Defective M9

[jonoslack]

HI Georgl

These effects can happen from one shot to another - most likely this particular sensor had no defect as it was tested. I was asking a Leica-engineer because I had the same problem with my M8 after about one year and got a new sensor-board, too. He told me that Jenoptik makes the sensor-board and therefore also selects the Sensors from Kodak, as I was told they reject 2 of 3 sensors from Kodak! Then the mapping is done and the finished board is send to Leica.

This particular effect seems to be more problematic because it effects the readout in a whole column instead of just one "dead" or "hot" pixel, it seems as mapping doesn't help - why would they replace a 2k$-board otherwise?

I think this was because in the past only Jenoptik had the mapping facility, so the boards had to be removed and sent back to Jenoptik. Now (as I understand it, and as Brad's mails suggest) Leica have the facility to do the mapping (and columns can be dealt with), so it's no longer necessary to replace the board.

 

[Seascape]

In spite of popular belief, ISO9001 has nothing to do with actual quality - it became even popular to use it as an "excuse" for cost-cutting by actually buying cheaper components. It doesn't define actual quality standards but only ways to test it - they could simply define that such banding is normal and therefore every single of these cameras would pass qualit control... In fact, I know of many smaller companies that rather save the money for the quite expensive ISO9001-certification and invest it into better materials/machines...

Exactly, ISO only requires companies to "define" their level of control, and then monitor to that level......ISO does not set industry standards that companies must comply to. Many companies have in fact lowered their level of control, to one that can be achieved, sometimes with less stringent quality standards.

Not saying this is the case with Leica, however an ISO certification only requires companies to meet self defined standards, regardless of how high the bar is set.

 

[f1point0]

Sorry Brad about your new M9...

Is there a trick to getting the column effect to show up more readily, e.g. underexposure, a certain ISO, etc. Also, does the column usually show up in a particular spot on the sensor or can it show up anywhere?

Just want to check mine and make sure it is OK. Thanks.

 

[Lloyd]

Sorry Brad about your new M9...

Is there a trick to getting the column effect to show up more readily, e.g. underexposure, a certain ISO, etc. Also, does the column usually show up in a particular spot on the sensor or can it show up anywhere?

Just want to check mine and make sure it is OK. Thanks.

On mine the effect was always visible. However, you had to look very closely to see it at lower ISO settings, especially if there were a mixture of colors or textures in the area. The effect (defect) was always in the same place (I had two lines), and was most visible at higher ISO settings, and especially against solid and/or dark backgrounds.

In the sample below, the lines are fairly apparent, and this was shot at ISO 2500, less so in the lower image.:
View attachment 24910

 

[f1point0]

Thanks Lloyd. Am I seeing correctly that the column/line on the right is lighter than the one on the left? In the second image I can't even see the second line.

 

[simonclivehughes]

Exactly, ISO only requires companies to "define" their level of control, and then monitor to that level......ISO does not set industry standards that companies must comply to. Many companies have in fact lowered their level of control, to one that can be achieved, sometimes with less stringent quality standards.

Not saying this is the case with Leica, however an ISO certification only requires companies to meet self defined standards, regardless of how high the bar is set.

The point of my ISO9001 remark was to suggest that Leica, who had previously enjoyed a reputation for fine quality mechanical workmanship, has entered into new territory, and obviously is at a bit of a loss with regard to quality management of electronic devices.

Cheers,

 

[Lloyd]

Thanks Lloyd. Am I seeing correctly that the column/line on the right is lighter than the one on the left? In the second image I can't even see the second line.

Yes, that's correct. The one on the left was more prominent. Interestingly, that line corresponded to a hot pixel in the upper left quadrant, and which was only visible on in-camera jpgs.

I started a thread about the line in the sensor of the M9, and posted a number of sample images. However, that thread seems to be MIA.

 

[jonoslack]

Yes, that's correct. The one on the left was more prominent. Interestingly, that line corresponded to a hot pixel in the upper left quadrant, and which was only visible on in-camera jpgs.

I started a thread about the line in the sensor of the M9, and posted a number of sample images. However, that thread seems to be MIA.

HI Lloyd
I think the defect is nearly always connected to a dead pixel - what I don't know (and can't find out) is whether, if you map out the dead pixel, the column defect goes away as well.

all the best

 

[jonoslack]

The point of my ISO9001 remark was to suggest that Leica, who had previously enjoyed a reputation for fine quality mechanical workmanship, has entered into new territory, and obviously is at a bit of a loss with regard to quality management of electronic devices.

Cheers,

How can you deduce that if the problems are happening after the cameras are shipped? Which is certainly the case some of the time, if not all of the time.

I don't see how they are able to do precognitive QA. Mind you - putting pixel mapping in the firmware (like everyone else does) sounds like a fine idea!

 

[GrahamWelland]

I wonder how different or effective 'in the field' remapping (i.e. in camera by the owner) vs 'back to base' sensor remapping by the manufacturer is.

The reason I ask is that there are a few cameras & backs that can remap the sensor but I have to assume that their effectiveness is somehow different to a more rigorous sensor mapping on the bench by the manufacturer. Is an in camera remap more aggressive and likely to map out more pixels over time?

It's interesting to note that only a few manufacturers provide a menu option or default remap feature in the camera (I noticed that my GF1 does have this). Most of the mainstream makers do not - in my case that's Nikon & Leica and in the past it was also true of my Kodak digital back. I don't recall it being available on Canon or my Ricoh either.

