Peak to peak noise detection circuit

[Teddy]

We are testing SMPS for peak-to-peak voltage noise and would like a
quicker and easier way to test for noise failures than reading the
noise level on an oscilloscope. The output is 12VDC, switching
frequency is 200kHz, with transistor turn-on and turn-off noise spikes
ringing at 20 to 60 MHz. We are mainly interested in whether the
peak-to-peak value of these spikes is less than or greater than
100mV.

Any suggestions as to how to do this?

Thanks.

TD

 

[Joerg]

Hi Teddy,

You'd have to give more details. Noise riding on a DC output? That's easy,
just run an AC-coupled AM detector with the threshold set at 100mV.

Regards, Joerg
http://www.analogconsultants.com

 

[Ban]

Teddy wrote:

We are testing SMPS for peak-to-peak voltage noise and would like a
quicker and easier way to test for noise failures than reading the
noise level on an oscilloscope. The output is 12VDC, switching
frequency is 200kHz, with transistor turn-on and turn-off noise spikes
ringing at 20 to 60 MHz. We are mainly interested in whether the
peak-to-peak value of these spikes is less than or greater than
100mV.

Any suggestions as to how to do this?

Thanks.

TD

The 4 transistors should be in an RF-array, opamp is not critical.

100k 10n 100k
+---+------------+---+---------o +5
| | GND GND | | GND
| .-. | | | .-. |
| | | --- .-. | | | .-.
| | | --- | | | | | | |
GND | '-' | | |1k| '-' | |
| | +---+ '-' | | '-'
.-. | | | | +--+ ___ |
| | | +--------)---)---)--|------|___|--+
| |1k| | | | | | ___ | |\
'-' | | | | | | +---|___|--+---|+\
|| | |/ |< | |/ |< | | ___ | >-+--o
o-||--+-|---| +-|---| +--)---|___|--+---|-/ |
|| |> |\ |> |\ | | |/ |
| | | | | ___ | ___ |
+---)------------)---)-----)---|___|--+--|___|-+
| | | | |
| | +---)-----+
+---+ | | | all 100k
10n| .-. | .-. |
--- | | | | | |
--- | | | | | |
| '-' | '-' |
GND | | | |
+---+------------+---+---------o -5

100k 100k
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de



--
ciao Ban
Bordighera, Italy

 

[legg]

On Mon, 26 Apr 2004 21:55:20 -0500, Teddy <name@usenet.com> wrote:

We are testing SMPS for peak-to-peak voltage noise and would like a
quicker and easier way to test for noise failures than reading the
noise level on an oscilloscope. The output is 12VDC, switching
frequency is 200kHz, with transistor turn-on and turn-off noise spikes
ringing at 20 to 60 MHz. We are mainly interested in whether the
peak-to-peak value of these spikes is less than or greater than
100mV.

To measure accurately and record the data, you need instrumentation
that can do this - I've only seen it done in GPIB setups.

For an indication, you can peak detect using a differential bipolar
pair - by measuring the dc imbalance required to cut off the
ac-coupled detector side of the pair.

If a fraction of the DC input is used as a reference for another
comparator or a simple AD, an indication of percentage can be
obtained. Indicator-only type circuits can be powered from the
terminals being measured (if correctly polarized), or additional
battery power can be used to power the AD.

Usually, a properly designed SMPS has sufficient margin on published
limits, so that an indicator-only measurement is sufficient assurance
of correct circuit assembly and function.

If the published limit is 100mV ppk, anything over 50mV ppk is highly
suspect, and typical performance at room temperature should reflect a
much lower value at room temperature. This lower value reflects the
typical performance needed of products demonstrating compliance over
the entire environmental range, as established in initial design
verification and validation records, and extending over the design
life of the product.

RL