Metal Bitcoin Seed Storage Stress Test (round VI)

BitcoinBitcoin
$20,119.86
3.12%
1M Ago
blog.lopp.net

Note: this is a recurring series; here you can read the full results of the entire series here.

The Contenders

This plate is made from 3mm thick marine grade 316 stainless steel.

Attenuo claims to protect against fire damage, water damage, corrosion, ink fade, accidental discarding, and being ripped up or chewed up by pests.

Bravo Seed claims to be fireproof up to 900ºC, corrosion resistant to humidity and the passage of time, and crush resistant up to 74 gigapascals.

Bravo Seed claims to be fireproof up to 900ºC, corrosion resistant to humidity and the passage of time, and crush resistant up to 74 gigapascals.

Claims to be:

Constructed out of 2mm thick highly polished stainless steel 304 stainless steel has a melting point of 1,400°C / 2,550°F Water proof Corrosion resistant

I'm a bit worried about the tiny screws because I often see tiny screws fail due to getting stripped or deformed.

Claims to be:

Constructed out of 2mm thick highly polished stainless steel 304 stainless steel has a melting point of 1,400°C / 2,550°F Water proof Corrosion resistant

Claims to be:

Constructed out of 1.8mm thick highly polished stainless steel 304 stainless steel has a melting point of 1,400°C / 2,550°F Water proof Corrosion resistant

Made from .071" thick, 304 Stainless Steel, the Codl.Co Punchplate claims to withstand fire, water, corrosion, and crushing.

All Coinplates are 100% made of superior grade German Stainless Steel (AISI 304 / DIN1.4301) and use ultra thick (5mm) stainless steel plates joined with stainless bolts.

Coinplate claims to be:

Fireproof up to 1,400°C / 2,500°F Waterproof EMP & X-ray proof Corrosion & rust resistant Pressure & impact resistant

One thing I'll note about the Coinplate devices - they are hefty. Each plate weighs over a pound; the joined plates weigh nearly 3 pounds.

All Coinplates are 100% made of superior grade German Stainless Steel (AISI 304 / DIN1.4301) and use ultra thick (5mm) stainless steel plates joined with stainless bolts.

Coinplate claims to be:

Fireproof up to 1,400°C / 2,500°F Waterproof EMP & X-ray proof Corrosion & rust resistant Pressure & impact resistant

All Coinplates are 100% made of superior grade German Stainless Steel (AISI 304 / DIN1.4301) and use ultra thick (5mm) stainless steel plates joined with stainless bolts.

Coinplate claims to be:

Fireproof up to 1,400°C / 2,500°F Waterproof EMP & X-ray proof Corrosion & rust resistant Pressure & impact resistant

I will note that the naming of this device is a bit confusing, because the "Coinplate Grid" is the one you use the center punch on while the Coinplate Punch for letter stamping.

This is essentially a paper thin sheet of copper. It claims to be:

Fireproof Waterproof Easy to use Requires no special tools — write with a pencil or pen

CRYO claims to be:

stainless steel durable rust-proof water-proof fireproof up to 2500°F each plate is 2mm thick for maximum durability

The outer CRYO plates protect the inner seed word plates from physical damage. The cover plate with the logo can also be flipped over to make the cold storage backup discreet and unremarkable.

I'm slightly concerned about the screws because they are a two piece design and one side doesn't have any indentations to grab onto; it could be tough to unscrew after being stressed.

Made from 304 stainless steel, CryptoNumeris claims to be:

Fireproof (Up to 1450°C / 2642°F) Waterproof (No rust/corrosion) Shockproof (14GA. sheet) Hackerproof (Digitally unreachable)

Made from 304 stainless steel, CryptoNumeris claims to be:

Fireproof (Up to 1450°C / 2642°F) Waterproof (No rust/corrosion) Shockproof (14GA. sheet) Hackerproof (Digitally unreachable)

CryptoTag claims to be:

Fireproof up to 1665°C / 3029 °F Waterproof

CryptoTag claims to be:

Fireproof up to 1665°C / 3029 °F Waterproof

DIYSeed is meant to be a cheap and private way to create your own backup from commonly available items. It claims to be indestructible and not easy to read without the associated templates.

Claims to be:

Fire Resistant up to 3000°F Corrosion resistant to acid, rusting, saltwater Crush proof: solid plate, deeply stamped

qreg has once again devised a novel encoding scheme. This one is also binary for line number of the word, but it's stored by drilling out holes in the plate. As usual, qreg does not make any durability claims.

