One of the earliest blog postings on this site was about the Cornice Storage Element, a 1.5GB micro hard-drive which was (at the time) slated to be an alternative to the more established IBM MicroDrive. The post tore down a unit which was extracted from a Leadtek My DigiBank which had failed due to a drop while moving house.
Fast-forward a decade, while rummaging around drawers, I found the second unit we purchased for my Dad. This one’s been sitting around for a while, complete with accessories, so now I can finally document the unit itself.
The Leadtek My DigiBank seems to be later to the party than I had thought. Looking back at my unintentional MSY price-list archive, I saw it listed for AU$135 back in 2005.
This actually makes it slightly less attractive than I had imagined – while it is larger than the largest 1GB USB 2.0 flash drive at AU$85, it is also a little more than proportionally more expensive and is built around “rotating rust” technology.
By the time 2006 had rolled-around, the flash drive market continued to reduce its price and now even 2GB drives were available for AU$63 (if you could get a hold of any stock).
The My DigiBank was hence on clearance at AU$55 – still more expensive than flash disks on a proportional basis, but it was at this point I decided to buy two units because I trusted rotating-rust a bit more than flash memory, especially as endurance and speed were concerns at the time. One unit went to me, the other unit went to Dad – it is the latter unit which we are looking at in this post.
The unit looks a bit like a 2.5″ external hard drive cut in half when it comes to size. The body is made of an aluminium shell with branding printed on it.
The approvals are on a label on the rear, with a serial label underneath. There is a notch in the casing on the side to ease grasping the USB connector.
The sides of the drive are ribbed …
… while the ends also follow this convention, it is made of plastic instead. One end houses the activity indicator, a high-intensity blue LED.
The other houses the swivel-out USB-A connector. The drive connects over USB 2.0 with this sturdy connector capable of supporting the weight of the whole unit.
As a result, the connector does protrude from the body, allowing it a better chance of plugging into a port without requiring a cable.
Just in case, the package also included a high-quality USB-A extension cable with ferrite bead suppression.
But perhaps the nicest part of the package – and something you don’t get with a USB flash drive – is an imitation leather case for the drive.
The case is finished in nice stitching with a magnetic clasp and velour interior.
The rear features a belt loop with velcro closure.
The side of the case has a cut-out for the swivel USB connector, allowing the connector to be swung out while still being in the case. The other side has a hole punched for the activity LED. Unfortunately, in this example, the opening for the USB connector is a little short – I recall the one I had fit quite well …
… thus you didn’t even need to take it out of the pouch (or off your belt) to access the drive.
The drive is recognised with VID 067B and PID 2507, suggesting it uses a Prolific PL2507 bridge chip. The drive comes preformatted as FAT with a FAT16 partition of 1,455,620,096 bytes capacity which is a bit shy of the 1.5GB rating on the box. This one had a few files stored on it … hence the used space.
As was customary in the Windows XP era, the partition offset was 63-sectors. The medium is recognised as a CORNICE Inc. Storage Ele with F/W 042R and a capacity of 1,463,648,576 bytes of total capacity.
I made a quick magnetic audio recording of the drive performing a sequential read and access time benchmark – digibank-magrec-cornice-se.mp4. The drive has a very aggressive spin-down policy perhaps to save energy and reduce chances of head-crashes. After the last access, the drive spins down in about 1.2 seconds.
Unfortunately, the drive is no longer healthy after all of these years. At first, plugging the drive in resulted in frequent hangs and disconnects/reconnects. Eventually, the drive stabilised enough after a number of repeated attempts to read the full surface that it could complete a full read without error (but this was a one-off fluke)!
Attempting to write the full surface was never successful. Instead, the drive dropped out and clicked, implying that it was having some difficulty.
Regardless, a “quick” read benchmark was still possible, showing some of the “interesting” behaviour that I had noted in the past – it seems that sector zero starts somewhere not on the outer edge of the disk which is relatively uncommon for hard drives. At a guess, it starts from this location and goes inwards, then jumps back to that track and goes outward resulting in this transfer rate curve. I’m not entirely sure why they chose to do this – perhaps it “games” certain benchmarks that only do a start/middle/end transfer rate check?
A few dips can be seen which are likely to be bad sectors or retries – but the average 4.5MB/s is about right for when the drive is in full health.
Random reads gave a peak IOPS rating of just 29 – limited by mechanical speed.
It was not possible to complete a full random access write test – but the partial result suggests that writes may be a little slower which is also common. Unfortunately, the drive was not in good enough health to do any other benchmarks.
Nothing lasts forever and it is very common to find equipment that was “working when last used” end up failing coming out of storage. It was nice to rediscover this relatively uncommon piece of portable storage technology, based around a micro hard-drive from a vendor that seems to be ultimately sued out of existence, but probably wouldn’t have had long to live either given what flash memory has achieved today. The benchmarks show a strange sector zero position and an access pattern that suggests the head goes towards the spindle, and also then from an intermediate track to the outside. The average sequential throughput of 4.5MB/s doesn’t seem all that quick now, but it was definitely competitive to “ordinary” low-cost flash drives of the day.
I found the old Cornice web site, but the information is sparse:
Your HD Tune access time graph has a width of about 15 msec which would suggest that the rotation rate is around 4000 RPM (15 ms is the total rotational latency, ie the time for 1 revolution). One seller on the web is advertising the RPM as 3600.
1.5GB Marketing Brochure:
2.0GB Marketing Brochure:
3.0GB Marketing Brochure:
8GB – 10GB Reference Design for OEMs: