Zener not working (the way I expect it to)

[bitman]

I'm testing a simple power-unit for a circuit I'm working on. I'm focused on testing the simple protections I've built into the circuit. You may recognize this from a prior post - Q1 is now different but the rest is pretty much the same. The zener is D2 - it's a 12V zener.

I'm attempting to test that over-voltage protection works.  When I measure on the output of Q1 (pin 2, source) - meaning the line to (1) on U2 - I always see the exact voltage I'm putting in - it's not capped at 12V. This was the whole idea, so something here isn't working (at least as I expect it to). When I measure the voltage potential on the anode side of the D2 zener, I see the voltage drop I would expect. Ie. if I feed 15V in, I see a voltage here of 3V. So the zener seems to be doing what it's supposed to be doing. But I'm not seeing the difference just using voltage potential measurements.

So I'm missing something - but I'm not sure what it is? I guess the potential will never change until there's a load?  (I'm measuring right now with no load at all).  I know the U2 can handle 15V - this is more an exercise for me to know how to do things "right" when I really need it.

[ledtester]

When I measure the voltage potential on the anode side of the D2 zener, I see the voltage drop I would expect. Ie. if I feed 15V in, I see a voltage here of 3V. So the zener seems to be doing what it's supposed to be doing.

The Zener is working - it's protecting the MOSFET by preventing the gate to source voltage from going past 12V. You want to make sure the gate-source voltage doesn't exceed 20V for this device.

Other than that, I don't see any other over-voltage protection. Once Q1 is turned on the voltage difference between (3) and (2) should be around 65 mOhms * the current through Q1. Since there is no load the difference will be zero.


DMP3099L datasheet: https://www.diodes.com/assets/Datasheets/DMP3099L.pdf

Update: Q1 is providing reverse polarity protection. The other option is to use a diode which would drop .6V even at low currents.  With Q1 the drop will be much smaller - e.g at 1A the drop will be around 1/10th the size.

[bitman]

When I measure the voltage potential on the anode side of the D2 zener, I see the voltage drop I would expect. Ie. if I feed 15V in, I see a voltage here of 3V. So the zener seems to be doing what it's supposed to be doing.

The Zener is working - it's protecting the MOSFET by preventing the gate to source voltage from going past 12V. You want to make sure the gate-source voltage doesn't exceed 20V for this device.

Well if that's the case, then I didn't build what I thought I was building. I wanted a simple zener to protect against over-voltage. I thought the resistor would help keeping the load down but it looks very much like I was wrong.

But I don't get how it protects the Mosfet against anything. In normal mode the mosfet is pulled to ground. So are you talking about the Vgs variance being kept within tolerances? I picked this one because it had a 20 Vgs but I still want to limit the circuit to 12V max.  Sounds like i need a second zener?  (and back to the books to understand why this setup did not do what I expected it to do).

Other than that, I don't see any other over-voltage protection. Once Q1 is turned on the voltage difference between (3) and (2) should be around 65 mOhms * the current through Q1. Since there is no load the difference will be zero.

Well, I thought the zener would do that even though it's connected to the gate too. With just a zener it seems like it would wear out quickly if the over-voltage is too large.  Granted, the risk of getting more than 24V is very small given the plug I'm using - most of my power supplies are 12V and below so it's is almost just academic for me. But that's the point - I'm trying to improve/learn and It hought I understood this simple protection, but alas obviously I don't.

Thanks for the quick reply. It's often better to fail than to succeed when it comes to progressing. Still, it's a bit annoying.

[ledtester]

What exactly do you want to happen if >12V is applied?

[bitman]

What exactly do you want to happen if >12V is applied?

Pretty simple - I've used a zener across the input before to do simple protection and over-voltage, straight out of a lot of tutorials like https://components101.com/articles/designing-an-overvoltage-protection-circuit-using-zener-diodes. But I don't like the little resistance the diode provides for that over voltage. I assumed the zener in the diagram I provided would do the same, but as you laid out it doesn't - well, as I measured too.

So in short - I would want the load voltage to be "regulated" in the sense that if the input power is too high it flattens out. At least avoiding high over-voltage like 9v when you expect 5v or in my case wanting to ensure that no voltage higher than 12v as my cooling/thermal needs were all calculated with 12V as a maximum.  So when 15V is fed in, the circuit past this only recieves 12V.

[T3sl4co1l]

Right, that's a polarity protection circuit (in the OP), not an overvoltage protection circuit.

What overvoltage are you protecting against?
What load current?

Tim

[ledtester]

That configuration is called a zener shunt regulator, but they are limited by the wattage of the zener which is usually around 1 to 2 watts. A 1W 12V zener, for instance, is limited to regulating 8mA 80mA of current.

Your original schematic shows a very hardy voltage regulator - the 7805. It can deliver 1.5A of output current, handle input voltages up to 25V, has current limiting and thermal shutdown protection "essentially make them immune to overload" (from the datasheet). The LM317 can handle input voltages up to 40V with the same current and thermal protection features. It also comes in a high-voltage version which works with input voltages up to 60V.