Looks like they have already gone out of business...

Secret Seed claims to be designed from quality titanium, which is lighter and stronger than steel. It is discreet, won't corrode, has a high temperature tolerance, is safe to use, and non toxic.

Personally I'm not a fan of it being two separate plates with no fastener. This is touted as a security feature because you can split the seed and hide it in 2 places, but of course that makes your backup more brittle and susceptible to loss. Also, the seed words are stored as hex which supposedly makes it less appealing for thieves, but I'd argue that is less appealing for anyone to use.

Don't let the close-up photos fool you; this really is a tiny device. It's 3.5 cm x 2.7 cm x 1 mm and weighs 13 grams.

Tinyseed claims to be:

Heat resistant to 1660°C corrosion resistant shock resistant inconspicious

This is a very straightforward stainless steel plate measuring 100 x 100 x 4 millimeters. It's quite hefty.

Willi claims the plate will protect your seed phrase from:

water fire corrosion mechanical damage other hazards Device Setup / Assembly

As usual, the straight punch method is one I find to be fastest and most user friendly.

Easy, straightforward setup.

Similarly simple setup.

One unique feature of the Bravo Seed products is that they come with perfectly sized stickers that you can use to cover the data. The idea being that you could glue a magnet on the other side and the casual observer would just assume that they are a fridge magnet. I put the stickers on for posterity, but clearly they will be destroyed by the fire test.

These plates already had some pre-etched markings though they weren't as deep as a center punch would be. I center punched a few divots just to see how they would fare in comparison. Note that the written seed words are meant to be wiped off and are only supposed to be a temporary template. I'm not convinced that these temporary templates are worth the extra space they use on the device.

Same situation with the single plate; I center punched the left column and left the right column as it arrived.

I only punched divots for "safe" and "earn" on this device. The others were left with their original markings.

The first 4 letters of each word are encoded in a clockwise pattern starting from the top triangle, one word per column. This is highly compact but I would not recommend it for older folks. It requires precision to punch and pretty good eyesight to read.

I will note that the latest generation of this device has tiny "guide" divots where you see each dot in order to help you correctly align your straight punch.

The device is large and hefty, there are no problems punching the divots into it. Though trying to see the markings in a photo is pretty tough.

Punching divots in the Grid is similarly easy, though it's more legible because you aren't punching on top of printed letters.

Stamping was a pain as usual, even when using a jig provided by ImpressSeed.

You'll notice I had a few light strikes at the beginning as I was trying to find a good surface for hammering and by the end some of my letters were rotated incorrectly because I didn't notice that the stamp had changed position after being struck.

This was easy but slightly annoying because I wasn't sure how hard I needed to press down while writing on the copper sheet. My first few words were done at a fairly normal writing pressure while the rest were at high pressure to make the maximum indentation.

Straightforward straight punching on these plates; I had no issues.

Easy straight punching.

My biggest problem when stamping this plate is due to the small size. It was hard to get a good position on the plate without the plate itself bouncing around as I hammered it.

You'll want to use "anvil" that's included in the box while you're punching divots into the Odin. Otherwise, when you try to punch a divot near an edge of the device, the rod will rotate and you might stab yourself!

Setup was easy as expected.

Hard to screw this up!

It was really annoying to stamp the letters into this plate due to its small size. And due to the thinness, the plate itself warped and curved inward with additional stamping along the sides. One nice thing was that ImpressSeed includes their own jig which makes it far more manageable. However, I noted that after several hundred strikes the jig started to fail, similar to how I've seen other plastic jigs give in to fatigue.

qreg has once again devised a novel storage scheme. This one is also binary for the BIP39 line number of the word, but it's stored by drilling out holes in the plate. My initial impression is that I actually find their previous form of punching out holes to be more user friendly than using this drill. For one, the drill comes with 8 different bit sizes and it wasn't clear to me which one to use. I started with the second smallest only to find that the holes were not getting any bigger.

On the left side I've only used the 4th largest drill bit. On the right side I started off with that bit and then did a second pass with the largest drill bit. I think the qreg folks could improve the user experience by only providing those two bits and giving further instructions on how to make the best use of them. I really like the results, though.

I'm not a fan of the hex encoding, but the straight punching of data is easy enough.

