Sensor Corrosion Analysis and Fix [Merged]

[jaapv]

Advertisement (gone after registration)

Leather reattachment should not be any problem. Even if the camera returns with a flaw in that department -the original leather is normally  not  re-used-, recovering the camera is  a simple DIY project, using the leather from a number of reputed suppliers. One of them is http://www.cameraleather.com/

Personally I would request the camera be returned uncovered and apply the leather of my own choice.

[kristyansen]

vor 22 Stunden schrieb Einst_Stein:

That 0.2mm light path means the sensor location is off by about 0.1mm too close to the lens, assuming the glass refractive index is 1.6 and the air is 1.0.  This implies when you focus a relatively close object it will be projected to the image plain supposed to be for infinite (and infinite will always be out of focus).  This does not count the light dispersion effect due to the glass thickness. 

The light dispersion effect can not be fixed by the shim. 

The focus error will make the following object to be projected like the inifinite.

1: focal length 16mm:  Objects beyond ~2.5m can not be focused.

2: focal length 25mm: Objects  beyond ~6.2m.

3: focal length 28mm: object beyond ~8m

4: focal length 35mm: ~13m

5: focal length 50mm: ~25m

6: focal length 90mm: ~80m

I guess this issue is almost irrelevant given the M RF accuracy.

 

Dear all, 

My M9 experiences sensor corrosion now (announced to me by Leica today as I sent it for cleaning; I had the camera checked by Leica several times even during the free replacement period, and despite hard spots on the sensor it was several times confirmed by Leica that no corrosion existed, and I hoped it would be like this forever...). Either way I love the old sensor, which I think has a different color signature than the new sensor so I may consider such intervention. At the same time, in the future I intend to acquire a Leica Tri-Elmar 16-18-21, which of course I hope to have a high focus accuracy even at relatively close distances but which may go beyond 2.5m as per the lens focusing scale, and also I expect to have clarity edge-to-edge. For larger focus distances I see no issue. Two questions: 

1. What could be done to also ensure focus accuracy for super wide angle: glass change to 0.8 instead of 0.6? (I need no IR capability.)

2. How could such a service be offered also in Europe? I expect quite high time / cost / administrative / tax complications by sending my M9 to the US and back.

Thank you,

Cris

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[dllewellyn]

While a thinner ICF / Coverglass is beneficial for wide angle lenses, the M9 coverglass at 0.80mm is not particularly thick.  With a mirrorless camera that sets AF off the sensor, you have a certain amount of leeway to change thickness of the ICF.  With the M9, If you used a 0.60mm coverglass, then you would want to compensate by moving the sensor forward.  There are shims between the sensor and camera chassis so one could make a pretty good guess that you need to move the sensor about 0.10mm closer.  You would have to remove one or more of the stock shims and make your own to replace.  While I do have precision thickness metal sheets, I am really not sure that I would advise on going that route.  To make sure the sensor is in the right position, you need a special machine.  I think you are better off leaving the sensor in it stock position, but it is your camera to do as you see fit.

I don't know of anyone in Europe offering a fix.  As far as I know, it is just us (MaxMax) and Kolari though I would like to see photographic proof from Kolari

If you send a camera from outside the USA, you want to include a $1 invoice that says you are sending the camera in for repair.  Otherwise, we may get charged an import duty and would have to rebill back to you.  On return, I am sure your tax authorities will want to charge you VAT, but it should only be on the value of the repair - not the entire camera.

Regards,

Dan

[Einst_Stein]

1 hour ago, kristyansen said:

Dear all, 

My M9 experiences sensor corrosion now (announced to me by Leica today as I sent it for cleaning; I had the camera checked by Leica several times even during the free replacement period, and despite hard spots on the sensor it was several times confirmed by Leica that no corrosion existed, and I hoped it would be like this forever...). Either way I love the old sensor, which I think has a different color signature than the new sensor so I may consider such intervention. At the same time, in the future I intend to acquire a Leica Tri-Elmar 16-18-21, which of course I hope to have a high focus accuracy even at relatively close distances but which may go beyond 2.5m as per the lens focusing scale, and also I expect to have clarity edge-to-edge. For larger focus distances I see no issue. Two questions: 

1. What could be done to also ensure focus accuracy for super wide angle: glass change to 0.8 instead of 0.6? (I need no IR capability.)

2. How could such a service be offered also in Europe? I expect quite high time / cost / administrative / tax complications by sending my M9 to the US and back.

Thank you,

Cris

The distance table I gave in the early post is very pessimistic. It essentially assumes you have infinite enlargement and the lens is perfectly sharp (zero confusion circle). In reality, the "bottleneck" of the focusing beam at the focusing plane is a fuzzy circle. The off-focus blurring is far less than that idealized calculation. 

