What is the service life of a stepper motor working 24/7

[TimB]

I'm back on a really really old project

The system I last tried while looking very promising has turned into a pile of poop. Just way to inaccurate.

Just so you know. I'm trying to measure the amount of gas in a volume of water

Volume = ~100ml
Gas range = 0.0 to 10ml

Accuracy required
Water volume 0.5ml +/- 0.5ml for the absolute 100ml. In between 0 and the 100ml marks +/- 5ml

Gas volume 0.1ml +/-0.1ml across whole range

I have tried so far
Capacitive sensors
Floats and magnetic sensors
Inductive sensors
Flow meters with optical sensing of the bubbles as they pass in a tube.
Optical image measuring of tubes containing the liquids and gasses using a camera

Of those and other methods only the last, the camera system has proved to be accurate. Infact its accurate to 0.05ml in some cases. However and this is the kicker for this system the camera has to be mounted a fair distance from tubes so as to view the whole length.

Now I'm going to try another system that I know will be accurate and that is to use a stepper motor to move some optical sensors that look at the tube and can follow the water levels as they change.
This can be done in a much more compact space than the camera system as they are mounted mm away from the measuring tubes.

So back to my question. Any idea on the service life of a stepper working 24/7? Really there will be virtually no loading. So current drive levels can be kept low.

5 years would be good.

Edit, do what I should have done eg searching for an answer shows its just based on the bearings

Thanks
Tim

[trastikata]

Tim, this is not answer to your question, rather side question - have you thought about ultrasonic level measuring?

[ken_k]

I don't know the life of a stepper I've never worn one out.
I guess you will have to convert the rotary motion of the stepper motor into linear motion with zero backlash, it might pay to examine some old floppy drives and CD DVD drives for some inspiration, some sort of home position mark will be required for calibration or maybe some other form of absolute position measurement.
https://www.hackster.io/ingo-lohs/check-a-stepper-motor-from-the-dvd-rom-25ed1f

ken

[top204]

If the camera system worked well, why not try a different mechanism to an optical process?

A row of very, very small SMD photo diodes touching each other, side by side, on a PCB will give an excellent resolution. Then scan each photodiode to see which one has a difference. Or one of the dedicated optical sensors as used in digital scanners. The photodiodes or optical sensors can be almost touching the tube.

Stepper motors have a problem with heat build up because they cannot be released, otherwise, they will simply slip. Why not try a linear activator, as in the types used for CD players or DVD players. Although, they will also need to be continually activated, otherwise, they will slip down. Maybe a gear mechanism will stop slippage when the power is released to the motor.

You could create a ratchet system, so they will move and not slip when the power is removed, but you will need a release mechanism for it. I've seen some of your metalwork Tim, and I know you can do it very nicely. :-) This will give a stepper motor a long life span because it will be off more than on.

[TimB]

trastikata No not personally tried ultrasonic. I did look long and hard at it but the accuracy was not good enough and due to the volumes tube has an ID of 11mm


ken_k I have seen those Steppers on sale but they were too short however that is the idea. If thy came with 110mm shafts it would be great.

Les
First off below is an image of the optical version It uses a camera to view the water levels. Ruler for scale

I looked at changing the optics of the camera version but could not find a solution.
I have a IR opto diode design. (see other image)
I got the PCB made and all the parts but never got round to populating it. With the likes of JLPCB they will populate it for free. I purchased all the parts from them anycase.

As I need to measure 2 tubes I will need > 240 opto Diodes 

I am always looking for the simplest solution.

One thing is I got some 3D printed parts from PCBway and was very impressed with the quality. I'm thinking now to get not think machining from ali but do the whole thing in just 2 parts reduces chances of leaks.

Nothing is every easy with this project.









 

[trastikata]

trastikata No not personally tried ultrasonic. I did look long and hard at it but the accuracy was not good enough and due to the volumes tube has an ID of 11mm

Tim, here's another idea -  you can measure the pressure created by the gas in the volume tube using a ported analog pressure sensor - excellent ones can be found at Freescale, then your accuracy and resolution could be very high.

P.s. in both cases for such a high accuracy I think you will have to take in consideration the changes in the atmospheric pressure.

[TimB]

Hi

Tried pressure transducers. They drift too much

[ken_k]

Hi Tim
Lots of people make CNC machines these days it might pay to try a few component suppliers.
This part is cheap and it may be possible to use it in a proof of concept trial.
https://www.ebay.com.au/itm/324245991314?hash=item4b7e912f92:g:uwkAAOSwpJhfItVB
for a long life product a rotating ball screw would be good.
https://en.wikipedia.org/wiki/Ball_screw
If mounted vertically and there was enough weight on the carriage the load would always be on one side of the thread so a simple screw and nut should not have a backlash problem.

