EEVblog #908 - Zener Diodes

[EEVblog]

Fundamentals Friday.
A tutorial on Zener Diodes.
Avalanche breakdown, zener breakdown, zener effect, knee voltage, power dissipation, differences from regular diodes, calculating dropper resistor value, voltage regulation, clipping, clamping, transient surge overvoltage protection, characteristic curve, forward and reverse characteristics.

[KerryW]

Very nice introduction to Zener diodes, but you left out a BIG "trap for young players": Temperature compensated Zeners.

Temperature compensated Zeners have the same symbol as "regular" Zeners, but they DO NOT conduct in the forward direction!  So, for your AC clamp, you would need to put the Temperature compensated Zeners in parallel, not in series.  And in your protection circuit, they would not protect against negative voltage.  You would need a parallel diode for that.

[gemby]

I would also like to se some theory around zener and transistor regulators, for example: 

and


And how those regulators cope with modern lienar regulators in all spects, power discipation, speed, accuracy, power drop and so on.

[Wilksey]

If I had a 3v Zener acting as a protection device to a uC input, should I expect to see a voltage drop?

I had a voltage divider taking 16V down to 3V, I put this into a uC input, which worked fine, I thought about adding a ZD to protect if the input to the divider went above 16V but I was seeing a 0.3V drop with the ZD in circuit.  I didn't have a series limiting resistor, could that have caused the drop? It was used for V Monitoring rather than driving a current.

[Galaxyrise]

If I had a 3v Zener acting as a protection device to a uC input, should I expect to see a voltage drop?

I had a voltage divider taking 16V down to 3V, I put this into a uC input, which worked fine, I thought about adding a ZD to protect if the input to the divider went above 16V but I was seeing a 0.3V drop with the ZD in circuit.  I didn't have a series limiting resistor, could that have caused the drop? It was used for V Monitoring rather than driving a current.

Diodes never block current 100%; they have a "reverse leakage" that goes up as voltage increases.  Just under the breakdown voltage, that 3V zener is maybe conducting a couple hundred uA or something. Imagine putting a 15k resistor in parallel with that leg of the divider: the divider voltage goes down as a result. If you had used a 6V zener, you probably wouldn't be able to measure the shift--the diode will have been conducting some nA instead.

If you had included another resistor in series with the zener, the effect would have been reduced but not eliminated.

[Macbeth]

Bloody Zeners and their ACL reverse breakdown voltages right at the knee... 

[Wilksey]

If I had a 3v Zener acting as a protection device to a uC input, should I expect to see a voltage drop?

I had a voltage divider taking 16V down to 3V, I put this into a uC input, which worked fine, I thought about adding a ZD to protect if the input to the divider went above 16V but I was seeing a 0.3V drop with the ZD in circuit.  I didn't have a series limiting resistor, could that have caused the drop? It was used for V Monitoring rather than driving a current.

Diodes never block current 100%; they have a "reverse leakage" that goes up as voltage increases.  Just under the breakdown voltage, that 3V zener is maybe conducting a couple hundred uA or something. Imagine putting a 15k resistor in parallel with that leg of the divider: the divider voltage goes down as a result. If you had used a 6V zener, you probably wouldn't be able to measure the shift--the diode will have been conducting some nA instead.

If you had included another resistor in series with the zener, the effect would have been reduced but not eliminated.

So, you are saying that the ZD was causing some parasitic resistance to the output causing another divide?

[timb]

I would also like to se some theory around zener and transistor regulators, for example: 

and


And how those regulators cope with modern lienar regulators in all spects, power discipation, speed, accuracy, power drop and so on.

They can be good basic regulators that work much, much better than a Zener alone, in all aspects (plus they can dissipate orders of magnitude less quiescent current). All they require is a single additional transistor to implement, so cost is minimal.

Just make sure to take into account the transistor's VBE drop. For example, if you want a 5V output, you need a 5.6V Zener (since VBE is roughly 600mV).



That said, you can get linear regulators in SOT-23 packages that will do the same job, for the same price (or less) with better regulation, speed, and only uA of quiescent current, so there's not many places you'd need to use a Zener+Transistor regulator these days.

