measuring Idss and matching

[metebalci]

I have a number of JFETs (J11[1-3]) and I want to match them. So I was measuring Idss (=Id with Vds=12V, Vgs=0) without any other components (setting Vds with bench supply, reading Id with bench DMM, Vds=12V because that is what I was planning to use them with). Naturally the component temperature increases when voltage is applied and Id decreases, as the speed of change slows down, around a range it stabilizes (I was using this value for matching). The issue is particularly for high Idss components (J111 is ~80mA) the temperature is very high (>100C), I actually thought it is going to break (and maybe it will if I let it like that). This made me question at what condition exactly I should measure Idss. Is there a standard procedure for this ? Which Vds should I use ? Does the (possibly different) temperature of the components have no effect ?

[David Hess]

The temperature has an effect depending on exactly what the Vgs or drain current is.  Measurements at high drain currents should be pulsed to avoid temperature rise.

An alternative is to measure Vgs at constant drain current.  You might also select a drain current where the Vgs temperature coefficient should be zero.

[Kleinstein]

One should do the matching more at conditions closer to the actual use.  So this is usually with a lower current.
A simple circuit is using a resistor at source and gate connected to ground. So a fixed drain to gate voltage and than a resistor to select about the relevant current.

For the J111 to J113 the current with low temperature sensitivity is at about 1 - 1.5 mA.  The gate to source voltage for a current in this range would be a reasonable parameter for sorting / matching.

[metebalci]

One should do the matching more at conditions closer to the actual use.  So this is usually with a lower current.
A simple circuit is using a resistor at source and gate connected to ground. So a fixed drain to gate voltage and than a resistor to select about the relevant current.

For the J111 to J113 the current with low temperature sensitivity is at about 1 - 1.5 mA.  The gate to source voltage for a current in this range would be a reasonable parameter for sorting / matching.

I guess it is correct to assume that if components are sorted by Id at a particular Vgs value, the sorted list is always the same for all Vgs<0 ? If so, then it should not matter what value of Vgs<0 is used for matching (so there is no need for re-matching if operating conditions are different) ? and then the best is like you said to find Vgs for minimum temperature sensitivity, match and they will be matched for any operating condition.

[Kleinstein]

To a large part the scattering in JFETs is 1 scattering parameter. So mathching by Ids or R_on or Vgs_off or gm at a fixed voltage would give the same sorting of parts. However there are also small secondary variations.  So parts that are similar in one test may have a slightly different matching at a different test condition. The secondary effects can be small for parts from the same batch, but can be larger for parts from different sources.

The minimum temperature effect is usually at about Vgs some 0.8 V above the threshold or about a fixed current for the type of JFET.
So to get a low temperature effect a fixed (selected for the FET type / family) current would be more suitable. The matching parameter would than be the gate voltage needed to reach that current.
The current depends on the "size" of the JFETs: some 1.2 mA for the J111-113  and more like some  30 µA for 2N4117-9.

Good matching is usually only needed for low frequency amplifier use and this usually means a low threshold (e.g. 1 V range for a good performance) . So Idss is not that temperature sensitive in this range.
Ideally one would also use an amplifier in a current range where the thermal effect is small.