I must say that I'm a big fan of the straight punch tool included in this kit - it's by far the most comfortable I've used. It's also impressive how compact the device is at 3.5 cm x 2.7 cm x 3 mm, making it even more robust against deformation. And despite the small size, I didn't have any issues punching divots into it. This is a meticulously designed product, all the way down to the micro guide holes in each cell to help ensure that you have the center punch... centered. Switching to a binary encoding of the BIP39 line number further improves the compactness.

My biggest pain with this device was getting the camera to focus on it for detailed photos.

No problems to report here - very simple setup.

Heat Tests

I applied ~2000°F heat for 10 minutes. Then the device was submerged in a bucket of water to simulate a firefighter putting out a structure fire.

Some discoloration and negligible amount of warping. No data loss.

Obviously the sticker facade on the devices didn't last long, but that was just for fun.

This device had discoloration but no data loss. In this scenario the guide divots may actually be working against it, making it harder to differentiate the data from the blank template cells.

Same story with the coin format.

The device itself survived just fine, however my initial fears about the screws were proven justified. I was only able to unscrew half of the screws while the others got stripped. I'm sure the data inside is fine, but this is an annoying usability issue.

Here we can see that the words I punched divots for myself are still quite legible. The pre-etched words are mostly legible though I wouldn't rely upon them.

Similarly the two words I punched myself on the left column are still legible while the pre-etched words not so much. What's more concerning is the amount of rust that already seems to be appearing, threatening data loss.

No issues reading the data despite some expected discoloration.

Unfortunately another usability issue has arisen. I was unable to unscrew the fasteners holding the plates together using the provided star wrench; it was seized up pretty well, likely due to a small amount of deformation caused by expansion and contraction of the plates.

I experienced the same seizing up of screws on this device. Data is fine, usability notsomuch.

For whatever reason I was able to remove one of the screws on this device and rotate the plates apart. We can see the data is still legible.

What's concerning here is you can see the top left edge appears to have melted a bit. The lightly imprinted words in the top left are nearly illegible with some slight data loss. The discoloration does not make it easier to read. Though I will note that reading the back side of the sheet of copper was a bit easier for some of the words.

This device absorbs and dissipates heat like a champ; you can compare it to the qreg on the left that does not appear to react nearly as much.

Upon inspection of the cooled device we see it has no data loss.

While the divots themselves are plainly visible, the concerning issue here is that the template is not nearly as robust as the data itself. If you completely lose the template and can't find another copy in the far future, this could be problematic. This is why I recommend templates be deeply etched into a device rather than printed / lightly engraved.

We can see the same issue happening with the template on this device, but it's far less of a concern because the data is the fully stamped words rather than a dot-based encoding.

Odin also handles the heat quite well.

No data loss. It's hard to make out some of the template in the photo, but it's still viewable at the right angle.

Zeus is pretty much the same body as Thor (which I tested years ago) and it performed admirably.

One thing I particularly appreciate about CyptoTag is their fastener system. Unlike the many products that use screws, the slide-on fasteners that CryptoTag users are very easy to pry off with a simple flathead screwdriver even if device deformation occurs.

The paper template guide burned off immediately as expected.

The underlying data remained intact. And you could use a new template (pictured on right) to decode the data.

Slight discoloration around the edges but no data loss.

The qreg barely even changed color while being heated, in contrast to many other devices.

Upon reopening the sandwich we see the center plate with data was only discolored a bit.

Performed well, no data loss.

Barely even any discoloration.

Plenty of discoloration but no data loss.

Corrosion Tests

I submerge the devices in muriatic acid for 12+ hours and them pull them out once the bubbling stops. In this the Attenuo was too wide to fit flat in the bucket and thus was only partially submerged.

A nice acid bath (left side) actually makes the Attenuo more legible. Although the dark lines of the template get erased, it's etched deep enough that the template itself remains legible.

While there's no data loss we can see pitting all across the surface of the plate. It's interesting that you can see a progression from left to right where the data divots seem to be enlarging at the same rate. I'm not sure how to explain that, but it's slightly concerning.

Similar pitting as the card above, but without the divot variation. No data loss.

In both a good and bad twist, the acid completely dissolved the screws that had seized up and prevented me from opening the device after the heat test. Now we can see that the straight punched divots are holding up well, though the other types of markings did not. What's more concerning is that the template is completely gone. This would be very problematic if you were unable to find a replacement copy of this device's template.

