Shop OBEX P1 Docs P2 Docs Learn Events
Output pin protection — Parallax Forums

Output pin protection

I have a Prop output pin that I need to protect against accidental contact with +12V. Because it is an output and I need drive capability it's not feasible to use a series resistor large enough to limit the current through the internal protection diode to a safe value.

My plan is to use a 56 ohm series resistor and a schottky diode from the output pin to +3.3V. The series resistor would limit the current, in the case of applied 12VDC, to about 150ma, well within the rating of the diode.

The Schottky diode would essentially be in parallel with the built in protection diode. My thinking is that as long as the forward voltage of the external diode is less than the forward voltage of the protection diode, the external diode will clamp the voltage to a safe level.

Does this sound reasonable?

Comments

  • As I understand your connection is as in my attachment.
    Suppose your output is connected to 12V. The direct connection between 12V and your output pin will fry your propeller. After that the 155mA in to the output from the voltage regulator will give you one firework more.
    And finally you will drive your load with 12V instead of 3.3V!
    I can’t see any way to protect an in/out pin on the propeller against a direct hit by 12V.
    Can you provide a drawing of your output connection to the load?
    1185 x 581 - 182K
  • Cluso99Cluso99 Posts: 18,069
    Just use an npn transistor to switch the load
  • jmgjmg Posts: 15,173
    jstjohnz wrote: »
    Does this sound reasonable?

    No!!
    What stops your 3v rail being pulled up by that 155mA you hoped to clamp ?
    3V regulators are not batteries, and clamping ones can be found, but they need special design (and so cost more)

    Look for DDR termination regulators, for ones that can Source and Sink as required.


  • There are "crowbar" circuits that short circuit the fault current to ground. It would trip at about 0.1V above Vdd. The current must be limited either by the the source (your 12V source) or by blowing a fuse (regular or resettable).

    You say you need drive capability from the Prop output. How much, how so?


  • kwinnkwinn Posts: 8,697
    jstjohnz wrote: »
    I have a Prop output pin that I need to protect against accidental contact with +12V. Because it is an output and I need drive capability it's not feasible to use a series resistor large enough to limit the current through the internal protection diode to a safe value.

    My plan is to use a 56 ohm series resistor and a schottky diode from the output pin to +3.3V. The series resistor would limit the current, in the case of applied 12VDC, to about 150ma, well within the rating of the diode.

    The Schottky diode would essentially be in parallel with the built in protection diode. My thinking is that as long as the forward voltage of the external diode is less than the forward voltage of the protection diode, the external diode will clamp the voltage to a safe level.

    Does this sound reasonable?

    Do keep in mind that the propeller output pin is rated for about 30mA. For maximum safety the best choice would be an optically isolated driver circuit. Post a schematic of what you are driving and you may get better suggestions.
  • My vote for the NPN open-collector approach - most small NPN's can handle 40V or more. Use the 56 ohm
    resistor and make sure the NPN can handle 250mA. The 56 ohm resistor will act as a fuse then!
  • Since the OP did not specify whether his pin is a high-side drive, low-side drive, or both, or whether the pin also has to be used as an input, none of us can say anything useful here that's not just speculation.

    -Phil
  • I am very fond of the TC4427 and related parts. SOIC-8 package will switch two circuits, up to 1A, with output voltage up to 18V. I use them as level-translators/signal conditioners (e.g. servos pulses) all the time.
  • feng wrote: »
    As I understand your connection is as in my attachment.

    Sorry for the limited and poorly worded description. The idea is to put en external diode essentially in parallel with the prop's built-in protection diode, from the output pin to Vdd. By using a diode with a lower forward voltage than the built-in diode, the external diode would be the part handling the current. Basically a work-around for the 500us limit of the built-in diode.

    Good point about the fault voltage bringing up the +3.3v rail.


  • jstjohnz,

    What are your other pin requirements, though? High drive, low drive, both -- and input also? Answers to those will help to determine the optimal solution.

    -Phil
  • jmgjmg Posts: 15,173
    jstjohnz wrote: »
    Good point about the fault voltage bringing up the +3.3v rail.

    Provided you design for that (ie use a Current Sink/Source regulator!) what you describe can work, and could be useful if you need fast Bi-directional IO, or do not want to loose General IO choices.

    The 56R can also be replaced with a small PTC - USB ports often have such resettable fuse resistors, that are really just PTCs. Littlefuse have many.


  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2016-02-08 05:47
    Personally I don't like "crowbar" techniques. I think they are wasteful and much like a band-aid only fixing part of the problem.

    The circuit below uses an Op-Amp configured as a voltage follower encompassing a diode in the feedback. This way the diode acts as a perfect diode and there isn't a diode drop seen on the output. Consequently virtually any diode can be used. The first transistor in parallel with the diode serves to allow the GND or LOW to flow towards the output. The remainder of the circuit, *The protection circuit", utilizes a 3.3V zener an NPN transistor and two resistors. The Zener and 47k resistor form a trigger if the voltage at the input exceeds the Zener voltage. That trigger turns ON the NPN transistor which in turn turns OFF the other NPN transistor in parallel with the diode.

