SD Card Tests with and without M9

[Robert_M]

Advertisement (gone after registration)

I read the thread for awhile. I think I understand the conclusion, but I may be missing something, is there a way to avoid the crashing before Leica comes up with some big solution?

 

Excellent question. I don't have a reliable answer to that. Sorry. But I would like to come up with a work around or procedure avoidance that works for my shooting. And, I'm really not that demanding of a user.

 

RM

[Advertisement]

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

[k-hawinkler]

Jamie,

 

You are certainly right about the stability of Windows (or lack of). I will certainly try repeating these tests on another OS system when I find the right tools and the time. I hope to find some time this weekend to try that.

 

You are right that this test only examines the USB and SD interface and bypasses all the image processing. But that is the virtue (and intent) of the test. When diagnosing problems, it is much easier to isolate one parameter at a time for tests. That is why I can rule out some of the other functions in coming to my conclusion.

 

I don't think that K-H's procedure is testing the same thing, as I explained in a post above.

 

In post #17 of this thread, I summarized what I know about Leica being aware of the problem and working on the issue. Please take a look at that.

 

Regards,

 

RM

 

 

Hi Robert,

 

I am not so sure as you it doesn't. I am preoccupied right now with something else.

But I will work out a more detailed version with additional stuff I didn't get into for lack of time.

Please, give me a couple days.

 

With your approach I would like to learn what those error messages really mean, towards the end, when the memory cards are getting filled up?

 

On the Mac I get an error message too, namely, card full, can't write entire file, so the Mac doesn't. No big deal.

 

The first thing we really have to define is, what do we want to measure? I refuse to be limited by the peculiarities of a particular tool.

 

Best, K-H.

[Chipkahn]

Thanks, well my last crash was so bad that I had to send my camera back to Leica for repair so we will see what they suggest after they fix it. I have the email address of the technician in New Jersey who worked on my camera if you want it. Chip

[scrubs]

Thanks, well my last crash was so bad that I had to send my camera back to Leica for repair so we will see what they suggest after they fix it. I have the email address of the technician in New Jersey who worked on my camera if you want it. Chip

 

 

Hey Chip, could you PM that over to me please, will be helpful if I ever decide to do it.

 

 

Cheers

[Robert_M]

Hey Chip, could you PM that over to me please, will be helpful if I ever decide to do it.

 

 

Cheers

 

Chip,

 

Yes, me also, if you could. Thanks.

 

RM

[Robert_M]

Hi Robert,

 

I am not so sure as you it doesn't. I am preoccupied right now with something else.

But I will work out a more detailed version with additional stuff I didn't get into for lack of time.

Please, give me a couple days.

 

With your approach I would like to learn what those error messages really mean, towards the end, when the memory cards are getting filled up?

 

On the Mac I get an error message too, namely, card full, can't write entire file, so the Mac doesn't. No big deal.

 

The first thing we really have to define is, what do we want to measure? I refuse to be limited by the peculiarities of a particular tool.

 

Best, K-H.

 

 

Take your time. Don't worry, I sort of doubt the issue will die in a few days....

 

Best,

 

RM

[k-hawinkler]

Advertisement (gone after registration)

You're very welcome Karl-Heins (did I spell that right?), I hope they work as well for you as they do for me! If they made CF cards I'd buy those too.

 

As for other points of knowledge, this is more of a hunch from experience; I can't afford to have the M9 lose shots at a wedding.

 

As an aside, in truth, my Nikon D3 is giving me more trouble than the Ms these days. There's something about the older 85mm AF lens on the D3 that messes up electronic communication with--you guessed it--the cards in the camera, even though it has one of the fastest data processing paths ever! The camera locks, with a "CD error," in the LCD, and I have to pull the CF cards and sometimes the battery to get it to be useful again. The camera then starts numbering every shot at "0001"-- and it only happens with that one model lens, and then only intermittently. Now that's annoying!

 

Gotta love digital cameras...nothing like relying on computers for reliability (which is why there's always an M6 and M3 in my bag or nearby)

 

 

Hi Jamie,

 

Many thanks.