Similar problems with the single plate, though some of the laser etched data fared better, the template markings were pretty much all erased.

The CipherTag fared the worst of the 3 due to its thinness. These devices are clearly made of a material that has some reactivity to acid. You can see how much of the left side of this device was dissolved by noting that the hole on the left is no longer usable. Catastrophic failure; given enough time I expect this device would completely dissolve.

While the plate survived quite well, it's exhibiting a unique phenomenon I haven't seen before. The acid is clearly enlarging the divots a bit, and due to how compactly designed the device template is, the enlarged divots are covering multiple spaces in some cases. As such, the decoding of the data is now at risk due to the ambiguity.

Thankfully I was able to get the device opened again. We can see that although some of the parts of the template had their dark lines erased, both the template and data remain deeply etched and intact.

I was still unable to open this device but I'm confident the data is intact.

No data loss.

Not great. We can see more of the copper sheet is disintegrating while the legibility of the remaining words is getting worse. Partial data loss.

Holding up quite well; no data loss.

The data divots are doing well but the data template is even harder to make out than after the heat test. I'd say that this data is still recoverable with a bit of extrapolation from the legible parts of the template, but the trend is concerning.

The template is completely gone but that's of little concern because we don't need it in order to decode the data on the device.

Completely legible.

Somewhat concerning because you can see some parts of the template are fading. The data and template are still legible, though it would take a bit of work to recover this data. Cryptotag might want to take another look at how they are etching the template onto this device.

The divots are quite clear and can be decoded with a (separate) transparent template placed on top of them.

Like some of the other titanium devices we can observe substantial pitting of the surface. A tiny amount of data loss with regard to the more lightly stamped letters. This seed phrase is certainly still recoverable.

No remarkable changes.

No data loss.

Similar surface pitting as other titanium devices, but the template and data are fully legible.

Some surface pitting, but fully legible data and template.

Crush Tests

I use a 20 ton hydraulic press to deform the devices and simulate something like a large building collapsing on top of the device.

No data loss.

The card barely deformed; no data loss.

No data loss.

No additional data loss after previous tests.

No additional data loss compared to prior tests.

I didn't use my 20 ton hydraulic press because it was unnecessary. Rather, I used a couple pairs of pliers to deform the tag to the breaking point. Catastrophic failure.

No data loss or remarkable changes compared to prior tests.

No data loss.

I was finally able to get the device reopened after flattening it back out with the hydraulic press. We can see that while there has been some fading of the template, both it and the data are still fully intact and legible.

No data loss.

No hydraulic press required.

Catastrophic failure.

I hit another usability issue that I unfortunately saw coming. After deforming the stack of plates, I could no longer unscrew the fasteners - they were seized up.

I tried flattening out the device again in the hopes that I'd be able to unscrew the fasteners, but I had no luck. However, despite not being able to view the data I'm quite confident that it's still intact and could be accessed with more powerful tools.

No appreciable changes compared to previous tests.

No data loss.

Barely even bent under 20 tons of force.

No data loss.

No data loss. Would still require some work to decode the data from parts where the template is missing.

No data loss, though as the device becomes more deformed it becomes harder to put a template on top of it to read the data. In extreme cases you'd likely need to flatten out the device again to achieve legibility.

No additional data loss.

Device did not deform much under 20 tons of force.

Data is perfectly intact.

No data loss.

This device is so small it was tricky to figure out how to attempt to deform it. Eventually I settled on putting all 20 tons of pressure onto its side.

Tinyseed withstood over 15 tons of force without deforming, and it wasn't until nearly 20 tons of force were applied that it finally gave in. However, the data is still legible. Obviously, if the force is applied directly onto the face of the device then it won't deform.

No data loss, no surprise.

Scores

Remember that these aren't highly scientific experiments and my grading system is arbitrary - my goal is to get a sense of which devices are the most robust relative to the rest of the market. If I'm unable to cause any data loss then a device should get an "A" rating, while a device that suffers catastrophic data loss gets an "F" - anything in between is a grey area.

Final Thoughts

Of this batch of devices I'd have to say that my favorite is the Tinyseed with qreg-hr a close second. They're not cheap, however - each runs over €100 per device. If you're on a tight budget, look into DIYSeed or grab Willi's plate for €27.