Hello. For my application, I would like to develop IR line array sensor. The task is to detect a human hand within a 3D space 400x300x400mm.
Requirements:

1. Leaving no dead zone
2. Fast detection (PIR sensor would not work here)

I have tried:

Ultrasound sensor - does not have wide enough beam to cover whole area
Diffuse photoelectric sensor - not wide enough beam to cover whole area
PIR sensor - Does not hold its state reliably when hand detected, too slow.
Light curtain sensor - works ideally, except that it is really difficult to find the exact length for my application.
The light sensor such as:
https://www.amazon.com/GW43-4-10-Safety-Curtains-safety-protection/dp/B00K1VXITI



Since I have tested a light curtain sensor and it worked great, I am going to develop my own light curtain sensor the required length and beam distance.

Before I do that, I would like to ask some general advice on which configuration would be better :
1.

In this configuration, both IR emmiter and receiver would be sitting on a single PCB. I assume in this configuration, the Y distance is really important since I do not want to have the receiver constantly receive the transmitted beam. Am I right? Also, what should be used as a reflector surface ( red square at the bottom ) to be able to reflect the beams of the transmitter? Is any material alright?

2.




I think this configuration is easier to get it to work properly since Y distance becomes irrelevant. If any of the beams are blocked by hand, the receiver should generate a signal. However, this configuration requires more wiring since the transmitters and receivers are 2 seperate PCB's.

For this configuration, I am slightly confused on the working principle? For example If a person puts his hand inside, it should block one or more emitters ( One or more receivers cannot receive the IR signal anymore) but that might not be the case since it might receive a IR beam from another emmiter. Il add a picture to show what I mean:





I am hoping to get some advice from some person with hardware and IR transmitter/receiver knowledge. Thanks in advance.

 

Have you considered various LIDAR range and gesture sensors like this one https://www.adafruit.com/product/3316 for example?

 

Yes I have considered most types of available sensors. The issue I found with this sensor is that it is not capable of generating wide enough coverage for my application. I was not able to find ultrasonic with 180 degree coverage hence I have dead zone as illustrated:

Also, lidar sensors have issue detecting targets at very close range. Maybe this could be solved by using multiple lidar senors (one in the middle and one in each side left and right, however, that brings some new issues regarding hardware and wiring. I really believe the best solution is the IR array sensor here.

 

Does the opening have to be rectangular?

What if the side from which the hand enters is not rectangular?

 

Perhaps that is one option. However, that would require a whole new redesign of the boxes :/ I also have concerns regarding the sensor reliabilty. If the hand moves inside the box, the sensor must toggle back ( must hold a steady output ) and only toggle off when a hand is taken out of the box.


UPDATE:

I had one ultrasonic sensor lying around and tested it. The sensor that I have is MB1603 from maxbotix. I cannot say that it works reliably enough for me. Detecting my hand within short distance. I need 100% reliability - price is not so much of an issue. I dont think these type of sensors are meant for a hand detection within a very close range, especially when high reliability and speed is required

 

Are there any power restrictions?
For a curtain maybe a 5 cm spacing with alternating emitters & detectors so that a narrow beam
IR LED with an 8 deg. spread could be used like VSLY5940 @ 4 / 1 US$. Might only need about 22.
Detector not selected yet.
I've made a 1 cm curtain for water drop detection & 12' with 1' spacing for car detection.
Red would be easier but little external light noise with 940 nm IR.

 

Are there any power restrictions?
For a curtain maybe a 5 cm spacing with alternating emitters & detectors so that a narrow beam
IR LED with an 8 deg. spread could be used like VSLY5940 @ 4 / 1 US$. Might only need about 22.
Detector not selected yet.
I've made a 1 cm curtain for water drop detection & 12' with 1' spacing for car detection.
Red would be easier but little external light noise with 940 nm IR.

Thanks for advice! No power restriction.
It is important that the curtain detect a hand even when it is rotated 90 degrees at its most narrow point hence I am concerned about proper spacing of emmiter.
I am also very concerned about the phenomena such as :

If a beam is too wide, the receivers will pick up the beam from another emmiter.

Could you advice whether using emmiter/receiver on a single PCB or using one side of the box for emmiters and opposite side for receivers is better?


Also, you mentioned 940nm IR sensor. Are there any IR sensor that work on an unusual wavelength? Since in the room where I am building this system there will be many external sources of IR that might have an effect on my system .

Lastly, I wan to know whether tiny surface mount emmiters/receivers have any difference compared to big through hole emitterr/receiver?

 

In passing threw 50 pages of Digi Key IR emitters I do not recall any SMT narrow beam emitters but most other specs. can be found in either thru hole or smt.
I made a simple test with Tx a VSLY5850,, 6 deg. 850 nm, 5mm, 100mA, driven at 10 mATx @ US $ 1.35 & Rx a VEMT2020,
15 deg., 880 nm, smt. Tx mounted in a 5 mm 3 cm deep hole, Rx mounted behind a cone reflector 1 cm X 2 cm. Rx & Tx spaced 60 cm apart. Tx could be moved side to side 7 cm before Rx dropped out.

