Smoothing DC Output from Transformer

[JaredJames]

I have a power supply which takes 230 VAC input and gives 24 VDC output. I use it to drive some differential pressure sensors (24 VDC supply and 0-10 VDC signal output).

It is stated to have a noise of 80 mV peak to peak. This can equate to a 25 Pa variation in readings

I'm finding with certain instruments they can jump around a fair bit and I suspect the ripple is the cause.

Is it worth me trying to further smooth this in some manner, such as by putting a capacitor parallel to the load? I'm not sure if it will provide any useful reduction in the noise and don't want to tear the device apart without some confidence.

For reference, I also have a low noise supply with 1 mV peak to peak, where the readings stay stable to within 1 Pa. Hence my suspicion that it's due to the noise.

The only alternative is that it's some smoothing done within the software that I'm not aware of on the original device, that I haven't added to my own.

Any help / ideas appreciated.

 

[davenn]

I have a power supply which takes 230 VAC input and gives 24 VDC output. I use it to drive some differential pressure sensors (24 VDC supply and 0-10 VDC signal output).

is this a commercial one or one you have built ?

Is it worth me trying to further smooth this in some manner, such as by putting a capacitor parallel to the load? I'm not sure if it will provide any useful reduction in the noise and don't want to tear the device apart without some confidence.

the answer to this in some part depends on the answer to my first Q

But additional smoothing capacitors isn't an issue ... what sort of current is your circuit ( all of it) drawing ?

Dave

 

[berkeman]

For reference, I also have a low noise supply with 1 mV peak to peak, where the readings stay stable to within 1 Pa. Hence my suspicion that it's due to the noise.

If the voltage levels can work out, one common way to fix this is to follow your transformer, full-wave rectifier and smoothing caps with a low dropout linear voltage regulator. You need the minimum voltage input to the regulator (the troughs of the input ripple voltage) to stay above the dropout voltage of the linear regulator (hence why you use a low dropout regulator). If the troughs of the input voltage across the capacitors at full output current loading is 23.5V, you could probably manage about a 22.5V output voltage from your adjustable low dropout linear voltage regulator, and it should be nice and quiet.

http://www.williamson-labs.com/images/pwr-supply-vr-linear-446.gif

 

[JaredJames]

Thanks for the responses.

is this a commercial one or one you have built ?

This is a commercially bought one.

This one to be exact: http://uk.rs-online.com/web/p/din-rail-panel-mount-power-supplies/0428445/

Looks like the noise is 150 mV peak to peak (previously said 80 mV).

what sort of current is your circuit ( all of it) drawing ?Dave

The sensor draw is tiny. Only a few mA.

 

[JaredJames]

If the voltage levels can work out, one common way to fix this is to follow your transformer, full-wave rectifier and smoothing caps with a low dropout linear voltage regulator. You need the minimum voltage input to the regulator (the troughs of the input ripple voltage) to stay above the dropout voltage of the linear regulator (hence why you use a low dropout regulator). If the troughs of the input voltage across the capacitors at full output current loading is 23.5V, you could probably manage about a 22.5V output voltage from your adjustable low dropout linear voltage regulator, and it should be nice and quiet.

Thanks for the reply.

So would we be looking at a lower output voltage than 24V if I implemented smoothing? I may get away with a small drop, but a couple of volts may not be feasible.

 

[Guineafowl]

Thanks for the reply.

So would we be looking at a lower output voltage than 24V if I implemented smoothing? I may get away with a small drop, but a couple of volts may not be feasible.

My (fairly crude) approach would be to pull a cap, say 50V rated, the larger capacitance the better, from a scrap board (do you have a box of scrap boards? You should!) and rig it across the 24V output with some jumpers.

Ideally you'd have a scope to measure the ripple reduction, but functional testing would work well too. See if you get more stable readings. Check your output voltage is within limits.

If all goes well, buy a decent low ESR cap and find a way to fit it neatly into the supply enclosure.

I had a similar problem with an 12V electric fence energiser that refused to run from a ripply battery charger supply. A CRC filter, two parallel caps with a series power resistor inbetween, had a dramatic effect on the ripple, and the current draw was low so the resistor hardly dissipated any heat.

 

[davenn]

Thanks for the responses.
This is a commercially bought one.

