SPICE Amplifier Simulation

[G-Daddy]

It's too bad this sort of thing isn't more reliable. There's a neat application of simulation that's starting to gain some attention, where you apply evolutionary mechanisms to circuit design. It's a pretty simlpe idea, but it's hell in the details.

Basically, you generate a 'population' of circuits, each composed of randomly connected electronic components. Then you resolve all these random circuits into spice netlists (or some other sort of sim method) and run the sim for each specimen. Once all the population members are simulated, you rank their 'fitness' against some arbitrary scale (for an amp, you might combine dc gain, linearity, power consumption, etc to get some weighted scale). At the end, you just cull out the promising individuals, 'mate' and mutate them through any of an assortment of mathematical methods, and repeat with the new population.

Taken to infinity, this process gets you a circuit design that's optimized for the 'healthiness' test you used. Taken more modestly (say 1000 generations or so), you can still get a pretty good result. Several research groups around the world have played with this, and gotten some very interesting data.

But the important thing to note is that if the simulations aren't an accurate description of what really happens in the real world, then the design evolved as 'optimum' may be faulty and undesirable when actually built. If circuit analysis could be developed to the point where a simulation accurately described the real world phenomena, then perhaps all us DIY experimentalists could start playing with home-evolved circuit schematics, instead of adhering to more traditional designs. Its something to think about. If anyone out there has any knowledge of this field, please by all means, speak up.

 

[ftorres]

Mundi,

Here are some of your requested models. I didn't test'em, but hop they work

*SRC=1N4004;DN4004;Diodes;Rectifier <=5A;400V 1A
.MODEL DN4004M D (IS=5.86E-06 N=1.70 BV=5.33E+02 IBV=5.00E-07
+ RS=4.22E-02 CJO=5.21E-11 VJ=.34 M=.38 TT=5.04E-06)
* Motorola 400 Volt 1.00 Amp 3.50 us Si Rectifier Diode 07-01-1990

.MODEL DN4148M D (RS=.8 CJO=4PF IS=7E-09 N=2 VJ=.6V
+ TT=6E-09 M=.45 BV=100V)

*1N5250B MCE 6/2/96
*Ref: National Discrete Products Databook, 1996
*20V 500mW Si pkg: DIODE0.4 1,2
.SUBCKT 1N5250B 1 2
* TERMINALS: A K
D1 1 2 DF
DZ 3 1 DR
VZ 2 3 19.45
.MODEL DF D (IS=2.51N RS=84M N=1.7 CJO=38.8P VJ=1 M=0.33 TT=50.1N)
.MODEL DR D (IS=61.8U RS=7.5 N=4.2)
.ENDS

.MODEL MJ15001M NPN (
+IS=1.23312E-13 BF=115.914 NF=0.85 VAF=10
+IKF=1.263 ISE=6.43082E-13 NE=1.59704 BR=4.58573
+NR=0.930118 VAR=99.9721 IKR=1.20463 ISC=6.28301E-13
+NC=2.79011 RB=2.81251 IRB=0.1 RBM=0.1
+RE=0.0218983 RC=0.109491 XTB=0.1 XTI=1
+EG=1.206 CJE=1.33292E-09 VJE=0.40062 MJE=0.385557
+TF=9.99606E-09 XTF=179.464 VTF=1.74347 ITF=0.001
+CJC=5E-10 VJC=0.95 MJC=0.85 XCJC=0.797142
+FC=0.10001 CJS=0 VJS=0.75 MJS=0.5
+TR=1E-07 PTF=0 KF=0 AF=1)

.MODEL MJ15002M PNP (
+IS=5.9955E-12 BF=1255.86 NF=0.989791 VAF=10
+IKF=0.142606 ISE=1E-08 NE=2.01784 BR=9.03224
+NR=1.5 VAR=100 IKR=0.00777049 ISC=1E-08
+NC=1.68621 RB=3.45026 IRB=0.1 RBM=0.333146
+RE=0.012422 RC=0.0621099 XTB=0.1 XTI=1
+EG=1.05 CJE=7.68242E-10 VJE=0.4 MJE=0.389835
+TF=3.59815E-09 XTF=1000 VTF=1.16654 ITF=0.00860153
+CJC=5.08874E-10 VJC=0.95 MJC=0.415655 XCJC=0.1
+FC=0.467765 CJS=0 VJS=0.75 MJS=0.5
+TR=1E-07 PTF=0 KF=0 AF=1)