As long as your follow the guidelines in the datasheet a 78xx or LM317 regulator will provide a stable voltage to your application.

[BrianHG]

He just needs to add a PNP transistor into his circuit between gate and source, with the zener tied to the base with a 10k series resistor to GND and a 100k from base to emitter to get the over-voltage shut-off working.  The gate should still have the 100k pulldown.

2N3906 / MMBT3906 should do.

Correction, some zeners have quite a bit of leakage current, so on the transistor, change the 10k to a 1k and the 100k to a 1k.

Sorry, scrap that.  Read below T3sl4co1l's post.

[TimFox]

A 1 W, 12 V device can handle 80 mA, 8 mA is a simple mistake.

[T3sl4co1l]

He just needs to add a PNP transistor into his circuit between gate and source, with the zener tied to the base with a 10k series resistor to GND and a 100k from base to emitter to get the over-voltage shut-off working.  The gate should still have the 100k pulldown.

2N3906 / MMBT3906 should do.

Correction, some zeners have quite a bit of leakage current, so on the transistor, change the 10k to a 1k and the 100k to a 1k.

Check the direction of the body diode

The easiest way to add ride-through is a depletion NMOS, wired as a source follower, fed from a zener shunt regulator.  Downside, it doesn't scale up very big.  No clue how reasonable it is for this application.  OP hasn't stated source or load ratings.

Tim

[BrianHG]

He just needs to add a PNP transistor into his circuit between gate and source, with the zener tied to the base with a 10k series resistor to GND and a 100k from base to emitter to get the over-voltage shut-off working.  The gate should still have the 100k pulldown.

2N3906 / MMBT3906 should do.

Correction, some zeners have quite a bit of leakage current, so on the transistor, change the 10k to a 1k and the 100k to a 1k.

Check the direction of the body diode

The easiest way to add ride-through is a depletion NMOS, wired as a source follower, fed from a zener shunt regulator.  Downside, it doesn't scale up very big.  No clue how reasonable it is for this application.  OP hasn't stated source or load ratings.

Tim

Thanks, my mistake.  He could do it with 2 mosfets, 1 for polarity protection, 1 the other way around with my circuit for over-voltage protection.

Yes, depletion NMOS does make a few things easier.

[bitman]

That configuration is called a zener shunt regulator, but they are limited by the wattage of the zener which is usually around 1 to 2 watts. A 1W 12V zener, for instance, is limited to regulating 8mA of current.

Your original schematic shows a very hardy voltage regulator - the 7805. It can deliver 1.5A of output current, handle input voltages up to 25V, has current limiting and thermal shutdown protection "essentially make them immune to overload" (from the datasheet). The LM317 can handle input voltages up to 40V with the same current and thermal protection features. It also comes in a high-voltage version which works with input voltages up to 60V.

As long as your follow the guidelines in the datasheet a 78xx or LM317 regulator will provide a stable voltage to your application.

The final circuit uses both the 12V and the regulated 5V (the latter for a ESP8266 - the board it's on does provide 5v outputs).  The 12V must be able to handle up to 1.5A in bursts - it's connected to two high power LEDs over a mosfet and a PWM regulates light signals (for a display).  It's really odd - just a few hours before your post I actually did rethink the whole approach due to the low wattage - even if I go for 9V instead it's still 15W or there about. So I know I need to rethink this and thanks for making it clearer why   There's nothing secret about the other half - so I've enclosed it here; I just wanted to focus solely on why my protection circuit didn't do everything I expected it to do.

I'm also thinking my total design needs a rethink because my mosfet that controls the LEDs is just running way too hot for my liking - I really don't want to use active cooling.  For now I'll work on faster PWM which should lower the amps and hopefully make me able to make a closed enclosure for the control circuit.

[bitman]

The final circuit uses both the 12V and the regulated 5V (the latter for a ESP8266 - the board it's on does provide 5v outputs).

And the mistakes are piling up - that's wrong too. I definitely need to just regulate to 3.3v up front too. Not sure how I measured 5V on the output.  Well, this is what "prototyping' is for.  Thanks again for your suggestions.

[T3sl4co1l]

Replace each LED with three in series (they can be rated lower current, of course).  Use a series resistor to limit current, and PWM them for brightness control.

The LEDs won't mind the supply being momentarily higher.  Can even put a voltage sense divider from +12V to an ADC pin, and adjust PWM for the actual supply voltage.

Or you can still use the MOSFET and it'll dissipate just as much heat as the resistor (at nominal supply) but be independent of supply; it just takes more stuff to build.

Or you can use a switching LED driver, which is more efficient than anything, but is generally harder to use (good layout, maybe additional filtering, etc.).

Why are you worried about the supply again?

Tim

[bitman]

Replace each LED with three in series (they can be rated lower current, of course).  Use a series resistor to limit current, and PWM them for brightness control.

