latching push-button controlled MOSFET switch with IRF7319

[luca0074]

Hello all

I have integrated the "soft power switch" from Sparkfun (https://cdn.sparkfun.com/assets/5/6/6/6/0/SparkFun_Soft_Power_Switch.pdf) in my handheld device.
Due to the unavailability of the SIL2308, I used the IRF7319TRPBF as in the original design (http://www.mosaic-industries.com/embedded-systems/microcontroller-projects/raspberry-pi/on-off-power-controller).

Here is the schematic of my implementation:


I have the following problem:
I can only switch on the device via S1 (or ON_BUTTON_EXT) if I have vBus > 4.4V.

Is this due to the IRF7319TRPBF used (https://www.mouser.ch/datasheet/2/196/Infineon_IRF7319_DataSheet_v01_01_EN-1228230.pdf) or is there another cause of the problem?

Is there an easy fix for the problem?

Thanks for your help!

Greetings Luca

[eblc1388]

No. The problem is not due to the IRF7319TRPBF.

Try, in the following order:

1. increase value of C3 between R14 and R15 from 0.1uF to 1uF, then retest, if still fails, then keeping C3 at 1uF, do
2. increase value of C5 from 10uF to 22uF.

[luca0074]

No. The problem is not due to the IRF7319TRPBF.

Try, in the following order:

1. increase value of C3 between R14 and R15 from 0.1uF to 1uF, then retest, if still fails, then keeping C3 at 1uF, do
2. increase value of C5 from 10uF to 22uF.

Hi eblc, thanks for your suggestions!
1. didn't work.
2. this just made the time longer that is needed for shutdown the device.

[Peabody]

The threshold voltage of the SIL2308 has a minimum of 0.5V and a maximum of 1V.  But the IRF part has a minimum of 1.0V, with no maximum shown.  So maybe that difference is enough  to make it not operate properly at low voltages.

Would the processor power up fast enough that it could turn on the N-channel and keep the power on with just a brief button press?  If so, you could move all the latching stuff into the processor and greatly simply the circuit.

[eblc1388]

This type of circuits behave heavily on the nature of the connected load. Most hated load with large capacitor connected.

Can you also confirm the circuit does not work even if no output load is connected? This might point to a different approach in modifying the component values.

[luca0074]

The threshold voltage of the SIL2308 has a minimum of 0.5V and a maximum of 1V.  But the IRF part has a minimum of 1.0V, with no maximum shown.  So maybe that difference is enough  to make it not operate properly at low voltages.

Thanks for checking!

Would the processor power up fast enough that it could turn on the N-channel and keep the power on with just a brief button press?  If so, you could move all the latching stuff into the processor and greatly simply the circuit.

i wanted a solution, where i can disconnect the power via hardware completely, in case of bad programming of the microprocessor.

This type of circuits behave heavily on the nature of the connected load. Most hated load with large capacitor connected.

Can you also confirm the circuit does not work even if no output load is connected? This might point to a different approach in modifying the component values.

i have total 11.2uf capacitance on vcc. there is only my regulator.

the starting current is @5V 120mA. running current is @5V 100mA.
i disconnected the load; the circuit doesn't work below 4.4V.

[Peabody]

Do you have a scope?  It would be good to know whether the p-fet turns fully on when you press the button.  In other words, what exactly isn't working?  Is the p-fet not turning on at all, or is the p-fet turning on but the n-fet isn't?

Well, I think the problem almost has to be the different threshold voltages versus the original design mosfets.  It's possible that playing with the capacitor and/or resistor values could get it to work at lower voltages, but in the end you may need to use discrete mosfets with the right threshold voltages, if you can find them.  Still, it seems like there ought to be a way to make it work with 3.7V input.

Edit:  If you start it up at 4.4V or more, then lower the input voltage, at what input voltage does it turn off?

Edit2: This may be crazy, but I wonder if replacing the n-fet with an NPN might make it work.  Well, that would be a different circuit.  Anyway, I just wondered.

[eblc1388]

i have total 11.2uf capacitance on vcc. there is only my regulator.

