Are stock transistor models in LTspice accurate at RF?

[joeh100]

Just checking, I'm doing some experimenting of small signal RF design in LTspice.

 

[Nigel Goodwin]

No idea, I don't use simulators - but RF designs are heavily dependent on the layout, and a simulator can't simulate the layout - so I wouldn't have thought it would be very useful?.

 

[joeh100]

No idea, I don't use simulators - but RF designs are heavily dependent on the layout, and a simulator can't simulate the layout - so I wouldn't have thought it would be very useful?.

As an electronic technician with very little knowledge in RF theory and design, it's extremely useful for learning basic concepts. I can throw a circuit together and make perfect measurements in a few minutes, without owning thousands of dollars worth of RF test gear.

 

[alec_t]

I can throw a circuit together and make perfect measurements

Perfect measurements of an imperfect circuit model. You should regard the models as approximations of reality. Don't forget to take account of parasitic resistance, capacitance and inductance in your sims. These become more significant in RF sims than at low frequencies.

 

[joeh100]

Thanks for the concern but this is getting a little off topic guys. I have a descent understanding of "the components not on the schematic" as they say.

I need to understand the math used in RF design and verify my math through simulation.

 

[Nigel Goodwin]

Thanks for the concern but this is getting a little off topic guys. I have a descent understanding of the components not on the schematic as they say.

I need to understand the math used in RF design and verify my math through simulation.

The math is no different to lower frequencies - it's reality where things change (and can change massively).

 

[joeh100]

No idea, I don't use simulators - but RF designs are heavily dependent on the layout, and a simulator can't simulate the layout - so I wouldn't have thought it would be very useful?.

There are software simulation tools specific for RF design, such as Genesys (for professionals and probably not cheap). I just want to know if there are any limitations with LTspice at RF (as far as transistor modelling).

The math is no different to lower frequencies - it's reality where things change (and can change massively).

Yes I get that already.

Transistor models and the design process do vary with frequency though. Can I design an amplifier in LTspice at 100MHz with s-parameters?

 

[bloki]

Spice models neglect parasitics of real transistors. For low frequency say few tens of MHz they are not needed. Transistors for GHz frequencies may have in datasheet complete circuit with spice model and all parasitic inductances.
Usually UHF transistors has S parameters table for frequency range where they can be used which are measured and enable better simulation. Don't know if LTspice has this possibility.

 

[ronsimpson]

I have used LTSpice many times at 100mhz and got pretty good results compared with the real PCB.
I think much of the RF and (hobby PCBs) problems is that some people don't understand that a long trace has inductance and capacitance. (also resistance and maybe transmission line effect) So I am blaming the problem of SPICE verses PCB on the person that designed the PCB and not SPICE.

---edited---- LPSpice changed to LTSPICE. Problem typing.

 

[joeh100]

I have used LTSpice many times at 100mhz and got pretty good results compared with the real PCB.

Thanks, that answers my question.

 

[crutschow]

At high frequencies (but probably not until the GHz region) even the internal package bond wires can have capacitance and inductance that can have an effect on circuit performance. As long as you model those, along with any PCB impedances, then LTspice should be able to give reasonably accurate simulation results, even at quite high frequencies.