Hi all,
Our recommended tube lens supplier is ThorLabs, but some OpenFlexure kit vendors have been looking at getting better bulk deals directly from suppliers in China. Back when @r.w.bowman and I were experimenting with V6 kit sales we found Nanyang Jingliang Photoelectric who were able to make similarly specced lenses. Bulk buying from them could significantly reduce the lens prices.
@jorodeo has bought a sample from them. The lenses were:
- f = 50 mm
- D = Ø1/2"
- Type: Achromatic Doublet lens
- Materials: BK7 and SF2
- AR coating: 400-700nm
Compared to the ThorLabs lenses the new ones were thinner:
There was no obvious change in imaging performance:
(First image is ThorLabs, second is with new lens).
My initial thoughts are that it shouldn’t matter how thick the lens is as long as it has the same back focal length:
As we use the back focal length to calculate the lens placement
The ThorLabs lens has a back focal length of 47.2mm specified.
@jorodeo Do you think we can ask the supplier if they have a number for the back focal length.
If the back focal length is slightly different I think the 3 options are:
- See if they cam make ones with a 47.2mm back focal length
- Calculate the effect/measure of the lens being in a slightly different place
- Make a modified optics module.
We probably also want to devise some sort of lens quality checks. Any ideas @WilliamW @r.w.bowman. We do already have some nice work happening in the forum on resolution measurement.
3 Likes
I think if the thickness has changed, it’s likely the length of the optics module or placement of the lens should shift slightly, but probably only if you do more sophisticated calculations that we currently use. On the other hand, it should be very slight - being 1mm or so out in focal length would result in the focal plane shifting by a fairly small amount - for a 40x objective, it corresponds to working ~10um from the ideal focal plane, which won’t affect anything very much: I think our tolerance on where the focus should be is on the order of 1mm, which I suspect means we can use either lens without modifying the design.
Our current calculations assume an ideal thin lens, but then shift it by the difference between the nominal focal length and the back focal length (i.e. we position the principal plane where the ideal thin lens should be). By that token, if the back focal length is the same, the thickness doesn’t matter - and if it’s different, we should just shift the lens by the change in BFL, to keep the focus in the same place.
If we’re being really rigorous, we could raytrace - but that would require more data on the lenses and objectives than we typically have, which is why we’ve never done it. If we standardise on a particular manufacturer for lens and objective, and they’re willing to share enough detail, I’d be delighted if someone was keen to do some ray tracing. However, I think it’s very unlikely this will have a meaningful effect on imaging performance for now.
1 Like
To answer the specific question I was asked, I think the resolution measurement should be sufficient for both objective and tube lens. I guess the tricky part is if you get a bad result, how do you figure out which is the problem.
Using a 150mm long optics module (shoulder-to-sensor) with no tube lens will allow the resolution of the objective to be measured without the tube lens - so I guess that’s how we distinguish which is the problem.
1 Like
I had a look a while back at different manufactureres that did state their back focal length. They are all different by of the order of 0.5mm. This is in
a Gitlab issue Check focus of infinity corrected optics module (#158) · Issues · OpenFlexure / openflexure-microscope · GitLab .
I think I agree with Richard that this difference will not affect resolution or field curvature too much. We do now almost have the calibration analysis to be able to say whether there is a measurable difference if the distance is wrong. If we had that then the thing to do is to have a few optics modules with the same lenses but different positioning to see whether ‘correct’ is actually ‘better’. That would tell us what we actually need to worry about.
1 Like
This is a good figure to have in mind. From how we do the lens positioning calculation (scroll to Section 3), as said above this 0.5mm change will effectively change the lens-objective distance.
Currently this is 8.5mm. Running the calculation with 9mm and 8mm, the difference in where we put the sensor changes by only 34um. We don’t have anywhere near that level of precision with 3D printers. And we don’t exactly know the height of the camera sensor above the board to that precision.
So I think we conclude that a ±0.5mm change in back focal length is fine. I think even ±1mm should be fine as this is ±69um in sensor position.
1 Like
35um is the difference in the distance that we want to put the sensor from the principal plane of the lens. However, with a fixed lens mount the distance from the sensor to the principal plane will be 0.5mm shorter if the back focal length is 0.5mm longer. Taking that into the calculation and working backwards, it would effectively mean that we are using the lens with an image plane at 143mm instead of 150mm. For a 40x or 60x lens, that moves the object plane less than 10um. That sounds similar to the likely variation in thickness of cover glass, so does not seem to be a big problem.
Ooops, yes, my calculation was wrong.
But I think our conclusion is that it probably doesn’t make a huge difference, but we should probably spec a back focal length and do some sort of resolution test?
1 Like
Hi @j.stirling I have emailed the manufacturer to see if we can get that back focal length measurement and will post here once I get a response!
1 Like
The manufacturer said the back focal length is 47.3mm
1 Like
So the difference in back focal length is 0.1mm. This certainly won’t be the limiting factor in getting the camera sensor in the optimal plane.
There are a number of other specs that ThorLabs provide
We could ask about things like the Spherical Surface Power, we could also ask about what quality checks are done and are they batched or performed on each lens. I don’t have a good feeling for what we would need for these numbers.
In addition to that we should remember that the actual back focal length used in the microscope design is 47mm, which is already different from the actual back focal length of the specified Thorlabs lens.
3D printer motion is generally very accurate, but there is a particular extra uncertainty in the thickness of the first layer. This is dependant upon the bed position at the start. In the optics module that thickness variation will directly affect the camera to lens dustance.
There is some uncertainty from shrinkage, and some from how well components seat. I think the main uncertainty is that the camera sensor sits on a squishy pad some estimated distance above the PCB.
Good point about the 47 in the code!
1 Like
