Using CFexpress v4.0 cards with the Nikon Z9
1 Jan 26
For several years now I have used a pair of ProGrade Cobalt 325Gb CFexpress Type B version 2.0 (v2.0) cards in my Nikon Z9. These cards and the Delkin Black cards have always featured prominently as recommended cards for the Z9. The ProGrade cards have been flawless so far. I’ve experienced no issues and they have been fast and reliable.
However, I have been on the outlook for larger capacity cards for the Nikon Z9 recently, and when my local photo shop had the Delkin Black 650Gb v2.0 cards at 50% sale, then I pounced. But, alas, the shop sent me the wrong cards. Instead of the v2.0 cards they sent me two Delkin Black 650Gb CFexpress Type B version 4.0 (v4.0) cards.
Shot at 560mm @ f/4, 1/1600s, ISO 5600
The Nikon Z9 was released when the v2.0 cards were the fastest available, and it is built to take full advantage of these cards. The v4.0 card specification was announced in autumn 2023 and have on paper much improved performance characteristics. Thes cards have exactly the same form factor as the v2.0 cards, but utilise newer PCI bus standards, theoretically doubling the throughput performance of the v2.0 cards. The v4.0 Delkin Black have a marketed max read speed of 3530Mb/s and max write speed of 3250Mb/s. Twice the speed of the v2.0 ProGrade Cobalt cards I have up till now used in my Nikon Z9. The Cobalt card has a max read speed of 1700Mb/s and write speed of 1500Mb/s.
Shot at 400mm @ f/4.5, 1/800s, ISO 1800
The Z9 cannot take advantage of the improved throughput capability of the v4.0 cards. I was pondering sending back the Delkin v4.0 cards as the shop was very helpful and apologetic, but then I decided to keep the v4.0 cards and find out how they would work with my Nikon Z9 and my 4-year-old Sony CFexpress card reader (model MRW-G1). Were indeed these cards as backward compatible as claimed? Would there be any performance implications? After all there is an element of future-proofing in sticking with the v4.0 cards, as no doubt future Nikon cameras, such as the Z9ii will be built to take advantage of the latest cards’ standard and capability. So, I decided to do a bit of quick testing …
Shot at 560mm @ f/4, 1/500s, ISO 1000
Z9 speed test and comparison
Method used: First, I did a full format of the new Delkin Black v4.0 cards in the Z9. I only tested for photography, not video, as that is what I use the Nikon Z9 for. The Z9 had the latest firmware (v5.30). I put the Z9 in Manual mode, on ISO100, 1/2000s shutter and f/2.8 aperture. I put on a 28mm lens, with the lens cap on. I turned the camera shutter sound on. The frame rate was set to 20 frames per second (fps), which is the maximum for Raw shooting.
I tested both Full Lossless Raw and High Efficiency* (HE*) Lossy Raw. Using a stopwatch, I pressed the shutter and kept it down, listening to the rhythm of the shutter sound, waiting for it to slow down to less than 20fps as the card started to struggle to keep up with the data load throw at it. For each Raw format I repeated this continuous burst test 5 times. I did a full format of the card before each test burst. If the shutter speed did not slow down after 10 seconds I stopped. 10 seconds at maximum fps equates to 200 images and more than enough for most wildlife action. Below are my findings:
ProGrade Cobalt 325Gb v2.0:
Lossless Raw: Continuous max burst 4.5 seconds (90 frames). Shutter speed slowed significantly after 4.5s but it kept the burst going.
High Efficiency* Raw: Continuous max burst 10 seconds (200 frames). Test stopped after 10 seconds as no drop in performance observed. Assume max rate can be sustained till card is full.
Delkin Black 650Gb v4.0:
Lossless Raw: Continuous max burst 4.3 seconds (87 frames). Shutter speed slowed significantly after 4.3s but it kept the burst going.
High Efficiency* Raw: Continuous max burst 10 seconds (200 frames). Test stopped after 10 seconds as no drop in performance observed. Assume max rate can be sustained till card is full.
Speed test conclusion: For all intents and purposes the cards perform similarly and very well. I mostly shoot High Efficiency* compressed lossy Raw which is a superb format with indistinguishable results in most situations from lossless Raw at 60% of the same file size. This is the default Raw setting in my camera. And with HE* both types of cards can shoot continuously at max burst rate for a long time. The cards’ write speed can keep up with the demand of the Z9. Even with full lossless Raw, ~4-5 seconds of action before the frame rate slows down is sufficient in most scenarios.
An interesting aside: When performing full card formatting in camera (which is what I do between shoots to ensure a clean reset of the card), the Delkin Black v4.0 was significantly slower than the ProGrade Cobalt v2.0. It takes ~1.5s to complete a full format of the ProGrade Cobalt card, but a much longer ~18s to complete a full format of the Delkin Black card. Not that this has much of a practical implication, as it is not something I do in the field.
Shot at 560mm @ f/4, 1/500s, ISO 720
Card reader test and performance
My CFexpress v2.0 card reader from Sony (model MRW-G1) had no problems with the newer v4.0 cards. I tested uploads of 200 HE* raw files (average size 29.5Mb) shot with both cards three times each. Upload speeds were similar for both type of cards using this reader. Clearly a v4.0 reader will likely provide improved upload speeds depending on the performance of your computer and ports/cables.
Shot at 560mm @ f/4, 1/500s, ISO 800
Overall conclusion
With the v4.0 cards now becoming prevalent, and they seem to be considerably cheaper in $/Gb terms than the fast v2.0 cards recommended for the Z9, it makes sense to invest in these v4.0 cards going forward. There does not seem to be any meaningful performance difference, and they provide a level of future-proofing.
Clearly, this test and its conclusion was limited to my cards, camera and my use, which is wildlife photography. I only tested two v4.0 Delkin Black cards. Delink Black has a very good reputation for performance and durability. There is no guarantee that other v4.0 cards will perform similarly, so do your own testing. Regardless of CFexpress versions, always buy fast and high-quality cards.
Shot at 560mm @ f/4, 1/5000s, ISO 8000. This image was shot using Nikon Z9 Pre-Release Capture function which only supports JPG files. At such a fast shutter speed the ISO is very high and the resulting noise significant. JPGs are less flexible in post processing than Raw files which really shows here. We need Nikon Pre-Release Capture for Raw ;-)