I did a simulation of your actual circuit, using 1uF and 22uF as stated above. I have chosen PMOS and NMOS with Vth=1V to match with IRF7319TRPBF and an external load consists of 10R and 47uF in parallel. After power is applied, a switch is close at time=2 and released a second later.

The simulation results show that the circuit can work down to 2.3V or lower. This indicates that the IRF7319TRPBF will most likely works without problem. You may have errors in pin connections or have defective components.

[Peabody]

Would it be possible to run the simulation with mosfets having Vth of 2V?  If that simulation still works well below 4.4V, then the OP may have some other problem.

[eblc1388]

Would it be possible to run the simulation with mosfets having Vth of 2V?  If that simulation still works well below 4.4V, then the OP may have some other problem.

In simulation with mosfets having Vth of 2V, the circuit won't work for Vin=2.3V. It works OK with Vin=2.5V or more.

[luca0074]

Do you have a scope?  It would be good to know whether the p-fet turns fully on when you press the button.  In other words, what exactly isn't working?  Is the p-fet not turning on at all, or is the p-fet turning on but the n-fet isn't?

Sadly i dont have a scope, but i could measure with a multimeter I have a PowerProfilerKit2, where i could measure up to 8 digital signals.  where should i measure what?

Edit:  If you start it up at 4.4V or more, then lower the input voltage, at what input voltage does it turn off?

It turns off at 2.8V.

I did a simulation of your actual circuit, using 1uF and 22uF as stated above. I have chosen PMOS and NMOS with Vth=1V to match with IRF7319TRPBF and an external load consists of 10R and 47uF in parallel. After power is applied, a switch is close at time=2 and released a second later.

Cool! Thanks for sharing the .asc file!

The simulation results show that the circuit can work down to 2.3V or lower. This indicates that the IRF7319TRPBF will most likely works without problem. You may have errors in pin connections or have defective components.

I have 4 assembled devices here and all show the same fault behaviour.
Maybe i made a mistake with the pcb design?

[eblc1388]

Edit:  If you start it up at 4.4V or more, then lower the input voltage, at what input voltage does it turn off?

It turns off at 2.8V.

The cut off voltage should be a lot lower.

I did the simulation with Vin at 4.4V and start decreasing at time=6s to see when the circuit will turns off. It did so at about 1.5V. The orange curve is the NMOS gate voltage during the voltage drop test.

Please repeat the same voltage dropping test starting at Vin=4.4V, then at 4.2, 4.0, 3.8..., decrease Vin at 0.2V interval and record the NMOS gate voltage(pin2 of the IC) until the circuit cut off. Post back the results.
 

[luca0074]

Please repeat the same voltage dropping test starting at Vin=4.4V, then at 4.2, 4.0, 3.8..., decrease Vin at 0.2V interval and record the NMOS gate voltage(pin2 of the IC) until the circuit cut off. Post back the results.

I did the measurements, but i have very different results.

Measurment	VBUS [V]	3.3V_EN (NMOS GATE) [V]
1	5.0	2.31
2	4.9	2.28
3	4.8	2.27
4	4.7	2.26
5	4.6	2.25
6	4.5	2.24
7	4.4	2.23
8	4.3	2.22
9	4.2	2.2
10	4.1	2.2
11	4.0	2.18
12	3.9	2.17
13	3.8	2.15
14	3.7	2.14
15	3.6	2.13
16	3.5	2.1
17	3.4	2.1
18	3.3	2.1
19	3.2	2.05
20	3.1	2
21	3.0	1.98
22	2.9	1.95
23	2.8	1.93
24	2.7	1.92
25	2.6	0
26	2.5	0
27	2.4	0
28	2.3	0
29	2.2	0
30	2.1	0
31	2	0
32	1.9	0
33	1.8	0
34	1.7	0
35	1.6	0
36	1.5	0

[Peabody]

Could you also measure Vcc at a few of these points?  Maybe I don't understand the circuit, but it seems that at VBUS of 5V, the gate should be at the zener voltage, which should be 3.3V.  Why is it lower?

