# .5-600 MHz LT5512 impedance matching

Discussion in 'Electronic Design' started by [email protected], May 23, 2007.

1. ### Guest

I'm working on a homebrewed receiver and am looking at using the
LT5512
as the mixer. I'm not sure of how to arrange an impedance match and
unbalance to balance conversion over a wide range. Perhaps people
can comment on the following so I can see if I'm on the right track.

The LT5512 datasheet says:

Frequency Impedance
10MHz 18.2 + j0.14
240MHz 18.1 + j2.8
450MHz 18.1 + j5.2

One simple approach seems to be:

50 ohm input - ----------- -----------|---------- IN+
) ( ) ( ___
) ( ) (- ground --- 6 pF LT5512
) ( ) ( |
ground - ----------- -----------|---------- IN-
3:1 1:1
Coilcraft Coilcraft
WBC3-1TL WBC1-1TL

The 3:1 transformer converts the unbalanced 50 ohm input to 16.67
ohms.
The 1:1 center tap transformer converts the unbalanced 16.67 ohms to a
balanced 16.67 ohms. The center tap also supplies a DC ground for
biasing the LT5512 inputs. The 6 pF capactor reactance is:

Frequency Reactance
10MHz 2652.58
240MHz 110.52
450MHz 58.95

placing it in parallel with the LT5512 inputs results in:

Frequency Impedance
10MHz 1/(1/(18.2 + j0.14) + 1/2652.58) = 18.21
240MHz 1/(1/(18.1 + j2.8) + 1/110.52) = 17.58
450MHz 1/(1/(18.1 + j5.2) + 1/58.95) = 16.7

Questions:

1) How off base is my simple analysis?

2) Is there a better way to do this?

3) How significant is the 1.6 ohm mismatch at 10 MHz?

-- John
-------------------------------------------------------------------------
| Feith Systems | Voice: 1-215-646-8000 | Email:
|
| John Wehle | Fax: 1-215-540-5495
| |
-------------------------------------------------------------------------

2. ### Tom BruhnsGuest

There's no need to worry about such a small mismatch. Especially if
you will be connecting the antenna directly to the input, you're
pretty much guaranteed that the source won't be 50 ohms at any but a
very few spot frequencies. It will have reactance and resistance that
varies all over the map. But even if it's driven from an amplifier, I
can guarantee that between the transformer(s), the layout, and the
amplifier itself, you won't have a match within 10% over very much of
that wide band.

You can use the freeware RFSim99 to play lots of "what-if" games,
including seeing how much loss there really is for various impedance
mismatches. RFSim99 also comes with a program called "spur search"
that you may find useful in indentifying mixer spurs (and intermod
products).

In any event, you likely will end up being more interested in the
noise figure than in the power match; you'll want to match for lowest
noise, which is essentially never the same as matching for best power
transfer.

Another thing you should be careful about is selectivity: if you let
everything from the AM broadcast band up through UHF TV into the
input, you will be letting in some big signals along with the
presumably small ones you will be interested in, and distortion
products will limit what you'll be able to hear. It's customary to
put some selectivity ahead of the first less-than-stellar-linearity
parts such as that mixer. If you had a +65dBm input-referred third
order intercept, I'd say you could probably do without the
selectivity, but much less than that and you'll benefit from
selectivity. There are times when you can't do that, like if you want
to digitize a wide bandwidth all at once, but it doesn't sound like
that's what you are after.

There are some good books on modern receiver design. Maybe your
library has one or could get one for you. Even a quick review of the
key issues would probably benefit you, if you haven't already done
that.

Cheers,
Tom