LU1014D Modulated Cascode

Jump to Latest
#101
woofertester said:
The measurement of the LU1014D gain parameter requires a 4-wire Kelvin measurement. If you rely on the power supply to measure the voltage, you are placing the cabling resistance in the drain and source. This resistance knocks down the gain.
Click to expand...
The resistance of the test leads is maybe 10 milli-Ohm. At a 3 A current, the error in the Drain-Source voltage would be 0.01 x 3 = 0.03 V. I'd say that's negligible. If you don't agree, it would be trivial to correct the numbers by calculating the voltage drop across the test leads.

The 0.03 V offset may be relevant for the Gate-Drain voltage, if the second power supply for the Gate voltage is referenced to the Source with a wire to the output terminal of the first power supply. This can easily be avoided by wiring the second power supply directly to the Source pin of the DUT, so the 0.03 V drop will not affect the Gate voltage.
 
#105
What amplifier circuit topologies benefit from triode behavior vs. pentode behavior? Here is my limited understanding.
  • Topologies that benefit from triode behavior:
    • Single Ended Triode (SET) common source with little or no feedback.
    • Triode common source mu-follower with little or no feedback.
  • Topologies where triode behavior has little or no benefit:
    • Source Followers
    • Common source with high feedback
I invite other knowledge or opinions. Hello Papa.
 
Last edited:
#108
I revised my sims from post #31 with a 220uF cap, the 10uF was too low and rolling off the low end early.

Yep, it'll be SET driving efficient loudspeakers when I try this in vivo... I think I'll like the sound even better at half the distortion. :cheerful:

I inverted the scope plots to show speaker polarity inversion, maintaining -H2 for the triode mod.

Triode Mod LuAmp with 220uF cap.png
 

Attachments

#112
Holy Grail turns into Fools Gold

I thought I had something really interesting when I first simulated the Vds range expanding LU1014D modulated cascode and discovered that it preserved the triode behavior of the LU. The Holy Grail?

What I didn't consider at that time were the resulting triode parameters and their effect when used in an actual power amplifier.

Recently I have run simulations of the LU1014D modulated cascode in a SIT-1 style amplifier, shown in folowing posts. As discussed in posts #105-#107, triode behavior most benefits common source amplifiers such as the FirstWatt SIT-1 and SIT-2. The simulations agree precisely with the circuit equations shown in post #63.

I have also performed bench tests of the same SIT-1 style circuit, obtaining triode parameter values that differ from the simulations in ways that are not unexpected.

The First Watt SIT-1 Owners Manual https://firstwatt.com/pdf/prod_sit1_man.pdf "Nominal Specifications" show a gain of 18dB (7.94) with an 8 Ohm load and an output impedance (Zout) of 4 Ohms. From these values, the Voltage gain with no load (Vgain[inf]) is estimated to be 11.9. Vgain[inf]=(Vgain_1*(Zout+Rload_1))/Rload_1. From Vgain[inf] and Zout the triode parameters are:

μ[SIT1]=11.9​
Rd[SIT1]=4R​
gm[SIT1]=2.98S​

The triode parameters derived from the simulated LU10146 modulated cascode amplifier are:

μ[casc] = 73.4​
Rd[casc] = 5.6R​
gm[casc] = μ[casc]/Rd[casc] = 13.1S​

The triode parameters derived from the bench test LU10146 modulated cascode amplifier are:

μ[casc] = 80.3​
Rd[casc] = 8.31R​
gm[casc] = μ[casc]/Rd[casc] = 9.67S​
​
Bottom Line:

Main problem with this modulated cascode is that the voltage gain μ[casc] is way to high, being magnified by a factor A = 1/k = 9.15 in these examples. The amplifier works, but the distortion and noise level is considerably higher than that of the FirstWatt SIT-1.

The only ways I know of for reducing the amplifier gain are global negative feedback and source degeneration, both of which defeat the purpose for having a triode. A FET with pentode behavior will work quite well, as with the First Watt F3 and F8.
 
  • Like
Reactions: wdecho
#113
I have performed both simulations and bench tests of the LU1014D modulated cascode in a SIT-1 style amplifier, shown in the first image below, where a constant current source (CCS) is used in place of a chain of resistors, primarily because I didn't have suitable resistors for the bench tests.
It is easy to measure the voltage gain of an amplifier driving different output load resistances. From those gain measurements one can calculate the output impedance (Zout), and the voltage gain without a load (Vgain[inf]) as follows:

Zout=(Rload_1*Rload_2*(Vgain_1-Vgain_2))/(Rload_1*Vgain_2-Rload_2*Vgain_1)

Vgain[inf]=((Rload_1-Rload_2)*Vgain_1*Vgain_2)/(Rload_1*Vgain_2-Rload_2*Vgain_1)

Measurements:

Simulation:
30.6 gain with 4R load​
43.2 gain with 8R load​
Zout = 5.6R​
Vgain[inf] = 73.4​

μ[casc] = 73.4​
Rd[casc] = 5.6R​
gm[casc] = μ[casc]/Rd[casc] = 13.1S​

Bench Tests:
26.1 gain with 4R load​
39.4 gain with 8R load​
Zout = 8.31R​
Vgain[inf] = 80.3​

μ[casc] = 80.3​
Rd[casc] = 8.31R​
gm[casc] = μ[casc]/Rd[casc] = 9.67S​
 

Attachments

  • LU-casc-CCS-test-jig-2-XXX-edit.asc.jpg
    LU-casc-CCS-test-jig-2-XXX-edit.asc.jpg
    95.5 KB · Views: 177
  • DSC_0772-crop-w1200.JPG
    DSC_0772-crop-w1200.JPG
    205.2 KB · Views: 163
  • Like
Reactions: wdecho
#114
lhquam said:
Holy Grail turns into Fools Gold

I thought I had something really interesting when I first simulated the Vds range expanding LU1014D modulated cascode and discovered that it preserved the triode behavior of the LU. The Holy Grail?

What I didn't consider at that time were the resulting triode parameters and their effect when used in an actual power amplifier.


...
​
​
Bottom Line:

Main problem with this modulated cascode is that the voltage gain μ[casc] is way to high, being magnified by a factor A = 1/k = 9.15 in these examples. The amplifier works, but the distortion and noise level is considerably higher than that of the FirstWatt SIT-1.

The only ways I know of for reducing the amplifier gain are global negative feedback and source degeneration, both of which defeat the purpose for having a triode. A FET with pentode behavior will work quite well, as with the First Watt F3 and F8.
Click to expand...
Isn't this because lack of source degen~ necessitates global neg fb, whereas with s-degen the triode behaviour disappears...
It should be self-controlled without s-degen or gnfb.
Here I posted a recent simulation of my alternative. There's only one other active element involved.
I'll make another one with a J-fet as input to match current sit's.
Will that be a N-ch or a P-ch? With a high voltage p-mos as protecting top cascode, the P-ch J version can become the complementary of a vacuum triode.
 
You must log in or register to reply here.
Top Bottom