in the hope to correct faster. Based on the sawtooths, it is the opposite of what you want to do. to a mount with large slopes :-).
By the way, although I did not buy the RST-135E to use as an auto-guided mount, I will be spending a few months with the encoder mount in place of my primary RST-135 to further study the problem. Conveniently, its encoder can be turned off. I do not expect the encoder to be able to smooth out the higher harmonics, and therefore not really improving the autoguided numbers by any substantial amount, if at all. I.e., just because the p-p periodic error is now 5 arcseconds instead of 50 arcseconds, I don't believe that it means that I can start using 2-second guide exposures. It is not the p-p error, but the slope of the error that determnes the needed frame rates. But we shall see.
Chen
Kevin_A I am still wondering how i can have perfect 90 degree calibration with so many clearing backlash pulses.
Oh, the angles are independent of the backlash.
Notice that when PHD2 calibrates, it only moves one direction at a time. It never for example moves RA by 3 steps and declination by 7 steps. When it is moving the RA, the decimnation is set to a constant 0. When it is moving declination, RA is fixed to zero.
So, all the backlashed movements all still fall in the same direction.
As long as you calibrate at the ceestial equator, and close to the Meridian, you should see more or less rthogonal read and blue lines.
However, backlash can cause other problems. It can for example cause the scale of the RA to be wrong. For example, instead of 3.3 arcsecond in RA per pixel on the sensor (that is what "calibration" tries to determine), you might end up with a scale of 3 arcsec per pixel instead, and that will cause the correction pulse durations to be 10% off.
Robotic arms (RainbowAstro's parent company build robotic arms for a living; the Astro subsidiary is just a hobby of the CEO of RainbowRobotics) like to use strain wave gears because of the small backlash. But they also manufacture the arms precisely enough so that other parts ot the arms (bearings, etc) do not introduce backlash.
Chen
- Edited
tbhausen I don't even bother to unpack them now.
I not only don't unpack ZWO cables; I dump them in the thrash right away so I wouldn't be temptent to use them.
I did keep a couple of the USB-C ones to use as charging cables for my Logitech mouse :-).
ZWO is buying from some low-bidder sources, and apparently do not QC them for parameters such as impedance. Given ZWO's track record, I do not think they even own a network analyzer to be able to measure cable parameters. I do not think they even own a milliohm meter (needs a 4-point probe to be able to measure down to the milliohms) since I found their externally fat power cables to be made up of 21 AWG or 22 AWG wires inside when I measured them. I had posted my measurements in a thread that I started a month or two ago.
The cables that ZWO included are just trinkets, just like their keychains.
Chen