Overspecifying your test system will directly impact cost
When a test system is designed for only one single product in mind, the specification of the needed instrumentation as well as the switching is easy since all the parameters to be tested are known (the product), but when we are talking about the investment on an Open Test Platform, where different circuit typologies, or frequencies, voltages and loads are to be tested, the situation is much more complex.
In any case, it will always be important to set realistic limits in the Spec for the instruments, having in mind that any excess in the limits will have its direct impact into cost.
A very good example of an excess in instrument spec is often seen in the selection of a DVM for your test platform: 8-digit resolution for a DVM is not a must for a production Final Test, because when in industrial environments the measurement conditions (Temperature, humidity,…) do not allow this kind of precision, and despite the case of calibrations, a 8-digit resolution will condition the product functionallity. Instead, a 4 ½ digit DVM will also be OK, having in mind that the more readings per second available, the more easy it will be to shorten the cycle time.
The case can be even worse when talking about frequency specs. Managing high frequencies will make the tester exponientially more complicated when going beyond GHz. When selecting the maximum working frequency, we are conditioning not only taking in account the cost of the instrument, but also the wiring, the switching and all fixturing components cost.
Another important point is the specification of the limits (accepted values for each measurement). A specification of 5Vdc at M1 is worse than a specification of 5VDC + 0,01%. The reason is obvious: The 5Vdc spec will potentially end with a conflict in between parts, since the tolerance is not set and therefore there are no PASS/FAIL limits. The 5VDC + 0,01%, even though is probably difficult and expensive to reach, will never end with a conflict.
Go to Document Menu
Move from Lab thinking to Production thinking
To define the parts that a system must process per hour or per shift during production is relatively easy.
Even to predict the ICT (In-Circuit Test) test time is easy as most can be done automatically by software.
Due to the nature of the tests it is much more complicated to determine the time that will be needed to complete the entire functional test process.
The time invested in stimuli, measurement, value comparison and data registering, will depend more on the test strategies selected or the design for testability incorporated, rather than on the chosen instrumentation itself.
Apparently, when talking instrumentation, higher Reading speeds to shorten time, will mean more inconsistency in your results or higher tolerances that need to be accepted. This is true up to a point, as we also have to take into consideration the fixed times coming from switching and stabilization of conditions for a proper measurement.
In the case of test systems with manual feeding (operator), the ergonomics of the test system must be considered as key factor for productivity. The more comfort the operator experiences in loading and un-loading the test system, the higher the throughput and productivity will be.
In the case of automatic test systems (in-line), the transfer times of the Unit Under Test (loading, make connections, testing, disconnection, un-loading) will be more important to shorten the cycle time.
Mass production lines, feature transfer times below 5s, as well as by-pass conveyor belts to avoid product flow interruptions caused by the products that do not need test.
A key factor for productivity is the number of parts that fail the test, since these parts will need to be re-tested after examination and/or repair. When the test fails it could be either because of a real failure of because of a system failure. Instruments rarely fail, but connections do.
Two important recommendations to minimize failures coming from the system itself are:
1. Fixtures prepared with cycles counter to be able to create an alarm for preventive maintenance of the test probes before they start failing (according to the recommendation of the manufacturer).
2. Re-Test action that will repeat connection action two or three times before starting the test. This action is to be implemented in those cases where due to the PCB morphology a good electrical contact is difficult (i.e. very low voltages and dirty contacts).
There are Self-test fixtures and software tools available to facilitate the detection of system failures even before the actually happen.
Go to Document Menu
Portability is Flexibility
Here we can talk about the transportability of the test platform itself inside the factory as well as the transportability of the test platform from the manufacturer to globally located manufacturing sites.
In general, the transportation cost is determined by the size and weight of the equipment as well as its internal wiring. When the instrumentation racks need to be separated from the DUT fixture, to move the set up will mean disassembling it, meaning that the cost and time will increase, and in addition, the risk associated to not getting the same results as before disassembling will be there.
In the nowadays world, the production is a global thing, and companies may have their production plants in different countries. In such cases, the requirements for the selection of the best test platform may consider seriously aspects as the technical service and spare parts close to the production sites.
To ensure the global transportability, there are some factors to be considered when selecting the vendor:
1. Detailed technical documentation in English (as most commonly used language when talking electronics), including spare parts lists and local suppliers. Commissioning instructions. Calibration instructions,…
2. Options for different requirements on input voltage depending on destination country
3. Graphic User Interfaces (GUI) where the language is a parameter.
4. Maintenance Manuals
5. Spare part kit (relays, mechanical parts,...)
6. Unpacking & transport instructions.
7. Remote assistance service available.
8. Failure diagnostic tools with repair recommendations.
Go to Document Menu
Mix & Volume
Depending on your needs...
