[HN Gopher] How to make your own CFexpress card for under $100
___________________________________________________________________
How to make your own CFexpress card for under $100
Author : zdw
Score : 49 points
Date : 2022-04-16 13:40 UTC (1 days ago)
(HTM) web link (photofocus.com)
(TXT) w3m dump (photofocus.com)
| LarsAlereon wrote:
| I worry that there's a hidden issue with maximum power draw. If
| the camera supplies less maximum current than the drive could
| potentially draw from an M.2 slot, you might see the card crash
| in certain situations. This could be the kind of very annoying
| thing that works fine in testing but fails in on specific
| scenario in real-world use.
| wtallis wrote:
| Ideally, SSD manufacturers would program their SSD firmware to
| provide multiple active power states accurately labeled with
| their maximum power draw, and cameras would select only power
| states that are within their power delivery constraints. All of
| this functionality is already defined and standardized. But I'm
| pretty sure neither side is taking it seriously enough to be
| relied upon.
| londons_explore wrote:
| I think the drive at least will be keeping an eye on its own
| power rail voltages and logging whenever they sag out of
| spec.
|
| Those logs will be used to rapidly deny your warranty claim
| when you try to return it for not working properly...
| wtallis wrote:
| The only logs of that sort that I'm aware of is the SMART
| counter for unexpected power loss. M.2 SSDs are not at all
| graceful about handling voltage droop, and I don't think
| I've ever seen one even attempt to automatically lower its
| power state (and thus peak power draw) in response to low
| input voltage.
|
| And since the maximum power draw information that SSDs
| provide to the host is unreliable, you need expensive
| measurement equipment to determine whether a SSD brown-out
| is more the fault of the host not providing enough power,
| or the SSD drawing more power than is reasonable.
| wmf wrote:
| Are we talking about the same drives that don't implement
| flushing properly?
| traceroute66 wrote:
| > _I built a 256 GB CFexpress card for under $100, and you can
| too._
|
| Probably because you valued your time at zero.
|
| At a minimum probably 30 minutes to buy the components - although
| realistically with the shortages at the moment might take longer
| because you'll have to look around for places with stuff in
| stock.
|
| Then let's say 20-30 minutes to get the components in place (I'm
| including packaging removal time here).
|
| Plus a bit more for testing, labelling etc.
|
| Sure it's not the most labour-intensive thing in the world, but
| the time soon adds up.
|
| And that's assuming everything works first time and you don't
| have to spend time troubleshooting or returning broken
| components.
|
| What hourly rate does your salary work out to ? ;-)
| ResNet wrote:
| CFexpress cards are considerably more expensive than other
| similar forms of storage like SD cards -- a comparable 256GB
| CFexpress card currently goes for $259.99 [0]. From a pure cost
| perspective, it could very well be worth it. As a hobby
| photographer I would consider doing this for the fun of it
| anyway.
|
| [0]
| https://www.bhphotovideo.com/c/product/1551702-REG/sandisk_s...
| traceroute66 wrote:
| You miss the point.
|
| The author links to (approx) $90 in parts.
|
| Your average HN reader is probably in a decently paid job,
| let's say it works out to $80-$100 per hour (at least !).
|
| So we're now talking about $190 all-in, i.e. parts _AND_
| labour.
|
| So we're now talking about a $60 price difference ($250 off-
| the-shelf vs $190 DIY).
|
| For the $60 difference you get: - A card
| that is camera manufacturer validated and so does not risk
| breaking your camera or invalidating its warranty -
| A card that has already been tested and labeled in the
| factory, so you don't need to do that - A card with
| a warranty
|
| I'd say that $60 is a bargain.
|
| People should not be valuing their time at zero when engaging
| in DIY projects.
| ResNet wrote:
| Many photographers don't earn what the "average" HNer does
| (if there is an average). My comment didn't really address
| the HN crowd.
|
| The warranty aspect is a great point though, if a
| commercial photographer did this and lost their client's
| work because e.g. the pins broke off, I fear their clients'
| response.
|
| The answer here comes down to what the individual is able
| to justify to themselves, be it via a cost-benefit analysis
| or merely "I'm a solo/hobbyist photographer, I could build
| this tomorrow evening during downtime and save some money,
| let's do it." :-)
| adam-a wrote:
| HN boasts an international audience and many of us earn a
| lot less than $100/hour. Plus you know it is called
| "hacker" news, not "bought it at the store" news.
| walterbell wrote:
| Candidate for alt HN tagline :)
| AverageHNreader wrote:
| 100 USD/hour translates to something like 325k yearly. Is
| "average" HN reader "at least" in a senior position in
| FAANG?
| JaimeThompson wrote:
| Don't forget that some people enjoy such things. Not every
| thing has to be reduced to a cost / benefit analysis.
| ResNet wrote:
| Absolutely. I value my time spent working on hobby
| projects and end up paying money per hour I spend on it
| -- as far as I'm concerned it's worth every cent!
| actually_a_dog wrote:
| Sigh.
|
| No, _you 're_ missing the point. For one, DIY projects are
| great for learning. Secondly, unless you already have a way
| to earn money with that time, time spent on stuff like this
| is "worth" $0. You'd actually have more money in your
| pocket after DIY'ing it and spending a couple hours than
| you would clicking "buy" on Amazon.com and spending a
| couple minutes or so.
|
| And, if you want to get really nitpicky, I'm pretty sure
| all those parts the author bought were also tested and
| warrantied by their respective manufacturers. And, show me
| a camera warranty that would be broken by using an adapter
| that's made for the card type the camera takes, anyway.
