Jump to content

NLCbanner2024.jpg.2478be509670e60c2d6efd04834b8b47.jpg

Preserving Autofocus on Modded DSLR's


michael8554

Recommended Posts

Yes James, those figures show a reasonable range of pitches, shows that it will be important to measure the pitch when doing the mod.

But my Merc has some bolts with Star heads, and the bolt diameter is one size bigger than the head !

Yes, Torx is 6 points, but there are 6 pointy  "dips" in between the points, thats what Gary means by 12 points.

I was slightly exaggerating when I suggested M12 .......

Michael 

 

Link to comment
Share on other sites

  • Replies 53
  • Created
  • Last Reply

Top Posters In This Topic

10 minutes ago, Croz said:

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

Me too !

I found a Canon 5D Repair Manual online, one of the sensor screws is M1.7 x 0.3, the other two aren't specified.

But the 5D isn't a floating sensor model.

Michael 

Link to comment
Share on other sites

Ah, right, I see.  Counting indents as well as points gets you to the 12.

I can't see the screws having a 0.5mm pitch somehow.  If they're less than 2mm diameter that would make them look like a corkscrew!  If they're really M1.6 then I think I'd expect them to have a 0.35mm pitch, but I'd understand given that they're for quite fine tolerance adjustment if they used the fine 0.2mm pitch thread instead.

James

Link to comment
Share on other sites

3 minutes ago, michael8554 said:

Me too !

I found a Canon 5D Repair Manual online, one of the sensor screws is M1.7 x 0.3, the other two aren't specified.

But the 5D isn't a floating sensor model.

Michael 

I wasn't even aware that there was an M1.7 thread size.  And it seems an odd pitch too, as M1.6 coarse has a larger pitch than that, and M2 fine is smaller.  Did Canon just invent this stuff for themselves, I wonder?

James

Link to comment
Share on other sites

You gave me the idea of looking for the 600d parts manual.  And from there I was able to trace the part and find a picture of it. It certainly appears like the thread pitch is larger than what we would expect. Check it out and let me know if you agree.

Actually I'm editing the post now because if I count 10 threads and the size of the screw is 5.5mm then this might confirm your 0.5 thread pitch.

http://www.mk-electronic.de/eng/items/show/CB3-3959-000

Edited by Croz
Link to comment
Share on other sites

Wow, well done Croz !!

The thread certainly looks very coarse.

Now the convention is that "M1.7 x 5.5" SHOULD mean a 1.7mm diameter screw with a 5.5mm pitch, which this clearly isn't, so 5.5 is probably the length.

The screw in the image has a shoulder under the head, but without that I reckon there would be 11 turns in 5.5mm, or 0.5mm pitch.  

That's very pleasing to know.

Michael

Link to comment
Share on other sites

The aim is to restore Autofocus, and focus to infinity.

The Canon lenses I have tried on an unmodded 450D all go through infinity focus, so some latitude is built in, but the expensive "white" lenses may be more finicky.
Even wide open, perfect autofocus is depth of field symmetrical to the subject, which can be measured with the classic method described on the internet: an inclined ruler next to a target, the out of focus points on the ruler should be the same distance behind and in front of the target. Ideally the mod should mimic that, but some offset would still be acceptable, so long as the target is in focus ? Hmm....

Canon spec the distance from the lens mount to the image plane as 44.0 +/- 0.02mm.
For earlier non-floating cameras such as the 450D, there are 11 adjustment shims available, ranging in size from 0.03mm to 0.35mm.
I will be using cut-up feeler gauge material to correct the 450D I will be modding next. 
The calculated correction figure depends on the value of RI used, but the results using 1.517 (Gary Honis) and 1.545 (Schott) only vary by about 0.01mm, so fall within Canon's tolerance.
I checked a modded/autofocus not corrected 450D on a scope with offsets smaller and larger than the calculated correction figure, the results suggest that the calculated figure for the 450D of 0.23mm/0.22mm is acceptable.

For floating sensor models such as the 600D, the next problem is how many extra points to screw in the Torx screws ? 
All that depends on is the pitch of the Torx screws, which Croz has helped confirm is 0.5mm.
You need a good quality Torx screwdriver, an Allen key is not the same and could damage the screw.
Comment posted to the Gary Honis site: 
"I tried two different T7 drivers, and could not get a bite on the screws. I used a T6, and that worked fine. Canon could be using different sizes."
Well Croz has also confirmed they are T7, however a cheap T7 screwdriver with a misshapen tip might well be a poor fit.

