Low-radiated noise bench linear power supply

[Sariel]

Hello,

I'm seeking recommendations for a linear power supply without any constraints on budget or brand.
While I’ve come across older discussions on this topic, I’d like to gather new insights and suggestions as of 2024.

My focus is primarily on super low-noise analog circuits, such as amplifiers and receivers.
For my lab bench power supply, I'm looking for a solution with minimal radiated noise, as this has been a recurring issue when working with high-gain, low-noise amplifiers.
I’m less concerned about conducted noise like ripple, since I typically use LDO regulators on my boards.

The main issue I encounter is the radiated noise from the power supply itself, which tends to interfere with my sensitive circuits.
I can easily detect switching frequencies from test equipment, and I’m looking for a power supply that is well-shielded and has negligible radiated emissions.

Are there any specific models you’ve used that have performed exceptionally well in these aspects?

Thank you,

[tautech]

Would suggest only pre 70's linear PSU's with analog gauges......if noise sensitivity or your workflow is so subject to local radiated emissions.

[tautech]

We need specify standards that PSU's might need to meet.

FWIW
4ch SPD4000X specifies:
EMC Class A test and measurement products in compliance with European EMC Directive 2014/30/EU

[BILLPOD]

Battery, (ies)

[nctnico]

EMC compliance says very little as there can still be a lot of radiation emitted.

The suggestion for older linear power supplies (ones without a digital readout) is a good one. Otherwise existing PSUs can be placed in shielded housings and / or positioned far away.

And then there are batteries but I don't see how batteries would work for a circuit that needs to become a product at some point.

[bsw_m]

Classic Agilent/Keysight E3630A?
This PS low noise and have very low common mode current.

[thm_w]

OP: would be useful to provide some current and voltage ranges you want to use.

And then there are batteries but I don't see how batteries would work for a circuit that needs to become a product at some point.

"For my lab bench power supply, I'm looking for a solution with minimal radiated noise, as this has been a recurring issue when working with high-gain, low-noise amplifiers."

Maybe a 12V lifepo4 or lead-acid battery and a charger that can be switched off when measurements are being taken.

[Sensorcat]

I’m less concerned about conducted noise like ripple, since I typically use LDO regulators on my boards.

I was curious about this and arbitrarily picked a datasheet of a modern TI LDO with high PSRR. Looks promising:

TPS7A94 1-A, Ultra-Low Noise, Ultra-High PSRR, Low-Dropout Regulator

High power-supply ripple rejection (PSRR):
–102 dB at 100Hz
–110 dB at 1 kHz
–95 dB at 10 kHz
–78 dB at 100 kHz
–50 dB at 1 MHz


They provide a lot of PSRR plots on page 8; 9, up to 10MHz. Actual PSRR can be as low as 30dB, always falling towards 10MHz. I don't want to question your skills, but are you 100% sure that none of the noise you see gets past the LDO?

[Sariel]

Thank you for your responses.

I haven’t set any specific limits for voltage and current ranges yet, as I’m open to considering the most commonly recommended options.
Using a battery for power is a good idea and typically useful for final testing, but for most of the initial bring-up, I prefer using an adjustable power supply.

It’s clear that the noise in the circuit is being coupled (radiated) from the power supply.
As I move my DUT closer to the power supply, I can see the switching frequencies and their harmonics more prominently on the spectrum analyzer, which is connected to the output of my analog circuit.

The classic Agilent/Keysight E3630A is a solid choice! What are your thoughts on the GW Instek GPE series or the TTI PL series for this purpose?

[Someone]

And then there are batteries but I don't see how batteries would work for a circuit that needs to become a product at some point.

There are various extreme low noise amplifiers that use batteries for this purpose, they come with a charger that typically also operates the system with reduced performance.

[Someone]

I haven’t set any specific limits for voltage and current ranges yet, as I’m open to considering the most commonly recommended options.
Using a battery for power is a good idea and typically useful for final testing, but for most of the initial bring-up, I prefer using an adjustable power supply.

Perhaps a linear "lab" type power supply modified to run from batteries?

[Someone]

The classic Agilent/Keysight E3630A is a solid choice! What are your thoughts on the GW Instek GPE series or the TTI PL series for this purpose?

Given how much the TTI units audibly buzz, their emissions might be short of your needs. But the only real test is to measure in your frequency range.

[wilhe_jo]

The main issue I encounter is the radiated noise from the power supply itself, which tends to interfere with my sensitive circuits.
I can easily detect switching frequencies from test equipment, and I’m looking for a power supply that is well-shielded and has negligible radiated emissions.

Well, get yourself a shielding room. That has about 80-100dB attenuation to the outside world. A new emi shielding room (no absorbers; just a steel "room" with nice door) sets you back in the realm of 30k. Used max 10k

Every radio "service" (radio, TV, rfid door openers,...) has pretty high levels compared to PSUs.

Other than that LED lighting can be a problem, flourescent lamps will always be a problem.

Linear PSUs tend to have big transformers having relatively large stray fields.

In the end it's a distance game.

4ch SPD4000X specifies:
EMC Class A test and measurement products in compliance with European EMC Directive 2014/30/EU

This one would be particularly bad! Class A means way higher emissions and should not operated where you have non-industrial stuff running.

So I suggest getting a battery... That's the cheap option..
Getting a shielded room is the other extreme.

Btw: I found that usb-c power banks have low emissions... At least the ones I have... I power my active loop antenna for emc measurements with them.

73

[tautech]

4ch SPD4000X specifies:
EMC Class A test and measurement products in compliance with European EMC Directive 2014/30/EU

This one would be particularly bad! Class A means way higher emissions and should not operated where you have non-industrial stuff running.

Thanks, can you link us to the standard ?

[wilhe_jo]

Well, almost any standard uses "Class A" for stuff that's classified "industrial".
Usually you have about 10dB higher limits there.

I'm pretty sure, they don't respect the standard/directive entirely. Usually, the standards give you some text to warn people from Class A products.
If not, the directive 2014/30/EU does... But I can't find any "trigger-words" in the manual other than "Class A".

Anyhow, they don't tell you the standard they applied. Directive 2014/30/EU just tells you to go with the state-of-the-art.
Google couldn't find a Declaration of Conformity either...

But I found something for the SPS5000x. There they also have "Class A" in the datasheet, and google found some DoC.
They use CISPR11/EN55011, so it's clearly not intended to be used in the residential environment.

73

[tautech]

Well, almost any standard uses "Class A" for stuff that's classified "industrial".
Usually you have about 10dB higher limits there.

I'm pretty sure, they don't respect the standard/directive entirely. Usually, the standards give you some text to warn people from Class A products.
If not, the directive 2014/30/EU does... But I can't find any "trigger-words" in the manual other than "Class A".

Anyhow, they don't tell you the standard they applied. Directive 2014/30/EU just tells you to go with the state-of-the-art.
Google couldn't find a Declaration of Conformity either...

But I found something for the SPS5000x. There they also have "Class A" in the datasheet, and google found some DoC.
They use CISPR11/EN55011, so it's clearly not intended to be used in the residential environment.

Thanks for your effort.
Maybe others can hunt out more info.

SPS5kX is hardly a bench PSU and it's SMPS.
I have a 40V 60A unit for general high power use although haven't seen it push above 35A 28V while on 320Ahr SLA charging duties.
Will do some Power On tests for Lab Power Supply Turn ON and OFF Characteristics thread maybe later today after ticking some other stuff off.