Adjustemen for photo sensor with special P CB design

[r6502]

Hallo all,

I just have a question regarding PCB design:
I have to adjust the position of a photo sensor, sitting on a PCb board. We do not have space, for mechanical constructions, so we got the idea to place a cutout in the PCB board, and adjust the position of the sensor by pressing with a screw to the side of the PCB in the area, where the sensor is located.

The PCB is fixed on the bottom side with two M2 screws, refer to 1st graphic. The PCB board will be paced rough to its final position and than the fine adjustment is done, by adjusting it with the screw pushing against the PCB.

Our adjustment range will be 0.5 to 0.8mm, so not mutch.

I had drawn two cuts outs in the middle of the board, do you have an idea, what cutout will be better? Wit the 1st cut out bending will be cloce to the top / bottom of the coutout and with the 2nd  cutout the whole remaining part of the PCB will be bent.

Did somebody did something similar? Does it make sense, to use a thicker board, like 2mm or 3mm? Traces will be 150µm (ca. 6 mil's) and the PCB will be milled, so we will have a theoretical distance between track outline and PCB border of 425µm when we look at the smallest of the PCB.

Please note, it is no final design, just an idea for it.

Your comments are welcome.

Guido

[thm_w]

That is a very small adjustment range, it makes me suspicious about the reliability of the product. What happens if dust gets on the sensor? Is the item its being sensed perfectly positioned?

I would use either a thinner board 1.2 or 1.0mm maybe or less, and less cutout. I don't see why you'd use a thicker board, it just makes it harder to flex.

[Whales]

A flexural element! 

A few thoughts:

(1) There might be some better options for your screw pattern. Checkout yours (left) versus mine (right):



Spacing the screws further apart gives them more leverage against the PCB rotating.  This, along with the thicker (XY axis) sections of PCB material left may help avoid creeping and cracking.

Additionally: your electrical pads are in a bad position.  If you imagine that your two screws are approximately one screw (because they are so close together): your PCB will rotate around them like a pivot.  This will then put stress on the soldered electrical pads.

(2) You might need to control the PCB's vertical position (Z axis).  It might want to rise up off the main board.

(3) Not all of the dimensions in your flexural arms are a tradeoff.

Thickness:  thinner is more flexible, but harder to manufacture.  Thicker or thinner may make them easier to break (there are multiple directions of stress and strain, it's complicated).

Length: longer is always better.  It reduces the stress and strain in any one element. 

To make the arms longer without needing more space: you can make them meander or curve.  Checkout my 3d printed flexural element to see what I mean.

(3.5) EDIT: Your first image shows overcutting in the corners of the arms.  This will focus all of the stress and strain in the thin corner bits of the arms.  You probably don't want to do this, it would be better for the whole length of the arm to equally share the strain so that overall strain levels are lower.  Your second image is better.

(4) Your wire-paths are suboptimal for noise suppression. The hole in your flexture is currently  the centre of a big wiring loop.  You probably want to put both the traces on the same arm, so that they are close together.

(5) "creep" and "plastic deformation" can appear later (days, weeks, months) and ruin your day. 

Reducing the stress and strain in the flexible elements will help avoid this (you want them to remain in the "elastic region" of stress and strain) but this is a whole other rabbit hole if you don't have detailed material properties from the PCB maker.

Instead it will probably be easier just to glue the element down once it has been adjusted.  Make sure to use a hard glue (not a flexible one), some epoxies might work but others not.

(6) You will need to prototype and test iteratively, so PCBs might be an annoying material to work with if you have to wait a week or two for each revision.

If you have access to 3d printers: a 3d printed carrier for the PCB may make things a lot easier (and as a bonus: it can provide anchor points for the screws too).  I would use a ribbon cable to connect the little pcb to the mainboard so that you don't have to worry about stress on electrical connectors.

[Whales]

I've probably gotten carried away in my excitement, this is a valid concern:

That is a very small adjustment range, it makes me suspicious about the reliability of the product. What happens if dust gets on the sensor? Is the item its being sensed perfectly positioned?

Is this optical sensor something like an endstop that you're adjusting? 

Can you use something like an inductive sensor instead?  In some scenarios they can be much more reliable.  Some have digital 0/1 output (instead of analog voltage levels) and you adjust their position using their thread.



Also: do you have a backup sensor if this sensor fails?  Or does the device self-destruct?

[aeg]

The way they do this in floppy drives is by making the mounting holes in the PCB obround. With the screws loose, there is play along one axis. You position the board for correct track 0 sensor alignment and then you tighten the screws.

[r6502]

Hello all,

Thanks for the replay's on this topic.

In our application we do not have so mutch space. we are detecting end positions of an electronic gripper for a pick and place application with a stroke of less than 1.5 mm. With the optical sensor I have a hysteresis of less than 0.1mm that is sufficient. We will use 2 sensors for each gripper detecting the positions fully closed and full open. There are 2 different grippers installed on an arm connected to on sacra robot. Space is a big issue here. We will adjust the sensors once, and never (hopefully) touch them again.

@Whales
No, inductive sensors do not work in the current design, we are using small inductive sensors with 4mm outer diameter, but it is difficult to adjust them in this special application, as the hysteresis is to large. Different screw patten is a good idea. I think, we will manufacture some samples and do some stress tests. I think, we will do a multilayer board and have GND on top and bottom and the signals in the 2 middle layers or we go with connecting  the electric contacts direct at the photo sensor so no traces are in the flexible part of the PCB.

@thm_w
The device will be enclosed and the whole setup is operating under clean-room conditions, so we do not have a dust problem. If the PCB becomes thinner, the sensor could "wobble" in the z-direction what is not wanted here, so I'd like to have the thickness of the board going more to 2mm or even to 3mm.

We will do some real tests - I think it will be possible to realise it. hanks for your thoughts / comments.

Guido
Edit: spelleing

[floobydust]

With that mounting, I would say you get a basic "diving board" or cantilever which will vibrate and give bad results with the detection tolerances you need. Not sure if thicker FR4 will stop it, the screw mounts need to be perpendicular as Whales pointed out.
I can't make out the mechanical but that might need to be improved as well to get such small detection range. Adding a shutter, fan, reflector, window kind of thing because most optical components have a wide angle response.