SatNOGS rotator Version 3

For the last two weeks myself and Agis ( @azisi ) have been trying to specify the required values for the third version of the SatNOGS tracker in order to design it. First of all we started by measuring the current version characteristics. The theoretical value of angular velocity is ω=150steps/s * 1.8deg/step / 60(ratio) = 4.5deg/s
and the one of torque, given that the steppers we are using have τ = 59Ncm, is τ = 354Kgcm (gear ratio 60:1).
However, practically we found out that torque was approximately 38kg
cm measured at 24V, 0.5A and the drivers’ Vref set at 1V. (At this point it is worth noting that the chinese pololus had different torque in each direction, and we had spent many hours before we understood why this was happening)
Some of the reasons we are not approaching those theoretical values are

  • The combination of stepper motors and stepper drivers don’t work
    properly.
  • Lot of friction between moving parts, the lack of bearings.
  • Oscillation is created because of non smooth movement together with
    elasticity of the box and the axis.
  • Worm gear is not optimal designed to have the maximum efficiency
    given the materials it is made of.

So we made a research on the net to find other commercial and DIY (only the documented) solutions. Because this forum doesn’t support tables, here is the link to an ethercalc with our results
https://www.ethercalc.org/v3specs

After this, and since all the information we could find about antenna mounting and tracking stations were scattered around the web and many of them were not so scientific, we made our analysis about torque, nominal and stall, angular velocity for ALT and AZ and accuracy.

Torque
The greatest force the tracker needs to withstand is the force created by strong wind. The worst case is when one antenna is elevated at 90 degs, facing the direction of the wind. We based our calculations on an article found online after comparing it to others. We “translated” the second table in metric (because we don’t understand imperial :stuck_out_tongue: and because we needed same units system in our calculations)
https://www.ethercalc.org/windLoad
and we applied the worst case model (EIA-222-F) in 3 different antennas: in the biggest one of our designs, and in two others, for which we obtained data from yaesu G800 rotator manual at page 3. We assumed that antennas are mounted in 1m away from the azimuth axis. For our antenna with 2m length (actual, not wavelength), made by 2cm square tube, the generated torque was ≈600Kgcm. For the 144MHz 10-elements Yagi from the article is ≈6000Kgcmand for the third 430MHz, 12-elements Yagi is ≈1800Kg*cm

Moment of inertia
Now for the moment of inertia: (for all installation methods we assumed that antennas are counterbalanced in the elevation axis) the worst case scenario here is to use two 3kg (our designs are less than 1kg) back mounted yagis with 3kg counterbalances both mounted in 0.75m away from azimuth axis. The torque you need in order to accelerate this system from ω=0deg/s angular velocity to ω=5deg/s (the math about angular velocity is below) in one second is about 60kgcm.
Note: we suppose that the mass of antennas is near to the altitude axis, so the torque of this axis that is needed to accelerate is approximately 0.
M1: torque of Azimuth axis
L: length of center of mass of antennas from azimuth axis (0.75m)
m: mass of antennas and of counterweight (3kg + 3kg = 6kg)
I: moment inertia
a: angular acceleration of azimuth axis 5deg/s^2
I = I1 + I2 = m
L^2 + mL^2 = 2mL^2 = 6.75 kgm^2
M1 = I*a = 6.75kgm^2 * 0.087rad/s^2 = 0.58 Nm = 5.8 kgm = 58 kgcm

Angular velocity
(How well do you remember trigonometry?)For the angular velocity max needed in altitude axis the things are straightforward. The closer is the satellite the larger the velocity. According to the wikipedia article about LEO, the lowest height limit is 160 km and the speed unit to orbit earth in this altitude is 7,8 km/s. As a result, maximum velocity in ALT axis is 2,8 deg/s. In ALT AZ rotator design there is a well known limitation: the closer something passes near zenith the biggest gets the velocity of the AZ axis. Therefore, we have analysed this problem to figure out the optimal velocity and how high we are allowed to track a target in relation to AZ velocity. The picture below illustrates a ground station B which tracks a satellite Γ in X degrees altitude. The satellite velocity at this point is vertical to the screen (page) plane.

