1TB CFexpress - ROLL-YOUR-OWN for $181.00 CND

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

Doug

Active member
Supporting Member
Marketplace
ROLL-YOUR-OWN- 1 TB CF express B for $181.00 CND



Using off-the-shelf parts, I have constructed a cost-effective alternative to today's fastest mass-market brands. No experience or tools are needed; anyone can assemble one for approximately $180

I've managed to assemble a budget-friendly alternative to the popular brands you see everywhere, simply by using ready-made parts. You don't need any experience or tools; anyone can assemble one for about 180 Canadian dollars. Easy-peasy!

I purchased these parts from Amazon for convenience. The 1TB M.2 30mm memory card uses memory chips produced in Taiwan. Despite being made in China, the case is CNC machined from aviation-grade aluminum, meeting the size requirements for CFexpress. The two halves are securely fastened together with tinny screws, a few extras are supplied..

1 x Corsair MP600 Mini 1TB M.2 NVMe PCIe x4 Gen4 2 SSD M.2 2230 – 4800MB/s – High-Density 3D TLC NAND ($115 CND)

1 x Zitay CF enclosure, aviation aluminum, and CFexpress Female/female plug-in connector. ($66.00CND)

1 x Terminal heat dissipation past: Not included, but is highly recommended.



Aside from being a DIY job, what distinguishes this DYI from leading CF card manufacturers is its assembly method. While market cards are soldered with the CF connector inside their casings, Zitay's casing employs a female-to-female connector to connect the card to the camera or reader.


Today, you can buy M. 2 2230 NVMe PCIe Gen4x4 cards from many manufacturers in 64 MB, 128 MB, 256 MB, 512 MB, 1TB, and 2 TB, with speeds ranging from 4000MB/s ~ 7800MB/s. These cards are used in 100s of thousands of computers and Steam Deck, Asus ROG Ally, Microsoft Surface Pro and all recent Laptops. If you need a 2Tb, Corsair, sells it for $209 CDN on Amazon.ca. Videographers may appreciate the cost/size and speed. Alternatively, used as a “backup” kept cool, dry, without extreme temperatures'life expectancy of 10~20 years and is cheaper monthly cloud storage of my 2TB photo catalogue. . My need for more than 1TB of full RAW images is virtually non-existent. However, technology will continue to become cheaper, and faster with higher capacities while using less power.

Heads up—I put a very thin layer of Terminal heat dissipation past on the memory chip, on both sides of the 1TB M.2 board.

Assembly steps:

1) Use a ground strap if you live in an area with low humidity and static electricity. It's rare, but a static charge can damage the memory chip.

2) Open the CF enclosure on a flat surface. Inside the plastic bag, you'll find the case, the internal female/male connector, a packet of tiny screws, and a screwdriver. I recommend assembling on a rubber mat, as the screws can bounce off hard surfaces and become nearly impossible to find.

3) Insert the M.2 into the supplied connector. Now, determine the correct orientation for the M.2 & connector inside the enclosure. It is correct when the top half fits snugly together. Apply the thermal paste to the chip, and place it facing down into the enclosure. Apply the paste to the top chip. Install the top screws.

4) Format the card in the camera and start shooting.

Now, you own a fast, inexpensive 1TB CF card. The impressive speed will become apparent when you upload the images onto your computer. The Z8 read/write speed is irrelevant, as the Z8 will handshake to the speed of the CFexpress cards.

I have been using this CFexpress card in my Z8 for a month. I've encountered no issues when inserting or extracting this card from my Z8, nor have I experienced any heat dissipation or communication problems. As a none videographer and as all SSD devices utilize this same memory technology. All CFexprees cards get hot as noted on the Nikon’s warning label, I anticipate videographers will also appreciate the size, and speed of using this card during long, high-resolution shots.

The aircraft-grade aluminum enclosure will effectively conduct heat away from the memory chips on the M. 2 SSD memory drive inside. The warning label also lets us know Nikon has taken heat into account in their design of the SSD compartment.

To reap the maximum transfer speeds depends on your computer's USB ports USB-C 3.2 or Thunderbolt port and the CFexpress reader. I use the Lexar Pro CFexpress RW550 reader/writer, which I use. There are other USB CF readers capable of USB 3.2 Gen 2x2 which also claim 1700MB/s speeds.

I've attached some pictures and disk read/write tests for your perusal. The host was a Dell 17 XPS 9720 with Thunderbolt 4 (USB Type-C), and the Lexar reader. I've also included the speed test of its boot C:/ drive C which is an M.2 2Tb that came from Dell. Your results may differ.