Edit:  Ok, it may be your meter dragging it down.  Instead of measuring the gate voltage, try measuring the intersection of the 10K, 300K and zener.

[luca0074]

Could you also measure Vcc at a few of these points?  Maybe I don't understand the circuit, but it seems that at VBUS of 5V, the gate should be at the zener voltage, which should be 3.3V.  Why is it lower?

Edit:  Ok, it may be your meter dragging it down.  Instead of measuring the gate voltage, try measuring the intersection of the 10K, 300K and zener.

I measured at VCC and POWERSENSE to GND. The measured values are practically the same...

Maybe I choose the false Zenerdiode?
I choose MM3Z3V3T1G https://media.digikey.com/pdf/Data%20Sheets/ON%20Semiconductor%20PDFs/MM3ZyyyT1G_SZMM3ZyyyT1G_Series_rev.11_Sep2014.pdf
It has the correct Device Marking (05) on it.

[Peabody]

The zener value is only valid at a specified minimum current, which you are probably well below.  So it will pass some current at a lower voltage.  As long as nothing is connected to 3.3V_EN or POWERSENSE, you could remove the zener and see if everything works better.  But if you really want this to work from 3.3V to 12V, you're going to need the zener to prevent high voltage from reaching the processor I/O pins.

[eblc1388]

Judging from the results, it is probably caused by the zener diode not getting sufficient rated current via R13(10K). The zener voltage will drops quite a lot from the rated voltage under this condition.

You have one of five possible choices.

1. Change one end of resistor R13 from connecting to VCC(input of the 3V3 regulator) into connect it to output of the 3V3 regulator instead. This will ensure the gate of the NMOS be supplied by 3.3V even the Vin falls from 4.4V to 3.5V. 

2. Put in series a silicon diode like 1N4148 with the 3V3 zener. The diode's 0.65V drop will add to the current 2.31V drop to a total voltage drop of around 3V which enable your circuit to work below VBus of 4.4V, hopefully at min. 3.8V.

3. Lower the value of R13 from 10K to somewhere between 1K~330R. This would gives 5mA zener current when Vin=5V if R13=340R but higher value reduces power dissipation in zener diode. The calculation for R13 is (5V-3.3V)/5mA = 340R.

4. Use a higher voltage zener diode, e.g. 3V9 instead of 3V3. With R13=10K ~ 1K, this would hopefully gives 3.3V when Vin=5V.
 
5. Use a 3.3V reference diode to replace the zener diode,e.g. the ZRC330 by DIODE INC. The ZRC330 is a two terminal device in SOT23 package and will give 3.3V with forward current from 20uA~25mA. However, this is the most expensive option.

 
ps. edited: added three more options for a possible solution. Corrected mistake in calculation of R13 value.

[Peabody]

The problem is it has to work at both 3.3V and 12V.    So if you tweak it so it works at 3.3V, the voltage may be too high at 12V because the zener will actually produce a 3.3V drop, or 3.9V.

I wonder if a white LED (about 2.7V drop) in place of the zener would give a sharper knee.  Or a white LED in  series with a regular signal diode (1N4148) to give about 3.3V in total.

[eblc1388]

The problem is it has to work at both 3.3V and 12V.

Why does it has to work at 12V?

The OP's circuit in reply#5 clearly shown a regulator with Vin of max. 6V.

[luca0074]

The problem is it has to work at both 3.3V and 12V.

Why does it has to work at 12V?

The OP's circuit in reply#5 clearly shown a regulator with Vin of max. 6V.

Sorry, it has only to work up to 5.5V. (Powered via 5V USB)

3. Lower the value of R13 from 10K to somewhere between 100R~33R. This would gives 5mA zener current when Vin=5V if R13=33R but higher value reduces power dissipation in zener diode. 

Thanks for all the choices!  i choose number 3, because i can fix that with just a change in the BOM.

The Voltage on POWERSENSE, when the Device is ON is with R13=48Ohm @ 5V now at 3.9V. With R13=10kOhm @ 5V it is at 2.3V (see attachment in Reply #14).
Sadly it dosn't fix the issue. Also id would be >3.3V and could damage my microcontroller; with 10k it would be sufficient for a logic High on most controllers.

