Author Topic: Overvoltage switch  (Read 9841 times)

0 Members and 1 Guest are viewing this topic.

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Overvoltage switch
« on: July 14, 2013, 05:01:04 pm »
For my current project I'm using a quad push-pull-driver with 5V supply. In each of the four outputs, there's a 5V suppressor diode and a polyfuse to protect the whole device against overvoltages. While this works, the suppressor diode only limits to ~6.7V and the polyfuse doesn't kick in immediately. Even worse, if one of the push-pull-outputs is switched high in this state, the 6.7V is connected to the internal 5V supply (via the high side FET) and thus reaches several ICs (CAN, MAX202) and this way also input pins of the CPU which are only 5V tolerant (max: 6V). The microntroller and other periperhals are supplied via a 3V3 linear regulator, so the problem is only the 5V supply.

As my quad push pull (Microchip TC4469) has additional negated inputs which I use as common enable/disable, I thought about disabling the outputs by pulling this disable line high if the supply voltage exceeds 5V.

I came up with a circuit using two BJT transistors and a Zener Diode (see appended screenshots).

The switch represents the pin of my microcontroller that I use for output enable/disable, Probe1 marks the net that will be connected to the disable pin of the quad dirver, the voltage on the right is the 5V supply voltage that is increased up to 6.7V due to external short to a higher voltage.

The "inactive" screenshot shows the case where the 5V supply is still 5V. As the micrcontroller switches the pin to ground to enable the quad driver's push-pull outputs, the disable pin (Probe1) is low and the quad driver would be active.
In the "active" screenshot, the 5V supply voltage is increased to 5.5V, so the PNP transistor switches through and the voltage at the base of the NPN transistor increases. So the NPN transistor switches through and applies the 3.3V potential to the disable pin (Probe1).

So does this make sense? Are there other approaches which need little space and effort (thought about using an OP/Schmitt Trigger, but I don't really have much space left).
« Last Edit: July 14, 2013, 05:02:36 pm by 0xdeadbeef »
Trying is the first step towards failure - Homer J. Simpson
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 19345
  • Country: gb
  • 0999
Re: Overvoltage switch
« Reply #1 on: July 14, 2013, 06:04:34 pm »
MCUs have built in diodes connected to Vdd so all you need to do is connect a high vale resistor (say 1M) in series with the input.
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #2 on: July 14, 2013, 06:24:33 pm »
Indeed I thought of this (maybe not 1M, but something like 20k), but there's nothing to be found in the LPC1768's data sheet regarding clamping diodes apart from the fact that most (but not all) of the pins are "5V tolerant" with a maximum allowed voltage of 6V. So I toyed with the idea of voltage dividers for all the 6 input pins, but all in all the problem stays that the 5V linear converter has a short circuit to 6.7V on its output and several ICs and one LCD which are not approved for 6.7V will be supplied with this voltage as long as there's a short circuit to e.g. 8V on one of the outputs and the polyfuses don't kick in.
Also e.g. the LCD's LED backlight would become brighter during an overvoltage condition etc.
So I really like the idea of switching off the quad driver and thus removing the short circuit to 6.7V from the linear converter's output.
Besides, I'd need to add 6 resistors for the 6 input pins in question, while for the proper (?) solution, I need only one Zener diode, two small BJT-Transistors and 4 resistors (the 100k is always there). As the diode takes the space of a resistor (and is around the same price), only the two transistors add costs and space, but one less resistor is needed. This makes the solution with series resistors even less attractive.
« Last Edit: July 14, 2013, 06:28:25 pm by 0xdeadbeef »
Trying is the first step towards failure - Homer J. Simpson
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 19345
  • Country: gb
  • 0999
Re: Overvoltage switch
« Reply #3 on: July 14, 2013, 06:53:05 pm »
You could use a crowbar circuit which short circuits the power if the voltage exceeds a certain point. When the zener's breakdown voltage is exceeded the SCR turns on and short circuits the power supply.

