Bed of nails prototype creation question

I am working on prototyping a bed of nails test fixture for a board we repair a lot of and would like some advice on the most accurate way to measure the test point lat/long.

There are 30 chips on the board and each chip has 5 test points so getting the lat/long of each test point correct is critical.

I basically need a way to measure precisely the location of each test point on my board so I can manufacture a bed of nails test fixture.

I am hoping there is software or a mobile app that can be used to help me accomplish this.

Does anyone have advice?

Terribly lazy, but maybe one of those PCB reverse engineering services?
https://dirtypcbs.com/store/pcbclone

Once you have the gerbers, you should have a drill file with coordinates.

Do you have 3d printer and a laser pointer?
You can scan your board with a scanner or a photo to get preliminary coordinates. Then you can use a laser mounted to the carriage to confirm and fine tune your coordinates.
You can also use a camera mounted to the carriage too. Similar to pick and place machines that look down at fiducials and pads.

I can probably make a number of suggestions, but I need some information to pare the response down to something reasonable.

  1. Is this a board you have Gerbers (or CAD source data) for?

  2. How large is the board?

  3. Through hole or surface mount? If surface mount, what pitch?

  4. Discrete stand-alone test point pads, or component pins as test points? If discrete pads, how big and what shape?

  5. how accurate are the board edges (scored or snap tabs?)

  6. How closely spaced are the two closest test points

  7. Test points on one side. or top and bottom?

  8. Does the board have fiducials?

As an example, for modest size through hole boards I don’t have data on, I will usually do a high resolution scan on a flat bed scanner. It’s usually good enough if you use raw uncompressed data and an image viewer that will give you cursor coordinates in pixels or mm.

Rich

Following this.

I would think if you have the design files, it should be easy to extract out the test point locations. I’m looking at doing a kicad plugin for something similar.

Great suggestion but it looks like they only work with 2 layer boards. I believe ours are 4.

Sounds like a reasonable approach but we don’t have that equipment. I wonder if there is a service out there that provides this?

Unfortunately no design files on this board.

Lots of great questions.

Does this image help? I’m on mobile and will reply to your questions when back to keyboard.

I guess I can’t post a link or image for some reason. Is there an admin here who can give me permission to post links or images?

Trying again https://share.icloud.com/photos/0hgNZUjy7pU42kELtN0VyYvcA#Prosper

I haven’t priced a job, but there are CMM services that can extract coordinates for you if you just need it just for this one project. Some machine shops can do that too.

Yea this is for a single job. I found some CMM services now that I know what to Google, thanks!

I’ll see what the quotes come back at.

I’d second that suggestion of finding a CMM service.

Both of the following are assuming those 5 vias around each device are the test points you’re speaking of.

As a mildly “out of left field” alternative, Try this:

  1. Try photo-copying the PCB. That should make a reasonably undistorted image of the PCB that you can then start measuring from. (You might need to de-solder that 330uF cap in the lower right.)
  2. Pick some reference edges, (I’d recommend the top and right edges from the photo you shared, as they are straight, perpendicular, and unblemished.) and have at it with a square.
  3. Pop those dimensions into KiCAD and place the test points.
  4. Make a 1:1 test print and double check your work. Adjust the values and repeat the check.
  5. Once you’re good to go, then order a PCB for that.

Alternatively, if those vias are not filled, what you could do is use them as a template and transfer the locations to something. I’m envisioning that you use a drill (if you can find one small enough) or a dowel or needle that will fit. Ideally, maybe use a piece of ESD delrin for the substrate. You can mark it directly, and then use the marks to come back around with a mill and drill the hole for the pogo-pin sleeve. Finally, use some sort of find adjustment to set your edges.

Hope that helps.

1 Like

I think Seth has a great plan.

I would suggest perhaps taking a good digital photo using a long focal length lens, a tripod and good lighting.
Measure an X and a Y reference on the board as accurately as possible (drop in on a machine shop, they will have nice ways to measure).
Then use a graphics program to import and scale the image.
Then just put the holes where you want them.

Good luck!

One thing I learned over the years - you can’t assume perfect rectilinear spacing of camera lenses or printed paper.

1 Like

Assuming (and you know what happens then) the through holes on the traces between ICs are the test points this should be pretty straightforward.

Based on the size of the hand (assuming you aren’t Andre the Giant) the board should be small enough to do a simple bed of mails without specialized locating features or floating pins.

You’ve got three locating holes for positioning the board. There are also optical fiducials near the holes.

The test points (if I have the right of it) are both large enough and open. A pyramidal probe will tend to self center.

I’d plop the board on a flatbed scanner and get the highest resolution raw TIFF output available. You’ve got fiducials near the locating pins, so you almost certainly have some on the bottom. You can use the fiducials or holes as references and distortion checks.

Import the image into your favorite mechanical CAD package and ID the test point locations. You can correct for distortions once you’ve measured the references.

That’s some pretty impressive roasterization on the lower left.

Rich

Most photo copiers do few % shrink to make sure they get the very edge of the original. It’s rarely uniform in X and Y. It might or not be significant in this case.

The tooling holes are more likely to be accurately controlled, and fiducials will be as well. This board looks 100% routed so the edges are probably OK, the holes will be better.

If you can still find a printer or repro shop that can do photoplotted film, that should be accurate enough to check. I suspect anything laser printed will lie to you. I know photocopies will.

It might be easiest to just order a board and figure what needs to be fixed. Depends on how big a hurry you are in.

Rich

1 Like

Thanks for all the excellent suggestions. I will let you know how we get on.