Best J-Fet for RF

[Paul98]

Hello,i have a question, is a 2n5398 or a 2n4416 better to amplify a microvolt signal? I could use one as the first stage and the other for the second (cascaded).

2n5398

2n4416

I first use fet like this because they already exist of the Proteus database. Otherwise I have difficulties with the mmic because I can not load the model (spice I think).
It would be ok also a MMIC but i have trouble to find simulation data for my EDA software thus a i can't try on it. Thanks.

 

[dick_freebird]

At microvolts of signal the noise factor is
likely the most important attribute. Though
this is also impacted by amplifierr stage
design.

I'm not enough of a "RF guy" to say which of
these two FETs is better or if either of them
will give you, in any configuration, the signal
chain performance you want.

NE32S10 has been popular with some of my
customers, in the past. No good idea how that
one stacks up.

Good quality SPICE models for MMICs seems
to be a money making proposition for some
folks, you might have to dig to find any given
vendor's models (SPICE or otherwise). It can't
hurt to wade into their applications pages
and if nothing's there for Downloads, email
their central apps folks (you might have to
be vague about your plans, if you don't smell
line "money, someday". You can always say
you're under NDA as far as marketing plans
of your customer, make up some shizzle
about the end application and annual units
-"if the part can be evaluated and turns out to
be the right answer". The apps guys are there
to sell parts via design-in. The sales guys are
there to take the order. So let 'em try to sell
you, and you tell them what might do that.

 

[Paul98]

Hi, yes I tried to send an email to some suppliers in the hope that someone will answer me. From what I can understand, mmics are somewhat obsolete products. Especially when they already have several years of life. About OpAmp actually they should guarantee a better quality point in terms of noise (obviously). Furthermore, there is the possibility of connecting them in a differential way to further lower the noise if needed. I took a look at some models but finding one that obviates a good degree of sensitivity and that goes from DC to 500 / 1000Mhz becomes a problem especially with SPICE available. Unfortunately like NE32S10 I can't find anything. Are you sure about the @dick_freebird ? My need is not to create spice models and resell it but to have the possibility to verify the component in simulation if it can do for me. In the photo there is a very very simple example which, however, would allow you to see the behavior of the component before buying it.

 

[FvM]

Why not use an arbitrary wideband OP with good performance for a feasibility study, e.g. AD8099 which provides 550 MHz bandwidth at G=10 and < 1nV/Hz^0.5 noise?

You'll get LNA with lower noise, but they are not not designed for a frequency range down to DC.

 

[Paul98]

Hello @FvM , thanks for suggestion. I will take a look!. Yes all the LNA looked minimum start at some kHz!.

 

[FvM]

Another important point not yet mentioned is source impedance. LNA are generally designed for matching to 50 ohm source impedance. OP have high impedance with some amount of capacitance.

 

[Paul98]

Well, i created and simulated the circuit with OpAmp that you reported to me and it works at least on the simulator. I could not find the reference schema with an inverting input. Do you know if it is possible to use this OpAmp on inverting input configuration? I have already tried to manually change the signal input but the scope remains flat at the output. Maybe I'm wrong in something anyway.

 

[KlausST]

Maybe I'm wrong in something anyway.

Maybe it's a good idea to show your schematic...

Klaus

 

[FvM]

Consider that feasible resistance range for feedback network is very limited at multiple 100 MHz. Respectively input impedance in inverting configuration is very low. With n.i. config input impedance is independent of feedback.

 

[Paul98]

Guy m i've solved! and work on simulation. Input 100uV , Gain 10 and Output 1mv both inverted and inverted input with the same source gen. Now the next stage need to amp from 1mv Input. Is possible to use the same OpAmp for the second stage (cascadable?) or need another OpAmp?

 

[Paul98]

Hi everyone, I drew a circuit with AD8099 as suggested by @FvM and in simulation it seems to be fine. With 4 OpAmp in series (cascade) With a signal of 100uV I can have an output voltage of about 1Volt. The limit of the simulator, however, is the generator which has problems managing, I already signal at 100Mhz and I don't know in a real scenario if the circuit actually works or not. Further I added a stage with in LT1226 because it seemed to fit me well. The problem is that Proteus does not display the wave on the scope or gives me an error "[SPICE] TRAN: Timestep too small; timestep = 1.25e-019: trouble with instance d: u2: on1." . I would like to understand if it is a problem in the schematic or a limitation of the simulator.

Thanks