Calculate current through zener

[eel]

Hi all, forgive me if this is really obvious. I have never used a zener before.

I am building a circuit to drive a fuel injector. Previously I have used mosfets with suitable avalanche characteristics but it is hard to find them now with shortages.

So I am looking at clamps to allow a larger selection of fets to be used.

Based off Reply #14 in this thread https://www.eevblog.com/forum/projects/is-this-mosfet-driver-circuit-ok/, I've drawn a quick schematic (attached below) but am not sure how to calculate current through the zener to select the appropriate part?

Thanks

[CMTan]

What is the zener diode for?
I don't think you need it.
 
What you need are
1.  A diode across the coil (fuel injector) - route the self induced voltage
     Cathode to +Vcc, anode to  drain of the NMOS
2.  MCU drive to the gate must be push-pull drive with sufficient voltage (greater than the VT of the NMOS)
     meaning if your MCU can only drive up to 3.3V, then the VT (VGS) should be must less than 3.3V, say not more than 2.5V

[MarkF]

First, your link is bad.
   https://www.eevblog.com/forum/projects/is-this-mosfet-driver-circuit-ok/msg2870144/#msg2870144

As to your question, the current through the zener would be its reverse voltage divided by its internal impedance.  For the 1N5369B mentioned the voltage is 51V and the impedance is 27Ω which would be I = V / R = 51V / 27Ω = 1.8A

[xavier60]

The Zener current will initially be what ever the inductor current was at the time the MOSFET switched off. It will then ramp down to zero.
The time taken will depend on initial current, inductance value and the voltage that is allowed to develop across the inductor, Zener voltage minus 14.4V.
I'd say that the energy dissipated by the Zener would be relevant also. By placing the Zener across the MOSFET, some of the energy dissipated will be drawn from the 14.4V supply.
What sort of injector is it?
If the circuit is to be used for testing injectors then a diode across it will do.

[mikerj]

What is the zener diode for?
I don't think you need it.
 
What you need are
1.  A diode across the coil (fuel injector) - route the self induced voltage
     Cathode to +Vcc, anode to  drain of the NMOS
2.  MCU drive to the gate must be push-pull drive with sufficient voltage (greater than the VT of the NMOS)
     meaning if your MCU can only drive up to 3.3V, then the VT (VGS) should be must less than 3.3V, say not more than 2.5V

No, a diode across the injector would absolutely ruin the injector closing time.  The idea is to allow the voltage across the transistor to increase to some safe limit so the magnetic field in the injector collapses as quickly as possible.

The way to calculate the clamp requirements is to consider the energy stored in the injector when open (0.5LI^2) and the maximum rate of injector operation which will be related to engine RPM and the injector strategy used (batch, sequential etc).  You then have an energy value and a time value so you can work out total power.  Add a suitable derating to this.

[CaptDon]

All of our locomotive fuel injection circuits have a diode from the drain of the mosfet to the positive supply feeding the injectors or in other words directly across the injector coil. As for a slowed closing time that would cause a richer fuel mixture and if the air/fuel system has a closed loop feedback it will compensate for the rich mixture with a shorter 'on' time. The delay in closing with the diode across the coil could probably be measured in 10's of microseconds where the injector open time probably 5 to 20 milliseconds. I suspect the delay would be a small portion of total time. We just let the fuel control system do the compensating. The injector to injector variation and even the resistance of the shortest injector wire run vs. the longest injector wire run probably contribute a bigger time delay difference. I know it does on diesel locomotives where the short run is 12 feet and the long run is 25 feet. We control the current ramp up rate with PWM to get a precisely predicted opening time which was far more critical to controlling detonation, horsepower and pollution.

[langwadt]

All of our locomotive fuel injection circuits have a diode from the drain of the mosfet to the positive supply feeding the injectors or in other words directly across the injector coil. As for a slowed closing time that would cause a richer fuel mixture and if the air/fuel system has a closed loop feedback it will compensate for the rich mixture with a shorter 'on' time. The delay in closing with the diode across the coil could probably be measured in 10's of microseconds where the injector open time probably 5 to 20 milliseconds. I suspect the delay would be a small portion of total time. We just let the fuel control system do the compensating. The injector to injector variation and even the resistance of the shortest injector wire run vs. the longest injector wire run probably contribute a bigger time delay difference. I know it does on diesel locomotives where the short run is 12 feet and the long run is 25 feet. We control the current ramp up rate with PWM to get a precisely predicted opening time which was far more critical to controlling detonation, horsepower and pollution.

the short closing time is important when you need a big injector big for full power so you need accurate short pulse width for idle and low loads

with PWM you need the diode or you will cook the zener, with PWM it is also less of an issue because the current in the coil is reduced so it closes faster anyway

never heard of any one controlling the ramp up, just the holding current

[MarkF]

