Sense Powder flow

[k09]

Hello all Forum Members..

I have taken up a project which aims to automate the entire coating procedure (from start to end of the coating process including intermediate cooling pass). This is to reduce the inputs from the operator and even enable a semi skilled operator to initiate, observe and control the coating process(only if required). Operator override would only be required to refill powder or gas canisters.

However as I was sketching a rough flowchart incorporating PLC controls at various stages, I was unable to find a solution for sensing powder flow in powder hose(which to some extent would also help in identifying if the powder hopper is running low).

In addition, the sensor should also be able to sense all kinds of powders (ceramics, metals, alloys, cermets, etc..).

Would flow of ceramic or metal powder passing a charged coil lead to a difference in the field voltage or the coil current? (just a wild guess!)

All suggestions and inputs are welcome..

Best Regards

K09

[pulaunias]

1) Make a segment of the hose transparent and monitor the pass of light? Powder flow should block the light.

2) I always wonder how the autmatic faucet works? when your hands approach it, water comes out. I heard that it sensed some "capacitance change". Might this mechanism work for you?

[k09]

Hello pulaunias..

Using is ligh is an intereting option but are there any such sensors already available. The problem is that as per program logic, if the light is blocked = powder is flowing. However this logic would also be true if the powder is blocked or powder has been shut down and excess powder comes right in front of the sensor...

I am not sure about the automatic faucet. it could be IR sensors. As you take your hand closer it reflects signals back to the receiver and the water flow solenoid is enabled..

I am looking for options that would not only detect flow but also detect prominent variation in the powder flow...

[MichiganMan]

Why not use a load cell at the powder hopper? If W at t1 > W at t2, then you know that you have flow. You could also configure the cell to alarm at a low level to notify that filling is necessary. The variance in density from material to material may cause minor issues in programming, but it seems workable.

It might be tricky to retrofit to the equipment, but once you get it going I can't see any major problems.

[k09]

Hi,

Thanks for your inputs Michigan Man..

If I consider the Fluidized Bed powder feeder as in case of Metco 9MC system, this would be a very good option and I will surely try and find if this feasible.

However when it comes to a disc type powder feeder, how would it be possible to incorporate a Load cell??

[MichiganMan]

I think that for optimum control at this point in the process you may need to let simplicity of system instrumentation and controls dictate what the mechanical components selected are. A system that allows you to incorporate the simplest possible flow sensing (which in my estimation is probably a load cell) would likely be best for the type of system you are considering.

Honestly I hadn't though of the complication that a disc type feeder might add to the application. That is a good question, but not one that I am able to answer for you.

[djewell]

You could mount a light sensor on the end of your gun. Choose a sensor that is sensitive to red or yellow light. When the sensor sees light at the correct wavelength, ie, when the powder is flowing, it could open/close a set of contacts, giving an indication of powder on/off. You may even be able to measure the intensity to roughly determine flow.

[dcrawford]

K09,
If you look too far down stream in the process...as djewell suggests... than there is usually too much averaging to see small changes in powder flow.

If possible...looking inside the plasma as the powder is injected is much better. It is very easy to see inconsistent powder injection in this video:
https://www.youtube.com/user/ControlVisi...Sve6BalCLc

This is what it is supposed to look like:
https://www.youtube.com/user/ControlVisi...Vj4OywLb3Q

And this is what a plugged injector looks like:
https://www.youtube.com/user/ControlVisi...mMvI6GcI2k

Image processing software can be used to detect powder flow as seen in the following video, however it would be better in your case to detect the particles as they exit the injector (move the red box measurement volume).
https://www.youtube.com/user/ControlVisi...d5aXIIrJh8

If your processes is internally injected...for example HVOF... than you could look further down stream at the molten particles, however at this point a lot of mixing has occurred and you won't see fast fluctuations in powder flow.

I also really like pulaunias's idea of a clear tube, although you would probably want to expand the tube so that at normal flow rates the powder wasn't too concentrated. With some development work this could be a relatively cheap/robust sensor.

Thanks,
Dan

[Gordon]

I think I'm with MichiganMan on this one. Load cells measuring change in mass. It is not dependent on powder properties. Though, I would not dismiss some of the good ideas above, as I have seen some in operation. The major problem is that they need calibration to the specific powder being fed, as optical or electrical properties will vary depending on powder.