[FrankBuss]

Some interesting facts if you want to build an analog noise generator with diodes:

https://en.wikipedia.org/wiki/Noise_generator

So looks like when using the avalanche effect, the noise frequency spectrum is flat (white noise) from 1 Hz to 100 kHz for some diodes:

http://www.edn.com/design/analog/4420926/White-noise-source-flat-from-1Hz-to-100kHz

BTW, with a FPGA once I implemented a white noise generator with Rule 30. The noise looks good (if you are far away from the Nyquist frequency) , but needs more LEs than a LFSR.

[b_force]

Zeners are pretty noisy anyway. That's why it's not a very good idea to use them in a regulator.
It's fine for circuits which are not very sensitive off course.

[Whales]

I recently had to reverse-engineer my soldering iron and it used zeners and large resistors absolutely everywhere.  Perhaps it's cheaper for TH stuff than vregs?

[Tom45]

When I was a EE student in the 60s Zener diodes were relatively new, about a decade old. So when Dave said Dr. Zener came up with the zener effect long before any of us were born, I was skeptical.

But a little research came up with a date of 1934 for his paper, 11 years before I was born. So it was around 20 years before the theory in his paper surfaced in a useful device.

Also, Dr. Zener pronounced his name ZEE ner, not ZEN ner.

[zapta]

Very good video. I would like to see more like this.

The I/V diagrams of Zener diodes should be flipped in the horizontal direction IMO. That is, the Zener affect should be on the positive voltage. This is more consistent with their use case. That is, a 4V Zener diode that is at its Zener region has '4V, not '-4V'. Yes, I know, it's a matter of convention.

BTW, I always assumed that the Zener symbol was selected because it resembles 'Z' (and the Schottky symbol for 'S').

[gwideman]

Fundamentals Friday.
A tutorial on Zener Diodes.

Correction for 29:00 in the video. Summary: Not dynamic resistance, but rather RC effect, with Zener showing appreciable C.

Dave discusses using a 5.1V Zener diode as a signal clamp. He demos what amounts to a voltage divider having a 1k input resistor, and then 5.1V Zener to ground.  The scope shows 7V square wave input and ~5.1V output signal, as expected.  Zooming in to 50ns timebase, the scope shows leading edge of output waveform with slow rise time.



(And then Dave repeats the demo with 100ohm || 1k, so R = 90 ohm, and shows much faster but still slowed rise time. )

Dave attributes this to the dynamic resistance. I'm fairly certain the effect is due to the RC effect of the input resistor with the capacitance of the Zener (and scope probe, although that's set to 10X, so minimal).  Calculating: The rise time from around 1V to 5V (so about 63% :-) ) was 150ns, which with R = 1k gives a capacitance of about 150pF.  That aligns well with data in, for example: https://www.onsemi.com/pub/Collateral/HBD854-D.PDF page 25.



[Later edit] It occurs to me that where "dynamic resistance" does come into play is at the top of the rise where the output transitions from rise to horizontal at ~5.1V.

Dynamic resistance is really just a way of characterizing the slope of the I vs V curve, modeling it as an ideal Zener (either on or off with a knee at say 5.1V) in series with a resistance... but then taking into account that the curve above the knee voltage is not straight as a resistance would provide. Instead it's curved, corresponding to somewhat gradual turn-on of the Zener, hence "dynamic" resistance.

In the demo, at the rising edge, the output voltage rises, slowed by RC, until it reaches near 5.1V. At 5.1V, with an ideal Zener, the output voltage would abruptly transition to a horizontal line. But it does not -- it gently curves from rising to flat. That's due to the Zener's only gradual transition from high resistance (below 5.1V) to low resistance (above 5.1V).  Quite noticeable in the 90 ohm example:




So, the "dynamic resistance" is indeed displayed on the scope. However it's not causing the most conspicuous feature (slow rise time) but rather the non-sharp transition from rise to flat.
Graham

[lordnoxx]

Hi Dave,
thanks for that good educational video.
The graph from the component tester showed quite similar slopes of the
curve in the forward and in the reverse operation region. So why not using a couple of standard diodes in series and forward biased to have lets say a 5.4V (9*0.6V) voltage reference (or clamping voltage) instead of using only one 5.1V (lets neglect the difference of 0.3V here) Zener? OK sure it is more expensive using nine 1N4148 but hey....from a pure technical point of few? Why not!?!?
I also can think of some single transistor based current sources that use Zener diodes as references but also use LEDs or standard
diodes which are forward biased.