The polyfuse will protect against the SCR against the large current when it's activated.


http://www.onestopgate.com/gate-study-material/electronics/circuits/overvoltage.asp
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #4 on: July 14, 2013, 07:21:41 pm »
I'd prefer disconnecting over short circuit. And I certainly don't want to short circuit my internal 5V supply as this would also affect the (currently unaffected) 3V3 supply and lead to a system reset instead of just disabling the outputs.
So I'd really get some input on the circuit I suggested instead of discussing completely different ideas. Well, at least until someone tells me that my idea is completely stupid for some reason.
Trying is the first step towards failure - Homer J. Simpson
 

Offline c4757p

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: Overvoltage switch
« Reply #5 on: July 14, 2013, 07:27:06 pm »
Well, at least until someone tells me that my idea is completely stupid for some reason.

It's completely stupid because I didn't design it! :-DD

Looks good to me, but one thing I'd advise you to look at is the temperature coefficient of the Zener diode. You may find it kicking in a bit early if the temperature drifts. Doesn't mean it's unusable, but you may have to put some thought into temperature compensation.

Alternately, use a TL431 configured to 5V. Technically you shouldn't use them in positions where they don't always conduct current (at least 1mA IIRC), but in my experience they do work fine in this sort of configuration.

Edit: If you use TL431, cut R1 and R2 to 270 and 680. You can cut R3 too if you want consistent values.

If you use a low-voltage MOSFET like 2N7000(/2N7002) in place of Q1 (get rid of the base resistor, but add a resistor, about 10k, from gate to ground), the output signal makes a smoother transition and looks more "digital".
« Last Edit: July 14, 2013, 07:40:52 pm by c4757p »
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #6 on: July 14, 2013, 10:05:29 pm »
Using a 2N7000 sounds like a good idea. I'll need to order some of them to test this though.
About the TL431: I'm not so sure if temperature is all that important as the device will always be used indoor at "comfortable" temperatures, but shouldn't a 5V Zener be relatively temperature stable as the positive coefficient of the avalanche effect and the negative coefficient of the Zener effect should somewhat cancel each other at  ~5V?
E.g. a 1N5231B (5.1V) has a coefficient of 0.03%/°C. That's 300ppm/°C which sounds not so bad, does it?
Special references like a LM 4040 have a coefficient of 100ppm/°C. A LM336-Z5 would be another option. Bit cheaper. But I'm just not sure the effort's worth the additional cost.
Anyway, the TL431 is an interesting option. Yet wasting 1mA or using it outside the spec booth seems a bit "ugly".
« Last Edit: July 14, 2013, 10:33:57 pm by 0xdeadbeef »
Trying is the first step towards failure - Homer J. Simpson
 

Offline lewis

  • Frequent Contributor
  • **
  • Posts: 704
  • Country: gb
  • Nullius in verba
Re: Overvoltage switch
« Reply #7 on: July 14, 2013, 10:21:33 pm »
Is the overvoltage condition due to a transient on the load that the TC4469 drives, or will it be due to an issue with the supply?
I will not be pushed, filed, stamped, indexed, briefed, debriefed or numbered.
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #8 on: July 14, 2013, 10:32:05 pm »
Short circuit of the load to a voltage >5V (up to 30V). As I have suppressor diodes (nominal 5V) in the outputs, they will limit the voltage to ~6.7V (measured with ~50Ohm output resistance). However as the output driver is a push-pull driver, an active highside in the driver will short the 6.7V to the 5V supply.
Trying is the first step towards failure - Homer J. Simpson
 

Offline c4757p

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: Overvoltage switch
« Reply #9 on: July 14, 2013, 10:40:32 pm »
Running the TL431 like that should not be a problem, as it will begin to draw sufficient current (with my updated resistor values) before it's time to switch, avoiding the issue. I just tested it myself, and it also gives a cleaner switch than a Zener. Combine that with the MOSFET and the circuit actually works very well.
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #10 on: July 14, 2013, 10:45:47 pm »
Ok, thanks for the input. I'll need to order some parts then to test it in reality...
Trying is the first step towards failure - Homer J. Simpson
 

