Need help with understanding amplifier

polyfet lb401 amplifier

Hi,
I have the following broadband power amplifier 400M-1G that I need to reverse-engineer sort of and having problems calculating the impedance match.
I do understand that the coaxial cable transformers (Zo=10ohms) sort out the basic transformation ratio but still the transistor (LB401) has got capacitive input and inductive output. That is where I can´t see how it is properly matched.
Any hints to tackle this problem? I have got ADS available BTW.
[img=http://img147.imageshack.us/img147/7956/tb207schematicv2dy1.th.jpg]

Values for the in- and output impedance of the transistor are:
@500M
IN:2.0-j6.4 and OUT:6.2+j1.6
@750M
IN:1.4+j4.4 and OUT:4.2+j4.0
@1G
IN:1.4-j3.6 and OUT:2.0+j6.8

Thanks for looking
Jens

I think, it could be interesting to link the complete application note

Some part values are much more plausible there!

P.S.: Another interesting document

Yeah, I know the article but still don´t see how it calculates.
I´d like to actually calculate it on paper if possible to completely re-engineer.
I´m gonna contact Polyfet soon anyway, cause the version you linked to is different.
The schematic I posted is directly from one of the Polyfet engineers.

Danke!
Greets
Jens

Hello Jens,

viewing closely the circuit photo from the tb207 application note reveals, that it's more equal to the "V2" schematic from your link rather than the version from app note. You can identify the 100 ohms feedback wirewound "resistors" and also a 10k gate to ground resistor, although this value should have no noticeable effect and could be simply omitted to my opinion. This strange value was a reason why I regarded the other schematic more plausible at first look. Now I think, the polyfet engineer could be right. Probably the part values have been modified by trial-and-error, resulting in the "V2" circuit.

The most important difference is the 15 ohms semirigid for output transmission line transformer compared to 10 ohms in "V2" circuit, but this could be also a drawing error.

Generally I see, that the short (< λ/4) transformer line is inductive and can - in a simplified view - compensate a capacitive gate impedance. When comparing with the impedance data from LB401 datasheet, you must also consider the additional feedback RLC. Unfortunately, no S11/S22 data are given for the circuit, thus you can't know the quality of matching. Power gain however seems to be similar to the datasheet values.

Regards,
Frank

Many thanks for your advice.
I didn´know the RC-feedback would alter the impedance match.
I will look into that.

Regarding the versions:
Yeah, Looks like trial and error to me too.
The engineer said there was no simulation involved and I also reckon they tackle all of their amps in similar way.

The 10k gate shunt resistance is indeed not needed. Usually they use 10-20Ohms here for a lossy match. I tried the low values when I had problems with the amp but that was due to the wirewound resistors being too inductive and creating a colpits oscillator.

Grüsse aus Gladbeck
Jens

Hello Jens,

can't say, how strong feedback network influences impedance matching, but it certainly isn't neglectable. In the existing circuit, the inductance of wirewound resistor must be known.

Another point to consider: The transmission line lengths are different from values written in tb207. I see ≈ 1.2" for input and ≈ 1.55" for output transformer from the photo. This seems to be also among the empirically modified circuit parameters.

For circuit analysis, the 4:1 transformer lines can be regarded as an ideal transmission line in parallel with an inductor formed by the outer conductor loop above the ground plane. Coax to plane distance thus also has an influence on impedance relations.

Best regards /Gruß aus Bochum,
Frank

Hi Frank,
hope you don´t mind me asking more questions:

I guess you estimated the gain simply from the transconductance value in the datasheet and the impedance of T2 and neglect L2 (so ~27 or 14dB)?

Regarding the calculation of the matching:
Do you know literature that covers this?
I thought I have good books about the subject (Sevick, Grebennikov,Trask articles)
but none of them actually calculates it precisely enough.

Do you work here locally? You seem to be quite fit with the topic.

Best regards
Jens

Hello Jens,

I a'm not aware of having yet calculated a gain. I mainly was considering the operation of transmission line transformers and which parameters would would influence it. It can be basically simulated with spice or other tools, but particularly the common mode inductance must be estimated or calculated by a 3-D magnetics tool. Some parameters, e.g. wirewound resistor inductance must be also estimated or measured from a similar part.

The next thing is, that I' m not sure if the transistor parameters from the datasheet are accurate enough. You don't have non-linear model parameters, just small-signal parameters for a given operation point. But how accurate they have been determined at Polyfet? If they have been measured with a standard network analysator, the impedance level, particularly at the input is far-off 50 ohms matching, errors are heigh.

My conclusion is not to expect too much from simulation or calculation.

Finally, I must admit that I'm not presently working with RF power applications, my connection to RF is almost small signal and time domain, wide-band pulse and linear appplications. But I also had a magnetron source in an analytical instrument or was enaged with RFID test equipment.

Regards,
Frank

Thank you so much Frank.
Your insight really helped to correct my still naive views on the subject.

Best regards
Jens