Noise in power supply

[HF_ATL]

Hi all,


I've found a circuit online to reduce the voltage regulator output noise. Please see link below.
I wonder if I can use it with +13.8V instead of +15V and if a 25A load is acceptable (using a 50W wirewound 0.05R resistor). I'm aware of 1.25V voltage drop with 25A load but that is acceptable.
My idea is to filter out as much as possible the 13.8V switching power supply I'm using with my ham transceiver.


Link: https://ibb.co/JCS6YWW



Thank you in advance!

[moffy]

Hi all,


I've found a circuit online to reduce the voltage regulator output noise. Please see link below.
I wonder if I can use it with +13.8V instead of +15V and if a 25A load is acceptable (using a 50W wirewound 0.05R resistor). I'm aware of 1.25V voltage drop with 25A load but that is acceptable.
My idea is to filter out as much as possible the 13.8V switching power supply I'm using with my ham transceiver.


Link: https://ibb.co/JCS6YWW



Thank you in advance!

I doubt the circuit will do anything useful, you could always try an LTSpice simulation as it's quite simple.

[Wolfgang]

The circuit you have here is called a "Wenzel Cleanup Shunt" and is meant for cleaning up noise for sensitive circuits like VCOs and the like, all low power.

When you are prepared to spend 1.25V in voltage drop you might go for a "capacitance multiplier" or a low-drop regulator instead.

What also helps a lot against switching noise is a Pi filter with a large enough inductor.

[David Hess]

I agree with Wolfgang; a pi filter would be more appropriate.  A 25 amp pi filter is not trivial, but the alternatives are even more trouble.

[Wolfgang]

Probably some toroids (30A types are available) would do the job.
Example: HHBC24W-2R1B0065 from KEMET, thats about 50uH under load.
Filter caps must be either ceramic or low-ESR foil for pulse.

[HF_ATL]

Thank you all for your great contribution, much appreciated!
I'll try to find a suitable toroid and capacitor. Switching frequency is around 30KHz.
I'll probably set the fc to 3KHz, choose a suitable inductor and calculate the value for the capacitor.

[David Hess]

It will be tough to get a low enough ESR and ESL on the capacitors without using many in parallel, but given the physical size of the required inductor, there should be plenty of space for multiple capacitors in parallel.  I would check out plastic film DC link capacitors which have those wide spade terminals.

[HF_ATL]

I've used an oscilloscope from a friend to have a look of the ripple (current=6.8A).
Frequency was varying a lot from 10 to 50KHz.
Picture is attached.

[David Hess]

Be real careful about making that sort of noise measurement.  Test to see if your oscilloscope is picking up common mode noise by connecting the probe tip to the ground lead while the ground lead is connected.  A common mode filter may be called for.

[Muxr]

I would also limit scope bandwidth to 20 Mhz, when measuring power supply noise.

[moffy]

+1 for what David Hess said, the spikes might not be as large as they appear. For such measurements a ground spring for the scope probe is useful: https://www.ebay.com.au/itm/385601068128
you can roll your own quite easily.

[HF_ATL]

Thank you all for the tips!
I'll repeat the measurements asap with a common-mode choke in-line and with limited 20MHz bandwidth.

[Wolfgang]

Another good idea would be to put the filter in a shielded metal box and let it have a coax monitor output.

[Vovk_Z]

Filter caps must be either ceramic or low-ESR foil for pulse.

I have to add that now polymer capacitors exist. They have a large enough capacitance combined with ultra-low ESR (typically) and high-rated ripple current (several times larger than same size 'wet' low-ESR capacitor).  So, they are perfect for low-voltage power supply filters.

As for inductors: when the current is high enough you dont need a large inductance. Something like 1 uH (combined with a polymer cap) may be quite enough to filter out a voltage after AC-DC or noisy DC/DC. A good example of such an inductor is an inductor from the CPU voltage regulator of a usual old motherboard.

[David Hess]

Thank you all for the tips!
I'll repeat the measurements asap with a common-mode choke in-line and with limited 20MHz bandwidth.

I did not mean that the measurement needs to be made with a common mode choke, but that your filtering application may require a common mode choke.

The measurement may need to be made differentially, however excessive common mode noise could be just as bad as differential noise.