Low side Mosfet switching for current drived loads not working properly

Thread Starter

kiweered

Joined Dec 13, 2021
28
Hey,

I driving 3 small electrovalves. This model of electrovalve must be supplied with a 250 mA current in order to open.
Here is an extract of the valve datasheet :

Operating current : 250mA

1680013730074.png

I'm using 3 mosfet to drive these valves with my STM32 mcu.
1680013471005.png
The mosfet is LB8721.
Gate Threshold voltage : 1.8V < 3.3V of GPIO mcu output.
Continuous Drain Current V GS @ 10V is 62A which is largely enough.

I tested the valve alone and here is what I got :
1680015119025.png
So you can check that the valve has an internal resistance when working of approx 4 ohm.
But when I build the mosfet low side switch circuit above, the current in the valves decrease a lot and isn't the same in every valve.
From what I tested it seems that it's working better when I connect only low-side switch than when the 3 of them are connected.
With 3.3V supply :
- 0.150 A flowing in a valve when 1 of mosfet gate is supplied with 3V.
- 0.100 A flowing in a valve when 2 of mosfet gate are supplied with 3V.
- 0.050 A flowing in a valve when 3 of mosfet gate are supplied with 3V.

With 5V supply (i'm using supply to power the mosfets gate in the test not the 3.3V mcu output) :
- 0.180 A flowing in a valve when 1 of mosfet gate is supplied with 5V.
- 0.160 A flowing in a valve when 2 of mosfet gate are supplied with 5V.
- 0.140 A flowing in a valve when 3 of mosfet gate are supplied with 5V.

I tried increase the supply voltage (that's why I have a supply of 5V instead of the recommanded 3V with 10ohm resistor) to increase current but it's not really a good solution because i don't want some random 6.4V supply in my system.

I don't think the issue comes from my power supply because it should be able to output 3.5A and I need less than 1A.

Do you have any idea where the issue could come from ?
I saw that in some cases, you use an other transistor to drive the gate of the mosfet, would it be helpfull ?
What other circuits could I use to drive the valves properly with my mcu ?

Thx,
Adrien
 

sghioto

Joined Dec 31, 2017
5,420
In the above schematic what are the voltages on the drains of Q3-Q5 when activated?
Suspect the mosfets may not be getting enough gate drive.
 
Last edited:

crutschow

Joined Mar 14, 2008
34,692
Gate Threshold voltage : 1.8V < 3.3V of GPIO mcu output.
Continuous Drain Current V GS @ 10V is 62A which is largely enough.
3.3V is probably not enough to fully turn on the MOSFETs.
1.8V is the nominal Vgs(th), which can vary up to the maximum Vgs(th) of 2.35V (below) which is likely too close to 3.3V to fully turn on the MOSFET.
Note that at Vgs(th) the current is only 25µA.
1680017870077.png
That's why they specify the full-on drain current with a Vgs of 10V.

You need a logic-level type MOSFET that will fully turn on at a Vgs of 3.3V
 
Last edited:

ElectricSpidey

Joined Dec 2, 2017
2,839
Funny, but the INFINEON page states that the MOSFET "is" logic level.

From their information page.

"30V Logic Level Single N-Channel StrongIRFET™ Power MOSFET in a TO-220 package"

.25 A should be no problem with 3.3 volts at the gate.

Looking at this, I'm thinking a supply problem of some kind, maybe in the test board.

- 0.150 A flowing in a valve when 1 of mosfet gate is supplied with 3V.
- 0.100 A flowing in a valve when 2 of mosfet gate are supplied with 3V.
- 0.050 A flowing in a valve when 3 of mosfet gate are supplied with 3V.
 
Last edited:

ronsimpson

Joined Oct 7, 2019
3,145
We don't really know the Gate voltage. Many GPIO pins do not pull up well. The 10k ohm Gate resistor may be keeping the voltage from reaching 3V. It the voltage is low try 100k or 220k or ..... That might buy you a half volt or more.
 

Thread Starter

kiweered

Joined Dec 13, 2021
28
Thanks all of you for your answers, I'll look and try what you said.