For any clarifications, feel free. Cheers, Doug Stead






20240330_115653(1).jpg
You can only see EXIF info for this image if you are logged in.
20240330_122027(1).jpg
You can only see EXIF info for this image if you are logged in.
Screenshot 2024-04-13 130902.png
Screenshot 2024-04-13 131035.png
Screenshot 2024-04-13 131932.png
Screenshot 2024-04-13 183628.png
 
the last benchmark is interesting because it shows that this is a v4 card because it's faster than theoretical max for v2. that said, folks should realize it'll probably get about half that in a v2 system (~1250MB/s, maybe)
 
the last benchmark is interesting because it shows that this is a v4 card because it's faster than theoretical max for v2. that said, folks should realize it'll probably get about half that in a v2 system (~1250MB/s, maybe)
You are correct, rated speeds are only possible in a perfect test setup or the top of the line latest motherboard. Speed tests in the field, are as ±accurate after a negotiated a speed.
 
Pretty neat. Can you comment on operating temperatures? One of the reasons to pay the $ is for a thermally efficient and fast card.
I don't have the test software need to test maximum speed and temperature. What I can say is that when uploaded 6000 x 50/MB RAW images to my laptop, transfer rate started as expected speeds got to about 25% and speed throttled down to perhaps 50%. This according to the file manager. That was likely the CF card or the reader, self-protection. I don't think it was triggered by the laptop.
 
I like that it is aluminum as I have considered using these for backup purposes while traveling and was concerned I would damage the plastic cases the commercial ones reside in. Thanks for posting this.
 
Last edited:
For shooting with the adapter people should note the Q&A on the sale page. I believe these speeds exceed what the Z9 can do.

How fast should my SSD read and write?
It depends on your NVME M.2 SSD drive and your CFexpress Type B card reader.Our adapter support write speeds up to 1200MB/s and read speeds up to 1800MB/s.
 
For shooting with the adapter people should note the Q&A on the sale page. I believe these speeds exceed what the Z9 can do.

How fast should my SSD read and write?
It depends on your NVME M.2 SSD drive and your CFexpress Type B card reader.Our adapter support write speeds up to 1200MB/s and read speeds up to 1800MB/s.
These cards are down downward read/write compatible with slower speed devices as the Z9 and perhaps the Z8, if Nikon did not updated the cameras interface cameras. all the latrst CFexpress cards are going to talk at yhe max speed in the camera regardless of the speed they are capable.
 
I made my own CFE card using the Sintech aluminum case and a Sabrent Rocket 2230 NVMe 4.0 1TB m.2 drive, cost about $160. In camera the card just doesn't perform up to the Delkin Black, Prograde Cobalt, or even my Angelbird AV Pro SX cards on still frame bursts. This is not a scientific test, just a practical test. I really haven't tested or used the card much for video so I can't comment on that. I think it is really hard to design a consistent testing method to determine how the cards compare.

Delkin Black, Prograde Cobalt, and Angelbird AV Pro SX
Lossless RAW at 20 FPS - 80+ images until buffer is full and FPS starts to slow
HE* at 20 FPS - 200 images until the buffer slows

Sintech/Rocket 1TB Card
Lossless Raw at 20 FPS - 62+ images until buffer is full and FPS starts to slow
HE* at 20 FPS - 126 images until the buffer slows
 
I have tried this to see if I could save some $$. Although I didn't use the exact same components, I found the end product way to un-reliable to use for any pro or important work. I'll admit this can be done (no reason that it can't), but I wouldn't ever trust its reliability for my work. The reviews on Amazon bear me out. I'll stick to commercially available product. Prices have been coming down. Best of luck.
 
I understand the safety concern for those keepers we work so hard to obtain. That said, my working life was at a company that designed and manufactures small, power efficient embedded computers and their power supplies. Specifically for hostile environment, low power, temperature, g-force, vibration and shock. Trains, planes, mining, military and a couple in space. I could be wrong, and if my testing and use shows any sign of failure, I will post the failure here.

The cost of a 1TB, let alone a 2TB CF card solution VS card failure is worth this testing. I am comfortable that the one real risk is infant mortality of the chips, which is almost non-existent using product manufactured in the 1000 of millions. The one great advantage of this solution is the larger capacity! With gate wear-levelling algorithm running inside all SSD drives, will likely give this solution a life expectancy/reliability greater than 10-fold. This based on a ridiculous usage rate, shooting and uploading 10,000 full Z8 RAW images a per day.


Assuming that the write operation occurs once a day:
  1. Total bytes to be written: 10,000 files * 50 MB/file = 500,000,000,000 bytes (or 500 GB)
  2. Write endurance in bytes: 3000 TBW * 1024 GB/TB * 1024 MB/GB * 1024 bytes/MB = 3,221,225,472,000,000 bytes
  3. Number of days the drive can be written: write endurance in bytes ÷ total bytes to be written ≈ 6,442 days
  4. note: 3000 TBW of energy signifies technological advancements in energy storage materials and systems.
It would take approximately 6,442 days (or around 17.65 years) to write 10,000 x 50MB files to a 2TB drive, assuming a write operation once a day.
Keep in mind that this is still an estimate and actual performance may vary depending on the specific SSD model, usage patterns, and other factors that affect the drive's lifespan.

Since the write operation occurs once a day, we can simply use the result of the previous calculation:
Here is a graph showing the estimated lifespan in days for SSDs of various capacities, from 2TB down to 64 MB, assuming a daily write operation of 10,000 x 50MB files:
SSD Lifespan Estimate (days) vs Capacity (log scale).

