This is part two of the interview. Read part one here.
Give us some free cable advice.
I would tell you to insist that every cable be performance tested. I can’t tell you how many times we get a call and customers tell us their system doesn’t work. The pieces all work. The frame grabber works, the camera works. But no one looked to see if all the pieces are really compatible. Is the cable really up to spec? Too often we end up having to work to validate the whole system to see what went wrong. And it’s simply the fact that the customer chose a cable vendor who did not do any performance testing on the cables.
This used to give the cable guys a black eye originally, because there wasn’t an economical way way to test the cables in a production environment. Cabling back at the beginning without validation testing was a nightmare. A custom configuration can take months to make, and once delivered , if it doesn’t perform…that can delay the entire system 12-16 weeks. Today, you can get certification that the cable has been performance tested–that it’s been put in an actual system without error.
You always have to push the limit.
The biggest challenge is people keep adopting the latest technologies. People want to rush ahead with a new technology to get the performance and hope that the “cable guys will figure it out” when it comes to reliability.
As I mentioned, most of these interfaces are not made for industrial purposes. They are too delicate, and I find that it’s often a mistake to use them, even if they are faster. In comparison, the old analog cameras used a very robust Hirosi interface connection. That was really made for tough applications. For example, we struggled with USB3. The original USB3 standard was written for 2m cables. So we, the cable companies, had to find ways to build outside of the standard. In a year we took USB3 to 7m and later to more than 25m. On copper!
I think the smart thing to do, even with these consumer standards, is to have a durable industrial connector on the camera itself, where there is much more exposure, and you can have a normal connector on the computer side, which is more typically in an office or at least away from the action. We are developing standardized tests to help. The USB3 committee will have vision-compliant cables and vision-compliant hosts. Firewire used to have problems like that all the time.
Do you see a solution?
Going forward, we need to choose tough interfaces – that can move, handle vibration, and deal with similar tough issues. USB3 and GigE are fragile. It took a lot of work to add screws so you wouldn’t have mating issues with RJ45 or USB3 issues. We’ve made that mistake with interfaces over and over, adopting technology because it’s new and simple and fast. We need something fast and robust.
People always say they want cables cheap. I say choose a robust cable from the beginning, something that can survive in an industrial environment. We don’t want things to fail. We want to only sell a cable once.
Is that most of what you do, customizing and testing cables?
Our customization tends to be making cables more ruggedized, improving quality control and standards, or improving security.
But that’s only part of the work. When we joined the industry , we wanted to change everything, and the other competitors reacted. We found ways to work together. Today, there are only a few companies that do what we do, and actively participate in creating standards. We’ve become friends, and now we can make changes quickly. It took a while to develop that community spirit and stop working against each other. That’s part of how we add value. That helps move these standards forward.
So we have to work together to move things forward. It’s the only way. When I hear about an interface standard with a sole source vendor I am immediately skeptical. A standard that hinges on just one participant is such a bad idea. It takes multiple players to make the interface good enough and cheap enough.
What would you like to see in the future?
A new standard usually starts with a camera engineer adopting an existing communication format. . The camera and frame grabber guys are the ones that decide which ones make sense to use and say “build that.”
I want to see more melding of the minds. A ruggedized industrial connector should be one of the first criteria in choosing a new machine vision standard. We usually define these cameras by what’s at the back of them. That’s not going to happen in the future – I expect them to have three or four interface choices.
Give us a wish list.
What I would like to see is high-speed copper with ruggedized and standardized locking connectors. We like copper because it’s easier to flex. Fiber can be fragile – it’s made out of glass. It could transfer 25GB/s over 100m, it would be watertight and all for a price that doesn’t upset the system with the costs. On one lane. Maybe someday…