Measuring ripple current.

[Paul]

I have a circuit fed from a 14 to 37V DC feed. The circuit naturally uses
between 20 and 80mA however it musn't present more than a 20uA ripple
current to the feed incase it corrupts communication on the line. I have
used a shunt regulator to try and control the current and the normal array
of caps to help smooth the current usage.

My question is what would be a valid way to measure that I have achieved a
20uA range, Is just inserting a DVM set to AC current sufficient or is it
likely to distort the results with any internal resistance it may have. The
DVM I am using is an old TTI 1905a although it is still within calibration.

Paul

 

[Rene Tschaggelar]

I have a circuit fed from a 14 to 37V DC feed. The circuit naturally uses
between 20 and 80mA however it musn't present more than a 20uA ripple
current to the feed incase it corrupts communication on the line. I have
used a shunt regulator to try and control the current and the normal array
of caps to help smooth the current usage.

My question is what would be a valid way to measure that I have achieved a
20uA range, Is just inserting a DVM set to AC current sufficient or is it
likely to distort the results with any internal resistance it may have. The
DVM I am using is an old TTI 1905a although it is still within calibration.

Naa, a set of caps is the opposite to smoothen
out a ripple current. An inductor, as big as
possible is better suited. And no, unless you
want to measure line ripple, a DVM is sub-optimal.

What frequency would the ripple current be ?

Rene

 

[Paul]

Naa, a set of caps is the opposite to smoothen
out a ripple current. An inductor, as big as
possible is better suited. And no, unless you
want to measure line ripple, a DVM is sub-optimal.

What frequency would the ripple current be ?

Rene

They haven't specced the ripple current frequency but, data on the line is
current pulses of 18mA for a duration of 240uSecs and rate of 1265Hz,
The central controller has to detect the presence or absence of the 18mA
pulse to decode data from over 100 devices on the same line. When I switch
to the heavier load it would be for a period of several milliseconds but
other things are going on in a micro and display which could effect the
load. Also the power is regulated to my circuit by a switching regulator
down to 5 ( then a TL431 to 3v3 to provide a constant current draw) which
may have some impact on the ripple current.

What measurement setup would you use to determine if I have met the spec.

Many Thanks

Paul

 

[John O'Flaherty]

They haven't specced the ripple current frequency but, data on the line is
current pulses of 18mA for a duration of 240uSecs and rate of 1265Hz,
The central controller has to detect the presence or absence of the 18mA
pulse to decode data from over 100 devices on the same line. When I switch
to the heavier load it would be for a period of several milliseconds but
other things are going on in a micro and display which could effect the
load. Also the power is regulated to my circuit by a switching regulator
down to 5 ( then a TL431 to 3v3 to provide a constant current draw) which
may have some impact on the ripple current.

What measurement setup would you use to determine if I have met the spec.

Given what you said is the range in your load current (20 to 80 mA) and
the wide range of your supply voltage (14 to 37 V), the static drop
across the meter resistance shouldn't be a problem.
You might also insert a resistor (maybe 10 ohm) in series with the line
at the point where the ripple must be measured, and use a sensitive
differential input scope to measure the voltage across the resistor.
You'll see exactly what the current is, at 10V / amp; for 20 uV ripple,
you would have 0.2 mV on the scope. You can then see what's going on in
terms of frequency. If you don't have a scope that sensitive, you might
be able to rig up a battery operated differential amplifier to boost
the voltage across the resistor to a level where you could see it. Of
course, you'd have to watch out for things like the frequency response
and noise characteristics of the amp.

 

[Rene Tschaggelar]

They haven't specced the ripple current frequency but, data on the line is
current pulses of 18mA for a duration of 240uSecs and rate of 1265Hz,
The central controller has to detect the presence or absence of the 18mA
pulse to decode data from over 100 devices on the same line. When I switch
to the heavier load it would be for a period of several milliseconds but
other things are going on in a micro and display which could effect the
load. Also the power is regulated to my circuit by a switching regulator
down to 5 ( then a TL431 to 3v3 to provide a constant current draw) which
may have some impact on the ripple current.

What measurement setup would you use to determine if I have met the spec.

From the line, I' first have a LC- PI filter to get rid of the
switcher noise, then I'd connect your device through a coil of
several 100mH to a Cap of 1000uF and run the switcher from there.

To measure the current, insert a shunt into the line.
I doesn't generate ripple, just a little voltage drop.

Rene

 

[Paul]

Given what you said is the range in your load current (20 to 80 mA) and
the wide range of your supply voltage (14 to 37 V), the static drop
across the meter resistance shouldn't be a problem.
You might also insert a resistor (maybe 10 ohm) in series with the line
at the point where the ripple must be measured, and use a sensitive
differential input scope to measure the voltage across the resistor.
You'll see exactly what the current is, at 10V / amp; for 20 uV ripple,
you would have 0.2 mV on the scope. You can then see what's going on in
terms of frequency. If you don't have a scope that sensitive, you might
be able to rig up a battery operated differential amplifier to boost
the voltage across the resistor to a level where you could see it. Of
course, you'd have to watch out for things like the frequency response
and noise characteristics of the amp.

Would it be OK to use the same 10 Ohm resistor and digital storage scope to
measure the peak in-rush current. I have a current limit which should
prevent currents exceeding 1mA at power up. This is under control of the
micro so once it is up and running the power can then go to the full 80mA.

Many Thanks for your help

Paul.

 

[John O'Flaherty]

Would it be OK to use the same 10 Ohm resistor and digital storage scope to
measure the peak in-rush current. I have a current limit which should
prevent currents exceeding 1mA at power up. This is under control of the
micro so once it is up and running the power can then go to the full 80mA.

I would think it would be alright, as long as your scope can give you
adequate measurements at a level of 10 mV. All the resistor will do is
subtract 10V/amp from the voltage applied to your circuit, and you can
record that voltage to see exactly what is happening with current. If
your scope is sensitive enough, you might use 1 ohm. Since your circuit
works from 14 to 37 volts, it isn't going to miss 0.8 V for 10 ohms (at
your maximum load) or 0.08V for 1 ohm. Anyway, at a nickel a resistor,
it's a really cheap thing to try.

 

[Roy L. Fuchs]

I have a circuit fed from a 14 to 37V DC feed. The circuit naturally uses
between 20 and 80mA however it musn't present more than a 20uA ripple
current to the feed incase it corrupts communication on the line. I have
used a shunt regulator to try and control the current and the normal array
of caps to help smooth the current usage.

My question is what would be a valid way to measure that I have achieved a
20uA range, Is just inserting a DVM set to AC current sufficient or is it
likely to distort the results with any internal resistance it may have. The
DVM I am using is an old TTI 1905a although it is still within calibration.

Paul

Place a low value resistor in series with the power lead, and the
device under test, and read the voltage signature the resistor with a
scope. That signature can be read as the current signature. You
would need a scope to see a ripple that small.

It would seem that a simple cap across the input side of a device
that is only consuming 80mA would keep your comm corruption away.