SPICE models for big fat RF transistors

[Joerg]

Hello Newsgroup,

Does anyone know where to get models for the really big fat base station
transistors? Right now it doesn't matter which mfg. NXP doesn't seem to
have one for the BLF878 anymore, somehow gone, I'll have to ask where it
went. Anyhow, they only provide scant information and only at particular
bias points, like these:

http://www.nxp.com/models/spicespar/BLF548.html

I need something real, something I can plug into LTSpice.

 

[Jim Thompson]

Hello Newsgroup,

Does anyone know where to get models for the really big fat base station
transistors? Right now it doesn't matter which mfg. NXP doesn't seem to
have one for the BLF878 anymore, somehow gone, I'll have to ask where it
went. Anyhow, they only provide scant information and only at particular
bias points, like these:

http://www.nxp.com/models/spicespar/BLF548.html

I need something real, something I can plug into LTSpice.

Why? Did something go *Phut* again ?

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

I love to cook with wine Sometimes I even put it in the food

 

[Joerg]

Why? Did something go *Phut* again ?

No, but I want to make sure it's all kosher because a *phut* situation
could become rather ugly if you combine a dozen of those. Molten solder
splattering about, blaring sirens ...

 

[Joerg]

Hello Newsgroup,

Does anyone know where to get models for the really big fat base station
transistors? Right now it doesn't matter which mfg. NXP doesn't seem to
have one for the BLF878 anymore, somehow gone, I'll have to ask where it
went. Anyhow, they only provide scant information and only at particular
bias points, like these:

http://www.nxp.com/models/spicespar/BLF548.html

I need something real, something I can plug into LTSpice.

So, guys, got an answer from NXP: They do not create SPICE models for
their RF products. They only create ADS models.

I assume ADS only works with EEsof which probably still costs the
equivalent of a nice car and I bet models are not convertible. Great.
Cuts off much of the secondary users for their RF transistors. What were
they thinking?

Oh, and the support email says you cannot reply to the message. Oh man,
now that they've got their web site in order it starts to crumble
somewhere else.

 

[Jim Thompson]

So, guys, got an answer from NXP: They do not create SPICE models for
their RF products. They only create ADS models.

I assume ADS only works with EEsof which probably still costs the
equivalent of a nice car and I bet models are not convertible. Great.
Cuts off much of the secondary users for their RF transistors. What were
they thinking?

Oh, and the support email says you cannot reply to the message. Oh man,
now that they've got their web site in order it starts to crumble
somewhere else.

Sounds like it's a Curtice model... works in PSpice ;-)

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

I love to cook with wine Sometimes I even put it in the food

 

[Joerg]

Sounds like it's a Curtice model... works in PSpice ;-)

But LTSpice wants LGATE, RGATE, CG, CRSS, CISS, LS, CS ... and so on :-(

I plan to use it in a quite unorthodox application (like usual ...).

 

[Jim Thompson]

But LTSpice wants LGATE, RGATE, CG, CRSS, CISS, LS, CS ... and so on :-(

I plan to use it in a quite unorthodox application (like usual ...).

I would guess someone here with an IEEE membership can locate a
translator.

(The Curtice model is built into PSpice... primary use: GaAs FET's)

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

I love to cook with wine Sometimes I even put it in the food

 

[Joerg]

I would guess someone here with an IEEE membership can locate a
translator.

I am a member but they nickel and dime you for access to just about
everything not in your subscribed societies, so probably no access to
what's needed. But thanks for the info, gives me pointers how to search
on the web.

(The Curtice model is built into PSpice... primary use: GaAs FET's)

Somehow it doesn't look like the Curtice model on page 8-103 of this
Agilent doc though:

http://www.ece.uci.edu/eceware/ads_docs/pdf/netlist.pdf

 

[Joerg]

think I have seen somewhere (yahoo forum?) that someone got a model
like
that into ltspice but it wouldn't work for transient analysis.

Haven't found anything there but will keep looking elsewhere. However,
transient analysis is exactly what I am after, I am not using it as a
regular RF amp.

what kind of transistor specs are you looking for?, ...

