Any hardware from Passmark that supports USB3.2 speed test?

[Haur]

Hi guys,

Does Passmark produce USB loopback testers that are capable of handling USB3.2?
I've seen the current product only capable of handling up to USB3.0.
Please advise.
Thanks.

Haur

[David (PassMark)]

To be really pedantic, the current USB3 loopback plug does support 3.2.
But it only supports USB 3.2 Gen 1. Which is 5Mbit/s (and was in the past called both 3.0 and also 3.1. The USB naming is insane)

But I assume you are really looking for USB 3.2 Gen 2 or USB 3.2 Gen 2x2. Which we don't have yet.

We are looking at it. But at the moment there is no cost effective solution. Maybe by around the end of 2020 it is possible to build a fairly cheap solution.

[ekearney]

Hi There,
Does the new Infineon FX20 USB 20 Gbps Peripheral Controller offer you a cost effective solution to make a USB 3.2 Gen2 or gen 2x2 tester?

Regards
Erik

[David (PassMark)]

We purchased a couple of the new FX20 (20Gbps) hardware development kits for evaluation. They are so new the documentation is only half complete, but they seem promising in the initial testing we have done. We are optimistic that this might be the cost effective solution we've been waiting for.

As you can see from the time line of the posts above, it has taken way longer than expected for silicon to become available. We tried a few other solutions over the last few years. Some were technically impossible. Some sounded good on paper, but when we tried to order the chips they weren't actually available. One company refused to supply chips and documentation unless we ordered maybe 20,000 units up front and also allowed them to dictate the design. One worked really well but was crazy expensive and wanted ongoing royalties (maybe $1000 / unit by the time we got it to retail). Hardware development isn't easy.

[ekearney]

Hi David,
Look forward to seeing what you can achieve it the FX20.
I agree that some chip manufacturers do put unreasonable roadblocks in the way of people using their ICs, it doesn't make a great deal of sense.

Regards
Erik

[Derek@Sealevel]

Is there a timeframe for this effort? I know previous concepts did not pan out but this solution does look encouraging...

[David (PassMark)]

There are a number of technical issues we are working through. These include very high idle power usage (leading to high chip temperatures), the ability to re-enumerate down to lower speeds to test older USB modes, the lack of a usable boot-loader for in the field firmware updates, collection of accurate error counters, etc...
Most of these issues seem solvable at the moment.

Hardware development is a slow process however. So we need to design a printed circuit board (probably with 1 or 2 rounds of prototyping), decide what type of enclosure we need, get plastic moulds made (also probably with rounds of prototyping) order & test E-Marked cables, decide if we need a cooling solution as these high speed chips run hot, sort out packaging, sort out how the QA testing will work, get device drivers signed by Microsoft, then do bulk production, etc...

So no date yet for retail sale. But this year, unless something really unexpected happens.

[Derek@Sealevel]

If I can make a suggestion...Please consider making provisions for adding some extra load to stress power rails close to the port limit. Either built-in or through external connection.

[David (PassMark)]

We've already got an available solution for stress testing USB power rails. Up to 240W in fact.

[Derek@Sealevel]

Yes, I'm aware. We have one and like it. But I'd need one of those per port and for moderate load of a standard port it's an overkill. Software-controlled power rail connector addition to the loopback wouldn't add much cost but a lot of versatility.

[David (PassMark)]

The power rail in type-C isn't just a simple wire with a fixed voltage. It is complex. Very complex. There are multiple standards, multiple profiles, a wide range of different voltages, a negotiation process with the host and the cable to determine available profiles, there are also 4 voltage rails and two ground, etc...

Having exposed pins also greatly increases the risk of blowing things up. Either short circuits or dumb loads with too little resistance (i.e. accidentally connect a 200W / 1A load at 20V and pull 20A. So briefly draw 400W until the magic smoke leaves the device being tested). I think this is the main downside, we already have customers blowing stuff up and it takes a huge amount of support time to deal with it. Maybe in the medium term we can expose the pins and also have some semi-dumb external load. Dumb, but with some protection circuits.