[gwideman]

So why not use [...]  nine 1N4148 [in series, forward biased]

The extent to which a diode does not sharply turn on at a particular voltage (I vs V is not vertical at 0.6V say) is represented by its series resistance. (And that line being curved corresponds to "dynamic resistance".)

If you connect nine diodes in series you get nine times the series resistance, so nine times less vertical an I vs V line, so to speak.  So the regulation capability of nine 1N4148's in series at (say) 5.4V will be nine times poorer than a single 1N4148 at 0.6V. (As judged by deltaV/deltaI or reciprocal.)

[Tom45]

In Australia I always heard and assumed it was ZENner probably because of the British way we say Zed not Zee. I shall endevour to say ZEEner from now on. Although I shall continue to accept ZEN-er quite calmly too.

I suppose then also I should always capitalise the name Zener. Something I am notoriously sloppy about. Ohms Amperes Hertz and so on.

Until Dave's video I had no idea there was a person named Zener. A name like that sounds like it could have been created by a marketing committee.

For those interested, you can read Clarence Zener's obituary at:
http://www.nytimes.com/1993/07/06/obituaries/clarence-m-zener-87-physicist-and-professor-at-carnegie-mellon.html

[b_force]

Also, Dr. Zener pronounced his name ZEE ner, not ZEN ner.

In Australia I always heard and assumed it was ZENner probably because of the British way we say Zed not Zee. I shall endevour to say ZEEner from now on. Although I shall continue to accept ZEN-er quite calmly too.

I suppose then also I should always capitalise the name Zener. Something I am notoriously sloppy about. Ohms Amperes Hertz and so on.

Jeez, if we are going to fall over pronunciation, than we can make a huge list about how English speakers say things totally wrong and vice versa.
Voltage is for example all totally 'wrong' in English.
In general there are a lot of original French words in English, but not pronounced on a French way.

On top of that you just have difference in accents, slang and dialect.
(I can tell you that 'Murican English is way different than Kiwi/Aussie and British, even with technical jargon)

[David Hess]

Also, Dr. Zener pronounced his name ZEE ner, not ZEN ner.

In Australia I always heard and assumed it was ZENner probably because of the British way we say Zed not Zee. I shall endevour to say ZEEner from now on. Although I shall continue to accept ZEN-er quite calmly too.

Here in the US, I use ZED for Z in amateur radio call signs which gets a second look from people not used to it but ZEE is not quite as distinct.

For US watchers, Dave's videos have a lot of entertainment value just for his Australian English.

[Brumby]

Here in the US, I use ZED for Z in amateur radio call signs which gets a second look from people not used to it but ZEE is not quite as distinct.

For US watchers, Dave's videos have a lot of entertainment value just for his Australian English.

I hadn't noticed..... 

[zapta]

For US watchers, Dave's videos have a lot of entertainment value just for his Australian English.

I hadn't noticed..... 

You got used to it.

[Kleinstein]

Many normal diodes in series is not a good replacement. It's not so much the series resistor but the temperature dependence that is rather poor. However for rather low voltages like 2 V, something like 2 or 3 silicon diodes or an LED might be an alternative. Some of the low voltage zener diodes are actually forward biased diodes not real zener diodes.   

[g0hjq]

Thanks Dave,

It's great to see fundamentals Friday back. My favourite segment of your blog, and definitely the most educational.

 

[VK3DRB]

An alternative protection input is using two schottky diodes (often in the one SOT-23 package for example). D1: anode at ground, cathode on signal line; D2: anode on signal line and cathode on PSU rail. The advantage is positive transients are clipped to 0.3V or so above rail and the negative transients are clipped to 0.3V or so below ground. And they are extremely fast in switching compared to  zeners which have a greater junction capacitance. Zener breakdown voltages are not that accurate at lower voltages in particular where as with two schottkies that is not a problem for low voltage supplies on some microcontrollers.

[rrinker]

 Interesting you should mention that, I JUST saw that same suggestion, pair of schottkeys in SOT-23 format soldered to the traces on an existing circuit (in this case, the secondary terminals of a current sense transformer) to clamp the voltage developed across the 1K resistor which is also across the secondary terminals under higher than expected current flow in the primary. As designed, it generally worked fine, but higher currents cause the resistor to greatly exceed its power capacity. As in 2 watts peak power in a 1/4W resistor. Oops. No real room to put a big sandbar power resistor, so someone came up with the schottkey clamping solution.