Wideband Mixer Impedance Matching
I am using the LT5560 mixer, which can operate from 0.01MHz up to 4GHz.
In its datasheet, it mentions the way to do impedance matching for specify frequency only.
But, in my applications, the RF input range is from 0.01MHz to 4GHz.
So, how can I implement the impedance matching for this wide band?
If it is impossible, then what other alternatives I can use?
Thanks.
Mixers are funny devices. People want many features of them, and quite often they find meeting all is impossible.
I do not know LT5560 mixer, but in general, I think no mixer can be matched over 0.01 MHz to 4000 MHz as you wish.
Matching a mixer means achieving that its three ports, RF, LO and IF have the same, real, and stable impedance like the connected device(s). Often we work with 50-Ohm lines.
Mixers usually comprise a diode, a pair or more diode pairs, a transistor, a FET or several devices like that. Imagine trying to match one diode to three ports over a wide frequency range.
One problem is that by LO input signal, the mixer elements are "pumped" to switch on and off. This switching generates harmonics, and good mixers must be matched up to 6th harmonic.
All these problems led to a number of patented devices. Manufacturers can do wonderful things but they do not like to tell their know-how. Try manufacturer's advice, maybe you find a solution close to your problem.
The best approach is to check mixer manufacturers. e.g. by Google, and find if "your" device is available. Otherwise you will have to divide your extreme band into sub-bands. Try to look in "Mini-Circuits", they have a n ice selection of mixers, some may be good for you.
Can I know how exactly I can divide the extreme band into sub-bands?
I would recommend to look at the catalog at www.minicircuits. com. They offer many mixers covering your band of interest. According to the frequency specification you can select a minimum number of mixers for your wide band. You will then have to carefully design the local oscillators which will need to be switched for each mixer. Do not attempt to run all LOs, the mixer products would be not controllable.
Mixers need to be accompanied with filters, at RF input to define the RF input bands, at IF output to obtain only the wanted products.
As an example, try to find schematics of old-style communication receivers. Typical commercial receivers cover 10 kHz to 100 kHz, 100 kHz to 30 MHz, 30 to 500 MHz, etc. An easier solution is to use the 0.1 to 30 MHz as the basic module and add converters to cover the lower and higher bands. Or try to follow spectrum-analyzer design, with switched RF bands and a narrower receiver at ~110 MHz.
First of all, let's get realistic.
Impedance matching is oversold, unless you're dealing with a) high power, b) the requirement to match things to the external world via 50 Ohm coax or c) a FE (front end) LNA (and in which case you probably are looking for a specific case of mismatch or match for best NF anyway) and d) Filters (filters always work better working into their design Z).
Inter-circuit mismatches aren't the 'deal breakers' most designers and hobbyists are taught, and this is where simulation tools (and known or measured device S-Parameters) can shed some light.
Again, there are a few 'points' where Impedance (or Z) matching is required; everything else is negotiable.
Jim
---------- Post added at 11:42 ---------- Previous post was at 11:28 ----------
jiripolivka,
It is remarkable the progress made on all 'fronts' of electronics; even scanners (e.g. Radio Shack Pro-2006) made 20 years ago employed the 'up' conversion IF scheme to make use of 1 (or 2 in the case of the Pro-2006 on account of the channel step size they wanted) LOs to perform wide band reception from a few MHz through 1.3 GHz ... and this was consumer-level gear ... individual BP filters were used just to prevent overload, but each BP filt was PIN diode switched to the same WB (Wideband) LNA (made of several discrete bipolar transistors) then to the WB mixer ...
I think the proposed design might need to be taken to the 'spreadsheet' level with known device Z interstage and look at ML (mismatch loss) given the stack up. There will never be ideal components found, even test equipment uses pads and other 'tricks' to achieve flat BW and minimum SWR or mismatch. Note that Agilent's (HP) spectrum analyzers specify the input RL with minimum 10 dB (maybe 5 or 6 dB today) attenuation set, never straight into the mixer.
Jim