Need help on transistor FHX13LG(LNA design)

[Abhishekabs]

linc2 simulator download

Hello everyone
Few months back I designed a LNA for 1.88GHz using transistor FHX 13 LG
**broken link removed**
The transistor is from fuzitsu.
My circuit did not work rather it is oscillating.
I use ADS for the simulation.
I usually make nonlinear model of transistor & simulate my circuits.
Fuzitsu does not give non linear model of transistors so I can not use this transistor for any application such as frequency multiplier or oscillator etc.
The company only provides S parameters for 2 biasing conditions.
Since My LNA circut failed I can not trust S parameters provided.
What should I do?
I dont want to waste transistors I got.
I also Dont have S parameter measurement setup to practicly measure S parameters.
Does anyone have measured the s parameters for this transistor?

Thanks

 

[khouly]

fhx13lg

u need to build ur own test setup to charcterize the transistor. try to see how to make a calibration board for ur VNA

khouly

 

[BigBoss]

fhx13lg lna

khouly..
Nonlinear characterization is one of the most difficult task in microwave filed..

linear s-parameter may only give the stability factor and gain and reflection of the device but it's not sufficient to design a oscillator or multiplier...

The best way to use well known transistors...

 

[Abhishekabs]

bigboss

Thanks big boss
You are correct it is good to take device with known charactrization
Some of my friends bought this transistor & gave samples to me.
Thats why I wanted to make something good out of it

Khouly:
I don't have any VNA for measurement
I only have access to spectrum analyzer & SNA(old one)

 

[khouly]

lna transistor manufacturers

BigBoss
I know that nonlinear charcterization is very hard.
but he want to LNA , so he can use the VNA and some calibration boards to get the S paramters of the device then design the LNA.
but sure he can't use the S parameters in the multipler design.
but the S paramters can be used in linear steps in oscillator but sure he will need to do alot of iteration

khoult

Added after 6 seconds:

BigBoss
I know that nonlinear charcterization is very hard.
but he want to LNA , so he can use the VNA and some calibration boards to get the S paramters of the device then design the LNA.
but sure he can't use the S parameters in the multipler design.
but the S paramters can be used in linear steps in oscillator but sure he will need to do alot of iteration

khouly

 

[BigBoss]

Thanks big boss
You are correct it is good to take device with known charactrization
Some of my friends bought this transistor & gave samples to me.
Thats why I wanted to make something good out of it

Khouly:
I don't have any VNA for measurement
I only have access to spectrum analyzer & SNA(old one)

For an LNA that will operate at 1.88GHz, it's not a difficult to find very well characterized bipolar or MOS transistor.
For instance NXP might be very good choice and they can send you some samples free of charge.They have very wide band transistor portfolio including GSM, PCS, DECT and ISM etc. bands.S-parameters and nonlinear models are very accurate and if you use ADS you'll probably find them in standart library..

https://www.nxp.com

 

[khouly]

yeah , i agree with u in this issue
but i think abhi need a very low noise figure , he was talking about .3 dB NF , so he needs a HEMT or Herterojunction FET


khouly

 

[BigBoss]

What about this one ???
**broken link removed**

 

[Abhishekabs]

Let us talk about FHX 13 LG
I have good transistors for LNA which can give .25dB NF.
, though
I want to use the FHX 13 LG. I have 10 Transistors.
And I dont want to thow them.
I can give them to students for design, but for that I need S parameters.
If some one has used this transistor & measured S parameterd please give it to me.
Will appreciate a lot.
Thanks

 

[RF-OM]

Dear Abhishekabs,

I do knot know your application (you did not provided enough data), so I cannot tell you how good your transistors are for your application. The fact of having 10 of them is not always the most important consideration.

Regarding the LNA design I can tell you that S-parameters are more than enough to design the LNA quickly without any problem. There are at least two ways: you can use Gonzales book or good RF simulator that allows you to use the circle utility. For example LINC2 will do all the required work in a minute and automatically stabilize your LNA. Then it is your turn how to check all the voltages and currents in DC domain and do all the temperature calculations. This part usually is much easier then RF design, but may take longer time due to hand calculations. Simple Excel based worksheet can automates this task. When you made all DC calculations and checked the temperature issues you cannot kill your transistor in LNA circuit unless something is wrong. I believe that most of transistors suitable for LNA applications died in the prototyping process due to multiple soldering and de-soldering, not becuase current or voltage values were exceeded.