Or maybe ACME leadscrew
https://www.ebay.com.au/itm/372243673900?_trkparms=ispr%3D1&hash=item56ab73b32c:g:zh0AAOSw6~VbuFoB&amdata=enc%3AAQAGAAACkPYe5NmHp%252B2JMhMi7yxGiTJkPrKr5t53CooMSQt2orsSafTQYbq3L7RBVAMi0K9cw58sU0aXRVxGz9ZsjlIYjjUG6Izzhq03larRDZyVt%252BAGjGBHy%252F8ieSq2Ok6GO8UqMqB2rr%252BVq8rlBoIUvwzc185TTZYDfSPJdFEDDTn9BjoMyLBZZ%252BwEYOgLAlfJrn0KKhCHhtnQZkx3M11xKTqQN31bYyU%252FMIe4i213gKCbJBJ5freLYxQbbdtyBDwfzDEYnAnXR0AToSvmk2APAXY%252BXrbum4YjrEn%252FUiD38NakMEIpzP2%252BuXgAfb2HW2kbuC4%252FZ3o4U9mKHGZpp72ejU3ZDI4t9igkHazebZfpj2THavS0JLeIm7jhKoFdHgxFD5HIYCZsPUcNxXUzLWkK%252F%252FHlKqoHAF1Qwt9JHNCqvtDbbi1JGkSeoNbsatBiOmspbcc2I79Ho638kRg4%252FrBU%252FJIRi3mVreOVS%252FkuidknPRA6enyhUy9xtLT%252F%252BaGESIR8rd9fM41owtwsu2kqov1hjxMKfHrnPH55M3pYwNOK5%252B4IRDf8CCFd%252FnFohWMntib5mKCB%252FhoXgCf4GmkLXi1Cb6T0Ph8aMhgUtk%252BoZV1rLaIgJmjeRWBCl7bdOYrWXjj4sKwGGI9AlExDS5nVIKr3g5q2xYJZHNekKbU%252F%252FP8WVRr7hESTSMAED86Wn2CcAvpJERZ%252BGreOp8h8u00Zi0pIPXXp%252FdkaHqw4Ph06RqjLIGRxB5ZqEivScmPoFCC3COdUumHSPoaota9IuznEjxwJETT122v6DKZhsjj%252FUJ5sPAdKFscIVWCciFYuqq%252Fq%252BaPjIMEksStEd39rnlW8n5a4BDNcQuv6LWkFgy4K%252F425%252BQe6uD%252BT%7Campid%3APL_CLK%7Cclp%3A2334524&frcectupt=true

[TimB]

I dug out my PCB with the Ir detectors on it. It turns out that 0603 while 0.8mm wide have to be mounted >1.0mm apart so I have 96 over a 120m range which reduces the resolution to > 0.1ml

I really like the idea of a one PCB system but looking at the amount of parts eg > 240 ir detectors it is frightening.
(I have two tubes to measure)

The camera can view both at once and I really like the system. However you have to mount the camera far away to view the 100mm I need. I also have distortion.

Now if I could get a custom designed Fresnel lens....

I going to look hard at a stepper if I can get a motor and acme thread small enough.

[JonW]

Hi Tim

Is the fluid always the same ie clean water?  What about making a capacitor from copper traces on the tubing and connecting it into an oscillator then measure the frequency change? 

Jon

[TimB]

Hi

Yes tried that. There are some very sophisticated capacitive to freq to spi chips. I will have to check my design files as I used one. The issue as with the pressure is they drift too much. Work ok to measure a change but not to measure absolutes.

The only system that I have found that actually works is optical. There are a whole host of other issues with the system but I ironed them out.

BTW I have tried floats but the float sticks to the wall of the tube via surface tension so no good for small volumes.

I had pinned my hopes on a flow meter. They state +/- 5% to 10% (depending on flow) error. I build a system to work out the gas volume based on the flow and the time I saw the gas (optically) This worked well and as I'm interested in the ratio between the 2 so errors would cancel themselves out. However I found that the errors were way more than 10% and were not ratio metric some of the time.
The test rig is always an issue so tests take time to perform. I have given up on that method now.

I know camera image analysis works, and it works very well! 0.03ml resolution and 100% repeatable. However due to the distance you need to mount the camera away from the sample tubes it look bulky. I want nice and slim

The Photo Diode system will work but the population of so many photo diodes on a board is putting me off.

I could use a CIS scanner module from a scanner but it would be a bodge.

