Comments on: Where USB Memory Sticks are Born

By: Jarrow

Jarrow — Mon, 06 May 2013 05:15:20 +0000

You called it. The chip fabrication procedures for memory and CPU differ considerably. Trying to shoehorn controllers onto memory wafers dramatically decreases flawless yields. Given how cutthroat the commodity fab business is, that would quickly send the experimenter out of business.

That is not to say that someone with a large enough bankroll and a good enough fab couldn’t built a process that does it cost competitively, but it would probably take tens to hundreds of millions to build up, and would be money lost with the next drop in feature size.

Realistically, until we get to the point of widescale custom printing on an atomic scale, separate chips made with separate processes are more cost effective.

By: Flash memory through the ages | Parallel Points

Flash memory through the ages | Parallel Points — Sat, 04 May 2013 20:50:42 +0000

[...] was reading bunnie’s recent post on the manufacturing techniaues used in USB flash-drives… bare die manipulated by hand with a [...]

By: Alex

Alex — Tue, 30 Apr 2013 10:28:15 +0000

hi Bunnie,

how can i contact you in private ?

I would to develop a new idea with UDP’s manufacturers and I have seen pictures of those factories in your blog.

Thanks

Alex ( from Spain )

By: Willrandship

Willrandship — Sun, 28 Apr 2013 15:22:47 +0000

Unless they’re ordering in major bulk, the customer-specified preloading would be easier to do on finished products. Adding on a process is easier than modifying one in this case, IMO. You can just grab the number you need from the pile of drives in the nearest bin, instead of setting up a dedicated line to programming the chips during construction.

By: Jonam

Jonam — Sun, 28 Apr 2013 10:08:44 +0000

I am amazed that this sort of process is still being done for these USB Flash drives.

Way back in 1985 when I was doing my uni thesis on IC design, my University would get multi-project chips made at a local foundry. To make use of the relevant part of the chip for my work, I had to manually place (using Araldite) and then wire bond bare chips onto a PCB to allow me to test and use each die.

It was difficult work in that the wire-bonding machine was an old Kulicke and Soffa wedge bonder that the Uni must have got second-hand from somewhere. Often, due to problems with the machine, I would have to do a fair bit of rework using the finest tweezers and scalpel blades I could get from RS components. Happy days!