*MJE15030
*Si 50W 150V 8A 70MHz pkg:TO-220 2,1,3
.SUBCKT MJE15030 1 2 3
Q1 1 2 3 QPWR .67
Q2 1 4 3 QPWR .33
RBS 2 4 12.5
.MODEL QPWR NPN (IS=17.7P NF=1 BF=195 VAF=220 IKF=3 ISE=1.54N NE=2
+ BR=4 NR=1 VAR=20 IKR=4.5 RE=85.2M RB=0.341 RC=34.1M XTB=1.5
+ CJE=2.68N VJE=0.6 MJE=0.3 CJC=171P VJC=0.22 MJC=0.2 TF=2.27N TR=5.1U)
.ENDS MJE15030

*MJE15031
*Si 50W 150V 8A 86MHz pkg:TO-220 2,1,3
.SUBCKT MJE15031 1 2 3
Q1 1 2 3 QPWR .67
Q2 1 4 3 QPWR .33
RBS 2 4 12.5
.MODEL QPWR PNP (IS=22P NF=1 BF=195 VAF=220 IKF=3.5 ISE=1.85N NE=2
+ BR=4 NR=1 VAR=20 IKR=5.25 RE=97.7M RB=0.391 RC=39.1M XTB=1.5
+ CJE=1.34N VJE=0.6 MJE=0.3 CJC=320P VJC=0.22 MJC=0.2 TF=1.85N TR=2.38U)
.ENDS MJE15031

*MPS8098
*Ref: Motorola Small-Signal Device Databook, Q4/94
*Si 625mW 60V 500mA 150MHz Amp pkg:TO-92B 1,2,3
.MODEL MPS8098 NPN (IS=50.8F NF=1 BF=260 VAF=139 IKF=0.3 ISE=25.2P NE=2
+ BR=4 NR=1 VAR=24 IKR=0.45 RE=0.103 RB=0.412 RC=41.2M XTB=1.5
+ CJE=22.1P VJE=1.1 MJE=0.5 CJC=8.05P VJC=0.3 MJC=0.3 TF=1.06N TR=737N)

*MPS8099
*Si 625mW 80V 500mA 225MHz pkg:TO-92B 1,2,3
.MODEL MPS8099 NPN (IS=9.73F NF=1 BF=1.33K VAF=161 IKF=.16 ISE=10.9P NE=2
+ BR=4 NR=1 VAR=20 IKR=.24 RE=.257 RB=1.03 RC=.103 XTB=1.5
+ CJE=22.1P VJE=1.1 MJE=.5 CJC=7.57P VJC=.3 MJC=.3 TF=623P TR=1.53U)

*MPS8598
*Ref: Motorola Small-Signal Device databook, Q4/94
*Si 625mW 60V 500mA 180MHz Amp pkg:TO-92B 1,2,3
.MODEL MPS8598 PNP (IS=50.8F NF=1 BF=260 VAF=139 IKF=0.3 ISE=25.2P NE=2
+ BR=4 NR=1 VAR=24 IKR=0.45 RE=0.103 RB=0.412 RC=41.2M XTB=1.5
+ CJE=22.1P VJE=1.1 MJE=0.5 CJC=7.81P VJC=0.3 MJC=0.3 TF=883P TR=614N)

*MPS8599
*Si 625mW 80V 500mA 240MHz pkg:TO-92B 1,2,3
.MODEL MPS8599 PNP(IS=1.82E-14 BF=208 VAF=161 IKF=0.155 ISE=1.05E-11
+ NE=2 BR=4 VAR=20 IKR=0.232 RB=1.93 RE=0.482 RC=0.193 CJE=2.48E-11
+ VJE=1.1 MJE=0.5 TF=6.62E-10 CJC=9.47E-12 VJC=0.3 MJC=0.3 TR=8.5E-7
+ XTB=1.5 )

Hope this helps

 

[Mundi]

Thanks Francois!

I now have models or substitute models for almost all the amplifier stages . The remaining 'hole' is the 2nd gain stage:
2N3439 or 2N3440
2N5416 or 2N5416

Regarding accuracy, the amplifier does not use MOSFETS, so that thorn issue does not arise. The accuracy of the components that are used is, of course, an issue. Since I do not have the resources (time and equipment) to test the models, I can only rely upon the time and efforts of others and hope for the best. Having a simulation tool is, I think, better than not, and allows what-if to be done very easily.

Thanks,