Tim

[top204]

A few optical concepts that may give you ideas Tim.

1. A fingerprint scanner... Both optical and close range, but not a very large surface.

2. A photo diode bar as used in photo scanners... They come in several resolutions and have quite a large surface and can be placed close to the tube with a suitable lens or prism. They are also compact, with all the pins coming out of the unit itself, so easy to mount.

3. A small LIDAR... The rotating laser can detect extremely accurate resolutiuons. There are quite a few LIDAR and YLIDAR units available with varying minimum distances.

[TimB]

First off you have to understand the tube is 100mm long. Any sensor needs to see the whole tube all 100mm of it.

If you look at the image below you see an issue that you get even using the best system I have to date. That is that you endup viewing various edges at one time.

Any image sensor will have that issue.

Bar sensors are generally short and no longer available off the shelf. You can buy refurbished CIS scanner modules from scanners but they do not have data sheets to explain how they can be interfaced to.

The camera system kind of works.
The ideal would be a loads of ir sensors pressed against the tube with an ir light source behind it as the light level is reduced by water and can be detected.

However as can be seen from a previous image it requires loads of ir detectors/diodes multiplexers and I still do not get the resolution I need.

Moving 3 sensors to follow the water level is another good option hence my question on the life span of steppers.

[trastikata]

Moving 3 sensors to follow the water level is another good option hence my question on the life span of steppers.

Hope this helps.
https://www.applied-motion.com/sites/default/files/Stepper%20Motor%20Life%20data-110817.pdf
https://www.pmpl.co.in/pmpl_reliability.pdf

[TimB]

I'm going to try this board

https://www.robotshop.com/uk/jsumo-xline-16-sensor-array-board.html

Just to see if I can detect a range of values as the water level moves between them. It may be though it detects nothing. (normally what happens)

When I tried magnetic sensing using a hall detector I could work out where a magnet was based on the values given from the 2 nearest hall sensors.

[Tenaja]

Depends on the quality of the motor.
What typically wears out in good quality steppers is the bushings or bearings (depending on the style). If you have no side load, they can last a very long time. I had a jig that used 80's pc drive take-off nema 17's...it ran for almost twenty years, at around 100k seconds per year, before they needed replacing. The replacements only lasted about 2-3 years. Granted, this is far from 24-7, and the first 20 odd years it probably had more wear from bouncing around than from

Sometimes you can burn one out, but that would be the super cheap ones that you are running too hot. Or, just crap quality. There are no moving wires within a stepper, and no brushes, so if you burn one out you know it was garbage out of the factory.

Coincidentally, the original steppers outlived the original PC motion control by quite a long margin. I replaced the PC + motion card with an 18f2331.

[JonW]

can you access the top of the tube to look down on the water level?  Could look at time of flight sensors? 
https://www.st.com/resource/en/data_brief/vl53l5cx.pdf

[top204]

That board doesn't look like it will have a good resolution Tim.

To get the camera closer to the tube, have you tried lenses or prisms?

[TimB]

The tests I'm about to try is to see if you can determiner the an intermediary level by looking at the levels around each sensor. I tried to cancel my order for that one as it uses trans-reflective sensors, I was too late . I have on order some basic visible light sensitive transistors to try also.
However I'm not holding out much hope

There are a whole host of issues i.e condensation in the tubes and these could be detected as the water level. The gold standard I think is either

1, Ir transmission reduction due to the attenuation of the IR through the water. Set the level right and a small drop will be ignored
2, Camera scanning as you measure the total light source across the whole width of the tube. Only issues is the white balance on the camera and the distance you need to mount the camera to get the whole tube in.

I'm thinking again about using 120 ir transistors but to get the resolution I will stagger them. Getting the PCB populated though is expensive. Saying that I could build the whole PCB for less than the cost of the sensor I was planning on using.

[top204]

Before you order a whole bunch of the same IR photodiodes Tim, order a few of different IR light frequencies and try them out for the IR level they will respond too.

This can be a straigtforward test with a meter or an ADC, and moving the water level so it goes below, on, and above the photodiode. Then see which photodiode has the best reaction. i.e. Difference. Do multiple tests, and on a sheet of paper, write the level values and the distances of the water below and above the diode, then collate the values into a graph to see which one is better.

Then cloud the tube to imitate condensation, and try it again, and see which one reacts better.

I know it will take time for the tests, but you will have a responsive diode that you will have confidence in, which makes the coding process a lot smoother. With the tiny diodes lying close to each other on a PCB, you will have an excellent resolution, just like a scanner.

They still make scanners as part of printers, so the scanner photo modules should still be readily available, unless they have changed the principle they use for scanning.