Uuuuuuuuh? Capacitor in (simple) circuit always reads 0nF

[mayor]

Hi,

I am stumped. I have a simple circuit (microcontroller reset pin to 3v3 through a 10k pullup, and to ground through a 1n cap). I measured caps in turn (1n, 22n and even 1u), each check out, but when in circuit... no capacitance!

I checked for shorts (resistance across cap is in the hundreds of k ohms). Can anyone help explain what I am observing?

[slaterk93]

Did you check the resistor?

Edit: Well what are you doing? Are you measuring the capacitance in circuit with a multimeter?

[krivx]

Its sometimes difficult to test electronic components while in circuit.  a way to check, out of circuit, is to remove one lead of the capacitor, and then check across the capacitor.

This can be pain with SMD parts. Try wicking off the solder one pad.

[mikeselectricstuff]

If it's in circuit, all bets are off.

[danadak]

Unless the I/O pin Cap is tied to is in tri state you will not be able to measure cap.

That's because probably the output is driving low and shows a Rdson from the pin to
ground of at most a few 100 ohms.

Note if the output is tri state you then have the issue of how the C is measured, the
compliance range of the C tester of concern.

Regards, Dana.

[mayor]

Hi,

thanks for the responses. Some answers:

1. Checked the resistor: good at 10K
2. All caps have been tested out of circuit: they check out!
3. I am checking for capacitance while the circuit is off. I can't say for sure what the state of the reset pin is, but its resistance to VCC is the expected 10K and resistance to ground is in the hundreds of Ks.

I understand measuring in-circuit won't necessarily give me the "expected" value. But 0 is truly unexpected :-) I guess the next step is to unsolder the reset pin from its pad?

[krivx]

How are you measuring capacitance? Does your meter apply DC and measure the charge time? If so kohms to GND can throw this off, as current will never approach zero and the meter thinks the cap is never charged.

Does you circuit actually work?

[mikerj]

I understand measuring in-circuit won't necessarily give me the "expected" value. But 0 is truly unexpected :-) I guess the next step is to unsolder the reset pin from its pad?

It's not clear if you suspect a problem with the capacitor because the circuit isn't working correctly, or simply because the result from an in-circuit measurement is unexpected?  If the latter, then you are almost certainly worrying about nothing.  Leakage paths through esd diodes and other components are probably stopping your capacitance meter working properly.  If the cap was good when you soldered it in, then forget about it.

[wraper]

But 0 is truly unexpected :-) I guess the next step is to unsolder the reset pin from its pad?

Why it would be unexpected. and for FFS why do you need to measure them in circuit, 10k resistor is what kills the measurement. Also, using 1nF cap on the reset pin is insane.

[Monkeh]

Also, using 1nF cap on the reset pin is insane.

And what would you suggest using?

[wraper]

Also, using 1nF cap on the reset pin is insane.

And what would you suggest using?

If the mcu does not require extenal reset, there might be no cap at all. If it does, some higher capacitance is needed. If using 1nF cap together with 10K resistor, there might be no cap at all as it has almost no effect.
EDIT: I checked MSP430 programming interface, this cap might be to protect input from the noise, not for reset. So I stand corrected.

[mayor]

I understand measuring in-circuit won't necessarily give me the "expected" value. But 0 is truly unexpected :-) I guess the next step is to unsolder the reset pin from its pad?

It's not clear if you suspect a problem with the capacitor because the circuit isn't working correctly, or simply because the result from an in-circuit measurement is unexpected?  If the latter, then you are almost certainly worrying about nothing.  Leakage paths through esd diodes and other components are probably stopping your capacitance meter working properly.  If the cap was good when you soldered it in, then forget about it.

Hi, thanks for the information!

There actually is a problem with the circuit: it looks like the MCU is held in reset state. ie, nothing happens, but when I connect the reset/test pins to the launchpad to debug/program, execution is good. This leads me to believe there is an issue with power-on reset timing, so the first thing I checked was the capacitor. I am now at 1u X 10k, still the same issue.

[mayor]

Also, using 1nF cap on the reset pin is insane.

And what would you suggest using?

If the mcu does not require extenal reset, there might be no cap at all. If it does, some higher capacitance is needed. If using 1nF cap together with 10K resistor, there might be no cap at all as it has almost no effect.
EDIT: I checked MSP430 programming interface, this cap might be to protect input from the noise, not for reset. So I stand corrected.

This model needs reset-- TI actually suggests 47k / 1-2.2 nf if one uses spy-bi-wire programming, 10nf otherwise:

"The pulldown capacitor should not exceed 2.2 nF when using devices with Spy-Bi-Wire interface in Spy-Bi-Wire mode or in 4-
wire JTAG mode with TI tools like FET interfaces or GANG programmers."

