1
\$\begingroup\$

I've got the resistive network generating 1V from 3V, but when I connect two N-FETs on the output with gates tied to sources tied to ground (that should result in them closing, right?), the voltage drops to near zero. Works as expected with only 1 N-FET connected. I'm using BS170 specifically.

Works fine in simulation: Image

Why is that?

\$\endgroup\$
  • \$\begingroup\$ Can you post the voltages you've measured? \$\endgroup\$ – diverger Nov 18 '14 at 12:17
  • \$\begingroup\$ @diverger It's 20mV. \$\endgroup\$ – andrey g Nov 18 '14 at 12:36
  • \$\begingroup\$ As Michael's analysis, you may have a bad FET and a good one. If you have some more, you can replace the old one, and have another try. Be caution when handling them. \$\endgroup\$ – diverger Nov 18 '14 at 12:40
2
\$\begingroup\$

In your simulation you are using an idealized FET model. There is no leakage across the source to drain and so the voltage divider stays unloaded and produces the 1V like you expect.

When you apply a real FET that does have leakage it loads the divider to cause the divider to produce a lower voltage. Excessive leakage will make the divider output approach zero.

After all with the resistor values shown the current through the unloaded divider is only ~10uA.

Your BSS170 is speced in its data sheet to have a MAX leakage is 0.5uA with zero gate voltage. So if your two FETs were good you could expect to see a little bit of droop of the 1V from the divider.

Have you been careful in handling your FETs? Devices like the BSS170 have very high gate impedance and can be damaged simply in handling if there is static electricity build up on your body from clothing or carpet or padded chair. I suspect that one of your FETs may be damaged.

\$\endgroup\$
  • \$\begingroup\$ Um, then why "Works as expected with only 1 N-FET connected"? \$\endgroup\$ – diverger Nov 18 '14 at 12:11
  • \$\begingroup\$ @diverger: The one that works is good? ;) \$\endgroup\$ – EM Fields Nov 18 '14 at 12:20
  • \$\begingroup\$ @EM Fields: In the question, "Works as expected with only 1 N-FET connected", that is with two paralleling, it won't work, but use only one, it work, right? You mean, one is good, another is bad:)? \$\endgroup\$ – diverger Nov 18 '14 at 12:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.