More Barn Door
Relative to the post here, I’ve been doing some more work on the beast…
With the code as written with the additions to shut the system down at 60 minutes, as well as the modifications for interrupts (you are reading Nick’s site aren’t you?), on my rig after an hour I’d got a drift of 0.8 degrees over what I was expecting. For the hinge/drive centre at one hour on a right angle triangle I was expecting the height to be 61mm for a base of 228.5mm and an angle of 15.04 degrees.
Instead I’d got a height of 62mm and an angle of 15.84 degrees. That in itself isn’t a killer thing as just throwing one rotation per minute on a 1mm pitch rod would give 60mm height and an angle of 14.7 degrees… So what’s that saying?
It’s saying that the motor is turning slightly too fast. I’d got a couple of suspicions that this would happen after reading the Arduino forum about this motor (here) that these motors weren’t geared exactly 64:1 and in testing it seemed to be, by eye, a fraction out per rotation.
Off to recalibrate then… Using the calculations found on the Arduino forum, the doublestep rate seems to be closer to 2038 (it’d probably work out a fraction lower) than what the spec sheet says of 2048. And with that loaded in to the parameters, the starting position recreated exactly, it’s been popping away for the last hour to open the drive out and give me a nice, finished angle. And… Managed to get it down to 15.19 degrees at 2038 doublesteps per rotation.
A bit of maths on the gearbox of the motor would give a step ratio of 4075,77, not 4096 as expected, and stepping that down by two to 2037.8 and change.
As I’m stuck home at the moment, because of the virus (not that I’ve got it, but I’m trying to avoid it, given that my work stuff involves me being alive as long as possible when others could be faltering) and the government have curfewed us, I’ve a bit of time with the testing. Sure 15.19 degrees an hour is very tight for something loose as this, but why not see if it’s possible to get it tighter? Lets find out…
When that test had finished I was at a loss to why the measurements were calculating to about 16.8 degrees, so I set back the code to 2038 and went out on to the balcony.
First caveat is my balcony is south/west facing so with being under curfew, albeit sunset is just in front of 18:00 now and curfew kicks in at 22:00 I could go up the field, polar alignment isn’t possible and with not knowing if there is drift then trying to drift align was out the window. So it was compass and angle as best I could, and got reasonably close; enough to track the 200mm facing west at 1 minute for a few shots, and then 70mm for a few more but it would go out to lunch about 1 minute (hell, that’s still a lot longer time than using NPF).
First thing I noticed with the images was that every so often there’d be a noticable curve or the stars were double; perhaps one frame out of three or four, that was a bit curious and I couldn’t see any reason why this should be happening. But it was, and after checking there was nothing binding or catching (I initially thought it was the power cable out the door to the phone charger), nothing seemed to “fit” to what was going off… Ay well, more debugging later.
Another thing I did; I put a plate (actually it’s PCB but it does the job) on the bottom side of the top bar for the acorn nut to roll over. There’s a bit of talk in some areas to whether a bearing plate is needed or not as metal will move smoothly on flat cut wood. It’s not a bearing plate…
What I noticed as I was doing a bit extreme with things (a 70-200 on a large SLR, that’s perhaps on this thing’s limit if it wasn’t west facing) was that there was a small indentation forming at closed position where the nut was touching the wood and probably channeling down as the drive was moving (though the indentation was more noticeable with the finger); so I can only assume that this plate is more for distributing the pressure of the contact point over an area instead of having it pressed in to one small area of something that is a lot softer than hardened steel as an indentation would create a small jitter in the first few turns as the nut overrode it. It wasn’t this, though, causing the wobble problem but after a few rewind sessions, the first image(s) off were showing less in the line of malformed stars.
Next thing that hit me, after doing a run and rewinding the drive; if you don’t get the drive to the exact same point, the theta0 calculation in the code goes out. Not by much though, and probably not enough to affect timings in the slightest (if a reset starting point had 5º of angle on it over closed then perhaps yes, but a tenth or two of a degree not significant). It has, however given me an idea on creating a closed hard stop where the pressure is off the drive, could put a switch on it to tell the MCU to start the step interval proper, but then I’d need more pins… However, a hard stop would have two benefits over what it is now; one would be to split the weight over a larger area when the camera is being set up and the other would give some leeway to when rewinding as the motor straining isn’t a nice sound when it binds on the base nut. More on currently theoretical stuff in another post.
So, back to these odd blobs in the images, and the weird angles… In the previous post I mentioned that the hinge was a bit sloppy but when weight was applied it cleared up; these hinges were out of a box in the basement and they seemed to be bearing loaded hinges, but after an hour or so on the balcony my head was in a different thought pattern.
I took the device back in the office and with the weight on it I was turning the rod by hand and every so often there’d be a slight but a noticeable knee jerk across the hinge itself, certainly enough to cause image issues. Curiosity got the better of me and at just after 01:00 this morning I set about stripping the hinge down to have a look. It turned out it wasn’t a bearing hinge at all, and the stiffness in it was nothing more than axle grease. In the centre of the hinge was a bit of pig iron and the bearings purely decorative. After clearing the grease off the centre pin, it was noticeable that there was a clearly defined moulding ridge down one side and it wasn’t straight; so I reseated it and had another look without the decorative bits. Sure enough the hinge was jumping a little and and with the eccentric nature of the bar, the centre point was changing across the hinge and not smoothly as it was snagging on the ridge, albeit this change was hidden to start with because of it being wrapped up in decorative metalwork and useless bearings. So off to spend a few more notes on it at the local builders merchants today.
Alright, enough waffle… You were probably looking for some images, yeah?
Here’s a few off it from last night; not really focused and without an accurate polar alignment but…
First off, it’s 200mm for 1 minute straight off the camera with tracker.
Second is 70mm for 1 minute with tracker, notice the foreground blurring.
Third is 70mm for 1 minute without.
So whilst still not perfect, not by a long, long margin; it’s showing a load of promise and hopefully in the next few days, time and work allowing, I’ll be able to get on and tighten down the niggles.
Another thing that is on the niggle list, is the insert nut has a lot of thread play (these nuts aren’t designed for this type of purpose) and when there’s weight on the drive, the drive is angling back ever so slightly and causing some slight eccentricity in the rotation which in turn is making the top plate move up and down slightly as the angle changes. I think the solution to that one is angle the guide rods opposite to the movement over the drive rod so it is creating counter tension and keeping the rod straight.