Getting there

Michael

Link to comment
Share on other sites

8 hours ago, Croz said:

You gave me the idea of looking for the 600d parts manual.  And from there I was able to trace the part and find a picture of it. It certainly appears like the thread pitch is larger than what we would expect. Check it out and let me know if you agree.

Actually I'm editing the post now because if I count 10 threads and the size of the screw is 5.5mm then this might confirm your 0.5 thread pitch.

http://www.mk-electronic.de/eng/items/show/CB3-3959-000

Oddly though that's described as a "hex" screw, not Torx.  Hopefully that's just a translation error, given that it's a German site.

James

Link to comment
Share on other sites

Yes James, the Hex had me concerned as well. After googling this and looking for images I found the below link. It confirms these are torx.

https://picclick.fr/3x-CB3-3959-3x-CB3-5878-Canon-70D-202408421092.html

I'm still a little concerned about Gary's 0.147mm 1/12 turn calculation. On the yahoo group he has created he refers to this number several times. So it's no mistake as I mentioned earlier. 

Michael's calculations look acurate. We should should be just over 5/12 a turn on the 600d. This is a huge difference from Gary's numbers. If you take his .147 and 1/12 turn and you need to get to .22 all you need is 3/24 a turn, which seems to fine of an adjustment. 

Eager to figure this out!

Kevin

 

Link to comment
Share on other sites

Since the screws have to be removed, I think I'd get a thread gauge and measure the thread pitch before replacing them, just to be sure.  You can probably pick one up from Amazon for less than £10.

James

Link to comment
Share on other sites

 I reached out to someone I found that provides this service who seemed open to helping. He's been very kind in confirming our numbers. Not going to name him to protect the innocent. Without looking at our numbers he said removing LPF two would be approximately a half turn inwards.  My rough numbers show that we are just over 5/12 of a turn, which is close enough for me. No longer concerned about the difference in what Gary has stated.

Will be modding my camera in the next few days.

Kevin

Link to comment
Share on other sites

2 hours ago, Croz said:

He's been very kind in confirming our numbers

Well why didn't you speak to him weeks ago and save me all that effort?

Joking, I have enjoyed the challenge and your confirmation is the icing on the cake.

Please let us know how you get on with returning the sensor to the factory position, and the thickness of the 600D LPF2, and  how many extra points you screwed in.

Good luck 

Michael 

Link to comment
Share on other sites

Tolerances for the Torx Adjustment

Four ways of looking at it:

A)  In Gary Honis's shim adjustment for a 450D Full Spectrum plus Astronomik Clear replacement:

http://dslrmodifications.com/rebelmod450d6a.html

He found the basic spacer to be 0.35mm, and sensor tilt was adjusted by an extra 0.10 shim on one post, and an extra 0.05 shim on another post.

So don't expect the 600D screw heights to be the same, sounds like each sensor's tilt is adjusted during manufacture.

The smallest shim 0.03mm equates to 0.72 points.

Which suggests if you get your adjustments accurate to about half a point you should be okay.

B) Canon's tolerance for the lens mount to sensor distance is  +/- 0.02mm, or +/- 0.48 points - there's that half a point tolerance again !

So getting the 600D back to the factory setting by aligning the Sharpie marks (after setting the height as close as possible with your digital micrometer or marked piece of cardboard) should be straightforward.

C) Now my calculated correction for the 600D, if the LPF-2 is 0.5mm thick, is 4.08 points.

4 points is easy, but how many mm are those extra 0.08 points ? 

It's 0.0033mm, or about a 1/10 of the tolerance, so setting to 4 points in should be good enough.

D) Croz's contact said "half a turn inwards". 

Taken literally that's about 2 points more than calculated, which is 0.083 mm more.

IMO this is too far in to work.

Michael

Edited by michael8554
Link to comment
Share on other sites

Well I've modded my 600d. It was a very long night.  Everything is operational but I had a heckuva time getting everything right. Multiple things went wrong through the evening. The biggest problem I ran into was marking the screwheads. I didn't realize just how small the torx screws are. And I had purchased a very fine point black sharpie. I wish I had chosen a different colour that stood out better.

I had found some guides created by Gary on the three screws that gave the depths of 1.95mm for 2 left screws and 1.75mm for the one on the left. In the end I couldn't really see the marks on the screw which was a major problem. I had to go with the guides and then I went with a half a turn inwards. 

