2024 is nearing the end. Can you offer any status update since the last post?
Announcement
Collapse
No announcement yet.
USB 3.2 Gen2 loopback
Collapse
X
-
Bit of a sad story. We had a working prototype and were doing PCB design for mass production. It was going to be expensive however. We were going to use a FPGA. Initially we thought costs were going to be something like $150 per FPGA chip (plus maybe $30 for other parts, plus $25K for off the shelf "IP" for the FPGA, plus maybe $100K for our own engineering & software). Plus we needed to make a profit, or at least break even. Then we found the FPGA wasn't quite fast enough. The next one up was ~$300 per chip. So if we produce 500 units, the cost base was going to be ~$600 per unit. This would have give us 10Gbit/s, instead of the current 5Gbit/s. It was also slightly technically risky using 3rd party IP we don't control for part of the project and economically risky, in that we might not sell many units if they are $800 each retail. Plus we are aware that 10Gbit/s isn't all that fast (going to 20Gb looked possible in 2025, but maybe at $2000 per unit, as FPGAs aren't cheap at high speeds).
At the same time we became aware of another non-FPGA chip which was claimed to have twice the speed (so 20Gbit/s) for maybe 1/3rd the price. (to be confirmed as we are just guessing for the moment). This chip + it's pricing is going to be released in Dec 2024, it is claimed. So 6x better on a speed per $ basis. Too good it ignore.
So we made the choice to put aside the relatively slow expensive solution and wait for the cheap fast solution.
Note that if there was some customer out their who was willing to give a firm order for 200+ units, we could still do this FPGA solution, as a lot of the risk would be removed. Let us know.
So, disappointingly, we've got nothing for 2024.
But we are still committed to the project.
Comment
-
Yes, the new chip eventually became available. It was late, so instead of Dec 24, it was June 25.
But we bought a bunch of them. and the good news is that they work at 20Gbit/s, as advertised. But then they got recalled because of some low level bootloader issue. But we can work around that. There was a few deep technical issues, like being able to count and report protocol errors, but we think we solved them.
High level task list is
Test development board => Done
Write firmware => Mostly done
Write windows client software => Mostly done
Sort out device driver => Mostly done
Design new prototype PCB => Done
Manufacture prototype PCB & source all the other parts (LCD screen, flash storage, protection circuits, etc..) => Done
Design plastic case => Done (3D print)
Test prototype => Next week
Get mold made for plastics case
Design 2nd prototype PCB
Manufacture 2nd prototype
Test 2nd prototype. (Hopefully it all works at this point)
Go to mass production of board + case + cabling
Sort out compliance, packaging, documentation, bar codes & web site.
Shipping
Will all this get done before the Christmas break, probably not. But we are pretty confident now for an early 2026 release, as the risky part seems to be solved.
Comment

Comment