The requirements of the parts to be tested may be very different:
• ICT (In-Circuit)
• ISP (Software downloading)
• BS (Boundary Scan)
• HiPOT (High voltage test)
• LED Test
• FCT (Functional test)
o Working voltages above 50V
o Currents beyond 10A
o Frequencies from baseband to RF
o Communication buses, wired or Wi-Fi (LAN, CAN, Ethernet...)
o Radio electric emission or reception
o Optical Fiber
o Test of Electro mechanic Components
• Various supply voltages or loads
Generally, DUT requirements imply the need of more than one of the technologies above, therefore, a test platform designed for general use will have the need to include resources to cover the tests specified.
On the other side, each new DUT requires a customized fixture and test executive software. If a number of fixtures is foreseen, it will be imperative to look for technical solutions focused on lowering cost, fast implementation as well as re-usability of fixture components when the life cycle of the products is finished.
When companies need to manage big manufacturing volumes, or the same product has to be manufactured in different sites, another parameter comes into the equation: Fixture exchangeability. This is an easy topic for ICT fixtures, but more complex one in the case of FCT. In FCT, every cable is permanently in touch with the DUT and therefore can be influenced or can influence the neighbourhood cables. To get exact copies for FCT fixturing, it will be mandatory to have an extensive knowledge of the behaviour of all signals involved and make the wiring exactly the same way as for the reference fixture. A final test and characterization to guarantee that the copy fixture behaves identically as the reference fixture will also be a must.
Go to Document Menu
Signal switching might answer to the following system needs:
1- To connect the instruments to every Test Point that needs it
2- To connect the power supplies, loads or temporary used devices.
3- Connection operations required by the DUT itself; example: switch simulation, start-up conditions set up...
Switches can be solid state or relays. Necessary data to make a good selection are:
- Maximum voltage
- Maximum current
- Capacity between contacts or between contacts and coil
- Contact resistance
- Mechanical and Electrical live of contacts (number of operations).
- Response time when connecting or disconnecting.
To take into consideration, also, the type of measurements to be performed as well as the isolated power supplies implemented in the DUT. In the simplest case, a DUT powered up with only a single power supply, where measurements will be done with reference to the Common. A 1-wire multiplexer will be enough, with the Common directly wired to the measurement device.
When measurements must be performed with other references, different from the Power Supply Common, a 2-wire multiplexer will have to be used. With this kind of multiplexers, we can do alternative measurements as the circuit can be always fully isolated from the source.
Kelvin measurements (4-wire) are also possible and thinking about them when talking about general purpose ATEs is strongly recommended. The YAV90059 is a great example of a very flexible mux to cover the above needs, since it is software configurable in order to act as a 1, 2 or 4 wire mux.
HF switching requires specific multiplexers, where in addition to the switching power, the circuit impedance, attenuation, crosstalk or channel coupling, CRRM, etc...are taken into consideration. When talking high frequency, each and every cable must be considered as a genuine component to be characterized.
Another important point, will be to have individual cycle counters for each relay. This information has can be used to: Calculate the relays live and debug the test executives in order to have all the relays with same level of switching operations per cycle.
Go to Document Menu
Standard or Custom Platform
Watch the hidden costs!
At first sight, a standardized platform, ready to test many different products, usually has more components and its price might be higher than one dedicated to a single product. But this statement is false, since in a standard platform, the mechanical and electrical engineering, the machine safety and the signals mass interconnect,...has been designed and tested multiple times, its parts and components are optimized and modified to get a better performance and endurance.
Manufacturing of standard platforms will always be in larger quantities, more efficient and with better quality processes. This is why the assumption that the cost of an open platform will be higher, is only valid when the special system does not take into account all non-previewed costs that will reveal themselves while debugging your test system.
When talking about the software, there's also a large benefit when you standarize. A software engineer can generate code based on whatever programming language, but it would be far from the goals of the test system user. The robsutness of a complete test system (hardware and software) depends mainly on its documentation, test results standardization, easy maintenance and reliable hardware, all of this can only be achieved when using a commercial software product.
Go to Document Menu
The cost is relative
The test system controller is the computer that executes the test executive software, and manages the instrumentation. The power and cost of industrial PCs in all their form factors have made them the best controller for a test system. They have free PCI and PCIe slots, as well as USB and Ethernet ports to control LXI instruments.
When the demand increases, either in terms of instrumentation or in terms of long warranty, PXI controllers are a great option. Since they are sharing the controller bus with the instrumentation, data transfer capacity is much higher and reliable.
PXI Embedded controller manufacturer's (i.e. National Instruments) choose processors from the embedded (long-term) roadmap of Intel, and have a wide range of offering: from best performance to best value; in addition, desired features as the one for monitoring the temperature of the PXI chassis are very welcome.
Beside that, another great advantages of using a PXI platform vs the rack & stack standard approach is that the test system size is reduced drastically, in terms of cost, footprint, weight and facility.
Go to Document Menu
S.A. Sistel . C. Solsonès, 87-89, Pol. Ind. Pla de La Bruguera, 08211 Castellar del Vallès · Contacto · Sitemap · Nota legal · CIF: A08768434
Tel. +34 937 270 074 · Fax +34 937 253 576 · www.6tlengineering.com · firstname.lastname@example.org