All of our locomotive fuel injection circuits have a diode from the drain of the mosfet to the positive supply feeding the injectors or in other words directly across the injector coil. As for a slowed closing time that would cause a richer fuel mixture and if the air/fuel system has a closed loop feedback it will compensate for the rich mixture with a shorter 'on' time. The delay in closing with the diode across the coil could probably be measured in 10's of microseconds where the injector open time probably 5 to 20 milliseconds. I suspect the delay would be a small portion of total time. We just let the fuel control system do the compensating. The injector to injector variation and even the resistance of the shortest injector wire run vs. the longest injector wire run probably contribute a bigger time delay difference. I know it does on diesel locomotives where the short run is 12 feet and the long run is 25 feet. We control the current ramp up rate with PWM to get a precisely predicted opening time which was far more critical to controlling detonation, horsepower and pollution.

the short closing time is important when you need a big injector big for full power so you need accurate short pulse width for idle and low loads

with PWM you need the diode or you will cook the zener, with PWM it is also less of an issue because the current in the coil is reduced so it closes faster anyway

never heard of any one controlling the ramp up, just the holding current

And you still haven't.

The purpose of the zener is the same as a diode across the coil.
The proposed zener clamps the back emf current faster than a parallel diode.
None of this is related to MOSFET turn-on, it is all for MOSFET turn-off.

Comparison from referenced topic:

[CaptDon]

I don't think anyone mentioned mosfet turn on time? I didn't. They are plenty fast, no problem. We had to control the current ramp up time do to the significant variation in cable length going to the injectors as well as the resistance variation going through multiple connectors from the ECU itself, the injector itself as well as several bulkhead connectors. Bosch gave us a MIN and MAX ramp-up trajectory to reach 26 amps cracking, then the 19 amp HOLD 1, and finally 13 amp HOLD 2 all of which are PWM controlled. At the end of the hold period PWM shuts off and the clamp diode catches the back EMF. Hoerbiger DID NOT control the ramp-up current trajectory and we had all kinds of injector problems including poor control of cylinder to cylinder exact timing. Simply slamming the injector with 55 volts steady state until peak specified current was reached and then PWM'ing the hold wasn't cutting it in a big way!! And the PWM solution brought about EMI penalties that also had to be solved.

[nigelwright7557]

Throw the zener out and use a flyback diode across coil.
Or a flyback diode in series with a zener and across coil.

A diode on its own tends to slow down coil turning off and can cause arcing.
The zener lifts the voltage across the coil and doesnt slow down relay turning off.

[srb1954]

First, your link is bad.
   https://www.eevblog.com/forum/projects/is-this-mosfet-driver-circuit-ok/msg2870144/#msg2870144

As to your question, the current through the zener would be its reverse voltage divided by its internal impedance.  For the 1N5369B mentioned the voltage is 51V and the impedance is 27Ω which would be I = V / R = 51V / 27Ω = 1.8A

This calculation is incorrect.

The quoted 27 impedance is the incremental resistance i.e. the slope of the voltage vs current curve as measured at the test current of 25mA. Under these test conditions that actual equivalent resistance is 51V/25mA = 2040 .

Attempting to use these resistance figures to calculate the Zener current in different circuit conditions is futile as the Zener impedance varies considerably over different operating currents. The correct way to determine the Zener current would be to read off the voltage vs current graph although this would not be very accurate due to production tolerances in the Zener voltage.

[eel]

Didn't expect so many replies this fast, thanks.

First, your link is bad.
   https://www.eevblog.com/forum/projects/is-this-mosfet-driver-circuit-ok/msg2870144/#msg2870144

As to your question, the current through the zener would be its reverse voltage divided by its internal impedance.  For the 1N5369B mentioned the voltage is 51V and the impedance is 27Ω which would be I = V / R = 51V / 27Ω = 1.8A

This calculation is incorrect.

The quoted 27 impedance is the incremental resistance i.e. the slope of the voltage vs current curve as measured at the test current of 25mA. Under these test conditions that actual equivalent resistance is 51V/25mA = 2040 .

Attempting to use these resistance figures to calculate the Zener current in different circuit conditions is futile as the Zener impedance varies considerably over different operating currents. The correct way to determine the Zener current would be to read off the voltage vs current graph although this would not be very accurate due to production tolerances in the Zener voltage.

If the calculation above is incorrect what would be the correct process for choosing a zener? Reading the voltage vs current graph?

[CMTan]

The zener diode or the diode approach is basically serves the same function of limiting the induced voltage, but it is a trade off you will need to consider

1. Speed of injector turn-off (affected by a single diode approach)
2. Risk of high voltage induction that would damage the NMOS

I would prefer to use a diode, because you can have better control of point 2 if you insert resistor in series with the diode which control the di/dt thus limiting the induced voltage which depends on the rate of induced current.  The value of the resistor determines the rate of change of current.  But do note that when a resistor is inserted, there will be induced voltage.

