Status
Not open for further replies.

dr pepper

Well-Known Member
A while back I interfaced a esp based webpage to a poly silo level guage system, it works well however the level detection hardware is 30 years old and unreliable so I'd like to replace this section.
I dont fancy building a guided wave radar system, I could put a plastic tube say 40mm diameter into the silo tank from top to bottom, then I could either have a sensor unit that could be raised/lowered to detect the level, or have an array of sensors to detect the level at certain points.
Any ideas how I could detect the presence of poly beads through a plastic tube, I was thinking an ultrasound device and 'listen' to the reflected sound and detect the difference when poly is present.
A magnetic system would be easier however I odnt think the beads would respond to a magnetic field, I'll try a test setup and see.

Last edited:

dknguyen

Well-Known Member
How far is the max depth? There are fairly cheap LIDAR system ($100-$200USD) now that could point straight down into the beads if that's possible (is that possible in a silo? Since it's laser it could be surronded by clear plastic if it might become embedded in the beads when the silo is full.
https://www.robotshop.com/en/single-beam.html

As for the having a sensor unit inside the narrow tube, you could use photodetectors (probably a phototransistor) and LEDs to detect beads through the tube. There are LED strips but no phototransistor strips so making an array that runs the length of the tube is difficult. It might be more feasible to have the sensor unit moving up and down the tube from a string or something so you only need to use one one photodetector and one LED. The tube material doesn't have to be visibly transparent either if your beads are opaque to IR, then you also have a bit more freedom of tube materials since some are visible opaque but IR transparent.

Last edited:

JonSea

Well-Known Member
I wonder if strain gages on the tube or tank wall would be sensitive enough to detect the pressure of beads at particular levels? A plastic tube may have reasonable deformation when beads are compressing it.

alec_t

Well-Known Member
How about a capacitive sensor? Dielectric constant of polystyrene is about 2.5.

gophert

Well-Known Member
I would use IR flashing at 38kHz and phototransistor with a filter to remove any low freq signals.

dr pepper

Well-Known Member
A bit more explanation is required.
The grounded aluminium silo is 10 meters tall and straight sided, about 2m dia and holds 51000 kg of poly beads.
Inside the tank has a very thick layer of poly stringy dust 'angels hair' as its called in the biz, I think this rules out long distance optical.
Strain guages on the outside of the tank would introduce wind loading issues, maybe a tube in the tank would work.
I was wondering capacitive myself, I'm not looking for distance here, just is there poly at this level or not.
Might knock something up with a 'duino, we use different types of poly, however for short range the difference in capacitance between poly and air is going to be enough to discriminate.
For the tube I could use flexible plastic, wire reinforced would be good but that probably wouldnt work with capacitive, 5mm thick wall ought to be heavy enough to take a 10m drop.

Last edited:

dknguyen

Well-Known Member
The silo is 10 meters tall and straight sided, about 2m dia and holds 51000 kg of poly beads.
Inside the tank has a very thick layer of poly stringy dust 'angels hair' as its called in the biz, I think this rules out long distance optical.
Yeah, that rules out ultrasonics in the open too, even water proof ones since it could cake on the sensor and mess up the frequency response (I presume).

gophert

Well-Known Member
A bit more explanation is required.
The grounded aluminium silo is 10 meters tall and straight sided, about 2m dia and holds 51000 kg of poly beads.
Inside the tank has a very thick layer of poly stringy dust 'angels hair' as its called in the biz, I think this rules out long distance optical.
Strain guages on the outside of the tank would introduce wind loading issues, maybe a tube in the tank would work.
I was wondering capacitive myself, I'm not looking for distance here, just is there poly at this level or not.
Might knock something up with a 'duino, we use different types of poly, however for short range the difference in capacitance between poly and air is going to be enough to discriminate.
For the tube I could use flexible plastic, wire reinforced would be good but that probably wouldnt work with capacitive, 5mm thick wall ought to be heavy enough to take a 10m drop.
Here, I just asked my ex coworkers - they suggest this resonant technology. They say nothing works better.

http://www.monitortech.com/product_p_vpzp.shtml

They manufacture plastic beads and support their customers (injection molders and plastics compounders.

dknguyen

Well-Known Member
Here, I just asked my ex coworkers - they suggest this resonant technology. They say nothing works better.

http://www.monitortech.com/product_p_vpzp.shtml

They manufacture plastic beads and support their customers (injection molders and plastics compounders.
The ad says that's for point level sensing right? Not for continuous measurement?

gophert

Well-Known Member
The ad says that's for point level sensing right? Not for continuous measurement?
They use two, one to trigger a refill and one to stop the filling.
Or a single tip and add a fixed quantity refill each time the unit senses no contact to product.