As big a honking RF power transistor as possible, good up to a GHz, or
at least lower UHF. I am prepared to live with whatever specs they've
got, pretty slim pickens up there anyhow because the market for this
stuff is small. But I'd need a somewhat proper SPICE model. This kind of
circuitry has to be simulated pretty well, trying it all on the bench
can lead to lots of molten solder splattering around, a loud bang,
smoke, ferrite pieces flying all over the place and so on.

... I've seen models for other
big rf transistors that work in lspice somewhere.

Yes, PolyFet has some but this looks like high-price specialty parts. I
was hoping for a more mainstream part from a bis semi mfg.

 

[Joerg]

Joerg, I don't understand what the problem is.

The parameters for Advanced_Curtice2 model are available - all you
have to do is enter them as a text field.

Ok, I'll try again but NXP told me you can't, only works for ADS.

But, if you need some BIG models for RF transistors,
you can try:

freescale website

Most of the GaN sites (Nirronix, Cree) have models
but they are in ADS format.

But AFAIK nothing above 50W. PolyFet has bigger ones though and much
better model support than NXP.

 

[Joerg]

Joerg, OK now I know a little bit more of what you are
looking for.

What frequency range are you looking for?

Devices that can do up to lower UHF, 500MHz or so, with decent
efficiency would be fine. Tons of power ;-)

Freescale has some LDMOS chips in the
500 - 1500 MHz range.

Cree and Nitronex have GaN chips which
use the MCM SPICE models

Yes, but even the Nitronex flagship NPT1007 can only do around 100W. I
am looking for something that can stomach a few hundred watts. PolyFet
has one, not quite like the big BLF but close and at least they give out
SPICE parameters.

 

[Joerg]

I looked into that some years ago when I still was interested in
fat RF transistors and the consensus was that Spice doesn't cut it
for this application. There were models from Moto/Freescale/On/MaCom
and Philips/NXP, but only for ADS, Microwave Office and Harbec/Genesys/Eagleware
(that is now Agilent, too). These simulators had been specially
enhanced to digest these models. IIRC it was a real physics problem,
maybe local themal effects.

Actually about an hour ago I simulated the first such power transistor,
a 300W VHF device from PolyFet because they furnish a fairly elaborate
SPICE model. It is a subcircuit with three different semiconductor
models and numerous discretes in there. So far it shows all the nasties
you see with VHF broadband stuff in the real world, meaning it can't be
too far from the real world. I have designed lots of power pulsers in my
life, all under 15nsec though and it seems that going to 5nsec make the
unwanted stuff grow 10x :-(

Sorry, that means $$$.

Yep, and that's not (yet) in the budget for this project. Also, those
simulators are geared towards RF stages that run continuous duty. Mine
will be pulsed pretty brutally, so hardcore time domain sims are needed.

Now I am trying to figure out a VSWR protection that acts within a few
nanoseconds ...

 

[Joerg]

IIRC, these fat FETs had a memory effect, i.e. they behaved differently
wrt what happened some us/ms before.

There are some local heating issues, definitely at the power level I am
dealing with. But it could be worse: The old BJTs had sort of a ROM
effect. Treat them too harshly and one of more of the emitter paths
would go phsst ... *phut* ... and then your transistor would deliver
only half the power from here on. Forever, it would not fix itself
anymore. I've used such a partially damaged transistor that was down to
60% of nominal in a 2m transceiver for over a decade, that used one was
all my student's budget would allow (dipping into the beer budget was
not an option ...).

 

[Joerg]

Joerg,

You might to reconsider the decision to build a switch to measure
VSWR.

You might want to look at gradually decreasing power as the VSWR goes
up.
If you switch everything at once, you will have a very large arc as
the power has to
go somewhere ( usually a place that you don't want it to)
There are companies out there that make VSWR protection circuits but
they
have be integrated into the PA in some fashion.

I can't go gradual because the switch action is what will actually
happen in this application. Near-shorts, full and sudden interruption,
and so on. It needs to survive that. The usual SWR protection circuits
rely on bridges, multipliers and whatnot, typically after rectifying the
phase-coupler outputs. By the time the rectified signals show up it'll
already be ... phut ... *KABLAM*