Question for Win Hill

Harry's right... 2N7000/7002. Want a bagful?

I've got a little circuit we just did; it outputs 0-60 volts into 50 ohms through a 1:1 isolation transformer, using 2N7002's. Tr is just under 2 ns at 60 volts, under 1 ns at 5.

John

Not yet. Right now all I needed was a number that was common. I'll throw it into PSpice and see how it works out, then I might build one up.

...Jim Thompson

Hey Jim, I can handle any of Win's questions. Depends how much you derate for HV but the 2N7002 in SMD or 2N7000 in through hole are the most popular. Happy Holidays, Harry

and John Larkin answered:

Jim Thompson responded...

Hah! It appears Harry and John can handle my questions just fine. Normally I'd respond by building on their fine answers, but my P4 WinXP home computer with its datasheet collection, is dead, despite considerable efforts to revive it, including an AMD Athlon64 3500+ mobo transplant.

The 2n7000 / 2n7002 parts are my jellybeans for low-power use, and they work well. The 2n7000 became popular after the Supertex VN01, a small MOSFET we featured in AoE. VP01 is its p-channel mate.

There are some larger alternates that may be preferred in production because they have higher gate capacitance and are therefore a little less vulnerable to static damage in handling.

Jim, I should warn you that spice models for these VMOS FETs are woefully inadequate, failing to correctly model the gate spreading resistance, the important sub-threshold region, or even the cascode JFET drain-capacitance scene for that matter. They also lack any reliable thermal model. Modest analytical calculations can often better predict a MOSFET's performance under unusual circumstances.

Plus the P4's longer pipeline was very damaging. I was running a 3GHz P4, which seemed pretty snappy at Spice. However I do expect my Athlon64 to do much better. I selected it in part because of the "no execute" feature implemented in SP2. And was getting tired of the increasingly power-hungry Intel processors. I have high hopes for my 939-pin Athlon64, and I look forward to upgrading to one of the other advanced 939-pin processors once they come down in price.

Jim, what are you working on with your little 50V 0.2W MOSFET?

I have a 6-month-old Athlon64 3400+ (2.2G clock), 1GB Ram.

It is more than twice as fast as my last 1.6G P3.

BTW: P4's are _terrible_ for Spice, since Intel did away with math-specific core... A 1G P3 is faster than a 2G P4 doing Spice.

...Jim Thompson

Will it have the obligatory "audiophile" explanations about how _your_ way of sacrificing a chicken to the audio gods is better than anyone else's, and that's why it sounds better?

Perhaps you should cryogenicaly treat the PC boards, so the crystalline structure of the copper on the traces is better aligned? Oh! and use teflon boards 'cause they're less dissipative of that important high frequency (1GHz) content?

Don't forget the LN2-dipped wall plugs for the listening room --

Or are you just going to make some lame claim that it's better 'cause you actually know what you're doing with the materials at hand?

Well, then surely you'll want viable Spice MOSFET models, because you'll be using the FETs throughout their useful linear range, and the output-stage crossover region is critical.

Sadly most standard Spice library VMOS models simply don't do the subthreshold linear region. For example, see the 10-decade plots on page 123 of our book. A jellybean 2n7000 is rather similar to the VN01 that we show in figure 3.14, and certainly a proper Spice model should be able to make that plot. But I'd be surprised if your standard Spice libraries work properly below say 5 to 20mA, which is not that far below the FET's maximum current. Keep in mind that linear power FET circuitry always operates well below the maximum rated FET switching current, to keep power dissipation junction heating under control.

The bottom line is you'll have to start your FET-amplifier design exercise by designing some decent FET models. Let us know what you come up with.

I see so much trash audio here, that, in my "spare time", I decided I would design an audio amplifier... class AB, but no bias pot required... and, contrary to my thinking BJTs are better, it will be power MOSFET output... the little device is for bias... all done just to prove I can do it better ;-)

...Jim Thompson

Will I do, I'm not Win?

I use the Zetex ZVN4210A, when I need an N-channel MOSFET. The ZVN4210G is similar, but SMT.

Leon

"Sounds" good to me ;-)

I may even use lamp cord to connect the speakers.

...Jim Thompson

Oops, I asked for that one! OK, good question, I'll retire and consider it. But *First* I have to get my Athlon64 running, so I can play with your selection of models, plus a few I have. It's a good and fair question you asked. Thanks, I think!

Win, Here's a bunch of models. Which would you recommend?