 So I take my 25mm ZM and M240 to check the more realistic impact of the 0.1mm light path. On my lens, near the infinite, if I turn the lens focusing ring to move the lens about 0.1mm, everything in 30m is blurring. This means the off-focus infinite is far more than the 2.5m. 

I then take the nearest object (0.5m) for the test, I check the blurring if I move the focus to make about 0.1mm focus-off, and check the image through the EVF with focus peaking  at the max enlargement, to my eye, I can't tell the difference. 

These tests indicate: 

1: at infinite distance, the 0.1mm light path difference cam make the object at X distance falls on the image plane supposed to be infinite. The X is far more than 30M. 

2: at nearest distance(0.5M), the 0.1mm light path difference is impossible to judge where is the optimal. 

Note that the above tests are based on the M240 EVF (or live view), which theoretically is applicable to M9. However, with M9 OVF, your focusing judge should be much less accurate. So that 0.1mm light path is even less a concern. 

I would assume the frequently seen OVF infinite off due to the OVF is much worse. Whoever really worry that 0.1mm light path, she/he should maintain the OVF accuracy all the time.  

[DandA]

Isn't a replacement 0.8mm ICF/coverglass for the M9 available instead of the 0.6mm, so that any concerns of focusing being even slightly "off" with say a 11mm or 15mm M lens (on the M9), is eliminated?

Dave (D&A)

[martinot]

On 9/14/2020 at 2:08 AM, Einst_Stein said:

I agree the best choice in the past was to trade-up, I paid $800 to trade it for M240.

That was a really good and generous trade in by Leica I think. Great improvement for just $800! Hard to make a better deal IMHO.

[BrianS]

Advertisement (gone after registration)

If Kolari intends to reposition the sensor it would be worth examining the impact of the new position on the offset microlens array that is special to the KAF-18500.

 

Leica+ONSEMI decided to use a 0.8mm BG55 filter for good reasons. This glass was not available when the M9 was introduced, Schott brought it out (re-introduced)  in 2011. The position of the sensor is critical, the KAF-18500 microlenses  just do not point straight up as those of other sensors. BG55 is a class 2 glass, with a sharp dropoff in transmission where you need it- at the IR. CCD's are much more sensitive to IR than a CMOS sensor, roughly 10x. Choosing an XNiteCC1 filter (a Class 1 glass that is very stable but has higher leakage in IR) might be fine for a CMOS sensor. On a CCD, the IR leakage will be more notable. Does anyone else remember how long it took Leica and ONSEMI to bring out the new sensor? Look back at the 2015 discussions on this forum.

[jaapv]

14 hours ago, dllewellyn said:

While a thinner ICF / Coverglass is beneficial for wide angle lenses, the M9 coverglass at 0.80mm is not particularly thick.  With a mirrorless camera that sets AF off the sensor, you have a certain amount of leeway to change thickness of the ICF.  With the M9, If you used a 0.60mm coverglass, then you would want to compensate by moving the sensor forward.  There are shims between the sensor and camera chassis so one could make a pretty good guess that you need to move the sensor about 0.10mm closer.  You would have to remove one or more of the stock shims and make your own to replace.  While I do have precision thickness metal sheets, I am really not sure that I would advise on going that route.  To make sure the sensor is in the right position, you need a special machine.  I think you are better off leaving the sensor in it stock position, but it is your camera to do as you see fit.

I don't know of anyone in Europe offering a fix.  As far as I know, it is just us (MaxMax) and Kolari though I would like to see photographic proof from Kolari

If you send a camera from outside the USA, you want to include a $1 invoice that says you are sending the camera in for repair.  Otherwise, we may get charged an import duty and would have to rebill back to you.  On return, I am sure your tax authorities will want to charge you VAT, but it should only be on the value of the repair - not the entire camera.

Regards,

Dan

0.10 mm is ten times the tolerance Leica uses for sensor positioning:  so the reshim will need to be 0.10 +/- 0.010 mm. The tolerance they use for measuring is ten times that again, so 1/1000th of a mm.

I would advise contacting UPS for the specific requirements to ships repairs from any country before actually sending the camera in.

[dllewellyn]

R: Brian  "CCD's are much more sensitive to IR than a CMOS sensor, roughly 10x. "

Nope.  100% wrong there.  Both made of silicon.  CCD's a cheaper to make than CMOS but use more power than CCD and have slower readouts.  Early CMOS sensors used a lot of the sensor space for the circuits so pixel size was smaller which means noise was greater (smaller photon wells = less dynamic range).  CMOS sensors cost more to design but are less expensive in the long run.  For a long time now, when you look at the chip under a microscope, you can see the pixels are edge-to-edge.  And with Back Side Illuminated (BSI) sensors, now even more so.

