Power Supply Test Jig Design

[Raynor]

Hi all,

I need to design a factory test jig for the DC-DC power supply board.
I need some idea and input on how to do it....such as how to measure
the power supply ripple, transient response, and digital loading, etc.
The users are the factory people so they can't use scope to measure
them so it has to be automated. I am a digital/firmware guy so I don't
have problem in doing the user interface and programming. However, I
am not quite familiar with the power supply testing stuff so I want to
hear some advices.

Please discuss. I hope it will be an interesting topic to everyone.

Thanks

Jim Raynor

 

[Dr. Anton Squeegee]

Hi all,

I need to design a factory test jig for the DC-DC power supply board.
I need some idea and input on how to do it....such as how to measure
the power supply ripple, transient response, and digital loading, etc.
The users are the factory people so they can't use scope to measure
them so it has to be automated. I am a digital/firmware guy so I don't
have problem in doing the user interface and programming. However, I
am not quite familiar with the power supply testing stuff so I want to
hear some advices.

<snippety>

A good way I've found to test any power supply, switching or linear, is to invest in
an electronic load, such as those made by Transistor Devices and others. They often turn up
on the used test gear market, Greed-bay, or at ham radio swap meets.

The big advantage an electronic load has over just plugging stuff into a power supply
is that you can apply precise load conditions in terms of voltage, current, etc. Most have
metering which will tell you at least the output voltage and current being drawn, and they
usually have test points to allow for connection of an O-scope for noise and transient
analysis.

Personally, I like the ones from Transistor Devices the best. They're very well made,
service docs are readily available from the manufacturer, and they Just Plain Work.

Happy hunting.

 

[me]

Hi all,

I need to design a factory test jig for the DC-DC power supply board.
I need some idea and input on how to do it....such as how to measure
the power supply ripple, transient response, and digital loading, etc.
The users are the factory people so they can't use scope to measure
them so it has to be automated. I am a digital/firmware guy so I don't
have problem in doing the user interface and programming. However, I
am not quite familiar with the power supply testing stuff so I want to
hear some advices.

Please discuss. I hope it will be an interesting topic to everyone.

Thanks

Jim Raynor

Are you hiring? Send me my first check and we can talk.

 

[Paul Hovnanian P.E.]

<snippety>

A good way I've found to test any power supply, switching or linear, is to invest in
an electronic load, such as those made by Transistor Devices and others. They often turn up
on the used test gear market, Greed-bay, or at ham radio swap meets.

The big advantage an electronic load has over just plugging stuff into a power supply
is that you can apply precise load conditions in terms of voltage, current, etc. Most have
metering which will tell you at least the output voltage and current being drawn, and they
usually have test points to allow for connection of an O-scope for noise and transient
analysis.

Personally, I like the ones from Transistor Devices the best. They're very well made,
service docs are readily available from the manufacturer, and they Just Plain Work.

Happy hunting.

These are good for applying precise loads, or step loads. But the OP
stated that he needs to measure ripple and transient response. Seeing as
how the DUT is a DC-DC supply, I'm guessing that the ripple will be at
typical SMPS switching frequencies and transients will require time
resolutions of microseconds or less.

This is probably going to require voltage and current measurements
beyond the capabilities of the electronic load. There may also be a
requirement to monitor input parameters. Its possible to build data
acquisition systems from scratch, but I'd question the OP's aversion to
using a scope. A software controlled scope with data capture capable of
measuring PS transients isn't cutting edge (i.e expensive) technology.
The controller just sets up the scope and test conditions and then grabs
a snapshot of the trace. Some simple s/w can compare the waveform to an
envelope limit and produce a go-no go indication.

 

[legg]

Hi all,

I need to design a factory test jig for the DC-DC power supply board.
I need some idea and input on how to do it....such as how to measure
the power supply ripple, transient response, and digital loading, etc.
The users are the factory people so they can't use scope to measure
them so it has to be automated. I am a digital/firmware guy so I don't
have problem in doing the user interface and programming. However, I
am not quite familiar with the power supply testing stuff so I want to
hear some advices.

Please discuss. I hope it will be an interesting topic to everyone.

A test jig will be needed to connect the dut to your test system. This
will have suitable connectorization, specialized switchgear and wire
routing to allow accurate measurements at a reasonable production
speed. The design of a new jig will be required for each iteration or
model of DUT hardware.

Test systems, however, have measurement and loading hardware,
switching matrices and some kind of software with operator interface,
reporting and statistical processing features. Building your own, even
using out-of-the-box subsystems, standard interfaces and
out-of-the-box software can be quite a project. Think in terms of
man-years.

Suggest you research costs of dedicated commercial power supply test
systems, gpib and other interface hardware and software.

There are some undedicated automated test systems out there that can
be applied effectively in this kind of job. In the 80s and 90s, I saw
the 'Pilot' automated test system used effectively, in conjunction
with a suitable load - this contained its own measurement system and a
PC serial inteface, running test programs in quickbasic.

RL