Problem in switch design with jfet.

[extrepito]

Hello,

I want to ask you for help to solve this headache. I have this design, which simulated in Proteus works correctly. But in practice it doesn't work. The jfets are j112 and work as switches and the 741 works as a comparator. Any idea what could be causing the malfunction?

Thank you!

 

[JohnRoberts]

what doesn't it do?

N channel JFET are not bilateral switches, those generally use CMOS technology (like CD4016/CD4066).

It is unclear from a quick look, what you are trying to accomplish.

JFET channel on resistance is controlled by JFET gate voltage WRT voltage of the drain/source channel. Generally a negative gate voltage will pinch off conduction in the JFET. Your schematic does not reveal a DC path for JFET drain/source.

JR

 

[extrepito]

Thanks for the reply.
Indeed, jfets require a negative voltage between gate and source/drain to open the conduction between drain and source.
It does not appear in the first image, but there is a 100k resistor between source and gnd, as it appears in the following schematic(r82). The function it performs is that Q14 does not drive when Q15 drives and vice versa. In simulation it works correctly but in the real test circuit it does not work (Q15 and Q14) they both conduct at the same time in any selection. What I have achieved is that by removing Q15 from the pcb, Q14 works correctly.
Thanks for the response John.

 

[JohnRoberts]

All of those JFETs have DC references, mostly to op amp outputs.

JR

 

[extrepito]

Could the reference to DC be the cause of the malfunction? The opams are no longer the rc4558 but the lm4562.

 

[JohnRoberts]

Could the reference to DC be the cause of the malfunction? The opams are no longer the rc4558 but the lm4562.

That is what I have been saying.

JFETs are simple devices. For N channel JFETS if the gate voltage is below the pinch off voltage the channel will not conduct. If the gate is close to the drain/source voltage the channel will conduct.

With a VOM you can measure actual voltages in circuit and perhaps get evidence about what is going on.

JR

 

[extrepito]

The measurements are the curious thing, when negative voltage (-5.8Vdc) is applied to the gate of Q14 and Q15 they remain conducting. But if I remove one of the two (or Q14 or Q15), the one that remains in the circuit acts as a switch correctly.

 

[JohnRoberts]

The measurements are the curious thing, when negative voltage (-5.8Vdc) is applied to the gate of Q14 and Q15 they remain conducting. But if I remove one of the two (or Q14 or Q15), the one that remains in the circuit acts as a switch correctly.

I just looked for a data sheet and J112 specifies 5V max for cut off, so -5.8V should be cut off (high impedance). Have you confirmed that drain and source voltages are 0V ?

JR

 

[extrepito]

Hi John,
when I measure the voltages without Q14 or Q15, at Gate they are +6.3Vdc (for driving) and -5.5Vdc (for opening the switch). In source we have a voltage that goes up and down in the order of mVolts. Then in Drain 0Vdc.

Once j112 (Q14 and Q15) are connected to the circuit, we obtain the values: Gate: +0.5Vdc /-5.5Vdc, Source: +2.5mV/-0.5mV, Drain: +2.5mV/+2.5mV.

Thanks for the attention. I've been thinking about this problem for 1 week without solving anything....

 

[JohnRoberts]

Hi John,
when I measure the voltages without Q14 or Q15, at Gate they are +6.3Vdc (for driving) and -5.5Vdc (for opening the switch). In source we have a voltage that goes up and down in the order of mVolts. Then in Drain 0Vdc.

Once j112 (Q14 and Q15) are connected to the circuit, we obtain the values: Gate: +0.5Vdc /-5.5Vdc, Source: +2.5mV/-0.5mV, Drain: +2.5mV/+2.5mV.

Thanks for the attention. I've been thinking about this problem for 1 week without solving anything....

A positive drive to a N channel JFET will cause the gate diode to forward bias and clamp at roughly +0.5V DC

With drain and source both within mV of 0V the -5.5V gate should be cut off (high impedance). The Gate with +0.5VDC will be turned on hard. In fact any current flowing into that gate will flow through the channel and out either the drain or source (likely source based on the +2.5mV). Ideally gates like to be driven to 0V to avoid DC gate diode current contaminating the audio path.

JR

 

[extrepito]

It is strange because with the voltage of -5.5Vdc I measure continuity between drain and source and I obtain about 25ohm, this in both (Q15 and Q14). Which I don't know what to try. Should I try another Jfet that is more similar to the tis75 or maybe a mosfet?

 

[JohnRoberts]

It is strange because with the voltage of -5.5Vdc I measure continuity between drain and source and I obtain about 25ohm, this in both (Q15 and Q14). Which I don't know what to try. Should I try another Jfet that is more similar to the tis75 or maybe a mosfet?

Measuring FETs can be funny, it doesn't take any current to switch states, they respond to voltage, if the gate is clearly below pinch off (-5V) the drain/source should be high impedance. Of course the drain and source also need to be proximate to 0V or higher while that is being measured.
===
No, for switching audio consider CMOS TG (transfer gates) like CD4016/CD4066... but I already said that.

Have fun.

JR

 

[extrepito]

I have already resolved the issue. The switches work correctly, what was failing was the continuity measurement between source and drain. I don't know why the ohmmeter does not detect that one of the njfets increases its impedance (opens) when the njfets are in parallel. But following the signal with the oscilloscope, that njfet does not allow the signal to pass through. Very curious. Thank you very much for the help! and the truth is that it is fun to work with jfets.