Les Jones

Well-Known Member
If you don't need fast response could you use a cord wound round a drum at the top and a weight on the end with a large enough surface area so that it did not sink into the beads. The drum would have an optical encoder to count revolutions and part revolutions. Normaly the cord would be hold the weight up at the top. The cord would pass through 3 pullies. Two would be fixed with a space between them and the third would be on the other side of the cord about half way between the first two and spring loaded against the cord with a microswitch to sense the position. To take a reading the cord would be unwound until the microswitch detected the drop in tension when the weight reached the surface of the beads. The encoder would indicate how much cord had been paid out and hence the depth.

Les.

AnalogKid

Are you sure that the dust would interfere with an ultrasonic distance measurement? You don't need a true level detection, but one would certainly give you a signal you could process. $2 modules on ebay. What is the range of bead diameters? An air bladder connected to a solid state pressure sensor would be squeezed to a trip value as the beads rose above it. 100% dust and water proof. ak dknguyen Well-Known Member Most Helpful Member Are you sure that the dust would interfere with an ultrasonic distance measurement? You don't need a true level detection, but one would certainly give you a signal you could process.$2 modules on ebay.

What is the range of bead diameters? An air bladder connected to a solid state pressure sensor would be squeezed to a trip value as the beads rose above it. 100% dust and water proof.

ak
Even if it could see through the dust, wouldn't the dust caking on the transducer cause issues anyways?

dr pepper

Well-Known Member
The vibratory sensor is interesting, it would need to be 10 meters long though, I wouldnt want to bash holes in the side of the silo.
Les you described pretty much exactly what we have now, only theres no encoder its a 10 turn pot driving a moving coil meter.
Vibratory?, that makes me think its some kind of low frequency thing esp if its self cleaning, I wonder if I could drive a piezo coin thing and then use another to 'listen', an msgeq7 chip is a 7 band audio frequency analyser which is simple to use with a microcontroller, an ordinary speaker or vibration shaker thingy and a acoustic mic might be better than piezo.
Probably way ott but maximum length sequence springs to mind for the excitation.

dknguyen

Well-Known Member
The vibratory sensor is interesting, it would need to be 10 meters long though, I wouldnt want to bash holes in the side of the silo.
Les you described pretty much exactly what we have now, only theres no encoder its a 10 turn pot driving a moving coil meter.
Vibratory?, that makes me think its some kind of low frequency thing esp if its self cleaning, I wonder if I could drive a piezo coin thing and then use another to 'listen', an msgeq7 chip is a 7 band audio frequency analyser which is simple to use with a microcontroller, an ordinary speaker or vibration shaker thingy and a acoustic mic might be better than piezo.
Probably way ott but maximum length sequence springs to mind for the excitation.
You mean like just sticking a huge pole into the silo and vibrating it? So fuller = less vibrating length = higher the frequency? That'd be one hell of a driver.

AnalogKid

Well-Known Member
You don't have to drive it continuously, just tap it periodically and measure the ring freq, like plucking a guitar string. Or sweep it with the mechanical equivalent of a medium-impedance source and watch the resonant peak shift, a variation of how traffic light sensors work.

ak

dr pepper

Well-Known Member
Nope, but that is an idea in its own right, like you say however not that practical.
My idea was to have separate speakers and mic's in the tank, one speaker/mic per sensing position, one audio amp to drive all the speakers and one amplifier/audio analyser with an analogue mux to switch mic's, and just process one at a time, at least that way unless the tank is all the way empty or full I'd have samples of both the presence and the absence of poly.
Still thinking of ideas at the moment.
In the good old days this was done by generating a mechanical vibration on a particular frequency, then a tuning fork which was mechanically coupled to the outside, and therefore the poly if it were present would detune the tuning fork, if there was no poly the fork vibrated actuating a microswitch, and if there was poly it wouldnt vibrate as much and it wouldnt actuate the switch.
Sounds just like the first Tv remotes, they didnt have batteries, they generated a frequency, a mic on the Tv amplified this and it was then applied to 2 or more tuning forks, if the freq's matched the fork vibrated and actuated a switch.