If you prefer to use zener diode than it is better for you to use TVS instead of zener, because zener are slow unlike TVS that are meant for transcient.

The third solution is to have both (diode and resistor) with TVS.  In this way, you can have full control of the rate of current change and yet set the upper limit of the induced voltage with the TVS.

[Terry Bites]

You wont get sufficent gate drive to fully turn the injector on properly. You need something like MCP1406/7. This will also significantly redcuce the turn off time- though its hard to believe electomagentic system is that sensitive. Gate protection with a properly rated TVS will save you future trouble.
Are you sure none of these easily available parts is a good substitute  nz.element14.com/c/semiconductors-discretes/fets/single-mosfets?st=irf101

[eel]

The zener diode or the diode approach is basically serves the same function of limiting the induced voltage, but it is a trade off you will need to consider

1. Speed of injector turn-off (affected by a single diode approach)
2. Risk of high voltage induction that would damage the NMOS

I would prefer to use a diode, because you can have better control of point 2 if you insert resistor in series with the diode which control the di/dt thus limiting the induced voltage which depends on the rate of induced current.  The value of the resistor determines the rate of change of current.  But do note that when a resistor is inserted, there will be induced voltage.

If you prefer to use zener diode than it is better for you to use TVS instead of zener, because zener are slow unlike TVS that are meant for transcient.

The third solution is to have both (diode and resistor) with TVS.  In this way, you can have full control of the rate of current change and yet set the upper limit of the induced voltage with the TVS.

Would a TVS be across drain to source as well or gate to drain?

[CMTan]

Treat the TVS like the zener.  Functionally identical, but the characteristic is different.  Same symbols too.  You just need to select the correct voltage and power rating for the TVS, just like you will do for zener.

[Andreas]

but am not sure how to calculate current through the zener to select the appropriate part?

Its not so easy.
It all depends on your inductive load of your Injector.

Lets assume you have a injector with ~12 Ohms and ~10mH inductivity. (may vary in your case).

the initial zener current on switching off is then the Injector current ~1.2A @ 14.4V.
So a Zener of 51V and 1.2A current would theoretically give a ~62W Zener diode.
But the zener is only loaded on switch off.
So you have to regard the switch off energy.
E = 0.5 x L x I x I  = 0.5 x 10 mH * 1.2A * 1.2 A = 7.2 mJ
At 6000 RPM and 4 stroke engine this happens 50 times / s giving a average power of 0.36 W.
So you need a zener with minimum 0.5W and pulse capability of 7.2 mJ (in warm condition).

The bad news: these pulse parameters are usually not specified for a normal zener.
So you will have to use a much larger zener as calculated for average power.

with best regards

Andreas

[eel]

Treat the TVS like the zener.  Functionally identical, but the characteristic is different.  Same symbols too.  You just need to select the correct voltage and power rating for the TVS, just like you will do for zener.

Excellent, I originally thought tvs but the thread I linked said zener so I assumed they knew something I didn't.

but am not sure how to calculate current through the zener to select the appropriate part?

Its not so easy.
It all depends on your inductive load of your Injector.

Lets assume you have a injector with ~12 Ohms and ~10mH inductivity. (may vary in your case).

the initial zener current on switching off is then the Injector current ~1.2A @ 14.4V.
So a Zener of 51V and 1.2A current would theoretically give a ~62W Zener diode.
But the zener is only loaded on switch off.
So you have to regard the switch off energy.
E = 0.5 x L x I x I  = 0.5 x 10 mH * 1.2A * 1.2 A = 7.2 mJ
At 6000 RPM and 4 stroke engine this happens 50 times / s giving a average power of 0.36 W.
So you need a zener with minimum 0.5W and pulse capability of 7.2 mJ (in warm condition).

The bad news: these pulse parameters are usually not specified for a normal zener.
So you will have to use a much larger zener as calculated for average power.

with best regards

Andreas

Beautiful, thank you that's the kind of explanation I was looking for, can I apply the same principal to a tvs?
I'll spec the tvs above whats calculated as a margin of error

[Andreas]

Hello,

main difference between TVS and Zeners is in specification.
Zeners are specified with operating voltage at a certain test current.

TVS are usually (not all) specified with a operation voltage where they shurely do not conduct. (the break through voltage is higher).
Aditionally with a maximum clamping voltage for a single 1 ms pulse (at room temperature and maximum 1 per minute).
Usually the parasitic capacitance is larger and thermal layout adapted to pulse peak capability.

with best regards

Andreas

[nigelwright7557]

No, a diode across the injector would absolutely ruin the injector closing time.  The idea is to allow the voltage across the transistor to increase to some safe limit so the magnetic field in the injector collapses as quickly as possible.

It's the same when using a relay, a diode across the coil will slow down action of relay and possibly cause arcing.
I use a diode and zener in series to limit back emf.