I've gotten interested in using the JFET-NPN hybrid Darlington connection shown in Figure 1 below. While a standard Darlington has an emitter follower which drives another emitter follower, this circuit has a JFET source follower which drives an emitter follower.

I think I'd like to try the "J105" device in this circuit, and as luck would have it, I bought 100 of them back in August of 2014. I keep a little snip of the shipping bag in my parts drawer, to remind me where and when I bought these units. The J105s came from DigiKey: Figure 2.

So I chose twelve J105 units at random, and put them on the curve tracer. I plotted IDS (vertical axis) against VGS (horizontal axis) with VDS held constant at 10 volts. The slope of this curve is d(Ids)/d(Vgs) which has a special name in circuit design: transconductance. Often written "gm".

The measured data is shown in Figure 3. There's quite a spread in the data!

Simplistic mathematical models of JFET behavior suggest that IDS = Param1 * (VGS - Param2)**2 where Param1 is a gain factor related to the JFET's channel width and length, and Param2 is the "Pinchoff Voltage". "**2" means raised to the power 2, i.e, squared. The data in Figure 3 suggests that the spread of Pinchoff Voltage for these twelve J105s, is about 700 millivolts.

To see this, draw a horizontal line at IDS= 50 milliamps. It intersects the leftmost J105 curve at VGS= -5.7 volts, and it intersects the rightmost J105 curve at VGS= -5.0 volts. A spread of 0.7 volts (!).

If a (very small) sample of just 12 parts, gives a spread of 700 millivolts, I imagine that a larger sample will give a much bigger spread. I hope to measure many more J105s in the coming days, and find the very leftmost curve, and the very rightmost curve. These represent the extremes of the sample and, I hope, they may approximate the extremes of the population of all J105s ever built and sold.

If by some miracle I can find SPICE model parameters which fit these two extreme transistors, then I can simulate circuits using the "best" J105 ever built and also the "worst" J105 ever built. If my circuit design works acceptably with both extremes of J105 then I can feel confident it will probably work acceptably with average J105s that are not so extreme. That's the hope.

Fairchild's datasheet is also attached.

_

I think I'd like to try the "J105" device in this circuit, and as luck would have it, I bought 100 of them back in August of 2014. I keep a little snip of the shipping bag in my parts drawer, to remind me where and when I bought these units. The J105s came from DigiKey: Figure 2.

So I chose twelve J105 units at random, and put them on the curve tracer. I plotted IDS (vertical axis) against VGS (horizontal axis) with VDS held constant at 10 volts. The slope of this curve is d(Ids)/d(Vgs) which has a special name in circuit design: transconductance. Often written "gm".

The measured data is shown in Figure 3. There's quite a spread in the data!

Simplistic mathematical models of JFET behavior suggest that IDS = Param1 * (VGS - Param2)**2 where Param1 is a gain factor related to the JFET's channel width and length, and Param2 is the "Pinchoff Voltage". "**2" means raised to the power 2, i.e, squared. The data in Figure 3 suggests that the spread of Pinchoff Voltage for these twelve J105s, is about 700 millivolts.

To see this, draw a horizontal line at IDS= 50 milliamps. It intersects the leftmost J105 curve at VGS= -5.7 volts, and it intersects the rightmost J105 curve at VGS= -5.0 volts. A spread of 0.7 volts (!).

If a (very small) sample of just 12 parts, gives a spread of 700 millivolts, I imagine that a larger sample will give a much bigger spread. I hope to measure many more J105s in the coming days, and find the very leftmost curve, and the very rightmost curve. These represent the extremes of the sample and, I hope, they may approximate the extremes of the population of all J105s ever built and sold.

If by some miracle I can find SPICE model parameters which fit these two extreme transistors, then I can simulate circuits using the "best" J105 ever built and also the "worst" J105 ever built. If my circuit design works acceptably with both extremes of J105 then I can feel confident it will probably work acceptably with average J105s that are not so extreme. That's the hope.

Fairchild's datasheet is also attached.

_