 Originally I was thinking about being very precise. The numbers of 5/12s at the end of the evening seem laughable. I was nowhere near that precise. 

 I ran into connecter issues with the camera not powering on, the card reader not working. And several other problems. I had to remove and re-seat the connecters at least a dozen times. And each time it seem like I put them back the same, only to have the same or different issues.

 But I do have everything completely operational. The concern I have now is do I have major sensor tilt. I have been testing my lens and it seems like I will be able to reach infinity focus manually.  My auto focus does snap the shutter but it appears like it is set to far forward. This was a nice to have for me, as I am only using the camera for astro.

im going to try and use the trial of CCDInspector which can check for tilt. Hopefully I'm close because the thought of opening up the camera again is depressing/frustrating. 

Good luck with your mods. The guys that do this as a service earn their money. 

Kevin

 

Edited by Croz
Link to comment
Share on other sites

Hi Croz

You've been very unlucky, I've never had to re-open a camera after modding to reseat ribbons, did you mark where the ribbon enters the socket with a Sharpie like Gary suggests, so that they go back in straight and to the right depth?

Yes the Sharpie marks on the screwheads and chassis are crucial if you want to avoid sensor tilt, you need to be within half a point of the mark. 

How did you end up thinking of moving  5/12ths in?  Is that what you calculated from your filter thickness measurement?

It's good though that you've confirmed one calculation for me,  in part D of my last post I thought that half a turn was too much and would leave the autofocus point too close to the camera. 

Could you send me the link to Gary's guide with depths of 1.95mm for 2 left screws and 1.75mm for right?

Interesting your Torx were T6, were they still 0.5mm pitch?

Michael

 

 

Edited by michael8554
Link to comment
Share on other sites

Micheal, 

 

I have attched the pic that Gary made for screw depth. I did not see anything about marking the ribbon cables for depth. Although this is a good sugestion and should be easy enough with everything working again. I get the feeling, unless I'm very lucky that I'll do this next time I open the camera up. Sky's are cloudy here. So it's a waiting game. CCDInspector works on the focus of a large field of stars. It calculates the differences in focus and gives a reading of x axis and y axis tilt. 

I should be able to level the sensor with some tedious work that would be best to avoided. Get a bright easy to see sharpie or 3 before you start the mod.

Kevin

IMG_0046.PNG

Link to comment
Share on other sites

Hi all

Here's a quote from page 1 of the mods:

"Comment by Oz: "Use a fine tip sharpie and draw a line on each of the ribbon cables where the ribbon exits the connector before disconnecting the cable. This will give positive visual feedback when reconnecting the cables to assure they are fully seated and square before locking the connector tabs."

Thanks for the diagram Croz.

For those of you puzzling over the diagram, the white "sticks" 1, 2, and 3, are a representation of sticks placed onto the sensor bed behind the Torx screws, with the screw height marked on them.

If you've also radially marked the head of the Torx screws and the chassis with a Sharpie, the sticks will probably get you close enough, the final adjustment being to match up the Sharpie marks.

Now I see this is marked as a "Baader Modification"

To me this means the LPF-2 being replaced with a Baader BCF filter.

I haven't discovered the thickness of the BCF.

Or whether it does preserve Autofocus, but Gary obviously feels it doesn't and gives it a small 1/12 turn tweak.

This is not the same as only taking out the LPF-2, you need 4/12 for this mod

Gary also says to tweak 1/12 turn for the "Full Spectrum + Astronomik MC Clear Glass".

So a retained LPF-1 and a Baader BCF are the same thickness as both filters removed and replaced with clear glass ?   Hmmmm .....

Michael

 

Edited by michael8554
Link to comment
Share on other sites

Today I modded a Rebel XSi (or 450D)

The sensor sits on 3 sets of shims, this one had:

Top Left 0.47 (0.35 + 0.08 + 0.04) mm

Bottom Left 0.39 (0.35 + 0.04) mm

Right 0.47 (0.35 + 0.08 + 0.04) mm

YOURS MAY BE DIFFERENT !!

The 450D LPF-2 was 0.65mm thick, as on all previous modded 450D's.

The correction for correct autofocus for that thickness of filter is 0.23mm

So I needed shims 0.47 - 0.23 = 0.24mm

and 0.39 - 0.23 = 0.16mm

I made new shims from feeler gauge material, two 0.20mm, and two 0.08mm, and reused two of the 0.04mm shims

0.24 = 0.20 + 0.04mm

and 0.16 = 2x 0,08mm

Your homemade shims will need two holes in, one for the screw and one to fit on one of the locating posts.

