Schottky diodes have higher V drop than PN?

Small Schottky's tend to be resistive.

...Jim Thompson

because they're small.

John

Thanks for the (brief) reply ;-)

What do you make of the difference between the BAT54 models?

Oh, I am not sure if you have LTSpice, though surely you must by now just for it's utility as the "PDF" of electronics forum discussions.

Good day!

Sure I have LTspice, though I use PSpice for work.

Schottky's are rarely modeled properly, because the model framework is built around junction-style devices, and the model makers don't know how to twiddle them to fit Schottky slope differences ;-)

...Jim Thompson

I hope you're talking pulsed current here. Fairchild specs theta at

John

That's why you should look at the LTSpice file I posted which models a typical ESD scenario. It gives 9-10A surge currents through the diodes. What can you do to protect from that except make some assumptions about the surge capability of the diodes at very short pulse widths.

So the question becomes, as the time duration of the surge current becomes smaller, how much more current can be tolerated.

I was given some very old Philips databook images for HC logic, which showed a graph of this. Now I don't know where it got to.

With any sort of input resistance that still allows fast data rates, one has little choice but to exceed the 1 second ratings of the diode. Although at least for the MMBD4148, they give a 1us rating of 2A.

But by putting a resistor in between the external diodes and the internal, as I said before, one can ensure that the chip's ratings aren't exceeded.

I sure wish the diode datasheets included more detailed surge ratings, such as a graph.

Good day!

It's about ESD.

See these:

Except that 'HC' parts don't use Schottky diodes for ESD ;-)

...Jim Thompson

Hello Chris,

BAT54 are resistive but not a whole lot more than a 4148. Bottomline, if you get much more than 450mV typical at 25C and 100mA for the BAT54 something must be wrong with the model.

Remember, this kind of diode isn't meant for several hundred mA. They can stomach non-repetitive peaks of half an amp but not several amps.

Regards, Joerg

Hello Chris,

Ok, just did. V(N004)-V(N005) (1N4148) gives 7V peak while the V(N010)-V(N011) (BAT54) results in 22V. Whether the BAT54 model is any good or not, with that 1.5k up front chances are that either diode will be toast. With such low resistance you need something larger. Even via the extra 100Ohms towards the substrate diode it'll still see 70mA. Too close for comfort for me.

If you need this much you'd have to ask the mfg but they most likely won't endorse it.

But I'd go higher than 100Ohms there.

We are lucky we have those graphs at all. Try to coax that kind of data out of uC manufacturers. I tried but it was like kicking a big oak tree.

Regards, Joerg

Jim, the debate is about which is better to put external to the HC device, or other device. My LTSpice model uses 4148 style diodes internally. I might not mind if you had any inside tips on the models of HC input diodes that you could share with us? I choose in my LTSpice model whether to use Schottky vs. PN external protection diodes depending on what the goals are (speed, etc.). The model also allows varying the ESD charge and voltage, various resistances, inductances, etc.

The debate has now shifted to whether the BAT54 models available are any good or not. The ones that I have tried predict much higher voltage drops than 4148 at the large albeit brief surge currents generated during an ESD event using the IEC ESD models.

Thanks for the input.

Good day!

It's >200mA with the BAT54 model, which is of course still in question. Odd, I get only about 57mA peak with the 4148.

Anyway, the RMS current over 1us for the internal diode is 20mA, which is the max allowed constant current. Note that the 4148 datasheet indicates that the surge current allowed for 1us is toward 2x the surge current allowed for 1s. Thus a safe assumption might be that the internal diodes might handle a similar 2x there constant current rating, for 1us. Thus, a safe assumption might be that if we get the peak current down to 40mA, then the 1us RMS will be less than that, let's see: With 150R before the internal diode (with 15kV and 1.2k external), I get 38mA peak internal diode current.

To avoid making ANY assumptions about the internal diode, and limit the current to 20mA peak, then 300R does the trick.

For the external, 3.9k limits to So the question becomes, as the time duration of the surge current

Yeah.

Ok, so I'd sure like to learn a better way to avoid exceeding any max current ratings while making no assumptions whatsoever (thus peak currents must be

Typical ESD for 'HC'... can vary from part-to-part and whether the device can stay on a bus with the power off...

Series resistor from input pin to circuit and ESD

Upper diode is the body diode of a PMOS.

Lower "diode" is an NMOS device, source AND gate grounded, drain to circuit-side of input resistor.

There is often some fancy additions to help prevent latch-up.

For a situation where the input comes from "off-card" an additional series resistor is all that is generally needed.

I haven't studied the models in question, but Schottky's are low drop only for moderately low currents.

...Jim Thompson

But at least the datasheets specifically state it can tolerate the IEC ESD standards.

This leads me to wonder, since so many semi datasheets state that the devices are not intended for use in life-support systems, etc., how do you medical equipment designers find any parts at all? Or do you make special legal arrangements with the manufacturers to absolve them of any liability then they'll sell you whatever you want?

Good day!

I looked at all the 1N4148 datasheets I have, and only one, the original GE one, shows any information about surge capability. The GE sheet has a transient thermal impedance curve, from which you can derive the surge capability. Shall I post it over on ABSE?

Hello Chris,

I was just eyeballing the 70mA. But 57mA is equally dicey to jam into an HC input.

Maybe but with currents of this magnitude I'd pick a large diode pair.

Since I work mostly in med there either is mfg data in writing on that or I can't do it. Similar to aeronautics and other critical apps.

Pretty expensive. It may be better to select a beefier pair of regular diodes. Slim pickings though and companies like ON Semi do not list current ratings on their parametric tables, for whatever reason.

Regards, Joerg

Hey Chris, Funny, I'm using 74HC14 gates as input line receiver in an avionics job. The inputs are from 28V/GND relays so I use a 100K input series protection resistors and a lot of filtering. The customer does not believe it is rugged enough to handle 32V inputs. I say +/- 2KV or until the 100K resistor smokes but he's not buying it. What a life!

cheers, Harry

Hello Chris,

Well, yes, but there are less expensive ways (big diodes).

We don't do any life support designs without a waiver. Companies that do often produce their own semiconductors. Seriously, I have seen a company run their own transistor and logic chip production. They test the dickens out of every single part. Of course, a 10c chip will then become a $10 chip. Now you know why pacemakers are so freaking expensive ;-)

Regards, Joerg

Hello Harry,

If the customer doesn't know electronics that will require a lot of s'plainin'. Just like my EMI sermons. Takes forever to sink in. It's like talking to marketing folks although the results can be very different there. The first time I had to do that in my career resulted in marriage ;-)

Oh, and we are still married.

Regards, Joerg