Offline DaveW

  • Frequent Contributor
  • **
  • Posts: 284
  • Country: gb
    • WattCircuit
Re: Overvoltage switch
« Reply #11 on: July 14, 2013, 10:48:47 pm »
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #12 on: July 14, 2013, 11:03:02 pm »
Hard to get and only SMD. Futhermore I'm not 100% sure if this fits my use case.
Besides, I thought about using a FET to disconnect the quad driver's supply from the 5V line. Then I realized that I could as well switch off the FETs inside the quad driver as the overvoltage only reaches the 5V supply through the driver's highsides.
Trying is the first step towards failure - Homer J. Simpson
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #13 on: July 17, 2013, 06:59:00 pm »
Ok, I did some measurements now but the results were not quite so good. In a static condition, the circuit worked like it should, but (as one could have guessed), I got quite a nasty oscillation. Each time the circuit triggered and switched off the quad driver, the voltage dropped and deactivated the protection, so the voltage rose again and so on. Even with some lowpass filtering, the mean voltage of that PWM-like oscillation easily reached >6V again.

So I got back to my original idea and placed a P-channel MOSFET in the quad driver's supply line (see screenshots).
I quickly reworked my test setup and used an IRF5305 as P-channel MOSFET (in the simulation, I had to use another one as I already used both allowed custom parts). This setup worked immediately. When applying a short circuit to power on the output with the highside switched on, the supply voltage rose to around 5.4V, then the overvoltage circuit switched off the P-FET, and the supply voltage dropped to 5V again. No oscillation there as the overvoltage circuit is not influenced by the switched off FET.
Trying is the first step towards failure - Homer J. Simpson
 

Offline Mr Smiley

  • Frequent Contributor
  • **
  • Posts: 324
  • Country: gb
Re: Overvoltage switch
« Reply #14 on: July 17, 2013, 07:07:30 pm »
Everything on the right of R3 should be on the left of R3 not  including the mosfet.

When your circuit trips, your turning off the supply to the 431, your 431 should be connected to the supply at all times and used to switch the mosfet further down the line.

 :)
There is enough on this planet to sustain mans needs. There will never be enough on this planet to sustain mans greed.
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #15 on: July 17, 2013, 07:19:12 pm »
I could move the supply of the TL431 to the 5V supply, but it doesn't really make much of a difference.
The supply is never really switched off as the body diode is always there. Also the overvoltage circuit only trips if the voltage on the right side is higher than on the left side, so the voltage on the right side is always >= 5V. So the TL431 is never without supply.
Trying is the first step towards failure - Homer J. Simpson
 

Offline c4757p

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: Overvoltage switch
« Reply #16 on: July 17, 2013, 07:21:46 pm »
Methinks your MOSFET is backwards...

And yes, the control circuitry should not be trying to switch itself.
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #17 on: July 17, 2013, 07:32:13 pm »
The MOSFET is perfectly right "backwards". That's about the idea of it. And as I said, the TL431 is never unsupplied. I could move it to the other side, but that has not much influence.
Trying is the first step towards failure - Homer J. Simpson
 

Offline c4757p

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: Overvoltage switch
« Reply #18 on: July 17, 2013, 07:57:08 pm »
Maybe I read your schematic backwards. Which side is power input and which side is output?
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline 0xdeadbeefTopic starter

  • Super Contributor
  • ***
  • Posts: 1570
  • Country: de
Re: Overvoltage switch
« Reply #19 on: July 17, 2013, 11:05:48 pm »
5V supply is on the left. The switchable voltage on the right is the overvoltage that comes from an external short circuit to power that propagates through the highside of my quad push-pull driver.
The P-FET (including its body diode) acts as diode with a switchable bypass. In case of normal operation (PNP is high-ohmic, gate pulled to ground -> V_GS=-5V), its activated channel is in parallel to the diode and thus the voltage drop is very small due to small R_DS (on). If an overvoltage is detected, the PNP transistor connects gate to source, so V_GS is ~0V and the FET's channel is deactivated. So only the body diode remains and it blocks the overvoltage on the right side from the 5V supply on the left side.

I appended a slightly improved version with TL431 on the "left" side with higher resistor values to get a more stable reference voltage during normal operation.
Trying is the first step towards failure - Homer J. Simpson
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf