lower frequency limitation on this HBT PA

Hi all,


I need to use this Power Amplifier,

http://www.datasheet.directory/datas...micro_devices_[rfmd]/RFPA2126SQ.pdf

but I need it to work down to around 450 MHz. The datasheet says its a 700MHz to 2.2GHz PA. It says its a single stage, 13.7-dB gain class AB PA, and the output biasing is off-chip. So I'm thinking its something as simple as a single device with an input DC block (a large capacitor), base biasing, and open-emitter output. Does that sound right?

So the main question is what would prevent me from using this below 700 MHz? Is it only the input high-pass formed by the DC block, or is it something I'm not aware of?

Next question is how would I go about modifying it? My thinking would be to simply analyze the source and load impedances, and synthesize new networks with the same impedance, but at lower frequency. BTW, I haven't been able to find s-parameters, or any kind of circuit models for this device.


thanks,
Aaron

Hi Aaron
I doubt if the input has a DC block. It costs to put a capacitor like that into a package. Also the application circuits show a low impedance series capacitor at the input, if there was already a DC block on the device thiswould not be needed.
The data sheetmentions pre-matching to about 5 Ohms at the input. This is probably what is limiting the datassheet bandwidth. The pre-match circuits are usually low pass structures, they can be made cheaply with bonding wires and simple capacitors to ground. If this is the case you should not have too mach of a problem maching to it at 450MHz. The probelm is without a device you would be making guesses at the impedances. Using the application circuits to estimate them at 450MHz is not going to be much use they are too far away from 450MHz to make it any better than a guess.
If you have to use the device, get one and measure its impedances.
At 450MHz you will be using wound inducotrs so tuning to find the optimum before fixing a value will not be too tiresome, particularly if you ar eworking aover a narrow bandwidth.

Peter

Hi Peter,


thanks for the help. Your logic sounds right, but anyway I can check tomorrow to see if there's a DC voltage at the input. Also, I was thinking that if the current gain is around 100, then the input resistance would be βnVt/I = 100*26m/0.5 = 5 ohm. I'm guessing that for this kind of device, the LPF which you speak of would not really transform the impedance much except to make it resonate around the operating frequency. Does that sound possible? Anyway if its true, then at 450MHz, I should be able to simply match my source to 5-ohms. I don't really have the tools to analyze the impedance, but I may be able to do some experimenting to figure out the input impedance at DC. Anyway, thanks for the help.


Aaron

Hi Aaron
Not having a network analyser is a bit of a handicap when it comes to this sort of thing, and trying to make up a bridge that would give meaningful results is a lot of work.
5 ohms is as good a start as any other, the impedances will be low and in that region. If you have can use variable capacitors and home made inductors that you can deform and change you should be able to tune it to work.

Peter

Hi Peter,


I checked the DC voltage at the input, and it turns out that the input really is AC coupled. So perhaps the high-pass formed by the AC coupling cap is really what's limiting the lower frequency range. As you say, tuning is going to take a lot of iteration.


thanks for the help,
Aaron