Wideband matching by lumped LC matching networks

hello all...

this is my first post and i'm new to RF design. hope to learn here. i assume something similar to this topic has been discussed but i would like to ask/verify some specific questions.

i need to match 75 Ohm source to very small load from 50 MHz to 1 GHz with discrete elements. can this be achieved by lumped LC matching networks? the load to match is approx 3-j*.088.

there are two concerns here: the S11 should ideally be -15 dB throughout the bandwidth but can degrade at higher frequency a little. the second concern is that the discrete elements need to be of reasonable size because i'd like to have on-chip matching if possible.

anybody has ideas how can this be done? is it possible? i tried a low-pass LC two section network but didn't get the intended results. do i need more sections? is there some other topology that would be better. thanks

Re: Wideband matching

If you care the chip size, on-chip matching at the frequencies bellow 1 GHz is impossable. Because large LC components will occupy very large chip size!

Even with off-chip LC components it's not a easy work with the so wide band.

Re: Wideband matching

Why do you care about the impedance match? What are you driving it with?

Since your load is more like a short circuit than a 75 ohm load, why not drive it with a current source. It will be broadband, and you do not need any whacky on-chip inductors. You just need on on chip transistor.

Re: Wideband matching

I guess you intend lossless impedance matching? This may be hard to achieve over the intended frequency range. You can get RF transformers (e. g. from www.minicircuits.com) but they don't offer 3 ohms impedance at GHz. May be two cascaded transmission line transformers could do the job.

P.S.: If this is a theoretical rather than a real design problem, there could be a periodical L-C-C, C-L-L ladder structure to achieve matching, I think, but nothing that would work in real life.

Re: Wideband matching

biff44 said:

Why do you care about the impedance match? What are you driving it with?

Since your load is more like a short circuit than a 75 ohm load, why not drive it with a current source. It will be broadband, and you do not need any whacky on-chip inductors. You just need on on chip transistor.

Click to expand...

Biff44,
Can you explain this technique in details please? How will you drive an impedance with a brodaband current source ? What about the matching ???

Re: Wideband matching

Well, you would have to do a detailed design, but one quick test would be to put the load at the emitter of a bipolar transistor, ac couple drive the base with the RF signal, and just have a collector to Vcc resistor to stabilize the bias somewhat. The transistor acts like a voltage to current converter.

Another way would be to have a differential pair of transistors, and have the load in the common emmiter/source to ground connection.

BTW, there IS no matching, except to the base of the transistor. The load just sits there, approximating a short circuit.

Re: Wideband matching

thanks guys for the replies....since i don't have much experience i wanted to see if others think this is possible to achieve. it seems like it's hard to match off-chip let alone on-chip.

the current source option won't work because there is VSWR requirement. This is an R&D effort (no customer) using HBT devices. There has already been a design with PHEMT but this is using HBT in SiGe. The problem is that the transistor is sized to handle 21 dBm of output power (hence the low impedance for HBT) and it is an LNA! Need to meet the noise requirement as well.

I can get the ~1 GHz BW but it is off centered (about 150 MHz to little over 1 GHz)...as was expected can't seem to match the low side (50 to 150 MHz).