Author Topic: Software question about ATE (Automated Test Equipment) platform on the cheap  (Read 2486 times)

0 Members and 1 Guest are viewing this topic.

Offline oncincTopic starter

  • Newbie
  • Posts: 1
  • Country: ca
Hi, I would like to build a ATE test system for a small volume production line.  The system would be used for measuring powers and doing a few functional test.  I attached a general block diagram.


My question is what PC software platform would be good for this purpose.  I am considering Python and sigrok.  But would like other suggestions, pros and cons, etc.
Thank you
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 27808
  • Country: nl
    • NCT Developments
I'd control everything over the serial port including the relay board.
If you configure all equipment for the same baudrate you can scan the serial ports using the *IDN command to figure out which equipment sits on a certain port.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline amspire

  • Super Contributor
  • ***
  • Posts: 3802
  • Country: au
Hi, I would like to build a ATE test system for a small volume production line.  The system would be used for measuring powers and doing a few functional test.  I attached a general block diagram.


My question is what PC software platform would be good for this purpose.  I am considering Python and sigrok.  But would like other suggestions, pros and cons, etc.
Thank you
I did do ATE back in the 70's and 80's.

It is not so much the language - it is how you implement it.

Just some basic rules:

You need a language/ide that lets you single step, do breakpoints and repeat execution of the current line more than once.
Write the program so a 5 year old can follow it. ATE is really about keeping it simple.
Have a part of the program that is in the form:

Code: [Select]
Set the relays for test 1
Do test 1

Set the relays for test 2
Do test 2

That is what you want to be able to step through. If you need to do the same thing 5 times, it is great for stepping and debugging if you just repeat the test 5 times rather then using a loop.

Also, just as a few recommendations:
Make it as quick and easy as possible to swap jigs.
Avoid the need of vacuum holddown if possible, but if you end up with a mass of sprung test pins under the board, you will need it.
Make a second jig that allows you to test all relays easily - relays stick or degrade in ATE and you want to be able to test all the relay contacts in a minute or so.
Put something on each jig - like a resistor - so that the ATE can recognised the jig and automatically load the correct program. If you run the wrong program on the wrong jig, you can damage the DUT or the ATE equipment. We just had a standard routine at the start of each test that measured a jig resistor and turned it into a number (1k = "0", 1.2k = "2", ... ,8.2k ="11", 10K = 12, 100K = 24 based on the E12 resistor values - a simple logarithmic progression). Our test programs were actually just called "0", "1", "2", "11", "12", "24" and so the routine knew what program to call just from the resistor value. This eliminated the need for a globally accessible jig resistor value to program name lookup table.

Richard
 

Online NorthGuy

  • Super Contributor
  • ***
  • Posts: 3240
  • Country: ca
If MCU has ADCs with acceptable accuracy, then you can use it instead of DMM. It can measure and transmit data back though USB. It'll certainly be cheaper, faster, and can have multiple channels.
 

Offline amspire

  • Super Contributor
  • ***
  • Posts: 3802
  • Country: au
If MCU has ADCs with acceptable accuracy, then you can use it instead of DMM. It can measure and transmit data back though USB. It'll certainly be cheaper, faster, and can have multiple channels.
DMMs are nice as they tend to be robust and they tend to have floating inputs. You are checking boards they may be faulty so the pins you are about to switch the MCU across may have the full DUT supply voltage across them due to a short.

If the jig/ATE is a little fragile or even just untrusted, then every time a board fails, the testers start wondering if it is the DUT or the ATE and jig that is the problem.
« Last Edit: December 20, 2016, 05:16:40 am by amspire »
 

Offline julian1

  • Frequent Contributor
  • **
  • Posts: 769
  • Country: au
How inexpensive?

For $300-$700 ebay/used, it is worth mentioning that the 34970a with a multiplexor board, would tick most of those boxes.

A standard configuration would be a 6.5digit DMM (voltage, current, thermocouples, freq) with a 20 relay multiplexor card for handling multiple DUT measurement inputs.

I/O can be simple rs232. So it's possible to hook up to a PC with a simple USB/serial connector that can be programmed in anything - C / python etc. That's assuming the builtin scanning is not flexible enough.

With a second multiplexor relay board, you can also control the DUT - turn it off/on, control the feeding of test signals etc.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf