EEVblog 1746 - Schottky vs PN Diodes & Measurement Traps

[EEVblog]

Demonstrating the difference between Schottky diodes and PN junction silicon diodes, and how your multimeter can influence your measurement when troubleshooting.
Answering an interesting Twitter question.

https://x.com/blind_via/status/2052223991839170919
https://www.tti.com/content/ttiinc/en/resources/blog/infographics/schottky-diodes.html
https://engineering.purdue.edu/~ee606/downloads/ECE606_f12_Lecture17.pdf

00:00 - Twitter question
01:21 - Recreating the problem
03:08 - Why doesn't it read zero volts? Let's try a PN junction diode
04:21 - Schottky vs PN Junction diodes, they are NOT the same!
06:06 - What about a diode tester?
07:15 - Why doesn't the Schottky diode work like a diode?
08:39 - It's a multimeter measurement TRAP!
10:37 - A closer look at Schottky diodes
14:21 - Datasheet inspection, and advantages vs disadvantages of Schottky diodes
15:21 - Reverse leakage current, PN vs Schottky
17:16 - THREE orders of magnitude difference!
22:29 - Digital vs Analog multimetrer, Analog WINS!

[isometrik]

I got bitten by the Schottky reverse current level issue in the past when I replaced the standard diode by a Schottky in an opamp based "ideal diode" circuit similar to the image below.



I spent a long time figuring out what was the problem, as I came up with all sorts of twisted theories that attempted to explain the misbehaviour... that obviously did not consider the Schottky,  the lesson has permanently been imprinted in memory once I realized the Schottky was the problem. 

[golden_labels]

Going back to the original surprise of seeing voltage on “disconnected” part of the circuit: there shouldn’t be any, even if that wasn’t a Shottky diode.

It’s convenient to approximate switching operation of diodes and transistors with a break. But unless there really is a physical hole between conductors, there is no break. It’s only a very high resistance. Megaohms, gigaohms range. But still a resistance. And that means voltage “passes” to the other side. Without additional load it has to be equal on both sides.

This route takes us back to the normal p-n diode, but from opposite direction. We may see the voltage it “passes” to be near 0 V, because the multimeter is a load that pulls anode to the ground. If the multimeter was 1 GΩ instead of 10 MΩ, the reading would be very different too.

[David Hess]

The actual "resistance" of a diode at zero volts can be quite high.  Leaky gold doped switching diodes can be only 30 kilohms or so, but schottky diodes can be even lower.  I have some transistor junctions in the teraohm range.

[ballsystemlord]

The actual "resistance" of a diode at zero volts can be quite high.  Leaky gold doped switching diodes can be only 30 kilohms or so, but schottky diodes can be even lower.  I have some transistor junctions in the teraohm range.

Tell me more about your transistors with teraohm impedance please. I'm curious.

[ballsystemlord]

@Dave , I'd love to see a follow-up video on this like you proposed about how diodes break down.
I'd also be interesting if you'd cover PIN diodes and zener diodes as well.

Thanks in advance.

[golden_labels]

To not confuse beginners: we are not talking about ohmic resistance, the inherent property of a conductor. We’re just talking about relation between voltage and current.

David Hess: to be honest I was not trying to be accurate with values. My goal is to give another perspective on the same problem. A mirror image of what Dave presented.

Yes, at exactly 0 V it certainly is very high: about infinity! But looking at 1N4001…7 datasheet, it should be like 25 to 200 MΩ beyond -1 V. So that was a good guess.

[Kleinstein]

One uses schottky diodes in switched mode supplies also because of the speed.  There are also fast PN diodes, but usually schotty diodes are still faster.
This is the reverse recovery effect that is also a trap for beginners, that could also be interesting to demonstrate (e.g. scope and square wave generator).

PN junction diodes can also have quite some variations.
The 1N4148 as a fast diode is actually relatively leaky for the PN diode - the gold doping used to make PN diodes faster also causes more leakage.
Slow PN diodes are usually lower leakage, but size of cause also matters.

The 1N400x (not the UF400x variants) are usually not that bad with leakage, though it may vary a little between manufacturers.

[David Hess]

To not confuse beginners: we are not talking about ohmic resistance, the inherent property of a conductor. We’re just talking about relation between voltage and current.

David Hess: to be honest I was not trying to be accurate with values. My goal is to give another perspective on the same problem. A mirror image of what Dave presented.

Yes, at exactly 0 V it certainly is very high: about infinity! But looking at 1N4001…7 datasheet, it should be like 25 to 200 MΩ beyond -1 V. So that was a good guess.

The resistance does not have to be high at zero or very low voltages.  The example that comes to my mind are differential pairs where the collectors are tied together with a pair of anti-parallel diodes for clamping.  Since the differences in the collectors are only millivolt and lower voltages because of feedback, the diodes should have no effect and gain should be extremely high with a high impedance collector load.

If gold doped (leaky) diodes are used, then the diode resistance around zero volts will dominate the gain, and the resistance of the diodes at zero volts can be measured with such a circuit by finding the open loop gain.  In this application, low leakage diodes must be used.

The actual "resistance" of a diode at zero volts can be quite high.  Leaky gold doped switching diodes can be only 30 kilohms or so, but schottky diodes can be even lower.  I have some transistor junctions in the teraohm range.

Tell me more about your transistors with teraohm impedance please. I'm curious.

I needed some low leakage diodes to repair multimeter front ends, so I designed and built a quick and dirty transimpedance (current to voltage) amplifier with a noise limited resolution of about 10 femtoamps.  (1) I then went through my collection of small signal transistors for collector-base junctions with low leakage at about 2/3rds of their collector-base breakdown voltage rating.  As expected, I found a lot of 50 and 200 milliamp NPN and PNP parts with leakages between 2 and 0.5 picoamps.  But what surprised me are a batch of BC337-40 (50 volt 800 milliamp) parts with leakages below 100 femtoamps, somewhat defying my attempts to measure them.

(1) I have done this before in a work setting, but not quite at this level.  Besides using a shielded enclosure, I have some ideas for considerably improving the performance even without better parts.

[MathWizard]

Dr. Behzad Razavi has a good 1st course online, on semiconductors and devices, I haven't gotten to exotic diodes yet, or reverse breakdown in detail.

So yeah I didn't remember Schottky diodes have much higher leakage, I guess that's bad news for a low power device on a battery. I wonder what the calculus would be, for when having the lower power loss on the lower forward voltage, would cancel any long term leakage losses. Or in switch-mode circuit when it's always going forward/reverse, and you have the recovery time to worry about for reasons I can't remember.

I just measured some diode, I can't remember the # of, from a SMPS, it's not much bigger than a 1N4007, but it has a leakage I of ~150uA vs ~6-7nA for a 1N4148. Other than my lowest 10pA range is fudged up somehow, I must try measuring some op-amp and JFET gate leakage currents from my best Digikey parts.

My Electrometer uses fairly rigid triax cables, and I just made a 1st-pass at a basic test jig on a protoboard, and what a difference that makes already. No more mangled leads from the grabber hooks, or PSU clips, etc.

But I did use big pieces of female pin headers, so  I suppose it will have a lot of leakage and capacitance the way I solder it together.

I need to order some proper hardware, and look up a good version to make in a big tin cookie can, w/ drilled holes and even various BNC type connectors.


Also, what textbook full of equations, would be considered the 'bible'/holybook on SMPS's and their parts ??