 

You have been very creative in inventing a new spelling of my first name that I hadn't seen before.

If you change the s to a z, then you've got it perfectly.

 

Concerning CF cards. I have a few SD/SDHC to CF Card Adapters. The ones I have are not made out of plastic but instead have metal covers. I ripped those off. One card still works, the other died in surgery. I also have a few unmodified cards. I was trying to increase the range of my Eye-Fi cards. Anyway, I stuck one of those SDHC Gold cards into an adapter and then into my Nikon D3. The good news is it works, the bad news is it's really slooooow. Typically it takes my D3 about a second to write a compressed .NEF image from buffer to an CF card. With the Gold card and adapter it takes about 5 seconds to transfer a single image. Bummer.

 

I have both the D3 and the older 85/1.4 AF Nikon/Nikkor lens. That combination seems to be working fine for me. Have you ever thought of sending both items in to Nikon for repairs? I once sent in a D40 on the last day of the warranty period. It came back perfectly, free of charge. Nikon ripped out the entire guts and kept only the outer shell. That D40 has been perfectly working ever since. I am sure Nikon would give you a cost estimate before you had to commit. Good luck.

 

Best, Karl-Heinz

[t024484]

I've not heard of any 4 quadrant designs. If you look at KAF-39000 on that list http://www.kodak.com/global/plugins/acrobat/en/business/ISS/datasheet/fullframe/KAF-39000LongSpec.pdf you will see that it is the size of 2 M9 sensors on a single chip. It is 2x the area of the M9 sensor, has about a 1 fps frame rate and still uses the 2 halves architecture. I could well imaging (but I don't know) that the M9 sensor uses some of the same development and tooling masks. One could speculate that a scaled down version of that ccd (by 1/2 area) would allow a 2 fps rate.

 

If you have any reference or firm information on a 4 quadrant design for a Kodak ccd sensor, I'd be interested in hearing about it.

RM

 

Hi Robert,

 

You are simplifying things by stating that the KAF 39000 39 MPixel sensor has a 1fps readout time, so an 18,5 MPixel sensor would allow for 2fps.

 

The 39 MB sensor has indeed a nominal readout time of 1077 msec, but on top of that there is a nominal flush time of 260 msec to empty all cells before a new exposure can be made, and an integration time to determine the dark current level of roughly the same time.

On top of that you have the exposure time for the duration of the shutter.

When you add the first 3 elements and divide them by 2, for having only 18,5 Mpixels, the time will be 800msec for the sensor alone.

Added to that comes some 250 msec for the time the shutter is moving, this only in case of exposure times shorten then 1/40 sec.

 

This brings us over 1 seconds per exposure for a 18,5 Mpixel sensor. It is also physically impossible to do 2 pictures/second with the technology that Kodak offers with just 2 quadrants.

As a matter of fact, all Kodak sensors with 6.8 micron sensors, have the same maximum frequency of 24 Mhz which already forces a A/D converter to make a conversion within a dazzling fast 40 nsec, so don't expect anything faster for the M9.

Also take into account that speed is the enemy of noise so to improve on noise you better reduce speed, which makes the conversion time even longer.

 

 

For all these simple physical reasons, I think it is a safe bet to assume that the M9 sensor is roughly composed of two M8 sensors with a total of 4 (vertical) quadrants.

 

Hans

[t024484]

In the Anatomy of the Leica M8's Power Consumption, Mark Norton showed that between firing the shutter and the moment the first curtain was open, some 65 msec was consumed. On top of that came the shutter opening time plus the time to close the shutter, in total some 100 msec for a shutter time of 1/30 sec.

 

If I am correct this was for an early M8 with the shutter still capable of gong to 1/8000 sec. The shutter used for the M9 is slower, and goes to a max of 1/4000 sec.

I kept on the safe side and took a time of 250 msec for the whole cycle because of this slower shutter.

After some thoughts, I think this figure is much too conservative and can be reduced to a lower figure.