[k09]

Dear Djewell & Dcrawford,

I truly appreciate your inputs and undoubtedly they are very good ideas but as Gordon mentioned all these equipments would not only require calibration but such systems are generally employed for high level analysis of powder flow. What i am looking for is a more simpler, cost effective means to enable the Micro-controller / PLC sense that there is powder flow. With this, we would be able to run a coating routine 'n' number of times to achieve desired thickness. When the control senses, less powder or no powder, the system will jump out of the routine, park the coating pistol in home position and set alarm for the operator to re-fill. This however would be one aspect of the complete automation project.

Considering this intially for the HVOF System which is already mass flow controlled and has decent PLC integration at various process levels, i need a simple technique to sense powder flow (especially carbides and cerments to start with). The load cell is indeed a brillliant idea and would have been a very simple yet effective technique however a disc type powder feeder restricts this application.

Regards

K09

[djewell]

I think you are wrong about the light sensor requiring calibration. If you buy a sensor which is only sensitive to yellow or red light, then it will see hot powder in your flame. If the powder flow stops, the flame will turn blue, and the sensor will change state. There are forms of LED technology which are sensitive to different colors of light and would provide the sensor you need.

[k09]

@ djewell: When I read the term wavelenght, I initially thought that system you were refering to was same as the particle flow HD Camera. Its much clearer now. And yes, this could be an effective option provided its not a bulky box sitting in front of the spray gun.

Could you please suggest a few suppliers for such sensors?

Thanks & Regards
k09

[Gordon]

Hi k09 and all

I think djewell's idea sounds good if you just want to detect when powder feed is on or off. To sense actual feed rate will involve much calibration. Load cells only need calibration against force, they don't care what powder your using or other system parameters and you can use the output to control system within the desired +/- limits of feed rate.

[k09]

Absolutely Gordon. Infact the main advantage of both methods (load cell & light sensors) are they are not dependent on the powder type.

@ Michigan man & Gordon: for the load cell in disc type powder feeder, would it work if i place the load cell just above the lower exit point? Probably this way it would sense the weight of the powder above it to a certain extent?! Does tht make any sense at all..?!

@ djewell: do you feel that this sensor has to be placed very close to the HVOF plume? If yes, would it withstand high temperatures? And would unmelted excess powder expelled from the plume cause any problems?

Regards
K09

[djewell]

The light sensor could be mounted in a bracket on the gun or nozzle out of the way of the jet. In this way, it will not see any high temperatures. It will also travel with the gun, so if powder flow stops while you are spraying, you will be notified.

[Harrison jess]

i once added a laser to the spraying gun to point the direction it really helped me in that case .

[dcrawford]

Sorry for the delayed response, but here are some thoughts:

Load Cell-While this is definitely the simplest and easiest to operate method but I think it gives you the least amount of information on the process. Here are the limitations as I see them:
1. Many powder feed problems happen downstream of the powder feeder.
2. To measure a rate of powder flow you would have to take two readings from the powder feeder and divide by a time interval. This has some inherent averaging in it and will not give you small changes in powder flow. Your measurement abilities will be limited by how noisy the load cell signals are and how much time is between measured weights. If these aren't an issue than this sounds like the best option.

Light Sensor on Torch (djewell's idea)-I think this idea has promise, however I don't think this is as straight forward as it sounds. Different powders produce noticeably different colored particles. In addition stray light from the plasma/flame could cause problems. I would have to spend some time looking at signals from some optical sensors before convinced myself this is an easy task. However on a positive note, I believe one could possibly pickup spitting form torches due to powder loading which is a common problem in thermal spraying. And the sensor could actually be quite small.

Transparent hose-I still think this is the best idea and would give most information and could be done relatively cheaply. It seems like a sensor similar to this should exist for another application. If I find one I will let you know. If properly configured this sensor would not be dependent on powder except on possibly on particle size.

[ives]

I would have to agree with what Michigan Man addresses here. Now that you are looking for the optimum control of the process, the first and foremost thing that you should take into account here is the minimalism as well as the mechanical components that you will be opting here!

[markspend01]

Hey Dje well i think that the information about light sensor is brilliant.Thanks!!

[Esteban]

Using is light is an interesting option but are there any such sensors already available. The problem is that as per program logic, if the light is blocked.