    Note: The NPN transistors and Diode can all be generic 2n3904, 2n2222, etc, and any generic general purpose diode. As far as the Op-Amp, there are several to choose from that provide decent current drive, or you can apply external techniques to increase the amount of current drive.

    bscircuitdesigns.com/Misc/Over%20Voltage%20Protection.pdf
  • jmgjmg Posts: 15,173
    Personally I don't like "crowbar" techniques. I think they are wasteful and much like a band-aid only fixing part of the problem.
    Properly designed, they can work very well.
    They are very simple, and allow bidirectional operation.

    To make a design even more rugged, you can use a PTC, and these are very popular - look at some of the stocking volumes at Digikey

    http://www.digikey.com/product-search/en/circuit-protection/ptc-resettable-fuses

    Cheapest look to be 1206, but you can get 0805 and 0603 for reasonable prices.

    If doing a lot of pins, another alternative to Schottky diodes is the TVS Diode Arrays used for USB etc, if used with a lower sinking supply regulator, the VF is compensated for when clamping.

  • That's clever Beau. But to echo--We really don't know what the OP is doing with the output.

    I use the Harris/Littlefuse SP720 crowbar in its 16-pin package to cover as many a 14 pins, both input and output. It works best if there can be additional resistance, even as low as 100Ω, in series with each pin and with the crowbar itself. I have the high side crowbar feed into the emitter of a PNP power transistor so that the high side fault is shunted to Vss rather than to the Vdd rail. That way the voltage regulator can be the standard kind that can only source but not sink current, even for micropower operation.
  • This particular application is sending a 1 mb/sec data stream through several hundred feet of coaxial cable. I'm looking for a simple, low cost protection method for the output pin.

    The prop output pins make very good line drivers as a result of their low nice source and sink capability. The problem is the susceptibility to damage from accidental application of voltage > Vdd.

    I was hoping the external diode would be an option, and I think it still is if the issue of holding Vdd is dealt with.

    The Sp720 looks interesting also. Thanks for the suggestions and comments.
  • Since you need source and sink drive, but no input capability, I would go with Jon's suggestion:

    TC4427 or, equivalently
    MIC4427

    -Phil

  • jmgjmg Posts: 15,173
    jstjohnz wrote: »
    This particular application is sending a 1 mb/sec data stream through several hundred feet of coaxial cable. I'm looking for a simple, low cost protection method for the output pin.
    The prop output pins make very good line drivers as a result of their low nice source and sink capability.

    Wow, that's a lot of Cable, have you actually tested that length and data ?

    You may need proper cable drivers / receivers, and even some pre-emphasis, over that length.

  • For something that exposed to the outside world and possible faults, I'd inclined to use a good measure of isolation between that and the Prop. For example, RS485 drivers are designed to withstand 12V or more applied to the long line and they are more or less designed for fault tolerance.

    That said, the Prop does indeed have good drive capability with about 30Ω output resistance. A resistor pi network at the output could lower the voltage some while still driving it at 50Ω, and the resistors would allow for better protection. If your signal is indeed only at 1MHz, you could even insert a suitable capacitor in series to pass the signal, and that would block the 12Vdc fault. That is often done on RS485 lines to avoid large DC currents.

  • If your signal is indeed only at 1MHz, you could even insert a suitable capacitor in series to pass the signal, and that would block the 12Vdc fault.
    I might still be afraid of that transient spike when the 12V is connected, though.

    -Phil
  • jmgjmg Posts: 15,173
    If your signal is indeed only at 1MHz, you could even insert a suitable capacitor in series to pass the signal, and that would block the 12Vdc fault.
    I might still be afraid of that transient spike when the 12V is connected, though.

    True, some clamping is always needed.

    I'd tend to use a proper driver, but it is still interesting to see what a chip alone could manage.

    With the Prop, you can connect pins in parallel, for more drive, and with a CAP coupling, you could auto-bias in input mode, to tolerate some cable loss.
    Given how the ADC works, I'd guess 500mV of centred signal would receive ok. (no DC component)
    Might work best in Duplex, as half-duplex would need time to bias the cap correctly on direction change.

  • jmg wrote:
    With the Prop, you can connect pins in parallel, for more drive, ...
    In theory, yes. But I'm always leery of any hardware circuit that might allow a software issue to cause damage. This is especially the case where multiple cogs can control the same pins.

    -Phil
  • jmgjmg Posts: 15,173
    In theory, yes. But I'm always leery of any hardware circuit that might allow a software issue to cause damage. This is especially the case where multiple cogs can control the same pins.
    Usually, but this OP example seems to be one where 'potential damage' is less of a concern ;)

    one detail we did find useful on older MCU designs with Cable/ESD events, was to include the Pin-mode config inside the main loop.
  • What I was thinking about was an additional resistor network like the attached.

    The extra resistance would make it much easier to protect against transients that might come from the far end. Additional protection could go in at the junction of the 70 and 100Ω resistors. Makes the load easier on the Prop, power supply loading etc. I'm assuming the Prop output impedance = 30Ω resistive. With the network, the output is 1.65V peak to peak instead of 3.3V, but still 50Ω. Back of envelope, the impedance of the 10nF capacitor is less than 1Ω at 1MHz.

    I don't know what goes on the other end of the cable, haven't heard that it is another Prop, or what drive levels it really requires.





    343 x 138 - 6K
  • jmgjmg Posts: 15,173
    What I was thinking about was an additional resistor network like the attached.

    The extra resistance would make it much easier to protect against transients that might come from the far end.
    Yup, the OP already mentioned 56 Ohms placed in series.
Sign In or Register to comment.