Preserving Autofocus on Modded DSLR's

[michael8554]

I believe that Canon DSLR models starting with the 500D now have floating spring-loaded imaging sensors, adjusted by Torx headed screws.  
Here are my (untried) calculations for adjusting the sensor to preserve Autofocus after LPF Filter removal.

Gary Honis gives superb instructions for modding almost every Canon DSLR at http://dslrmodifications.com/

Each LPF filter refracts the light passing through it, which extends the light path. 
So without a LPF filter the sensor has to be screwed in further towards the shutter to meet the autofocused image.
Gary's calculation for the correction figure is based on a Refractive Index of 1.517 for the LPF:

(0.517 x Thickness of Removed Filter) ÷ 1.517 = Correction in mm

I understand the LPF#2 filter on the floating sensor models are 0.5mm thick, but that can be checked with a micrometer.

                                                                       **** MY 600D / T3i WAS ACTUALLY 0.60mm THICK *****

So for a LPF#2 removal, retaining the LPF#1 filter, the correction figure is:

0.517 x 0.50 ÷ 1.517 = 0.17mm

                                                             EDIT    *** READ MY FINAL POST ****

I also understand the Torx T7 screws are M1.6 x 0.5mm pitch. 
To check, measure the length of 10 threads with a micrometer and divide by 10 to give the pitch.
0.5mm pitch means each full turn moves the screw 0.5mm in or out.

Gary's method of adjustment uses "points". 
He notes that the star-shaped sockets of Torx screws have 12 "points". 
So before unscrewing the Torx screws he makes a radial Sharpie mark on each Torx screw head and the sensor mount.
So on replacement, the Torx screws should be returned to the Factory "height", at which point the Sharpie marks should align. 

The final adjustment for 0.17mm Autofocus Correction is:   

                                                         EDIT           ( ****** READ MY FINAL POST  ***** )                    

12 (points) ÷ 0.5 (pitch) x 0.17mm = 4.08 "points"  (4.08/12ths of a turn).

                                                                  EDIT     ***** READ MY FINAL POST *****

So the Torx screws are tightened a further 4 "points" to preserve Autofocus.

                                                                    EDIT    **** FINALLY TIGHTENED 6 POINTS ******

I hope someone can confirm or refute these calculations before I get going on a mod !!

Michael
 

Edited December 1, 2018 by michael8554

[Grimbles]

Hi Michael,

I am afraid I can't help you with the measurements as the last camera I modded was a 450D which was relatively straight forward.  However I have just had a 500d modded for a very reasonable price by Juan at https://cheapastrophotography.vpweb.co.uk/

I used Juan because I wasn't confident with retaining the autofocus / keeping the sensor square and he has a lot of experience along with very good reviews on here.   I have had the camera back a few days now but not had the chance to test it on a scope yet due to weather, but with a normal lens it focuses great in auto and manual.

I know this does not answer your question, but just a thought if you do have any reservations about going ahead yourself.

Martin

[michael8554]

My previous camera was a 500d.

I too had it modded by Juan, because of the floating sensor.

But my current camera is a 6D that nobody wants to modify, only to sell ready modded.

I've already modded one high shutter count 6D, but I now want to do a newer one.

The Autofocus is actually no big deal for my solely telescope needs, but would be a bonus in any future sale.

Michael 

[Croz]

I'm very interested in an answer to your question. I'm going to be modding a 600d and would like to retain autofocus, or at least be able to manually focus a lens after the mod. After looking at the video of the 1/12 adjustment it's not clear if a 4/12 would be possible? Is there enough travel to tighten this much? 

Not sure if you have seen the threat below. The numbers are slightly different than yours. But perhaps this is because of different filter thicknesses on different models of DSLR?

https://www.cloudynights.com/topic/488823-ability-to-reach-infinity-with-lens-self-mod-cam/

Edited October 21, 2018 by Croz

[Merlin66]

If you want to manually focus the standard lenses on a modded camera you just have to add a clear clip in filter.

