The Card Failure Issue

Updated 

The fact that the new Nikon Z cameras have only one card slot seems to have precipitated a massive wave of negative commentary. I have a completely different view, apparently, than most of the folks bemoaning the loss.

First, a story. I was with Galen Rowell when he was on assignment and shooting with a new F100 body. Within minutes his camera was triggering "early rewind," a fairly common defect when the F100 was first introduced. Film cameras didn't have backup film slots ;~). 

Even though Galen had a backup camera with him, I simply handed him my F100 and he shot with that for the rest of the assignment (my camera didn't have the problem). I then dug out my backup camera so that I, too, could keep shooting.

The reason why I mention this story is that people don't seem to understand "point of failure." Whether your camera has one or two card slots, it has many potential point of failures. If you are truly professional you plan for all the ones that you can think of. And you make strong plans for points of failure that would be fatal. 

When I travel to Africa for workshops and my own wildlife work, pretty much everything I bring has backups, and many have backups to backups. Anything that fails in the middle of the Okavango Delta or the Serengeti means you stop shooting unless you have a backup. And when I write I have a backup for everything, I mean everything: cameras, lenses, cards, chargers, batteries, cables, readers, hard drives, even support systems. (I also bring a toolkit ;~). Everything is a point of failure.

Okay, now we're ready to begin.

Point of failure. Likelihood of failure. Condition of failure. Outcome of failure.

In the days of CompactFlash cards, we had a very serious point of failure: the pins on the card slot in the camera. They could be (and were) easily bent, particularly when people tried to change cards fast or with force. Sometimes you could, with great care, unbend the pins in the field. Indeed, if you couldn't do that, then that last bit I noted above—outcome of failure—was catastrophic; you could no longer shoot. 

SD cards are a little different. The slot doesn't have "pins" that can bend. I've seen three types of SD card failure over time: broken card casing, broken/loose write protect tab, and file table corruption. The first two are physical problems that are designed into the card (and made worse by poor quality control by the maker). Sometimes I can fix those in the field when they occur, sometimes I can't. Rough handling of the card tends to make the SD card more prone to failure.

Finally, XQD cards, which is what the new Nikon Z cameras use: I've never seen a physical failure with XQD. I've not heard of a physical failure from others. One of the reasons why I liked XQD from the beginning is that it "fixes" many of the weak points of the early card designs. 

Which really leaves us with one thing—already mentioned in the SD section—file table corruption. This occurs one of two ways: (1) the camera caused the corruption by writing incorrect data onto the card; or (2) one of the card's memory modules failed.

Early on in the digital era, #1 was fairly common, particularly when you deleted images on the card during shoots, and particularly so when you were nearly out of space on the card. I still don't recommend doing deletions to clear up space for a couple more shots because of this, but over time every camera maker managed to sort out the bugs in their file management. It's fairly uncommon to see a camera-caused corruption these days. The exception to this is moving cards between cameras of different manufacturers. While each camera maker is 100% consistent with their own file table and image file structures, they're not 100% consistent with each other. 

To basically reduce #1 to the point where you don't worry about it as a likely point of failure: (a) always do an in-camera format before using the card; (b) never delete in camera; and (c) don't move cards between cameras.

It's #2 that a lot of people seem to worry about. In theory, the NAND flash cells used in XQD cards have MTBF (mean time between failure) measured in the millions of aggregate writes, and are managed by a fairly high-end memory controller (because of the speed at which the memory management must be done). It certainly is possible for an XQD card to have an internal memory error occur during shooting, but it must be extraordinarily rare, as I've not heard of any. Indeed, even in SD this kind of error is rare. 

It's been pointed out to me that NAND does age (limited number of reads/writes), and this will eventually produce failure. That's true of all storage cards consumers can buy. I probably should note that I regularly retire my cards. It's probably one reason why I don't generally experience these types of card failure issues, even on SD or CF. People need to start thinking of cards like they think of tires on their autos: they have a limited read/write life span and should be replaced with some regularity if you care about data integrity. SD, in particular, is particularly prone to this problem because it's been around for so long. I keep encountering people who are still using their original SD card in their latest camera. Not only is that card slower than the camera can manage, but if you've been shooting long enough, you're going to hit cell degradation. 

Wedding and sports shooters who are pushing tens of thousands of images through their cameras every weekend need to be particularly concerned about NAND cell overuse. And just shooting with two cards in the camera doing backup can lead to a false sense of security if those cards are getting equal loads and have equally long histories. 