I did indeed have wrong information with regards to the pullup, but the timing for the RC I have now is 0.01s (1u x 10k), which to my novice mind is way enough time for VCC to ramp up and stabilize (?).

[rsjsouza]

Hi,

thanks for the responses. Some answers:

1. Checked the resistor: good at 10K
2. All caps have been tested out of circuit: they check out!
3. I am checking for capacitance while the circuit is off. I can't say for sure what the state of the reset pin is, but its resistance to VCC is the expected 10K and resistance to ground is in the hundreds of Ks.

I understand measuring in-circuit won't necessarily give me the "expected" value. But 0 is truly unexpected :-) I guess the next step is to unsolder the reset pin from its pad?

To measure capacitance the DMM (or a simple LCR meter) will put an AC voltage into its probe terminals, thus also injecting it into the circuit in your case. Given the cap is connected to an input terminal (the reset pin),  the DC measurement with the ohmmeter will most probably give you a different result.

[suicidaleggroll]

Which model MSP430 is this?

If it's one of the bigger ones, with a big image you're loading into it, it's possible to exceed the watchdog timeout before code execution begins.  If this is the case, you'll need to add a special instruction to shut off the watchdog timer during loading, rather than during execution.  I've never run into this myself (all of my MSP430 designs use the small G-series with small images), but I see mention of it it all the time when googling around for other questions.

[mayor]

Which model MSP430 is this?

If it's one of the bigger ones, with a big image you're loading into it, it's possible to exceed the watchdog timeout before code execution begins.  If this is the case, you'll need to add a special instruction to shut off the watchdog timer during loading, rather than during execution.  I've never run into this myself (all of my MSP430 designs use the small G-series with small images), but I see mention of it it all the time when googling around for other questions.

It's a G2553, so the image is pretty small. I will definitely keep this in mind in the future, and will check whether this could be an issue even with this version. Thanks for sharing.

[rsjsouza]

Just FYI, a good thread about this:
https://e2e.ti.com/support/microcontrollers/msp430/f/166/t/221152

[wraper]

This model needs reset-- TI actually suggests 47k / 1-2.2 nf if one uses spy-bi-wire programming, 10nf otherwise:

"The pulldown capacitor should not exceed 2.2 nF when using devices with Spy-Bi-Wire interface in Spy-Bi-Wire mode or in 4-
wire JTAG mode with TI tools like FET interfaces or GANG programmers."

This cap is certainly not for generating the reset signal. The thing is, power on switch on will rise slower than delay such RC circuit can provide. Although I never used MSP430, I guess reset is generated by internal Brown-out detector.

[mayor]

This model needs reset-- TI actually suggests 47k / 1-2.2 nf if one uses spy-bi-wire programming, 10nf otherwise:

"The pulldown capacitor should not exceed 2.2 nF when using devices with Spy-Bi-Wire interface in Spy-Bi-Wire mode or in 4-
wire JTAG mode with TI tools like FET interfaces or GANG programmers."

This cap is certainly not for generating the reset signal. The thing is, power on switch on will rise slower than delay such RC circuit can provide. Although I never used MSP430, I guess reset is generated by internal Brown-out detector.

No-- not what I meant, though poorly expressed. I understand it is a timing device for releasing reset! Thank you for taking time to help.

[suicidaleggroll]

FWIW - I do a lot of designs with the MSP430G series, I never put a cap on the reset line, and I've never had an issue with power-up or sporadic resets while running.  If it is necessary, I'm not sure what for, and I've apparently never encountered any case where it mattered.

Unless your cap is faulty and has an internal short, I'd look elsewhere for the source of your problem.

[mayor]

FWIW - I do a lot of designs with the MSP430G series, I never put a cap on the reset line, and I've never had an issue with power-up or sporadic resets while running.  If it is necessary, I'm not sure what for, and I've apparently never encountered any case where it mattered.

Unless your cap is faulty and has an internal short, I'd look elsewhere for the source of your problem.

Hi there,

I've only done 10 or so designs with it, and this is the second time this happened.

Thing is, it seems code-dependent. For example, I was working on a simple project to transmit data over an NRF2401, and everything was totally fine without a cap on reset.

That is, until I started making subtle changes to the code, mostly just reorganizing it. The device just stopped working. Took me a while to realize that there was no error in the code. I started probing the circuit, and when I touched the reset pin, it started working. In that case, adding the recommended value cap just solved it.

That brought me back to the project we are now discussing: I thought I just needed to add a bit of capacitance to resolve my issue, but no joy yet. I will look at the initialisation function and stop the watchdog timer early and see what that does in this case.

From the feedback here, there doesn't appear to be a point trying to read capacitance while the cap is in-circuit.

Thanks for taking time to provide feedback.