The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> question on input impedance matching
https://designers-guide.org/forum/YaBB.pl?num=1226301421

Message started by flywill on Nov 9^th, 2008, 11:17pm

Title: question on input impedance matching
Post by flywill on Nov 9^th, 2008, 11:17pm

Hi,
Forgive me if the question seems dummy.
I know impedance matching in RF design is important for maximum power transfer and min reflection. So that we need input/output Z match. It is a bit easier to understand for out Z match to have max power delivered to load. For input impedance match, it is not very clear.
1. in RX, after 1st stage (LNA), we care only voltage (assume no external components after LNA), so why max power transfer matters as long as max voltage sensed? From LNA until input to ADC, we are amplifying signal voltage, aren't we?
2. with matched input Z, no reflection in the source point, but between matching network to input of chip(assuming matching network done on pcb instead of on-chip), reflection sustains, why it does not matter?
3. if there is input Z mismatch, if max voltage sensed at input stage and amplified in following stages, what is the matter to have matching requirement?
4. what is the bad effect of input mismatch or reflection?

Thanks in advance for any clarification.

Title: Re: question on input impedance matching
Post by RFICDUDE on Nov 10^th, 2008, 6:30am

Your question is not a dumb question at all.
Why do we need to consider impedance matching?

High frequency circuits to do not have infinite input impedance because there is always some shunt capacitance, so we can not try to think about signal transfer as voltage only. If the reactive impedances are resonanted out, then you are left with the condition that the real part of the source impedance and the real part of the load impedance are not generally the same. This is where conjugate matching comes into play because maximum power transfer from the source to the load happens when the real parts of the source and load impedances are equal and the reactances are resonanted out.
Voltage mode signaling requires that either the real part of the source impedance be very small or the real part of the load impedance be very big. In general neither of these conditions are true (i.e. the real part of the source and load impedances are not extremely different). There are also bandwidth problems when the real parts of the impedance are widely different (i.e. narrow band high Q matching is required).

At low frequencies voltage mode signaling is used because the input impedance of MOSFET transistors is very high due to the high input capacitance reactance.

As a practical matter, antennas and filter/duplexers are designed with particular terminating impedances such as 50 ohm. You cannot change the impedance and if you try to present a different real part the response of the antenna or filter may be altered.

Also, antennas and filters are connected with finite segments of transmission lines. An impedance mismatch establishes standing waves on the line that creates voltage rises which are higher than the terminated signal. This could be bad for LNA dynamic range.

The other question you did not ask is why doesn’t conjugate matching (optimum source load power transfer) result in optimum noise figure performance.

Title: Re: question on input impedance matching
Post by flywill on Nov 11^th, 2008, 9:57pm

Thanks for your reply.
I agree with all your points :)
What still puzzles me is that why internally we do not care impedance matching. Say from LNA to down-mixer, it is still at RF signal, but no impedance matching considered. Is it because internal connection is much shorter than wave length of the RF signal processed?

RFICDUDE wrote on Nov 10^th, 2008, 6:30am:

Title: Re: question on input impedance matching
Post by RFICDUDE on Nov 12^th, 2008, 7:03pm

Yes, internally we are not trying to maintain a controlled impedance like 50 ohms between blocks.

However, impedances and matching are a consideration if optimum noise performance is important at high frequencies. Just resonating out the capacitances between stages does not provide optimum noise performance.

All of this starts to fall apart above 10-20GHz since the wavelengths are short enough that the interconnects are considered to be transmission lines.