To repeat myself a bit - these aren't traditional LEDs - but power LEDs - 5W and about 1200mA with a forward voltage of 2.2V.  I've been able to run them at around 7-9V but find 12V provides the best control. At 12V for a single LED that would need a 10 Ohm (14.5W) resistor - I have two in parallel so that would be a 5ohm resistor (28W - not even sure I have a resistor that can handle that). Much much larger than the mosfet and it too would get rather hot.  I have a pot to help set an overall brighness (wattage) so help keep the power consumption down (and another thing my prototype shows me is that my calculations are off on the opamp setup so I can provide way too much power via the mosfet - I think this is partly my problem).

I need to control the switch - and I don't have any OPAMP that can drive that much current either so I went for the MOSFET (IRF510 in this case). If I keep the maximum current below 1A and generally avoid turning the light too full brightness for long periods of time, with a beefy heat sink I have a reliable mosfet running - meaning it levels out the temperature well below it's maximum tolerances. A code error however kept the brightness at full much longer than I  wanted it to, and at 1.5/2A it got well above 100^oC too close to comfort. And that's with a heat sink twice the size of the prototype PCB.

This needs to basically just flash in certain patterns - it is not meant to run the LEDs at full power for very long periods - just in very short bursts, so I'm hoping that with software setup I can adjust things than a smaller heat-sink can handle it.

The LEDs won't mind the supply being momentarily higher.  Can even put a voltage sense divider from +12V to an ADC pin, and adjust PWM for the actual supply voltage and I would definitely need active cooling.

Or you can still use the MOSFET and it'll dissipate just as much heat as the resistor (at nominal supply) but be independent of supply; it just takes more stuff to build.

Right - active cooling is definitely needed with my current design. Still, I'm still attempting to make passive cooling enough but as the above calculation shows, 28W is quite a lot of "cooling" to deal with.

And yes, he watts/power is the same regardless of device. However I can find heat sinks to handle that with a mosfet. I think my biggest power resister is 5W and it doesn't allow me to control/provide the PWM (again, I've not found using the OPAMP ideal if feeding the output straight into the LEDs).

Or you can use a switching LED driver, which is more efficient than anything, but is generally harder to use (good layout, maybe additional filtering, etc.).

That's pretty much what I'm doing right now? All these ICs do is have a hardware frequency and can adjust the PWM width based on this - my cheap Wemo does all of this in software - but in the end, it's a frequency to a gate (mosfet in my case).  And it too would drive the same watts if just inserted instead of the mosfet.

I did try a much higher frequency last night for my PWM and while it helped a bit it wasn't significant. I still need to find a way to keep the average load small enough not to heat things up too much.  Well, there is of course using a aluminum project box and use it as a heat-sink too - but now we're up in a huge size that I was hoping to avoid; and it feels like quite a big waste of power->heat this way.  When I started, I thought I only needed heat-sinks of the LEDs themselves - boy was I wrong

Why are you worried about the supply again?

My power/wattage -> temperature is based on a 12V feed. I want to ensure that isn't exceeded.

[T3sl4co1l]

To repeat myself a bit - these aren't traditional LEDs - but power LEDs - 5W and about 1200mA with a forward voltage of 2.2V.  I've been able to run them at around 7-9V but find 12V provides the best control. At 12V for a single LED that would need a 10 Ohm (14.5W) resistor - I have two in parallel so that would be a 5ohm resistor (28W - not even sure I have a resistor that can handle that). Much much larger than the mosfet and it too would get rather hot.  I have a pot to help set an overall brighness (wattage) so help keep the power consumption down (and another thing my prototype shows me is that my calculations are off on the opamp setup so I can provide way too much power via the mosfet - I think this is partly my problem).

Yeah that's why I asked, what about smaller LEDs in series?

And yes, he watts/power is the same regardless of device. However I can find heat sinks to handle that with a mosfet. I think my biggest power resister is 5W and it doesn't allow me to control/provide the PWM (again, I've not found using the OPAMP ideal if feeding the output straight into the LEDs).

Well that's the thing, with a resistor you only need a tiny transistor to switch it, you can get SOT-23s that handle several times that current.

That's pretty much what I'm doing right now? All these ICs do is have a hardware frequency and can adjust the PWM width based on this - my cheap Wemo does all of this in software - but in the end, it's a frequency to a gate (mosfet in my case).  And it too would drive the same watts if just inserted instead of the mosfet.

Nah. Switching regulator.  Needs an inductor and diode at least.  Transforms power, so you can run a single 2.2V 1.2A LED from 12V and under 300mA.

Why are you worried about the supply again?

My power/wattage -> temperature is based on a 12V feed. I want to ensure that isn't exceeded.

Okay but you just defined it as "12V", why would it ever be exceeded?

Tim

[ledtester]

You really should look at constant current drivers - there are tons of chips designed for that purpose.

Just some examples:

https://www.homemade-circuits.com/how-to-make-1-constant-current-led/

https://youtu.be/_5nCSPLxdAQ