Post by thread:[EE] Using Shottky Diodes as 'Overvoltage Protecti

So, I have a shunt input on a device I'd like to protect from human stupidity.... Basically +-100mV input, from a shunt resistor which is typically 1 ohm or less... And it occurs to me that the maximum usable expected forward voltage is around 165mV on this... ever (the ADC won't read more than that due to the 20V/V gain). I happen to have B1100 schottky diodes (1A 100V) in vast supply, since they are basically considered a jellybean part. It also occurs to me that the forward voltage on such a diode is around 0.4V, according to the datasheet. So obviously I immediately come up with the idea of just putting a pair of these - one each way - across the input (after some current limiting resistors). Two questions: 1) Reverse Leakage is rated for like lots of reverse voltage... aka 100V.. Intuitively it would seem this would be less at a few mV... Is this the case? I have a feeling this might be the killer, especially with current limiting resistors in place. 2) Is there some other thing that I'm missing? -forrest

To reply to my own post... I also have bat54s'es as part of the jellybean collection. Their reverse leakage rating is *much* lower (2uA instead of the multiple mA rating of the B1100's).... so this might solve that problem, however, looking at the forward voltage curve, it looks like if the forward current is close to zero - so is the forward voltage... If this is similar, I might just have to put a lower-voltage tvss diode on this input. It should be noted that this drives a high-side instrumentation amp (INA196), which has a lowish common-mode input absolute maximum rating. -forrest On 8/16/2011 7:35 AM, Forrest Christian wrote: {Quote hidden}> So, I have a shunt input on a device I'd like to protect from human > stupidity.... Basically +-100mV input, from a shunt resistor which is > typically 1 ohm or less... > > And it occurs to me that the maximum usable expected forward voltage is > around 165mV on this... ever (the ADC won't read more than that due to > the 20V/V gain). > > I happen to have B1100 schottky diodes (1A 100V) in vast supply, since > they are basically considered a jellybean part. It also occurs to me > that the forward voltage on such a diode is around 0.4V, according to > the datasheet. So obviously I immediately come up with the idea of > just putting a pair of these - one each way - across the input (after > some current limiting resistors). > > Two questions: > > 1) Reverse Leakage is rated for like lots of reverse voltage... aka > 100V.. Intuitively it would seem this would be less at a few mV... > Is this the case? I have a feeling this might be the killer, > especially with current limiting resistors in place. > > 2) Is there some other thing that I'm missing? > > -forrest > >

On 17 August 2011 01:43, Forrest Christian <spam_OUTforrestcTakeThisOuTimach.com> wrote: {Quote hidden}> To reply to my own post... > > I also have bat54s'es as part of the jellybean collection. Their > reverse leakage rating is *much* lower (2uA instead of the multiple mA > rating of the B1100's).... so this might solve that problem, however, > looking at the forward voltage curve, it looks like if the forward > current is close to zero - so is the forward voltage... > > If this is similar, I might just have to put a lower-voltage tvss diode > on this input. It should be noted that this drives a high-side > instrumentation amp (INA196), which has a lowish common-mode input > absolute maximum rating. > > -forrest > > On 8/16/2011 7:35 AM, Forrest Christian wrote: >> So, I have a shunt input on a device I'd like to protect from human >> stupidity.... Basically +-100mV input, from a shunt resistor which is >> typically 1 ohm or less... >> >> And it occurs to me that the maximum usable expected forward voltage is >> around 165mV on this... ever (the ADC won't read more than that due to >> the 20V/V gain). >> >> I happen to have B1100 schottky diodes (1A 100V) in vast supply, since >> they are basically considered a jellybean part. It also occurs to me >> that the forward voltage on such a diode is around 0.4V, according to >> the datasheet. So obviously I immediately come up with the idea of >> just putting a pair of these - one each way - across the input (after >> some current limiting resistors). >> >> Two questions: >> >> 1) Reverse Leakage is rated for like lots of reverse voltage... aka >> 100V.. Intuitively it would seem this would be less at a few mV... >> Is this the case? I have a feeling this might be the killer, >> especially with current limiting resistors in place. >> >> 2) Is there some other thing that I'm missing? >> >> -forrest >> >> Check reverse leakage with temperature. Schottky diodes can get very leaky at even moderate temperatures. RP

On Tue, Aug 16, 2011 at 8:35 AM, Forrest Christian <.....forrestcKILLspam@spam@imach.com> wrote: > So, I have a shunt input on a device I'd like to protect from human > stupidity.... Basically +-100mV input, from a shunt resistor which is > typically 1 ohm or less... > > And it occurs to me that the maximum usable expected forward voltage is > around 165mV on this... ever (the ADC won't read more than that due to > the 20V/V gain). > > I happen to have B1100 schottky diodes (1A 100V) in vast supply, since > they are basically considered a jellybean part. It also occurs to me > that the forward voltage on such a diode is around 0.4V, according to > the datasheet. So obviously I immediately come up with the idea of > just putting a pair of these - one each way - across the input (after > some current limiting resistors). > What is your input circuit like? How many volts will it take to damage it? It seems like you are clamping too low, and might have accuracy problems due to leakage. Maybe the current limiting resistors are sufficient, with clamps to supply rails. Regards, Mark markrages@gmail -- Mark Rages, Engineer Midwest Telecine LLC markragesKILLspammidwesttelecine.com

On 8/16/2011 11:49 PM, Mark Rages wrote: > What is your input circuit like? How many volts will it take to damage > it? It seems like you are clamping too low, and might have accuracy > problems due to leakage. Maybe the current limiting resistors are > sufficient, with clamps to supply rails. Regards, Mark markrages@gmail I provided a bit more information in my other post, but in short this is going to a current sense amplifier with a vss to input (common mode) range of many volts (like 80). The differential input voltage is a lot lower (like 18V), and is also a lot more likely to be accidentally exceeded due to things like a wiring error. The correct way to do this looks like a couple of series resistors, and some sort of bidirectional TVSS diode across the input, say at 12V or less. Clamps to supply rails won't work because of the high common mode input range. I basically have to clamp across the input. -forrest

If polarity can be guaranteed, a super-low-threshold MOSFET with gate connected to drain would suffice. Reverse polarity sees body diode so probably still OK. Put P & N channel back to back for bidirectional protection. Check datasheets to see if off and reverse leakages are good enough. P Channnel. 400 mV Vth at 100 uA gate to drain http://www.onsemi.com/pub_link/Collateral/NTJS3151P-D.PDF N Channel 400 mV at 1 mA http://www.nxp.com/documents/data_sheet/BSH103.pdf Neither looks very convincing at 400 mV as a FET but they will probably do a good enough job. Somebody makes 0 volt gate MOSFETs - or whatever other voltage you want. Buried gate is charged to bias FET to whatever ste is required. Russell McMaho