Update: Matching JFETs – Revisited here.
Update: Kits available in the shop here.
Update: Matched sets of JFETs available for sale here.
We’re talking here about N-channel depletion-mode junction field-effect transistors.
- “Field effect” because the current flowing between source and drain is controlled by the electric field strength at the gate. Impedance at the gate approaches infinity so for practical purposes we assume that no current flows into or out of the gate – and therefore drain-source current is controlled by an electrostatic field rather than a current.
- “Depletion mode” because if there is no electrostatic field at the gate maximum current flows between drain and source.
- “N-channel” because the gate has to be negative with respect to the source to turn the drain-source channel off.
Quite interesting facts about JFETS:
1. Current can flow in both directions in the drain-source channel unlike a bipolar junction transistor which only conducts in one direction.
2. A lot of small-signal JFETS (of the types used for stompboxes and the like) are symmetrical. This means that the drain and source are interchangeable. This property is of practical use when designing a printed circuit board.
3. When the gate-source is at zero volts a current can flow between drain and source. In this state the JFET is acting as a constant-current source (and therefore the current flowing between drain and source is independent of drain-source voltage). This current is termed Idss and is one of the parameters of interest when we match JFETs.
4. To turn the JFET off we make the gate negative with respect to the source. At some voltage – called Vgs(off), the “pinch-off voltage” – current in the drain-source falls to zero and the JFET is fully off.
5. In between fully on and fully off there is a linear region where we can use the JFET as a transconductance amplifier. Transconductance? Well in a BJT (bipolar junction transistor) a small current flowing between base and emitter controls a larger current flowing between collector and emitter. This makes a BJT essentially a current amplifier. In the case of the JFET the voltage at the gate-source controls the current flowing between drain and source – and a voltage-controlled current-amplifier is called a transconductance amplifier.
6. The nature of the manufacturing process for JFETs means that two JFETs with the same part number and from the same batch may vary wildly in important parameters. Consequently JFETs don’t get used as amplifiers much in commercial designs if the manufacturer is not prepared to pay extra for tightly-matched (i.e. specially selected) parts. Even when used as a switch the manufacturer may need to select devices with a Vgs(off) within a suitable range for the design.
I took a batch of twenty Fairchild 2N5458 JFETS and tested each sample for Vgs(10kOhm), Vgs(10MOhm) and Idss.
Firstly, R.G.Keen’s improved JFET matcher design was used to find the value of Vgs where Rds (resistance between source and drain) is 10k. Here is a partial schematic of R.G.Keen’s design:
R1 and R2 form a potential divider that fixes the op-amp’s non-inverting input at 1/2 Vcc. The JFET and R3 also form a (variable) potential divider. The circuit will balance when the op-amp’s differential input voltage is zero. So if the voltage at pin 2 is the same as pin 3, the JFET/R3 potential divider must be equivalent to the R1/R2 potential divider. Hence the output of the op-amp biases the JFET so that its Rds is 10k Ohm.
This test is useful for matching JFETs in phaser designs where a bunch of JFETs all get the same bias signal and need to be the same drain-source resistance for a given bias voltage.
Secondly, a digital multimeter was connected between gate and source to measure Vgs when the resistance between source and gate is equal to the input impedance of the voltmeter.
Here we are measuring the gate-source voltage when there is a very large resistance in the gate-source circuit. Digital multimeters (DMM) typically have an input impedance of 10M Ohm so the DMM in the above circuit (left) is equivalent to putting a 10M Ohm resistor in the source leg and then measuring the voltage across the 10M Ohm resistor with a perfect voltmeter (i.e. a voltmeter with infinitely high input impedance). This technique gives a value for Vgs(10M) which is practically the same as Vgs(off).
This test is useful for making sure that there is enough voltage to fully switch off a JFET when it is used as a switch (e.g. in an Ibanez-style effects pedal).
Thirdly the gate and source were connected together to set Vgs at zero volts and Idss was measured.
Idss is the current which flows in the JFET when there is no negative voltage on the gate with respect to the source. If we are planning to use the JFET as a constant current source, it is important to know what the value of that current is.
The results of this testing on our twenty 2N5458 JFET samples revealed that:
- Vgs(10K) varied between -0.99V and -1.75V
- Vgs(10M) varied between -1.61V and -2.38V
- Idss varied between 3.24 and 5.99 mA
According to the Fairchild datasheet for the 2N5458, Vgs(off) ranges between -1V and -7V and Idss ranges between 2mA and 9mA, so – whilst my batch of twenty is fairly well clustered in the middle of the range, the lowest Vgs(off) is still 50% lower than the highest and the highest Idss is 84% higher than the lowest. These are significant differences.
Here’s an integrated JFET tester/matcher to speed up the process of testing or matching JFETs:
The finished article:
Here is the PCB overlay and etch-resist pattern. The etch resist pattern should be printed at 300 dpi. The board is 55 x 38.5mm.