Also, an engineer who works on data systems for the military pointed out something to me that isn't being considered: there's an additive risk in putting in more hardware and logic to do a parallel redundant system, and that risk actually gets bigger with mismatched "slot" speeds (ala the D500/D850, Sony A7/A7r/A9). The notion that "just having an extra slot solves all data problems" may be what consumers think, but it's not what trained engineers think. Complexity adds potential failure paths.

That leaves one thing to examine: the card slot mechanism itself. Those that think that two slots are physically different and would limit failure, haven't seen how the internal designs actually work. I'd be more worried about the digital board to write mechanism connection than the slot itself with XQD (and probably even SD). 

So, we're ready to examine the four things I noted earlier:

  • Is a single card slot a point of failure? Yes. But so are two card slots, particularly when made into one part, as Nikon would have likely had to do to fit two cards in the Z series cameras.
  • What is the likelihood of failure? Low. And not particularly higher than if you had two card slots with XQD and current Nikon card management.
  • Is there a condition of failure I should be looking for? Yes. Any report from the camera of a card failure—it does basic structure checking every time it accesses the card—should mean you immediately remove that card from the camera and set it aside for potential review/repair/recovery. If you see issues on image review, same thing.
  • What's the outcome of failure? You could lose images. If you saw the failure happen and took the card out immediately, there's a relatively high likelihood of recovering most, if not all of the images on the card. That could cost you time and/or money. 

Finally, I should mention the most common "card failure" that people report: a band of color—often purple— where image pixels should be. That's actually a write mechanism failure. The card slot and card did not synchronize the write process correctly. A camera that is triggering this, whether it has one or two card slots, generally needs to be looked at by the manufacturer, as this error occurs because of timing issues that could live much further back in the chain than the card slot itself.

All this is a long-winded way of saying this: I'm not particularly worried about having one slot. Indeed, I've been shooting my D500 and D850 as one slot cameras for quite some time now. Why? Because the second slot in those cameras is an SD slot, and even the fastest SD card in that slot has a tendency to slow down the camera in a number of operations (also reduces the buffer if you're shooting Backup). 

You Sony users rejoicing that your A7 model cameras have a second slot have the same problem: the second slot on the Sony is much slower, and that has impacts on camera performance if you have it active. Indeed, Sony makes the second slot inactive by default.  

Personally, I see a second XQD slot mostly as a pacifier. I've yet to hear any report from someone who was shooting Backup to the second slot where the first card was corrupted and the second wasn't (again, this is for XQD; the potential physical failures I outlined with CF and SD are a different story).  

All that said, Nikon did take something away that they had given us in their prosumer and pro DSLRs (second slot). There's always a pushback when that happens. 

Some are comparing what Nikon did to Apple taking away the old USB slots or headphone jacks. That's about right. As technology moves forward in reliability most companies want to move away from physical point of failures (a headphone jack is a bigger point of failure than a Bluetooth chip to Apple). 

The lack of a second XQD slot on the Nikon Z series doesn't make me any less likely to use and enjoy the cameras. I'll be honest: I'm more likely to misplace or lose a card(s) than to have a critical error in camera now. The actual most likely point of failure now is me, not the camera's card slot(s). 

Update: one last thing. I'm seeing people use the word "failure" in a very generic way when describing why they want two slots. To me, a data failure means I can never recover the data under any means. Yes, having a corrupted file/sector/table on a card is annoying. But as I noted in my interview with a data recovery specialist back a couple of years at NAB, it's extremely rare that data can't be recovered using extreme measures. Yes, that often takes time and money, and some shooters—particularly wedding shooters—would have issues with that. But for the bulk of us, we can deal with these issues as they arise. It'll be annoying, take time, and perhaps money to solve, sure, but it's difficult to describe that as a true "failure." I'll tell you what a true failure is: dropping a card in the sand or ocean and not noticing or being able to find it. That's why I absolutely hate microSD cards: too easy to fumble and lose. But SD cards aren't far behind on my list. XQD cards are a very substantive size and physically able to stand abuse. I haven't lost one yet, nor have I managed to crush/crumble/bend one.

text and images © 2018 Thom Hogan
portions Copyright 1999-2017 Thom Hogan-- All Rights Reserved
Follow us on Twitter: @bythom, hashtags #bythom, #sansmirror