That’s what I use on the modded 1000D. Works perfectly.

 

[michael8554]

I agree with Gary's calculation in Cloudy Nights, except all the 450D LPF-2 filters I have removed are 0.65mm thick, not 0.60.

Now one of the mod services says "0.21mm to compensate for removal of 0.7mm".

0.7mm is Gary's figure for the RETAINED filter thickness, not the difference of 0.6mm

 But taking 0.7 as correct, 0.21mm suggests a RI of 1.43, not 1.517.

Somebody's got it wrong, not sure who.....

Merlin, from memory only I thought we'd established your images did not look to be autofocused correctly with the clear clip in ?

Michael 

[Merlin66]

Yes, but I was replying to Croz, Re manual focusing.......

 

[michael8554]

5 hours ago, Merlin66 said:

Yes, but I was replying to Croz, Re manual focusing.......

 

Thanks for clarifying Merlin.

Michael 

[michael8554]

I have just ploughed through the Schott Filters catalogue, the RI of their bandpass filters is between 1.54 and 1.55.

So I discard the 1.43 figure above.

I have another 450D I can mod, I will try making replacement thinner shims for this from feeler gauge material, based on a RI of 1.545, and report back.

Michael 

[michael8554]

On 21/10/2018 at 14:57, Croz said:

it's not clear if a 4/12 would be possible?

That's only a third of a turn, if the Mod companies can do it, so can we....!

Michael 

[Croz]

Lol, looks simple enough. But watching the YouTube video it looked like he was cranking it down to get 1/12.  I get the feeling no one wants to reply, because this will make the mod just that much easier. Less reasons to pay for someone else to do it.

[michael8554]

19 hours ago, Croz said:

it looked like he was cranking it down to get 1/12.  

That's right, the Torx screws have to be tightened, to move the sensor closer to the lens.

19 hours ago, Croz said:

this will make the mod just that much easier. Less reasons to pay for someone else to do it.

I have no problem with the professionals keeping their secrets. However, few owners have the nerve to mod, so discovering the right setting isn't going to be a world changer.

Michael 

[michael8554]

Isn't it amazing how your brain works out problems while you're sleeping ?

Attach modded 450D (autofocus wrong) camera to ota, critically focus with LiveView and Bahtinov mask.

Attach unmodded 450D, critically focus by moving out of focuser tube, trickey so repeat, measure the gap each time between end of focuser tube and T ring with feeler gauges.

This will give the figure the modded camera's sensor has to be moved inwards.

That figure is only correct for that particular filter thickness, but does allow me to work out the correct RI of Canon's filters, which can be applied to any camera once the thickness is measured.

It also tests wether the "move 0.20mm" statement is correct or not.

Watch this space.....

Michael 

[michael8554]

Did a trial measurement this afternoon.

6 hours ago, michael8554 said:

critically focus by moving out of focuser tube

Easy to say, impossible to do I'm afraid.

Michael 

[Croz]

Michael - thanks for all your hard work and sharing your information. I'm hoping you can work it out and that my filter thickness would be the same? I would think that the lpf1 filter from the 450 and 600d would be the same or close. 

This looks to be extremely precise, I'm hoping my lack of fine motor skills doesn't hamper my mod ?

[michael8554]

Hi Croz

You can't hope they're the same thickness, you have to measure or you may end up with wrong Autofocus.

You've seen that 450D filters could be 0.60mm or 0.65mm.

And I hope you're thinking of removing LPF-2, not 1........

What bugs me is that Gary Honis and I are using the same formula to work out the correction, but are so different in our figures for Torx turns.

I'll sleep on it and have a fix in the morning.....yeah right!

Michael 

[Croz]

Michael, 

Yes, I mention lpf1 because that is the filter that is not being removed. I plan on removing lpf2, and whatever changes that I make either allows autofocus to work, or at least let's me manually focus without an additional filter.