 

[simonclivehughes]

How can you deduce that if the problems are happening after the cameras are shipped? Which is certainly the case some of the time, if not all of the time.

I don't see how they are able to do precognitive QA. Mind you - putting pixel mapping in the firmware (like everyone else does) sounds like a fine idea!

Agreed, Jono, as I mentioned in my original post, if it is an intermittent problem then it can be a no-win situation. I also agree that pixel-mapping should be in firmware.

In many industries, it is the customer that establishes and drives the QA goals. For example, my company services Set-top boxes and our customers (and the manufacturers whose equipment we are authorized to repair) demand a zero defect level.

One would have thought that Leica had learned a lesson from the issues that had with the M8 series to implement a more stringent QA policy, including pixel-mapping so the customer can cope with issues that occur after the sale.

Cheers,

 

[jonoslack]

It's interesting to note that only a few manufacturers provide a menu option or default remap feature in the camera (I noticed that my GF1 does have this). Most of the mainstream makers do not - in my case that's Nikon & Leica and in the past it was also true of my Kodak digital back. I don't recall it being available on Canon or my Ricoh either.

Well, actually I think almost everyone does it . . . it's just that you don't necessarily have an option - I'm pretty sure Nikon and Sony have it, almost certain that Canon have it, and I know that Olympus and Pentax have it! They don't necessarily have a menu option though - I think most cameras do it automatically ever period of time (week, month, whatever). Try doing a bit of googling on this one and I think you'll find that pretty much everyone does it.

 

[f1point0]

There is a reason most of the big companies have moved away from CCD's and adopted CMOS as the sensor of choice. CMOS chips are easier/cheaper to produce with much fewer defects and none of the temperamental problems that come with manufacturing CCD's. I think the issues that we are seeing is the "con" of going with a CCD, but perhaps for technical reasons, the CCD is necessary in the rangefinder design.

The "pro" with going with the CCD is that I can clearly tell an M9 image apart from my Canon 5D Mk II images. There's just something about the way that a CCD draws that looks better, IMHO. Of course we need to give credit to amazing Leica glass, but even when using Leica R lenses on a Canon, you can still see the difference because of the CMOS chip versus the CCD in the Leica M9/M8.

 

[carstenw]

There is a reason most of the big companies have moved away from CCD's and adopted CMOS as the sensor of choice.[...]

The "pro" with going with the CCD is that I can clearly tell an M9 image apart from my Canon 5D Mk II images.

Putting those two parts together, one comes up with the statement that there is a reason the very high IQ companies (MFDB, Leica) have stuck with CCD. There is just something about that low ISO CCD look. Maybe it can be done with CMOS, but it hasn't yet, to my knowledge.

 

[GrahamWelland]

pixel mapping:

Try doing a bit of googling on this one and I think you'll find that pretty much everyone does it.

I did. They don't (Nikon/Canon at least).

I know with my Nikon gear that it really has to go back to Nikon service to get rogue pixels remapped. I assume that there's either a hidden service option or that they have remapping software that reloads the sensor mapping table. I have a few hot pixels on my D3x that appear occasionally but the raw converters will often map these out as part of the rendering and they also appear sensitive to the lighting of the pixels surrounding them. Touch wood - I've never had this happen with the Leicas so far, although the M9 raises the probabilities for this just due to pixel count.

Btw, there is some credibility to the cosmic ray effects on the sensor (and you too!). Something to be aware of if you travel a lot with your camera.

 

[jonoslack]

pixel mapping:


I did. They don't (Nikon/Canon at least).

Odd that - I did - and they do . . . not necessarily in the menus (as I said). Usually on a time basis, with a service shortcut.

But . .. the point is that everybody does it, whether it has to go to a service point or not.

 

[GrahamWelland]

Odd that - I did - and they do . . . not necessarily in the menus (as I said). Usually on a time basis, with a service shortcut.

But . .. the point is that everybody does it, whether it has to go to a service point or not.

The last point I absolutely agree with. It's the ubiquity of this function being available to the end customer where I think we disagree.

Anyway, the point is that it certainly would be a valuable option on the M9 that we could use as the camera ages without having to send the camera for servicing. :thumbup:

 

[bradhusick]

Sorry Brad about your new M9...

Is there a trick to getting the column effect to show up more readily, e.g. underexposure, a certain ISO, etc. Also, does the column usually show up in a particular spot on the sensor or can it show up anywhere?

Just want to check mine and make sure it is OK. Thanks.

In my case the line showed up immediately and for all of the frames. I only took about 15 photos total when I discovered the problem, packed up the camera and sent it to Leica.

 

[Lloyd]

In my case the line showed up immediately and for all of the frames. I only took about 15 photos total when I discovered the problem, packed up the camera and sent it to Leica.

Same with mine, there from the get-go. I had the same with my first M8, which I bought the first day of the M8 release.

 

[Peter Klein]

Ouch, Brad, that must have been so frustrating. It's too bad that one bad pixel creates an entire line in the DNG files. And too bad that it requires a trip back to Leica (hopefully NJ, not Solms!). My humble Olympus E-510 has a hot/dead pixel mapping routine that takes care of such issues.

Looking at a couple of the samples, I had one thought. We used to have issues like that in the film days. We called them "scratches."