The equations that lead to maximum altitude at which we can track in relation to AZ angular velocity are
ω : angular velocity of AZ DOF in rad/s
H = ΑΕ + ΕΓ : Minimum Height of LEO, 160 km
R = ΑΕ : Radius of Earth, 6500 km
u : linear velocity of satellite that rotates in 160km height is 7.8 km/s
ΒΔ = u / ω : ΒΔ in km
α = atan(ΒΔ / R)
δ = π - α
γ = asin( sqrt(R^2+ΒΔ^2) * sin(δ) / (H+R) )
ά = π - δ - γ
ΓΔ = (H+R) * sin(ά) / sin(δ)
χ = atan(ΓΔ / ΒΔ)

Below you can see the plot of the equations mentioned above, where horizontal axis represents angular velocity (ω) in deg/s and vertical axis shows the max track altitude (χ) for lower bound of LEO.

After studying this diagram, we came up to the conclusion that an angular velocity of 5 deg/s is adequate. For this decision, we took into consideration the main lobe of antenna (Δ3db) which in most situations is about 20 deg.

Together with the above mentioned specifications, we would also like for the 3rd version of SatNOGS rotator to be:

  • inexpensive (less than €300, if possible)
  • lightweight and portable
  • rigid and durable
  • easy to build and fix (try to use easily available materials)
  • weatherproof
  • electromagnetically shielded, so that noise in reception is reduced
  • accurate (backslash reduction and use of encoders at the axis)
6 Likes

The value given for torque is holding torque, e.g. the dynamic torque is somewhat less.
Have you tried using 1/16 microstepping? This greatly improves possible speed and reduces ringing. Then we can realize higher speeds and higher gear ratios at no higher cost.

I think portable and rigid are kind of opposite demands.

I added the Itallian Prosistel to the list of rotators for comparison. I added tax to the prices for equal comparison.
The backlash of them is about 1 or 2 degrees, but the position feedback is 0.1 degrees. Accurate feedback is only useful with very high gain antennas with narrow beam width.

I can assist if required as I own a prosistel, and a SPID X-Y along with a friend who has a yaesu G5500.

It might be a good idea to add braking torque to the list of items for comparison.

There is a comparison of some azimuth only rotators here but it may be biased to this supplier.
http://www.prosistel.net/home-e/tabcomp/tabcomp.html

In relation to position feedback I have done a lot of work in this area. There are basically two systems.

  1. Absolute position feedback, where the controller is told the actual position. (Normally a pot or digital)
    The POT is very common, but the digital seems to be only used in very expensive commercial systems.

  2. Non absolute
    a) Stepper guesses how far it has moved and assumes it has not missed a step
    b) DC or AC motor with a reed switch that sends 1 pulse per motor rotation before gearbox.
    c) DC or AC motor with a electronic switch (normally hall) that sends 1 pulse per motor rotation before gearbox.

The problem with non absolute is that if you ever miss one pulse (it does happen in the real world) then the position is wrong from then onwards, until the unit is manually calibrated again.

There could be a way to recalibrate automatically, as performed by many 3D printers with stepper motors, is to have a limit switch at the very end, and when it reaches this it is at a known position. With 3D printers they do this each time they turn on, so they can start at a known position.

Regarding design, if you are using absolute position, or automatic calibration, then the best motors to use would be DC motors for cost, torque etc. I do not think the accuracy of steppers is needed for VHF/UHF use, but they are useful for non absolute feedback based on counting steps. DC motors also have the benefit of less RF interference when operating.

I think it is very important to keep the design simple, but if you intend leaving the system operating unattended for many months, then it will need absolute position or automatic re-calibration.

As already said a portable system and an automatic ground station are different things with different needs. For portable you can manually calibrate it, and probably would HAVE to manually calibrate it, as you would not know which position it has been put on the ground. (The north would have changed)

Great work!!

PS My algorithm for achieving maximum coverage (taking into consideration physical constraints) is almost finished and I hope that it will be published till Sunday, so your measurements will come in hand!