I'm using the modded 600d only for astro, but want to occasionally do some Milky Way shots with a lens. Looks like I need to buy a micrometer.

Croz 

[Croz]

I've joined the yahoo group, and found this quote from Gary. This seams a difference from your calculations. If we are trying to get to 0.20 or 0.17 inwards turn, then we are just slightly over 1/12 like...3/24? I'm trying to help figure this out. Not at all being critical. And I might be way off as math was not ever my strong subject.

Thoughts?

A 1/12 turn clockwise moves the sensor 0.147 mm.  You use the paper strips to return the the sensor to its original position, then you do the 1/12 turn for the Full Spectrum modification

 

[michael8554]

17 hours ago, Croz said:

If we are trying to get to 0.20 or 0.17 inwards turn, then we are just slightly over 1/12 like...3/24?

I showed my calculations to derive 0.17mm is 4.08/12ths of a turn, but the whole point of this post is for people to chip in with corrections to my calculations.

So how do you get 1/12th turn is  0.17mm ?

Michael

[michael8554]

Latest is I have mounted an unmodded 450D on a 1280mm scope, and set focus on a barcode target.

Then without altering the focus, tried the modded 450 at different offsets in the focuser of 0.10mm,
0.20mm, and 0.30mm, measured with feeler gauges.

Best result was 0.20mm, not a million miles off the calculated value of 0.23mm.

So I feel the formula using an RI of 1.545 (the average of Schott Band Pass Filter thicknesses) is looking
suitable.

But don't rush off modding until I've modded the other 450D, using feeler gauge material to give me new
thinner shims under the sensor, and report back !

If that works out I'll try a 600D with the calculation for the "points"

Michael

[Croz]

I'm focusing on the thread pitch and the number of 0.147 that Gary has given for a 1/12 turn. This did not make any sense. I think I may have figured it out. If you divide .5mm by 12 you get 0.0417. Is this an error in that the numbers are mixed up?  Dyslexia? 

In any event, it looks like you are getting close. I have found some people stating that the new floating systems don't have much travel inwards. From your calculations above, it looks like just short of 5/12 of a turn to get to .20mm. Let's hope there is enough space to turn this much on all of the screws.

Kevin

[michael8554]

I agree that 0.147MM = 1/12th turn makes no sense to me, suggests the Torx screws have the same pitch as an M12 bolt you'd use on a suspension bridge.

Now I've only read that the Torx Pitch is 0.5mm, but there's only a small range that's physically possible for such small screws -you'll have to measure.

But there's Nagging Doubt that he knows something we don't know....

For a screw with 0.5mm pitch:

0.5mm = 1 turn = 12 points

1mm = 2 turns = 12/0.5 points = 24 points

0.147mm = 12/0.5 x 0.147 = 3.528 points - so where does his 1 point come from? A touch of Fake News to protect his Intellectual Property ?

I've tackled Gary on his calculation, he wouldn't be drawn........

Michael

[JamesF]

1 hour ago, michael8554 said:

suggests the Torx screws have the same pitch as an M12 bolt you'd use on a suspension bridge.

Seems odd to me, too.  As I read it, the ISO standard that defines metric threads gives a coarse pitch of 0.35mm, or a fine pitch of 0.2mm for M1.6 threads, whereas M12 is 1.75mm for coarse and 1.25mm for fine.  For those nominal diameters, any other pitch wouldn't be considered genuinely a metric thread as I understand it.

James

[JamesF]

Oh, as far as I was aware, Torx is only 6-point, too.  If you have a 12-point head, I don't think it's Torx.

James

[Croz]

The torx screw is 6 points. Gary uses divides this into 12 because he needs to do a 1/12 a point turn. See link.

http://dslrmodifications.com/650Dmod/650Dreinstallsteps2.html

I wish I could find something that documented the thread pitch of these screws to confirm that these are 0.5.

Edited October 28, 2018 by Croz