But even if it would be 100 msec, with a now reduced cycle time of 900 msec it still wouldn't allow for 2 pictures per second with two quadrants.

 

Hans

[Jamie Roberts]

[quote name=k-hawinkler;1785924{snipped}

I have both the D3 and the older 85/1.4 AF Nikon/Nikkor lens. That combination seems to be working fine for me. Have you ever thought of sending both items in to Nikon for repairs? I once sent in a D40 on the last day of the warranty period. It came back perfectly' date=' free of charge. Nikon ripped out the entire guts and kept only the outer shell. That D40 has been perfectly working ever since. I am sure Nikon would give you a cost estimate before you had to commit. Good luck.

[/quote]

 

Karl-Heinz (you know, I meant to type a z but it was just a typo that I didn't even catch ,

 

My D3 & 80 1.4 have already been back to Nikon once... no joy. So I know I have to send it again, but I'm not the only one with the problem and that camera lens combo, so watch out for the issue if you have the lens. Evidently the new 80 1.4 doesn't have this problem,,,

[k-hawinkler]

Karl-Heinz (you know, I meant to type a z but it was just a typo that I didn't even catch ,

 

My D3 & 80 1.4 have already been back to Nikon once... no joy. So I know I have to send it again, but I'm not the only one with the problem and that camera lens combo, so watch out for the issue if you have the lens. Evidently the new 80 1.4 doesn't have this problem,,,

 

 

Hi Jamie,

 

What a bummer. Thanks for the heads up.

I will then keep a careful eye on that D3/85 combination.

It's 85 mm not 80, right?

 

Best, K-H.

[scrubs]

OK I just encountered a problem I have not had before. After just reading a post by

someone who had lockups at a wedding. It made me think that when I have had lock

ups it seemed down to buffer problems, Even though I was not rapidly shooting,

but my slow pace may be even too fast for the camera to handle in it's tiny

tortoise slow buffer.

 

When shooting in single mode DNG uncompressed + JPG it takes 15 seconds

for each photo to clear the buffer. I have a choice of waiting a full minute for buffer

to clear or wait 15 seconds in between shots I believe this makes the buffer crap out.

In single mode I can only shoot 4 shots to fill the buffer. So I switch to continuous mode

which I know can handle 7 shots in the buffer, with the intention of timing how long it takes

the buffer to completely clear. I fire of seven shots.. buffer takes it's time, but eventually clears..

 

Press play and last image is displayed fine. Go to delete image and I get a

'Message cannot be deleted' error... That's new to me..

 

So I scroll back to previous image all I get is a blank black frame displaying

shutter speed = 1 ISO = 0 goto previous image all the same - approx 50 images

only two were displaying correct shutter speed and ISO settings even though

there was no photo displayed. I should note that these 50 photos were not

all photos I had just taken - I had taken some prior and were already on the card.

by the time I had scrolled back round to the only visible photo, that had dissapeared

also.

 

Turned camera on and off (did not pull battery) and all images re-appeared

and was now able to delete.

 

Could someone experiment filling the buffer in continuous and single mode

to see if you experience or can induce something similar.

[k-hawinkler]

Scrubs,

 

Many thanks. Excellent observation.

Question:

So, this happened once, right?

My experience is that this behavior can't be repeated on demand, right?

What did you use, JPG fine?

Or did you use JPG basic?

 

Best, K-H.

[jaapv]

This is very interesting information, as it is definitely not normal M9 behaviour. The camera should take about three seconds to write an image from its buffer. Please do the following test: In C mode press down the shutter and hold it down after the buffer is full. The shooting rate should slow down from 2 fps to about 1 frame per three seconds.

If the cameras that are misbehaving do indeed write far slower than they should under some circumstances, or do so intermittently, it may indicate part of the cause.

[scrubs]

Scrubs,

 

Many thanks. Excellent observation.

Question:

So, this happened once, right?

My experience is that this behavior can't be repeated on demand, right?

What did you use, JPG fine?

Or did you use JPG basic?

 

Best, K-H.