Anasoft-1:

..SUBCKT 2N7000/PLP_XN _ssi_pin0_1 _ssi_pin1_2 _ssi_pin2_3 Cgs 2 3 12.3E-12 V_ssi_pin2 _ssi_pin2_3 3 0 V_ssi_pin1 _ssi_pin1_2 2 0 V_ssi_pin0 _ssi_pin0_1 1 0 Cgd1 2 4 27.4E-12 Cgd2 1 4 6E-12 M1 1 2 3 3 MOST1 M2 4 2 1 3 MOST2 D1 3 1 Dbody ..MODEL MOST1 NMOS(Level=3 Kp=20.78u W=9.7m L=2u Rs=20m Vto=2 Rd=1.186) ..MODEL MOST2 NMOS(VTO=-4.73 Kp=20.78u W=9.7m L=2u Rs=20m) ..MODEL Dbody D(Is=125f N=1.023 Rs=1.281 Ikf=18.01 Cjo=46.3p M=.3423

• Vj=.4519 Bv=60 Ibv=10u Tt=161.6n) ..ENDS

Anasoft-2:

..SUBCKT 2N7000_XN _ssi_pin0_3 _ssi_pin1_4 _ssi_pin2_5

• Nodes D G S V_ssi_pin2 _ssi_pin2_5 5 0 V_ssi_pin1 _ssi_pin1_4 4 0 V_ssi_pin0 _ssi_pin0_3 3 0 M1 3 2 5 5 MOD1 RG 4 2 343 RL 3 5 6E6 C1 2 5 23.5P C2 3 2 4.5P D1 5 3 DIODE1
• ..MODEL MOD1 NMOS VTO=2.474 RS=1.68 RD=0.0 IS=1E-15 KP=0.296

Supertex

..MODEL 2N7000 NMOS (LEVEL=3 RS=0.205 NSUB=1.0E15

Philips:

..SUBCKT 2N7000/PLP 1 2 3 Cgs 2 3 12.3E-12 Cgd1 2 4 27.4E-12 Cgd2 1 4 6E-12 M1 1 2 3 3 MOST1 M2 4 2 1 3 MOST2 D1 3 1 Dbody ..MODEL MOST1 NMOS(Level=3 Kp=20.78u W=9.7m L=2u Rs=20m Vto=2 Rd=1.186) ..MODEL MOST2 NMOS(VTO=-4.73 Kp=20.78u W=9.7m L=2u Rs=20m) ..MODEL Dbody D(Is=125f N=1.023 Rs=1.281 Ikf=18.01 Cjo=46.3p M=.3423

• Vj=.4519 Bv=60 Ibv=10u Tt=161.6n) ..ENDS

Ancient MicroSim:

..model M2n7000 NMOS(Level=3 Gamma=0 Delta=0 Eta=0 Theta=0 Kappa=0.2

• Vmax=0 Xj=0 Tox=2u Uo=600 Phi=.6 Kp=1.073u W=.12 L=2u Rs=20m
• Vto=1.73 Rd=.5489 Rds=48MEG Cgso=73.61p Cgdo=6.487p Cbd=74.46p Mj=.5
• Pb=.8 Fc=.5 Rg=546.2 Is=10f N=1 Rb=1m)

Zetex:

..SUBCKT M2N7000/ZTX 3 4 5

• Nodes D G S M1 3 2 5 5 MOD1 RG 4 2 343 RL 3 5 6E6 D1 5 3 DIODE1 ..MODEL MOD1 NMOS VTO=2.474 RS=1.68 RD=0.0 IS=1E-15 KP=0.296

...Jim Thompson

Weird. Last few power amps I designed, I didn't simulate anything.

John

Right, it's more fun to do the various calculations than to run a boring simulation. But next comes the point where one has to either get out the drill and soldering iron, or procrastinate by firing up Spice to test the design. I assume that's Jim's position, because he wrote, "I'll throw it into PSpice and see how it works out, then I might build one up." Well, that's when Spice becomes interesting. If it can be believed, that is. Hmm, does the tool test the design, or does the design test the tool?

Wow Tim... You really have no idea who Jim is, do you?

A photo of Jim, 35 years ago:

;)

Actually the last audio amplifier I designed was doodled on paper, built on a hand-painted PCB, and worked first time powered-up.

That would have been 1977 when I bought my first 280Z.

I bought a good Pioneer tuner and added my own design bridge amps to give a little "boost" to the audio ;-)

...Jim Thompson

I'm of the simulation-tests-the-design camp. "Design" is done by math, art and experience, NOT with a simulator. I'm troubled by the current vintage of designers, who seem to have no hands-on experience whatsoever. Don't universities have lab courses anymore?

Simulators are only for cost-saving. Gets expensive doing breadboards on-chip ;-)

...Jim Thompson

If you can do it better, will you kindly post your results? I've been dicking around (excuse me, heuristically engineering) with FET amplifiers to replace my thirty year old comp-symm bipolar designs and as yet, I haven't been able to make anything better. Equal, yes, but FETS are a bit more pricey than their bipolar equivalents, and unless I can justify going to the cost of new devices, the old 5-bipolar current amplifier fed by an opamp with feedback around the whole mess will continue to be the standard.

My constraint is a 12 volt supply (13.6 when running) but I'm allowed 5% distortion before anybody even bats an eye. Both speaker (8 ohms) and headphone (32 ohms) amplifiers.

Jim