 

Another small test to allow some aiming of the Tx. 3 cm mounting strip drilled with # 24, 3.8 mm & counter drilled with 1/4" to a depth of about 7mm so that when IR LED is inserted the back is flush
with the surface. An adjusting strip of thin plastic- circuit board is drilled for a snug fit with leads and a second hole of about 1/4" offset about 2 cm--- must run.

 

The issue about receiving light from adjacent emitters can easily be overcome by activating the emitters sequesntially - 1 at a time. Then you know which light path to check while all others are considered irrelevant - or contribute to e.g. size assessment of the interrupting object.
Re reflector for reflective curtain: cat's eye style reflectors should be your choice. These are available in rigid forms as well as as sticky tapes (e.g. from 3M).

 

Thank you for your answers! Can you guys comment on which setup would be more suitable for this type of application?:

or:



And what would be the key differences between the two?

 

Also, I have ordered some of those Vishay narrow beam emitters. In terms of a receiver, I am slightly unsure whether I should look at narrow viewing angle ones or wide angle as I am unsure how will affect the system

 

A reflective light barrier eases the installation (only 1 active element). Put this one on top and the only thing collecting dust would be the reflector - easily wiped clean with a damp cloth.
The downside of the reflective barrier is that the signal is just half of the "see-through" barrier - which shouldn't matter much for such a short distance.

BTW: just looked up the reflector required: the exact and full denomination of its properties is "prismatic retroreflective".

 

Also, I have ordered some of those Vishay narrow beam emitters. In terms of a receiver, I am slightly unsure whether I should look at narrow viewing angle ones or wide angle as I am unsure how will affect the system

If you energize a single emitter at a time (or some emitters with sufficient space between), the characteristics of the receiver wouldn't count much. But a narrow-angle receiver might prove better with respect to sensitivity, immunity vs. ambient light etc.

 

Also, I have ordered some of those Vishay narrow beam emitters.

I made a chronograph sensor of almost the same dimensions using 12 Vishay TSAL6100 emitters and 12 BPV10NF photo diodes. The configuration was the same as your second drawing, emitters opposite the receivers
The circuit produced a brief pulse as the projectile passed through.
This 3D space is it a box open on both ends?

 

I made a U frame of 3 cm X 3 cm X 26 cm spaced apart 45 cm. RX & TX alternated @ 5 cm. Rx with a cone concentrator 1.1 cm X 2.6 cm giving about 3 X sensitivity improvement . No output on adjacent
Rx s. Adjustment tab on Tx works. Rx Rohm RPT-38PB3F , 3mm, +- 36 deg. ,800nm, obsolete, Jameco US $ .45. Reading Rx with Heathkit MM1 on X 10k ohm scale.

 

With this link and comment, I’d like to say, make sure your approach services the need.

https://www.amazon.com/GW43-4-10-Safety-Curtains-safety-protection/dp/B00K1VXITI


UPDATE:

I had one ultrasonic sensor lying around and tested it. The sensor that I have is MB1603 from maxbotix. I cannot say that it works reliably enough for me. Detecting my hand within short distance. I need 100% reliability - price is not so much of an issue. I dont think these type of sensors are meant for a hand detection within a very close range, especially when high reliability and speed is required

 

Maybe a good match for VSLY5850 is WP7113SF6BT-P22, 860 nm, 5 mm, Digi-Key #754-2224-ND @ US $ 6.79 / 10.

 

I made a chronograph sensor of almost the same dimensions using 12 Vishay TSAL6100 emitters and 12 BPV10NF photo diodes. The configuration was the same as your second drawing, emitters opposite the receivers
The circuit produced a brief pulse as the projectile passed through.
This 3D space is it a box open on both ends?

Thanks for all the tips! The space will be 3D. it is going to be box with one end open (where they put their hand) and the other side will be closed. They will be picking items from the box and I must record how many items they have picked ( They are only allowed to pick one items at a time).

The only important thing is that once the hand is through, the sensor must not toggle off ( loose its state due to blind zone or whatever ) because that will cause me to count incorrectly. It must hold the state while the hand is in the box does not matter how he moves and where he places his hand.

 

I made a U frame of 3 cm X 3 cm X 26 cm spaced apart 45 cm. RX & TX alternated @ 5 cm. Rx with a cone concentrator 1.1 cm X 2.6 cm giving about 3 X sensitivity improvement . No output on adjacent
Rx s. Adjustment tab on Tx works. Rx Rohm RPT-38PB3F , 3mm, +- 36 deg. ,800nm, obsolete, Jameco US $ .45. Reading Rx with Heathkit MM1 on X 10k ohm scale.

Thats cool. Maybe you have got some pictures? Also, can you confirm that there is absolutely no blind spots across whole 26cm? Like when you move your hand from one side of the box to the other side of the box, I would assume that a receiver should hold a steady output ( either active LOW or active HIGH ) and not pulse.