Effect of power supply ripples on RF PA/LNA

[Simo.Da]

Hello all,

I was using a bench power supply with 1mV of ripples to power up some RF power amplifiers/ LNAs.
If I used a different power supply with nearly 100mV of ripples, will that affect the PA/LNA performance? How to know at what level the PA/LNA's response will be affected? Does the phase noise test give any indication?

Any advices please?

 

[KlausST]

Hi,

I´d look for a "power supply rejection ratio diagram" (PSRR)
Or for power supply (filter) recommendations .. and keep on them.

Talking about 1mV or 100mV ripple is useless without knowing the frequency (range).

In detail you also should know whether the 1mV or 100mV are for a single freqency, or a whole frequency range, how weighted, if it is RMS or PP...

Klaus

 

[Simo.Da]

Hi,

I´d look for a "power supply rejection ratio diagram" (PSRR)
Or for power supply (filter) recommendations .. and keep on them.

Talking about 1mV or 100mV ripple is useless without knowing the frequency (range).

In detail you also should know whether the 1mV or 100mV are for a single freqency, or a whole frequency range, how weighted, if it is RMS or PP...

Klaus

Hi Klaus,
Thank you for your reply.
Usually, the PA manufacturers don't reveal the PSRR. Also, I would think of having filtration at the output of the PSU as a last option.
That's true that it's frequency dependant. One of them will be between (100MHz to 450MHz) and the other a bit wider (4GHz to 12GHz).
Is there any test I can conduct on the PA/LNA to see when that PA/LNA will be upset? If so, what exactly do I need to measure? phase noise?

 

[KlausST]

Hi,

Generally, possibly, maybe, I´m guessing...

Why not giving exact informations:
* The exact type and manufacturer and device name
* a link to the datasheet

.. then we don´t need to guess and you get fast answers as precise as can be.

Klaus

 

[ferdem]

Is there any test I can conduct on the PA/LNA to see when that PA/LNA will be upset? If so, what exactly do I need to measure? phase noise?

The first thing to check AM distortion. Power supply ripple mostly manifests itself as AM modulation on an input. You can apply a single tone input and look for sidebands (at the ripple frequency and its harmonics) in the vicinity of the tone frequency in a spectrum analyzer. Consider that it may be down to -80 dBc or below depending on the sensitivity of the DUT. Check the entire spectrum, check the ripple frequency and its harmonics.
The ripple may cause phase distortions/noise as well, it is another aspect, you should decide what kind of distortion is important for your application.

[moderator action: corrected spelling]

 

[Simo.Da]

Hi,

Generally, possibly, maybe, I´m guessing...

Why not giving exact informations:
* The exact type and manufacturer and device name
* a link to the datasheet

.. then we don´t need to guess and you get fast answers as precise as can be.

Klaus

Here is the LNA https://www.minicircuits.com/pdfs/ZFL-500LN+.pdf
and here is the WB amp https://www.minicircuits.com/pdfs/ZVA-183-S+.pdf

 

[KlausST]

Hi,

how comes WB amp into play now? Did you mention it before?

Good that you posted the link. Now I know that this is a complete amplifier device (or module) and not a single IC.

However, the table on top of page2 shows that the gain varies with the input voltage. Thus you get some kind of amplitude modulation.
This may be ture for very low frequencies above the supply, but it does not state how it behaves at higher frequencies.

What I´d do:
* Check their web site whether there are application note or other addtional informations that answer your question.
* In doubt you may contact their technical support.

Klaus

 

[Simo.Da]

The first think to check AM distortion. Power supply ripple mostly manifests itself as AM modulation on an input. You can apply a single tone input and look for sidebands (at the ripple frequency and its harmonics) in the vicinity of the tone frequency in a spectrum analyzer. Consider that it may be down to -80 dBc or below depending on the sensitivity of the DUT. Check the entire spectrum, check the ripple frequency and its harmonics.
The ripple may cause phase distortions/noise as well, it is another aspect, you should decide what kind of distortion is important for your application.

Should the ripple's frequency range between 60 to 120Hz?
The LNA frequency range is between 0.1-500MHz, while the band of interest is something around 100-450MHz. Here I can understand the ripple's frequency may upset the LNA biasing point!
The PA frequency range is between 700-18000Mhz, while I am interested only in 4000-12000MHz. Has the ripple's frequency anything to do with the PA range?

 

[dick_freebird]

EVery nonlinear element is mixer and every switching power supply is a frequency comb generator. Biggest problem may be the fundamental if the channel frequency mask extends to center plus PS fSw, harmonics more likely step on adjacent channels, none of this is good.

Use of LDOs to knock down ripple requires high loop BW which is not a feature of older types, but a specialty application attribute.