Lifespan in days:
+-----------------------------------------------+
| 2TB | 1TB | 512GB | 256GB | 128GB | 64GB |
+------------------------------------------------
| 6442 | 3221 | 1610 | 805 | 402 | 201 |
+-----------------------------------------------+
  • Larger capacity SSDs (2TB and above) have a very long lifespan, exceeding 6,000 days (~16.4 years)

  • Medium capacity SSDs (1TB to 256GB) have a lifespan ranging from 3,200 to 800 days (~8.8 years to ~2.2 years)

  • Smaller capacity SSDs (128GB to 16GB) have a relatively shorter lifespan, ranging from 400 to 50 days (~1.1 years to ~0.14 years)

  • Minimal capacity SSDs (8GB and below) have a very short lifespan, less than 50 days (~0.14 years)

This graph is a rough estimate and actual lifespan may vary depending on the specific SSD model, usage patterns, and other factors that influence a drive's lifespan.
Here's a rough interpretation of the graph: Obviously none of us are ever going to shoot 10K images a day, to say nothing of culling! I will leave that math to you.

Cheers, Doug
 
Doug, all those statistics are very nice (I am a trained engineer). The proof, however, is in real life. While I have used M2. 2230 SSDs for other things without issues, I do have issues with those cheap, imported CFe enclosures and I wouldn't trust them with my images, even for a minute. I understand that you might enjoy the savings but I enjoy my confidence in the name brands. (I am very selective even amoung the name brands and their various models). Given the value of my equipment, and my work for clients, I choose to stay away. I am happy to report that every XQD and CFe card I have ever bought, including 32GB, 64GB, 128 GB, 256GB and 325GB still work as they did when new and I would guess some are 10+ years old. All have been "hammered" while being used for sports action work. Just my personal experience, yours may vary. Best of luck.
 
It's really interesting. I've toyed with doing this myself, until Hasselblad launched the X2D with a 1Tb internal SSD. I'd been saying to expect this for some years now! But now its proven tech, and so I'd not be surprised to see the other manufacturers start including internal SSDs - as soon as they get reliable, stable production.
 
I have tried this to see if I could save some $$. Although I didn't use the exact same components, I found the end product way to un-reliable to use for any pro or important work.
i think the primary potential areas for issues with diy solutions like this are:

1) the controller being used on the module
2) the fw version on the controller

basically when we buy cards, someone else has done the component selection/test. when we diy, we have to do it ourselves.

that said, when we find a reliable recipe, the result should be fine.

so my basic takeaway here is the Corsair MP600 looks promising for those so inclined

the one thing i am curious about is to see what fw it's running and if it's up to date, because that can be a factor as well
 
Doug, all those statistics are very nice (I am a trained engineer). The proof, however, is in real life. While I have used M2. 2230 SSDs for other things without issues, I do have issues with those cheap, imported CFe enclosures and I wouldn't trust them with my images, even for a minute. I understand that you might enjoy the savings but I enjoy my confidence in the name brands. (I am very selective even amoung the name brands and their various models). Given the value of my equipment, and my work for clients, I choose to stay away. I am happy to report that every XQD and CFe card I have ever bought, including 32GB, 64GB, 128 GB, 256GB and 325GB still work as they did when new and I would guess some are 10+ years old. All have been "hammered" while being used for sports action work. Just my personal experience, yours may vary. Best of luck.
I respect that professionals like you always balance risk and best practices. Your position is valid.

I also have 4 of the best cards money can buy. I rotate these after each day of shooting. This reduces the potential loss of images due to card failure. Every so often, I also enjoy experimenting. When off the grid, I will likely use this card as a backup for my pictures after loading them onto my laptop. For day-to-day shooting, this card will be in rotation. I take images locally. To be clear, I have no dog in this race. My risk tolerance is mine alone.

For whatever it may be worth, Zitag has updated its Amazon page with a new video which claims to show how many images can be stored when a Z9 is continuously shooting. I am not concerned about this happening because I have yet to fill the Z8 buffer.

 
I respect that professionals like you always balance risk and best practices. Your position is valid.

I also have 4 of the best cards money can buy. I rotate these after each day of shooting. This reduces the potential loss of images due to card failure. Every so often, I also enjoy experimenting. When off the grid, I will likely use this card as a backup for my pictures after loading them onto my laptop. For day-to-day shooting, this card will be in rotation. I take images locally. To be clear, I have no dog in this race. My risk tolerance is mine alone.

For whatever it may be worth, Zitag has updated its Amazon page with a new video which claims to show how many images can be stored when a Z9 is continuously shooting. I am not concerned about this happening because I have yet to fill the Z8 buffer.

Like you, I rotate my cards and I do not erase (format) them agaig until placed in the camera. This not only comforts me by formatting in the exact camera being used, but also provides an additional backup for recently shot images as I do not MOVE my images when importing to LrC. Best of luck to you!
 
Back
Top