|
| SMH.
| sdoering wrote:
| I have a very decently paid job here in Germany. After
| taxes, health care and all the other deductibles I take
| home somewhat around 30 - 35 Euros per hour. As overtime is
| included in my job (I work hard not to do too much, but
| still) having fun building something like that on a weekend
| would also be an economical bargain.
|
| I still am totally in line with my sister comments in
| valuing the fun building something and understanding a few
| things a bit better in the process quite higher than the
| economic aspect. This would only be the icing on the cake.
| jeroenhd wrote:
| The difference between the parts and shipping (+-$90) and a
| comparable CE card (+-317) is $227. Assuming three hours of
| work including research you'd need to earn about $75 per hour
| to break even. Assuming a 40 hour work week, 4 weeks of holiday
| time and no other benefits, that amounts to a yearly wage of
| $144000, placing you firmly in the top 1% of the world, and the
| top 10% of the USA (according to
| https://www.compareyourincome.org/)
|
| After taxes, my current salary comes down to about EUR12,28 per
| hour ([?]$13.27). Not a great rate, but higher than 36% of the
| rest of my country assuming a full time working week. That
| means I can work at least 17 full hours on such a project and
| end up making a profit. The 30-60 minutes of work that this
| project takes saves me over $200.
|
| Most of the world doesn't work for the ridiculous rates FAANG
| pays. Good for you that you're too rich to even bother about
| spending $200 extra for a memory card. However, your skewed
| expectations don't detriment the value of this article in any
| way.
| spiffytech wrote:
| > What hourly rate does your salary work out to ?
|
| If I work at a fixed salary, my effective hourly rate isn't
| relevant here. I can't spend the time I saved to earn more
| money. My only choices are spend less on DIY or spend more on
| ready-made.
|
| The right discussion here is opportunity cost: what did I have
| to give up to use my time or money this way? That depends more
| on my bottom-line discretionary funds and free time than on my
| top-line salary.
| Zak wrote:
| I think you're overestimating the time required to do this. It
| looks like 30 minutes to me, all-inclusive.
|
| The cheapest no-name 256gb cards are $130 on Amazon, so that
| works out to $60/hr assuming the DIY version is no better than
| no-name. Big brands cost twice that, so if the DIY version is
| equivalent quality to those, $320/hr.
| traceroute66 wrote:
| > It looks like 30 minutes to me, all-inclusive.
|
| Really ? Are you honestly and truthfully including all steps
| : - Online shopping (*INCLUDING* time
| spent looking around for alternative components/stock)
| - Opening up packaging - Getting tools together
| - Putting it together - Testing and labelling
| - Clearing up afterwards
|
| Sure I'll grant you that _IF_ all your ducks line up
| absolutely perfectly, you might be able to scrape by 30-45
| minutes. I suspect the reality will need a bit of extra time
| margin though.
| walterbell wrote:
| _> There are several companies that produce CFexpress NVME
| adapters._
|
| Neat. Is there a comparable NVME 2230 enclosure with USB-C
| external connector for use in laptop USB4/Thunderbolt ports?
|
| That could combine SSD speed/reliability with the portability of
| a flash drive.
| adam-a wrote:
| I hope CFexpress replaces SD card slots on laptops. I'd love to
| boost or swap the storage in my in mine but sd cards are so
| slow and.
| beardedscotsman wrote:
| This needs more camera manufacturers adding CFExpress over SD
| cards. However, there are a large number of other devices
| such as audio recorders using SD, so this could be wishful
| thinking for now at least.
| ghostly_s wrote:
| The latest SD standard has speeds comparable to CFExpress.
| davidgay wrote:
| IIUC, CFExpress is just a PCIe interface in a different form
| factor, so the adapter in the post should basically just be a
| pass-through. Conversely, a USB-C adapter would require some
| level of protocol adaptation/electronics/etc.
| Nextgrid wrote:
| USB-C connectors can carry Thunderbolt which is essentially
| external PCI-Express, so the adapter might have to be very
| minimal if you're fine with it only being compatible with
| Thunderbolt-capable devices.
| ResNet wrote:
| A Thunderbolt flash drive in the form factor of a USB-C
| YubiKey 5C Nano [0] would be the ultimate solution for me.
| I trust that some day we'll get there if it's not
| superseded by some other standard by then.
|
| [0] https://www.yubico.com/us/product/yubikey-5c-nano/
| walterbell wrote:
| Is there a USB-C flash drive smaller than the 2022
| Samsung USB 3.2 Gen1,
| https://www.samsung.com/us/business/computing/memory-
| storage...? For USB-A there is Sandisk Cruzer Fit similar
| to Yubikey Nano, but most USB-C flash drives are larger.
| nrp wrote:
| Yep. We considered building an M.2 2230 NVMe USB-C adapter,
| but it was more efficient to just use a USB-C NAND Flash
| controller that interfaces the same types of flash that NVMe
| SSDs do. One less layer of translation.
| anonymousiam wrote:
| I don't know about 2230, but I have used this with the 2280
| form factor.
|
| I have used this: https://www.amazon.com/gp/product/B07Z8Y85GL/
|
| It works, but it runs hot so I added one of these:
| https://www.amazon.com/gp/product/B07WJVFQQ9
|
| I've tested it with 8TB nvme flash. Don't know if it can
| support anything higher.
___________________________________________________________________
(page generated 2022-04-17 23:00 UTC)