This one to be exact: http://uk.rs-online.com/web/p/din-rail-panel-mount-power-supplies/0428445/

Looks like the noise is 150 mV peak to peak (previously said 80 mV).[/QUOTE

ok

The sensor draw is tiny. Only a few mA.

I did ask for ALL of it, not just the sensor

what sort of current is your circuit ( all of it) drawing ?

and speaking of ALL the circuit, it's probably way past time you showed us ALL of it, specially after comments like this ...

So would we be looking at a lower output voltage than 24V if I implemented smoothing? I may get away with a small drop, but a couple of volts may not be feasible.

If we knew what your circuit looked like, type of sensor etc ... better suggestions could be made to help youDave

 

[JaredJames]

Here is an sketch of the circuit: https://www.dropbox.com/s/wc85l75a56m8ia0/2017-08-24 14.26.10.jpg?dl=0

Highly complex as you can tell.

Current draw of the sensor is 20 mA according to the datasheet. What other current draw should I be looking for in that circuit? There's about 30 cm of wire in total if you wish to include that.

 

[berkeman]

Here is an sketch of the circuit: https://www.dropbox.com/s/wc85l75a56m8ia0/2017-08-24 14.26.10.jpg?dl=0

Your sketch shows 24V shorted to GND in two places...

So would we be looking at a lower output voltage than 24V if I implemented smoothing? I may get away with a small drop, but a couple of volts may not be feasible.

That's why you use a low dropout linear regulator -- to minimize the voltage drop. You should be able to get away with about 0.5V drop at your low currents, but check the datasheet for whatever part you choose to see what the dropout voltage is.

This one to be exact: http://uk.rs-online.com/web/p/din-rail-panel-mount-power-supplies/0428445/

That looks like a switching power supply, not a line frequency transformer with rectifier and capacitive output filter. What frequency is the ripple voltage? Probably a few 100kHz? Be careful just adding capacitance to the output of a switching power supply -- there could be stability issues. Check the power supply datasheet to see if it specifies a maximum output capacitance...

 

[Guineafowl]

That IS a switching power supply. The ripple is 150mV p-p, as you say, but is this not a maximum? Under low current draw, I suspect it'll be a lot less. Is the voltmeter/sensor really being affected by kHz range ripple?

Perhaps a better solution would be a 0-30V supply, put through a linear regulator as suggested.

 

[jim hardy]

Switching power supplies are inherently noisy. They make RFI at switching frequency that shows up as 'ripple' .
For your light load you might look for a "Wall Wart" transformer based linear supply. Linears are electrically much more quiet.
they look like this

A Linear weighs about three times as much as a Switcher, so they're easy to distinguish
Rating is always printed on the case. .
Most thrift shops have a basket full of them for around $1 .. Get a couple heavy ones and see what they do ?
Some are regulated some are not. An unregulated 24V one will likely put out 27 volts or so at light load.

It's not impossible to build a 'quiet' switcher, just difficult to do it cheap enough for the consumer market.

Good luck with your project.

Have you a a link to your pressure sensor's datasheet ? Could be some filtering on its output would solve your trouble.

old jim

 

[Guineafowl]

It's not impossible to build a 'quiet' switcher

How would a money-no-object, quiet switching supply be made? Or is this a subject for another thread?

 

[berkeman]

How would a money-no-object, quiet switching supply be made? Or is this a subject for another thread?

Generally you follow the switcher with a linear regulator (usually an LDO regulator). That's what we often do for RF transceivers that can get desensitized by a noisy power supply.

 

[jim hardy]

Filters on all wires entering and exiting and lots of RF shielding.

Then you put it in an RF emissions test rig and measure what RF leaks out, keep fixing leaks until it meets your standard..
Here's an outfit offering that service
http://www.curtisind.com/files/pdf/RFI-Filters-Conducted-Emissions-Testing.pdf

Take apart some laptop power supplies. You'll probably find better ones have copper foil RF shielding inside.

I'd just stick with a linear supply.

 

[JaredJames]

Thanks all for the information. Didn't realize there was a difference between the supplies (switching / linear - I'm not electrical).

I'd just stick with a linear supply.

I bought a fairly cheap 24 VDC supply as suggested and tried it. Worked very well. The fluctuation in values is a lot lower.

I also tried the low noise supply that was available (the device was now in pieces so figured why not) and it was even better.

Reading the comments here, it's surprising that I would register such noise on a voltmeter.

 

[Svein]

If it is just a question of reducing the ripple, this simple circuit does so.