 

 

Hi mate here are all my settings:

 

Not all necessary most likely but at least you have in case. JPG set to fine.

 

I shot in Aperture priority mode

 

Lens detection = Auto

Advance = Standard

Auto ISO setup = 1/30 - 2500

Sharpening = Standard

Color saturation = Black & White

Contrast = Medium High

Bracketing setup = 3/-/0/+ / 0.5 EV

Exposure comp setup = SET menu only

Monitor brightness = Standard

Histogram = STD/ Clip +/-

Auto review = 5s / Histogram

Auto power off = 2 min

Flash sync = 1st curtain

Auto slow sync = lens dependant

Color management = sRGB

DNG Setup = Uncompressed

 

White Balance = Manually set to 6000k

Compression = DNG & JPG Fine

Resolution = 18 MP

Exposure comp = +0

EXP Bracketing = Off

[scrubs]

This is very interesting information, as it is definitely not normal M9 behaviour. The camera should take about three seconds to write an image from its buffer. Please do the following test: In C mode press down the shutter and hold it down after the buffer is full. The shooting rate should slow down from 2 fps to about 1 frame per three seconds.

If the cameras that are misbehaving do indeed write far slower than they should under some circumstances, or do so intermittently, it may indicate part of the cause.

 

 

 

Definitely not the case here.

 

I just did your test it took a total of 50 seconds while holding down shutter for another

frame to fire, I released after than one shot (so 8 shots in buffer at this stage)

and red light did not stop flashing until a total of 2:15 seconds.

 

Are those times you quoted for DNG + JPG Fine?

[scrubs]

These tests were done with class 4 card so just repeated same test with a class 10.

 

 

7 shots in continuous. all shots after the seventh fired one frame every 10 seconds.

 

 

Firing 8 frames took the buffer a total of 1 minute, so much faster, that sound acceptable?

 

 

 

 

Scrubs,

 

Many thanks. Excellent observation.

Question:

So, this happened once, right?

My experience is that this behavior can't be repeated on demand, right?

What did you use, JPG fine?

Or did you use JPG basic?

 

Best, K-H.

 

 

Sorry forgot to answer your other questions, Yes it just happened once

literally 10 mins ago but not able to repeat it on demand by doing exactly

the same thing which is frustrating.

[jaapv]

Definitely not the case here.

 

I just did your test it took a total of 50 seconds while holding down shutter for another

frame to fire, I released after than one shot (so 8 shots in buffer at this stage)

and red light did not stop flashing until a total of 2:15 seconds.

 

Are those times you quoted for DNG + JPG Fine?

The shots after the buffer has filled which appear to slow the camera down are NOT extra data in the buffer. The camera starts clearing the buffer immediately and fires another shot when there is space for the new data. So you are actually seeing the writing speed out of the buffer onto the card through the whole system this way.

If your camera behaves anomalously the implication is clear.

 

I just tested my camera on all compression settings. With slight variations in time due to different settings, my camera just slows down to about 2 to 3 seconds per frame when you hold the shutter depressed in C after filling the buffer. Then,when carrying on shooting for a while and then releasing the button it clears the buffer in less than half a minute. Consistently.

I used a Sandisk Ultra 16 Gb newest type.

[Advertisement]

Hallo,

Look here: Leica M

[k-hawinkler]

Scrubs,

 

Okay, please do the following now

 

Use full battery, necessarily battery has been removed from compartment.

Format memory card in M9 with overwrite, how long does that take?

Set to DNG compressed only

Use C mode not S

Fire 7 shots, time it

What do you get?

 

Please let us also know exact description of the memory card(s) used.

 

Many thanks, K-H.

[k-hawinkler]

BTW, Jaap's timing numbers are what I measured as well some time ago.

There are threads around in which folks did these measurements for various cards.

I participated as well in those.

A slower card might take a little longer.

However, previously folks reported some variability in the timings.

 

If your timing measurements don't improve significantly with compressed DNG only then I would conclude that your combination M9 + memory card behaves abnormally.

 

Good luck.

 

Best, K-H.