When you buy glass from Schott (and I do), you can specify, within reason, whatever thickness you want.  If Leica/Kodak/Truesense/On Semi wanted to by 0.5mm glass, it isn't a problem.  And they could have easily added coatings to change the UV and IR response.  It's not that big of a deal.  Plenty of companies that offer coating services and you can model some pretty complex response curves.  And if they had added the coatings, they wouldn't have had the 'corrosion'.  Maybe the datasheet specified coatings, but, at least for a lot of cameras, the coatings aren't there.

We were the first company that figured out how to re-purpose a consumer camera for vegetation remote sensing by making a filter that passed plant absorption (blue or red light) to one color channel and a plant reflection channel (NIR) to another color channel.  Plants reflect the NIR because the longer, lower energy wavelengths don't have enough energy to support photosynthesis.  So if you look at how a plant is absorbing and reflecting light, you can determine things like biomass, plant health, etc.  We also knew how to measure the camera spectral response so we knew how the RGB channels opened and close in the visible and NIR.  It turns out that the red channel is open in the red and stay open throughout the NIR while the blue and green channels don't open up until around 800nm.  So we designed a filter that passed visible blue and green light, blocked red light, passed NIR from 680nm to 800nm and then blocked light above 800nm - and that was one filter on one piece of glass.  That way, only the red channel saw NIR.  Once you had a picture with that information, you could run the picture through a computer program to calculate things like NDVI (Normalized Difference Vegetation Index).

[BrianS]

> CCD's a cheaper to make than CMOS but use more power than CCD and have slower readouts. 

This is 100% wrong. CCD's are more expensive to make than CMOS, which is why CMOS mostly took over. You can mix the light-sensitive areas with the A/D circuits and processing- all on board the chip.

 

Dan- as you stated on petapixel, you don't believe in looking at Data Sheets for performance.

You do not know what you are talking about. If you bothered to look at the formulation of the sensors, you would see that the chemistry changed over the years to reduce IR sensitity. BUT- go ahead and use XNiteCC1 as a cover glass for the M9 sensor, post some color images without using a Hot Mirror filter over the lens.

But- you don't believe the Data Sheets from the manufacturer, and don't look at spectral response, and don't look at the chemistry of the individual sensors. I have, you have not.

I don't know how many sensors you have designed, how many types of glass that you have designed. I know that you don't bother reading data sheets. 

Why anyone would trust their M9 and M monochrom to you is a mystery to me.

Edited September 15, 2020 by BrianS

[dllewellyn]

No Brian.  The CCD chip itself is a simpler, less expensive chip to make but you need other support chips making the CCD design more expensive in the long run. For small manufacturing runs, CCD can be less expensive.  The CMOS chip is more complex and more expensive to make, but it needs fewer support chips so less expensive for volume runs.  CCD's also use more power, are slower and have read-out issues that CMOS does not have.  Where you get your CCD's are 10x more sensitive to IR light is a mystery.  Maybe you can show me a data sheet on that?  Sounds like you don't know much about sensor design.  I am no expert, but I know a few things.

I never said I don't believe in data sheets.  I just don't believe them when they are incorrect.  Usually the data sheets are correct.  For that particular Kodak sensor and at least for a lot of sensors, the data sheet doesn't match the actual data.  Is what it is. 

Take your cheap shots elsewhere.

Edited September 15, 2020 by dllewellyn

[dllewellyn]

RE 0.10 mm is ten times the tolerance Leica uses for sensor positioning:  so the reshim will need to be 0.10 +/- 0.010 mm. The tolerance they use for measuring is ten times that again, so 1/1000th of a mm.

Yes, but there is a difference between measuring and the actual shim.  I don't know how far down their shims go, but I doubt below 0.01mm.  I measured the thickness of typical office copy paper and mine is 0.10mm thick.  A shim that was 0.001 thick would be so thin you could almost see through it.  Even handling it would be a challenge.  A shim 0.10 +/- 0.01mm is no big deal - I have made them lots of times.  Precision stock doesn't cost much.

[jaapv]

3 hours ago, dllewellyn said:

No Brian.  The CCD chip itself is a simpler, less expensive chip to make but you need other support chips making the CCD design more expensive in the long run. For small manufacturing runs, CCD can be less expensive.  The CMOS chip is more complex and more expensive to make, but it needs fewer support chips so less expensive for volume runs.  CCD's also use more power, are slower and have read-out issues that CMOS does not have.  Where you get your CCD's are 10x more sensitive to IR light is a mystery.  Maybe you can show me a data sheet on that?  Sounds like you don't know much about sensor design.  I am no expert, but I know a few things.