 

[Abhishekabs]

Thanks RF -OM
I was helping few students to make LNA for kind of interferometer project.
It's application was to find mobile phone direction.
so the frequency was 1.88 GHz.
NF up to 1 dB was sufficient.
May be you are right, I should give one more try & use s parameters given by manufacturer.
Just One question.
Most of the transistor has 2 source leads provided but in simulation model we dont use it rather models provided by company themselves dont include 2nd source lead/terminal.
suppose during process of making any circuit say LNA,suppose you left that Pin open or soldered to small pad & other source lead you grounded(common source amplifier), will that affect performance of circuit?

I hope my question is clear.

Added after 6 minutes:

Please see example.jpeg for more calrification of my question.

 

[RF-OM]

The source or emitter impedance is crucial for LNA performance. This is the major reason why manufacturers are including the second lead for this pin. If you will leave it hanging in the air it will kill your LNA. What you need to do it to connect it to the ground plane with at least two via in parallel. If there is any kind of emitter (source) degeneration, these two pads must be connected together with as wide trace as it possible. The main idea is to to have MINIMUM INDUCTANCE between emitter or source and connected element (resistor or inductor).

 

[Abhishekabs]

suppose i want to add reactance in common source terminal to make transistor stable or unstable like i shown in figure,
so should i place the same value of reactane on other source terminal also ?to balance?

 

[G4BCH]

To stop your amplifier oscillating add a series resistor, 10 to 20 ohms should do, as close as you can get it to the drain of the device. Without it in my experience you will be very luck to get a low frequency LNA to be stable.
You can add some source inductance to get a reasonable input match without degrading the noise figure too much. Be careful because the device may become unstable in the 10 to 20GHz region.
You do not show any input matching, at this frequency you will not need much more than a series inductor.
The FHX35, a similar device can be used to make good amplifiers from below 1GHz to above 10GHz using only the manufactures S-parameters.
As with all LNA designs check the stability circles every step of the way.
Peter

 

[Abhishekabs]

Thanks G4BCH
Have you used this transistor FHX 35LG or FHX 13 LG before?

Added after 1 minutes:

RF-OM:
I did not understood the term
"emitter (source) degeneration"

 

[G4BCH]

I've used the FHX35 a few times at around 2GHz and at 10GHz, one of my colleagues has used it quite extensively between 1 and 10GHz. It is quite a good device, not as low noise as some of the latest offerings, but can be made to give quite a high output power that can be useful for higher dynamic range applications.
It does as I said need some resistance in the drain to calm it down, but this is not unusual, all high frequency PHEMTs I've used are unstable when used in low frequency applications.

I think RF-OM was referring to adding source inductance to give some negative feedback. This can improve the input match.

 

[RF-OM]

I finally was able to download your picture (my office machine has a problem with downloading from EDA site). There are at least to problems with source network. First you are right: second source pin must be connected. Secondly you did not include via and parasitic into source network. They may change simulation results a lot. Emitter degeneration is commonly used term for including resistance or inductance into emitter or source network. It provides negative feedback and reduce gain while increase input impedance. The problem here is that reactance in source circuit will rotate impedance on Smith chart and may move it into unstable zone as well as change matching. Also there some parasitic resonances are possible. Often designer wants to keep any reactance into source circuit as small as it possible. The best way to stabilize LNA is to include series resistor as G4BCH told you. Such resistor can be installed into input network too, but it will increase the noise. Usually Circuit Utility in RF simulators can do all design work quickly and without any problem, especially when you use good component models.

Again G4BCH is absolutely right: you did not include any input network, so this simulation do not tell you the true. Via must be included in all locations where they are used in actual board.

One more point. You use radial stub to have ground point in drain network. Are you sure that this stub will good for 1.9GHz? Probably properly chosen bypass caps will do better job for so low frequency. This also can affect stability.

 

[Abhishekabs]

Thanks RF-OM for feedback
The redial stub was not necessary.
Next time I will include only capacitor insted of radial stub.
But I still dont get why you need second source pin when 1 can do your job?
It confuses a lot.

 

[RF-OM]

Second pin allows to decrease parasitic inductance in source-to-ground connection. Two lines in parallel will provide almost two times less inductance.

 

[ramaraju]

hi,i understand your problem.i thinh this transistor model is not available in library. send me transistor model ,biassing circuit length of w,l sothat i will simulate in my system.i know ads well.send me those detatils to my mail id ramaraju.bandari@gmail.com.
i will tryit