Optimizing Production Lines: Offline Programming vs. In-Circuit Testing

Colin:Thanks for having me. Great to be here. 

Colin: If I start, I guess, by saying to understand that all of our customers are effectively subcontractors. So, whether that's a tier one manufacturer who's designed their own parts, in production the economics of what they do is, as a contractor, they make many parts across many lines in the factory and they're servicing either external or internal customers.

They're trying to optimize throughput and costs in every case and they're dealing with very complex supply chain and logistics environments. They make a series of printed circuit board assemblies and for any printed circuit board assembly, at the end of the line, when you've done all your chip loading and the part is built, whatever capital you've expended in doing that, you then must optically and electrically test the device. You have usually an AOI device that inspects that there's no blobs of solder or any other gross errors of that sort and then you have a specific to each board test fixture that tests whether it works. Before you can test a part, you must load the software that that part uses on in order to test if it does what it's supposed to do. In every case, and I think I can say this almost without exception, the bottleneck on every line I've ever seen is a test. You make whatever number of boards you make on a shift, the test time is almost always the longest item. You can only reduce test time or reduce that bottleneck by doing one or two things. You can add testers and the testers are expensive, they're very high capex items and the fixtures that you need per board are expensive. You can add that but it's a capex hit or you can reduce test time.

And what EPS Global does in a nutshell is we take the loading of a software out of test and by doing that offline and presenting the chips to the production line pre-programmed, we reduce test time. In nearly every case that improves the throughput of the line and that's sort of in a nutshell what we do versus programming and service. Wherever you get a case where the programming time of the chip is a significant part of the overall test time, there's a business case for using EPS Global. 

Colin:  This is an example, I can't use the customer's name. For example, we came across with a client in Romania, they had a particular card they were using. It was a component of a base, a GSM base station. They were able to make 600 pieces per shift and, you know, there was a significant lag in test as a number of devices were being programmed prior to test.

We sat down with them, we said, okay, we can, we will charge you per chip for the pre-programming, but what do you get for that? And, I mean, if you think of in a very simplistic notion, that's 600 pieces that they make per shift, every piece of overhead in the entire factory is divided by that output. It's the only thing the factory produces. Every buyer, every plant manager, every ops guy, every, the guy who's the security guys in the car park, all of that cost is divided over the 600 units. We sat down and by the time we had offloaded the pre-programming, the line, the beat rate went to 650. There's a significant difference in the cost per unit when you can do that. We do charge per unit. You must factor that back in. It's not all one way again, but there's a net benefit to every customer in the cases where there's a significant reduction in test time.

Now, if the test time is a trivial, if the programming time is a trivially small part of your overall test time, it's probably not a factor, but that's not often the case. And we have a very cost-effective way of programming because it's usually quicker to program at a chip level than at a board level. Sometimes even where we would see fast programming times, that's not what the manufacturer sees in circuit. That's typically the difference between the two are a good example as I could give or where you'd see a benefit. 

Colin: It does. There's different types of security and there's many different levels, but generally speaking, you're going to add test equipment when you add security. You might add a HSM or other items. The CapEx piece gets bigger. If you can take that offline and you can, for example, if a HSM is required, HSMs have very high throughput. They're almost never consumed in a line that produces hundreds per shift.

For us to use the same HSM, we can apply it to dozens of different customers because it's perfectly capable of producing tens of thousands of units per hour and we could use it to run many programming centers. If you start to apply one per line, it could be relatively expensive compared to what we do. And there's other versions of security. I mean, in the most secure applications that we've seen, you'd actually have your tester online and you'd be doing certificate signing within the chip online. In that situation, which is the most secure way to sign certificates into a chip, but it requires that your production line would be connected to the internet at all times. In a lot of places, our customers have backup generation and power is not a guarantee in every place that we would support our customers.

The idea that the internet is always on would be a risky prospect for a lot of our clients in a lot of the countries that we operate in. In a situation like that, offline programming could also be a big advantage. 

Colin: Of course. It's a hardware security module and it's the device that's used to basically either encrypt the information on a chip or to help you sign certificates into a chip, which are things you need for that chip to subsequently identify itself as a valid program chip to whatever our customer is using. 

Colin: For example, Amazon would be a good example. It could be your wireless charger on your phone that the certificate might allow you to say, hey, I'm a very high power phone and there could be a certificate in the charging chip that goes, that's good. I'm a high power charger. We can talk to each other and it saves a high power charger burning out a low power phone, for example.

That type of exchange is what we would use certificates for. When you want to put certificates in, you can't put them in offline and that's one way of doing it. It's not the most secure way of doing it. Depending on the customer's requirements, that could be another reason why offline programming is a good idea. 

Colin: If I start by saying, in terms of the case for using offline programming, it's one of the sweetest solutions. If you can do stuff in ICT, most customers will. We sit in a solution set that's an other offering, When I go out talking to our security partners and some of the major chip vendors we work with, they're not suggesting to every customer to use EPS Global. They're suggesting for the guys where this is the fit for what you're looking for, here's a really good solution. And there may be other ways of doing it to fit your application better.

So, we're somewhere in that spectrum. If you've decided to program offline, and if you get to that decision and you've said, hey, I'm going to definitely program offline, how would you do that? Generally speaking, what we see as our major competitor is our customer buying their own equipment. That would be the case where somebody decides to go out into the market and buy a machine, set that up in-house, and program themselves. The things I always say, generally speaking, that can be successful for some customers depending on their experience, but quite often we have a number of baked-in advantages that we try to bring to people's attention if that's something they're thinking about.

Firstly, we're not in the machine selling business, we only sell services. We sell programmed parts and charge per part. That's the only offering we have. But behind that, we have a significant infrastructure, a significant operation, a big R&D center. We make our own handlers. We'd have a significantly lower cost of capital than any of our customers, and our per piece price programmed would be competitive with them buying a machine in almost every case. It has much more advantages as well. We have a number of ways of implementing security that are much more versatile and accessible, and with much more resilience. Because of our cost of capital, we don't have any centers in the world where we would only, for example, have one machine and there'd be a risk of failure. We'd have backups, or in many cases, multiple backups, that you would have no risk of machine downtime to your production line. So for anyone in a... and we operate in a number of relatively remote places.

I'm thinking maybe of Brazil or India, Thailand, Eastern Europe, where you're not very, very close to the service guy for a machine. If you buy one machine and it breaks, you have exposures, whereas you don't have that with us. We will have an adjacent site, we'll offer you the service. If a machine breaks, obviously our machines break as regularly or as irregularly as anybody else's, you're not depending on a single machine, you're dealing with a factory that has a network of machines, one's backing up the other. You're never going to see downtime from that perspective. We generally speak, and we'll oversubscribe capacity as well. If you have capacity lifts, we're going to be able to service you better than if you have a single machine.

If you have capacity downtimes, you're not going to be paying for our service where you will be paying rental on a machine. There's a number of reasons why I think we would set ourselves up from a cost point and from a service point to compete directly with you buying your own machine, and I think we have a fairly compelling argument in most cases in that respect. 

Colin: Great. It was nice to have a great talk with you.