I never said I don't believe in data sheets.  I just don't believe them when they are incorrect.  Usually the data sheets are correct.  For that particular Kodak sensor and at least for a lot of sensors, the data sheet doesn't match the actual data.  Is what it is. 

Take your cheap shots elsewhere.

Whatever else, the Leica M9 cost - Leica, not customer- 1500 Euro at the start of production and ran up to 1800 in the end Why? Special microlenses and small production runs. The price of the material is irrelevant - the machine it is made on bt ASML is a multi-miliion piece of gear, not to mention the costs of setting it up for a few sensors.

[dllewellyn]

A friend who once owned a M9 Monochrom told be once that he knew some upper Leica management that said, at least for awhile, the Monochrom was net loss for Leica.  Small manufacturing runs are hard to do for something complex like a camera.  I also talked to a guy at a government research agency in New Mexico who told me they were partially responsible for instigating the Monochrom by giving them some financial (?) encouragement.  I am just repeating what I have been told and have no way of knowing if there is truth to it.

Edited September 15, 2020 by dllewellyn

[Chuck Albertson]

4 minutes ago, dllewellyn said:

 I also talked to a guy at a government research agency in New Mexico who told me they were partially responsible for instigating the Monochrom by giving them some financial (?) encouragement.  I am just repeating what I have been told and have no way of knowing if there is truth to it.

Los Alamos. They wanted a b&w camera for the MiniMax portable x-ray machine they developed (it won a few engineering awards). Leica were thinking about an M Monochrom before then, but it was that order that gave them the last little push over the cliff.

[dllewellyn]

Hi Chuck,

Yes, that was it. I just wasn't sure I wanted to name the place.  I talked to the guy a few times because I have also converted cameras to see X-Rays directly and also sell monochrome converted cameras to a company in Poland that makes large X-Ray machines.  In fact I have an X-Ray generator here and also sell X-Ray phosphors.  Small world I guess!

[dllewellyn]

A bit off-subject, but here is a picture of an electrical cord strain relief with a camera I converted to see X-Ray's.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[BrianS]

5 hours ago, dllewellyn said:

No Brian.  The CCD chip itself is a simpler, less expensive chip to make but you need other support chips making the CCD design more expensive in the long run. For small manufacturing runs, CCD can be less expensive.  The CMOS chip is more complex and more expensive to make, but it needs fewer support chips so less expensive for volume runs.  CCD's also use more power, are slower and have read-out issues that CMOS does not have.  Where you get your CCD's are 10x more sensitive to IR light is a mystery.  Maybe you can show me a data sheet on that?  Sounds like you don't know much about sensor design.  I am no expert, but I know a few things.

I never said I don't believe in data sheets.  I just don't believe them when they are incorrect.  Usually the data sheets are correct.  For that particular Kodak sensor and at least for a lot of sensors, the data sheet doesn't match the actual data.  Is what it is. 

Take your cheap shots elsewhere.

Okay- you are  a hack. The Data Sheet is what an engineer goes by. The last time I worked on a VNIR Sensor, the CCD had 10x the response in IR than did CMOS. The dopants and specific design of the sensor affects the spectral response. To state that both are made of silicon shows a lack of understanding of sensor technology. The filter that you plan to use has too much IR leakage for the M9, and will give results closer to the M8.

So-  you announced that Kodak and Leica were "Stupid, Stupid, Stupid"- and want me to take cheap shots somewhere else while you hang out on the Leica User Forum.

Edited September 15, 2020 by BrianS

[Chuck Albertson]

1 hour ago, dllewellyn said:

Hi Chuck,

Yes, that was it. I just wasn't sure I wanted to name the place.  I talked to the guy a few times because I have also converted cameras to see X-Rays directly and also sell monochrome converted cameras to a company in Poland that makes large X-Ray machines.  In fact I have an X-Ray generator here and also sell X-Ray phosphors.  Small world I guess!

Oddly, the MiniMax was one of the lab's few "products" that was developed for public sale. Most of their other stuff, you can't buy in the stores (probably just as well).

[Advertisement]

Hallo,

Look here: Leica M

[dllewellyn]

Okay- you are  a hack.

OK - I guess a hack with 25 years experience, well over 10,000 customers and one of the few companies that that can actually measure camera spectral response.  Give me a break.  I have been measuring camera spectral response for 15 years including CMOS and CCD.  I have been hired by the US Army to measure camera absolute irradiance spectral response, including various lenses down into the UV turning the cameras into calibrated spectrophotometers.  I suppose you have done that as well.

Your clear lack of understanding is brilliant - just brilliant.  BTW, still waiting on those data sheets that show CCD  and 10x the IR response of CMOS.  *crickets".