Testing LNA with battery as low noise supply
I want to test an LNA using battery as low noise VDD?
I have VDD1=0.7V and VDD2=3.3V.
Commercial batteries are 1.25V, 1.5V, 9V etc.
What is the best way (least noisy)to get my VDDs? May I use regulator or resistive network?
I think regulators will be noisy, right?
Maybe mix different chemistries to get the desired voltage? Like two alkaline cells in series to get ~3V, then use two NiMH cells to subtract ~2.3V from that to get 0.7V. Not very precise, but probably the lowest noise method. Maybe regulation could be done by varying the cell temperature? I wonder if anyone has ever manufactured such a thing...
The impedance from DC to GHz depends on source impedance, ESR, ESL and can be made low milliohm scale using a large ceramic and RF cap. Decoupling each stage is important with low ESR RF Caps.
THe LDO has low ESR due to feedback gain bandwidth and is specified at DC with droop Voltage vs current.
As any LNA amplifies only RF frequencies, regulator noise is easily suppressed by DC power filter. Use e.g. 100 uF in parallel with 0.1 uF and a ceramic 1 nF, and a series resistor to drop to a required voltage for your LNA.
Using a battery is OK but some dry batteries are also noisy. DC filter helps in any case.
Even the very low noise cryogenic cooled amplifiers (Noise Temp ~10K) use regulated (and well filtered) DC power supplies.
Not the same thing is in the case of the very accurate Phase Noise oscillator measurements, where using batteries instead of regulated DC power supply is mandatory.
You can search the net for various examples how making a clean of noise DC power supply, as these examples provided by Wenzel:
http://www.wenzel.com/documents/finesse.html
There are many commercially available cellular phone batteries that work 2.7, 3.6 V etc.
Power supplies can be noisy but a high quality power supply will not affect too much your LNA since the circuit has a well filtered supply path.
Thank you everyone for your comments! It is great to have all experts together.
My circuit is a receiver with fRF=2.42GHz, fLO=2.4GHz, fBB=20MHz. So, I want to decouple 2.4GHz & 20MHz. If I use 10nF for 20MHz and 5pF for 2.4 GHz then there will be antiresonance peak formed in between. My question is:
1. Is it a problem if there is no useful signal content at antiresonance? Can noise get boosted to cause ringing?
2. I may use several caps from 20MHz to 2.4 GHz and suppress the anti-resonace peaks. How should I space the caps? Some say 2X but that will lead to huge number of caps.
I do not see anything useful in employing "antiresonance". What you need is a good pass-band filter that rejects fLO, and a low-pass that passes 20 MHz IF.
In case of 2.4 GHz devices I am afraid using a very good LNA only makes the receiver more prone to interference.
Therefore, use good filtering and care less about LNA low noise figure.
It is possible to decouple 2.4G for 2.42G unless using some cavity filter, who made such frq plan?