 

[biff44]

if it is a concern, why not add some honking big capacitors to the DC bias lines? Add 100 uf in parallel with say a 1 uF ceramic cap, and that ripple will be much less.

and yes, ripple WILL show up as additive phase noise on a signal passing thru the amplifier. it will be a small effect, but sometimes you do not have that much tolerance for things like phase noise.
If the ripple is more sinusoidal in nature, the noise added will be more like a single tone phase modulation than actual pseudo random phase noise. Systems respond to a single tone phase modulator spur differently than they do to broadband phase noise

 

[G4BCH]

My experience of MiniCircuits packaged amplifiers is that they are one or more of their MMICs on a connectorised housing with application note type bias i.e. usually just a resistor to the supply and simple decoupling. I would add your own regulated supply with additional filtering to reduce high frequency noise to acceptable levels. In addition tot he capacitors recommended by biff44 I would consider adding some series inductance to reduce the HF noise still further. That would depend on your noise requirements. When I used those sorts of things the requirements were quite tight.

 

[Simo.Da]

if it is a concern, why not add some honking big capacitors to the DC bias lines? Add 100 uf in parallel with say a 1 uF ceramic cap, and that ripple will be much less.

and yes, ripple WILL show up as additive phase noise on a signal passing thru the amplifier. it will be a small effect, but sometimes you do not have that much tolerance for things like phase noise.
If the ripple is more sinusoidal in nature, the noise added will be more like a single tone phase modulation than actual pseudo random phase noise. Systems respond to a single tone phase modulator spur differently than they do to broadband phase noise

I'm thinking of adding a filtering capacitor as the last solution.
So, depending on your comment, the only way is to capture our fundamental tone of the PA response experimentally before and after adding the new power supply and see if there is any change! Is that true?

 

[KlausST]

Hi,

for me a filtering capacitor is the first solution to think of.. because it´s simple.

to capture our fundamental tone

What means "our" and "fundamental tone".
If this means "your wanted signal", then I think your approach is not the best, because it focusses on the "wanted signal". But if you are concerned about noise, I´d rather focus on all except the wanted signal.

Thus doing a spectral analysis / FFT is what I´d do.

Klaus

 

[biff44]

well, back to the original question, do you have a phase noise test system?
if so send a clear signal through this amplifier, and measure the phase noise at the output of the amplifier (be careful to put a high power load on the amplifier, and measure the output after a directional coupler so you do not blow up your test equipment).
then come back here and post the phase noise result, AND tell us exactly what type of system you are trying to use this in, and we can comment further.

or another way to do it, if it is a communications system, run the system with the transmitter amplifier first run by the good lab supply, and then run by the power supply with ripple, and see if there is a significant increase in bit errors with the 2nd power supply.

 

[Simo.Da]

well, back to the original question, do you have a phase noise test system?
if so send a clear signal through this amplifier, and measure the phase noise at the output of the amplifier (be careful to put a high power load on the amplifier, and measure the output after a directional coupler so you do not blow up your test equipment).
then come back here and post the phase noise result, AND tell us exactly what type of system you are trying to use this in, and we can comment further.

or another way to do it, if it is a communications system, run the system with the transmitter amplifier first run by the good lab supply, and then run by the power supply with ripple, and see if there is a significant increase in bit errors with the 2nd power supply.

Yes, I have a SignalHound that is able to measure the phase noise.
Unfortunately, I am still waiting to receive the switch-mode power supply as the lead time is a little bit long.
Once I have it, I will post the phase noise results here to get you advice.

 

[biff44]

good.
realize too that what you are using this PA for matters.
if you are sending low data rate digital communications, the close in additive phase noise is a big deal. If you are sending 200 MBPS data rate digital communications, the phase noise from your power supply will be filtered out by the carrier tracking loop in the receiver

 

[Simo.Da]

Right. Sorry for the delay in response as I was waiting to loan an SMPS.
Here is a phase noise measurement using linear PS and SMPS
The setup was the following: The power supply is parallel connected to the DUT amplifier + DC electronic load to run it nearly at the maximum available full load (around 10A)
I can't see any significate difference as you can see below

But, when I used the same setup connecting the oscilloscope instead of the spectrum anaylser, I got something terrible as you can see below
Linear PS

SMPS

I used this passive probe to connect between the scope and the PS!
Is there anything that I am doing wrong!

I really get confused. Please help!

 

[G4BCH]

But, when I used the same setup connecting the oscilloscope instead of the spectrum anaylser, I got something terrible as you can see below

Did you really connect the oscilloscope to the output of the amplifier?

 

[Simo.Da]

Did you really connect the oscilloscope to the output of the amplifier?

Sorry. It has been connected to the PS output

 

[biff44]

First, oscilloscopes are NOTORIOUS for adding their own noise on top of what is actually there. your power supply might be much better than you think it is.

one thing you could do is make a phase noise and AM noise test with the power supply--use multiple measurement averages. then replace the power supply with a battery and a bunch of capacitors, and retake the phase noise test. Compare the two. then you know the effect of the additive noise of the amplifier. you might need to float the 3rd pin earth ground on your ac wall outlet on the power supply

also, it really depends on what MODULATION you are using (if a communications system), or what sort of subclutter visibility you need (if its a radar system). i do not see you mentioning the system application.

a 4FSK sytem is a cake walk compared to a 4096 QAM system!