That's because it would be very difficult to get the tiny shims under the sensor plate and aligned with the screw holes.

Also you can't just make one hole and place the shim over one of the posts, because the bases of the location posts that the shims fit on are lower than the post the screws go in.

Here are Before and After shots showing Depth of Field, 135mm lens wide open.

You can see that the Autofocus range on this Canon 18-135mm lens is not perfectly centred on the inclined ruler on the Before shot.

But it is out by about the same amount in the After shot.

In both shots the barcode target is inside the Autofocus range.  

That IMO has proved the calculations used to correct Autofocus, next task is a 600D and those pesky Torx screws !!

 

***** EDIT:  after all these fancy tests I have neglected to try the obvious one - focus on a distant object.

        Autofocus was spot on, but manual focus through the viewfinder gave out of focus images.

       The focal point was closer to the camera, like the close-up tests below, but unfortunately the distant object was not in the depth of field window.

        The LiveView focused shot of the distant object was of course in focus.

        I will have to investigate what's going on here.... ******

****** EDIT 2:  Based on the 600D Floating Sensor mod I did next, the 450D sensor could be usefully moved closer to the shutter by using even thinner shims.

The correction figure for the 600D turned out to be 0.25mm for a 0.60mm thick LPF-2.

So for the 0.65mm thick 450D filter the shims need to be reduced by 0.65/0.60 x 0.25 = 0.27mm ******

 

Before.jpg.77db16d6621931fe355c6a53466f4fa1.jpg

Michael

After.jpg.2e0cc663da6e9682aeb61a7cf3af5abd.jpg

Edited by michael8554
Link to comment
Share on other sites

The autofocus midpoint on my Canon 18-135mm lens is nearer to the camera than the target, before and after modding.

Ive read an unconfirmed suggestion that this is a deliberate move by Canon.

The point and click user focuses on the target and expects everthing in front to be in focus too, not bothered about the background.

Are your lenses like this or was mine made on Friday?

EDIT:

The User Guide says: "In the Basic Zone modes, the camera will normally focus THE CLOSEST SUBJECT" - (my caps)

But my test were done in the Creative Zone on the rotary dial......

Michael 

Edited by michael8554
Link to comment
Share on other sites

I have now modded a Canon with Floating Sensor adjusted by Torx Screws, a T3i aka 600D.

I found it well worth while taking the front off, so that the top could be slid up to give better access to the Torx screw to the left of the viewfinder, and to the topmost ribbon.
It's 3 screws on the bottom, 2 screws at the front under the "Canon" sign, and the diopter adjustment for the viewfinder.
Then ease the front over the tripod boss and remove, there are no cables.
The top panel will then ease upwards, still attached by some cables.
I missed the holes in the ribbon cables that allow you to pull them in and out with a toothpick, only a couple on this model.

I found the LPF-2 filter to be 0.60mm thick, not the 0.50mm I had found in my research. Read my very first post with this in mind.
This means the adjustment to preserve Autofocus is 0.21mm.
Or tighten 5 more Torx "Points", after returning to the factory setting.
This is much closer to the "half a turn" that Croz's contact suggested for the 600D. I had rejected this as too far, based on my 0.50mm thick filter calculations.

Tests after reassembly showed that 5 Points was not enough. 
The Autofocus point was closer to the camera than the target.
So I opened her up again and moved to 6 Points, or half a turn - Croz's figure.


This figure gave correct Autofocus !!


So much for all the maths I had applied to this problem ;-<

However it does give a figure that can be applied to other cameras that may have different LPF-2 filter thickness.

If 0.60mm Thickness = 6 Points

Then apply 1 Point per 0.10mm of Filter Thickness.

Over and Out.

Michael

Link to comment
Share on other sites

This project has taken so long I'd forgotten why I started it......

I wanted to mod a 6D and preserve Autofocus, an expensive camera so worth doing once it's open.

And a good selling point if I ever resell it, daytime shots are usually okay with a Custom White Balance.

But have I used up all my enthusiasm?

Watch this space !

Michael 

Link to comment
Share on other sites

  • 1 year later...
On 01/12/2018 at 19:48, michael8554 said:

I wanted to mod a 6D and preserve Autofocus, an expensive camera so worth doing once it's open.

 

Hi Miachel, have you done 6D mod ? Did you manage to determine what is the focus distance connrection ? I have made 0,18mm but it is not perfect for closer objects. 

Edited by Sesjusz
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.