A friend, that wants to apply for Socis at Satnogs, and I have given the position Sensor some thought too. However absolute Sensor are either Bad (Bad precision, limited cycles) or quickly become really expensive.

Basic potentiometers work quite well in this application. You can buy slightly more expensive 3 turn or 10 turn precision potentiometers

If the potentiometer goes faulty you can detect that, and stop and give an error. Many commercial units do that, basically if they should be moving, and there is no change in feedback, they stop. The same is true for any feedback system.

There are some fairly low cost digital precision units for pedal sensors for car accelerator pedals, and others based on just a digital hall chip near a magnet. For digital it would work out as 20 to 50 Euro for a set. The more you gear them down afterwards the more precision they gain, but they would need to be multi-turn.

Absolute position feedback: Youre right. Although “rather” expensive (5-10) Euro per good multiturn potentionmeter it looks like the best way. I found them cheaply in one store (http://www.reichelt.de/Mehrgang-Potis/534-1-0K/3/index.html?&ACTION=3&LA=2&ARTICLE=2540&GROUPID=3138&artnr=534-1%2C0K) however on digikey, rs, farnell, they start at 12 Euro, and for some weird reason the 10 turns are cheaper than the 3 turn ones. I am looking into the eletrical design, to adjust the measured signal with an opamp so you can basically use any potentiometer and easily adjust to reach best precision with arduino. Anyone wants to to take on the mechanical design for this?

Rigid is something that the tracking box regards, portability is something that regards the Tripod. I think we can observe them as two different systems. For a permanent setup I would directly attach the mast somewhere and not the tripod, then the setup can become rigid.

The conclusion I came to is for a rigid setup we must use solid booms and masts. I would say that a 40mm galvanized aluminium diameter would be a reasonable decision for starters for the mast and it doesnt push the price very much. If you want to be lightweight you may still use a PVC pipe on the trackingbox.

Normally they just connect one side of the pot to 5V, the other to ground, and connect the wiper to an analogue input. For a 3 turn or 5 turn an op amp might help but normally they do not use this. We do not need very high accuracy as the beam width of the dish is so wide. Most a/d converters have at least 256 levels. Some of the fancier systems can use look up tables as the pot may not be linear.

With a wide beam width dish it should be good enough.

Yes, thats what I am looking into at the moment. Basically most 1 turn potentiometers fall out, and for some weird reason I find 10 turn potentiometers are more available for me and cheaper than 3 turn ones.

If someone is into doing mechanical stuff for the potentiometer drop me a message, I really dont know FreeCad or else.

Adding to the discussion stepper vs dc motor:
Of course with stepper you have worst EMI noise, but through commutation you also have the issue with dc motors. I find the key difference is that dc motor forces you to have a closed loop, while when sticking with steppers you can do both. I think there is some potential we can still use with dc motors (1/16 step mode) which should bring improvements in speed, ringing effect and torque.
I am really curious about your results of dc motors vs steppers, looking forward to that!

Doesn’t related to mechanics, but general suggestion to stick to single sizes for bolts. Just ordered bunch of hardware from aliexpress…It wasn’t easy as it might seem. Just see how many standards are used in v2 guide. See “relevant parts” section.

Hello, we try to fix this in the third version but sometimes is difficult to avoid all those different lengths.

Hello, this is my first post in this community. Exist documentation for rotator ver.3? Im printed 3d parts yet worm gear (involved part for 3D printers :slight_smile: ) and axis gear all in ver. 3 and I do not know yet how passed in other parts. Would help me either snapshot fully assembled rotator box ver. 3 or drawing assembly. More thanks. Dalibor OK2JKD

Sneak peek! More designs and info soon :smile:

mmm…I see…new version worm gear in git repo…previous will need sharpen :smile:

thanks of photo, Dalibor Ok2JKD

More sneak peek!

Things are looking great overall. @azisi & @manthos are working non-stop on fine-tuning the design.

That looks new and fancy :smile: Keep up :slight_smile:

1 Like

Moar SatNOGS rotator v3 sneak peek.

:smile:

very good boys :), wholly different construction than with “ABB” box. It is not the E tube loose in beering?