I did more testing and here is what i have :

With the power supply of Vcc=5V, when I connect the gate of the mosfet to Vcc, the power supply voltage decrease drasticly :
1680078229068.png
- with 1 mosfet gate connected to power supply -> Vcc drop to 3.57V and Vdrain = 0,370V
- with 2 mosfet gate connected to power supply -> Vcc drop to 3.35V and Vdrain = 0,310V and 0,470V
- with 3 mosfet gate connected to power supply -> Vcc drop to 3.14V and Vdrain = 0,370V and 0,407V and 0,318V

With the power supply of Vcc=5V, when I connect the gate of the mosfet to my MCU 3.3V output, the power supply voltage drop also :
1680078196342.png
- with 1 mosfet gate connected to mcu 3.3V -> Vcc drop to 3.7V and Vmcu drop to 3V, Vdrain = 0,3 V
- with 2 mosfet gate connected to mcu 3.3V -> Vcc drop to 3.36V and Vmcu drop to 2.9V, Vdrain = 0,3-0.4 V
- with 3 mosfet gate connected to mcu 3.3V -> Vcc drop to 3.09V and Vmcu drop to 2.8V, Vdrain = 0,3-0.4-0.35 V

I guess my power supply isn't great because when you need to output current the votage drop but I don't understand why the 3.3V output of the mcu also drop.
 

Irving

Joined Jan 30, 2016
3,992
I guess my power supply isn't great because when you need to output current the votage drop but I don't understand why the 3.3V output of the mcu also drop.
And the 5v to the valves is independent of that? Check that the 5v from the USB is holding up. If not, something is very wrong. Sounds to me you have a faulty wiring somewhere...

What happens if you short drain to source of a MOSFET via an ammeter? Does valve behave and you get expected volts/currents? If so MOSFET is wired wrong or faulty. BTW breadboards are really bad at currents greater than a few mA but that doesnt explain why 5v supply drops, unless you're measuring voltage on the breadboard somewhere rather than at supply.

Whats the pinout of the MOSFET, should be 123 = GDS. Check G to S is very high impedance (several Megs or open) when MOSFET out of circuit.

Spice model suggests MOSFET is fine for this application.

1680083922738.png
 

Thread Starter

kiweered

Joined Dec 13, 2021
28
And the 5v to the valves is independent of that? Check that the 5v from the USB is holding up. If not, something is very wrong. Sounds to me you have a faulty wiring somewhere...

What happens if you short drain to source of a MOSFET via an ammeter? Does valve behave and you get expected volts/currents? If so MOSFET is wired wrong or faulty. BTW breadboards are really bad at currents greater than a few mA but that doesnt explain why 5v supply drops, unless you're measuring voltage on the breadboard somewhere rather than at supply.

Whats the pinout of the MOSFET, should be 123 = GDS. Check G to S is very high impedance (several Megs or open) when MOSFET out of circuit.

Spice model suggests MOSFET is fine for this application.

View attachment 290980
Now I see, when I measure the voltage on the breadboard I measure this voltage drop but when I put my voltmeter directly on the output pin of my power supply, the voltage doesn't drop. I don't know why ...

1680093850552.png
 

sghioto

Joined Dec 31, 2017
5,420
Move the 5 volt supply wires as close as possible to the circuit. Acts like a breadboard issue..
Same with the negative of the 3.3 supply
1680094515338.png
 

ronsimpson

Joined Oct 7, 2019
3,145
There is resistance in wires. There is resistance in the breadboard connections. As current flows a voltage develops across the wires.
I see long wires from supply to bread board. Then the power reaches the breadboard it does not go directly to GND and 5V but the power enters in the middle and must go through more jumpers to get to GN and 5V. Every connection, every wire adds resistance.
 

Audioguru again

Joined Oct 21, 2019
6,759
The simulation shows a "typical" Mosfet that turns on well when its Vgs is 3V. But you cannot buy a typical one, some have a gate-source that need 4.5V to turn on "fairly well".
 

Irving

Joined Jan 30, 2016
3,992
The simulation shows a "typical" Mosfet that turns on well when its Vgs is 3V. But you cannot buy a typical one, some have a gate-source that need 4.5V to turn on "fairly well".
True, but its a high probability that at least one from 3 would be in that ballpark. What we're seeing suggests thats not the issue.
 
Top