EDIT: The following informations are false, because I measured wrong. see the correct measurements in Reply #22

I dig a little deeper and recorded the current consumption of the device in state OFF. I discovered a hysteresis on the issue.

when VCC goes under 4.3V the current goes from 65uA up to a average of 1-2mA with 15mA peaks, till VCC is again greater than 4.8V.
The device can only be turned on, when the current is at the low value (65uA).
(i measured with R13=48 and also R13=10k. The results are the same)



also shouldn't the current consumption overall not be much smaller? or is my measurement not precise enough for that little currents?
Quote from: https://www.nordicsemi.com/Products/Development-hardware/Power-Profiler-Kit-2:

200nA to 1A measurement range
Resolution varies between 100nA and 1mA depending on the measurement range

[eblc1388]

Thanks for all the choices!  i choose number 3, because i can fix that with just a change in the BOM.

The Voltage on POWERSENSE, when the Device is ON is with R13=48Ohm @ 5V now at 3.9V. With R13=10kOhm @ 5V it is at 2.3V (see attachment in Reply #14).
Sadly it dosn't fix the issue. Also id would be >3.3V and could damage my microcontroller; with 10k it would be sufficient for a logic High on most controllers.

Sorry. My bad.

R13 value should be (5V-3.3V)/5mA but I mistakenly used 0.05 as the divisor instead of 5mA. Therefore the correct value of R13 for 5V should be 340R or more.

I cannot see any reason why the supply current will increase with the decrease of Vin. When the circuit is OFF, it should consume no current. Are you referring to the 3.3V microcontroller consumption only instead of this power On/OFF circuit? Have you follow the instructions on how to setup the power profiler to measure current? You need an extra 5V supply to power up the device in order to measure its current consumption. The power profiler cannot be both the source and as a current meter.

[jfiresto]

... I wonder if a white LED (about 2.7V drop) in place of the zener would give a sharper knee....

I am not sure where, but I came across the attached current vs. voltage curves for a blue LED over a wide range of currents and temperatures. Perhaps they might help address your question.

[luca0074]

Thanks for all the choices!  i choose number 3, because i can fix that with just a change in the BOM.

The Voltage on POWERSENSE, when the Device is ON is with R13=48Ohm @ 5V now at 3.9V. With R13=10kOhm @ 5V it is at 2.3V (see attachment in Reply #14).
Sadly it dosn't fix the issue. Also id would be >3.3V and could damage my microcontroller; with 10k it would be sufficient for a logic High on most controllers.

Sorry. My bad.

R13 value should be (5V-3.3V)/5mA but I mistakenly used 0.05 as the divisor instead of 5mA. Therefore the correct value of R13 for 5V should be 340R or more.

320Ohm works perfect! now it looks much more as in your simulation.

I cannot see any reason why the supply current will increase with the decrease of Vin. When the circuit is OFF, it should consume no current. Are you referring to the 3.3V microcontroller consumption only instead of this power On/OFF circuit? Have you follow the instructions on how to setup the power profiler to measure current? You need an extra 5V supply to power up the device in order to measure its current consumption. The power profiler cannot be both the source and as a current meter.

Sorry. My bad. next time I rtfm
The current draw looks now better. 1uA@5V and 100nA@4V


but i still can't turn on the device below VCC of 4.4V

[eblc1388]

but i still can't turn on the device below VCC of 4.4V

If that is the case, there must be some loading pulling down on the "3.3V_EN" and/or "POWERSENSE" connections. This is possible as the MCU had no supply voltage yet and the above two signal pins could be powering up the MCU via port pin internal diodes. This affect the operation of power On/Off circuit badly.

A good circuit designer should make sure this situation could not happen, i.e. there must be no loading on these two signal pins until the MCU has properly powered up, reset and running.

[Peabody]

If VBUS is, say, 4V, and you press the button, the first thing that has to happen is the P-fet turns on.  Is that happening?  If you put your meter on Vcc, do you get any kind of blip when you push the button?

"A scope!  A scope!  My kingdom for a scope!"  - Richard III