Very basic Zener Diode/LTSpice simulation question

[eranrund]

Hello.

I have set a very simple circuit in LTSpice: Voltage source of 12v connected to a resistor which is then connected to a 8.2v Zener diode.
When measuring the voltage between the resistor and the Zener I notice two things I cannot explain:
1) The measured voltage is around 8.7v (i was expecting 8.2)
2) From 0ms to ~1.5ms I see the voltage slightly dropping. Why is that?

Attached is a screenshot.

Many thanks

[janneman]

Originally Posted by eranrund

Hello.

I have set a very simple circuit in LTSpice: Voltage source of 12v connected to a resistor which is then connected to a 8.2v Zener diode.
When measuring the voltage between the resistor and the Zener I notice two things I cannot explain:
1) The measured voltage is around 8.7v (i was expecting 8.2)
2) From 0ms to ~1.5ms I see the voltage slightly dropping. Why is that?

Attached is a screenshot.

Many thanks

Hi,

The nominal zener voltage of 8.2V is specified with a specific zener current, often 10mA, which should also be in the data sheet.
Your circuit has much larger zener current and that, with the non-zero differential zener impedance, gives a higher zener voltage.

Can't explain the 40uV (!) initial drop - possibly some numerical initialisation in LTspice?

jan didden

[eranrund]

Hello,

I have changed the resistor value to 730ohm in order to have a current of 5.03ma and not fry the Zener and now I measure 8.32v.
The datasheet specifies a minimal value of 7.7v and a maximal value of 8.7v, nominal being 8.2v for 5ma current.

How does the Zener's model come up with 8.32 instead of, for example, any other close value?

[janneman]

Originally Posted by eranrund

Hello,

I have changed the resistor value to 730ohm in order to have a current of 5.03ma and not fry the Zener and now I measure 8.32v.
The datasheet specifies a minimal value of 7.7v and a maximal value of 8.7v, nominal being 8.2v for 5ma current.

How does the Zener's model come up with 8.32 instead of, for example, any other close value?

I would check the model text; you'll probably find a small series R.
Possibly there is also a current dependent voltage source.

jan

[DF96]

Does the model include temperature?

[bill_a]

Its hard to read your screen capture, but I just duplicated your description, with a 720 Ohm resistor and it works fine. Zener voltage of 8.269 @ 2.4mA. That's about the best you can hope for with a Zener - they're just not that good a regulator.

The 'droop' you see is due to the junction capacitance charging up - you'd see something similar in real life if you could set up a good enough experiment. If it bugs you, change your simulation time to say 1mS - you hardly notice it at all....

Bill

[sbrads]

Originally Posted by bill_a

Its hard to read your screen capture, but I just duplicated your description, with a 720 Ohm resistor and it works fine. Zener voltage of 8.269 @ 2.4mA. That's about the best you can hope for with a Zener - they're just not that good a regulator.

The 'droop' you see is due to the junction capacitance charging up - you'd see something similar in real life if you could set up a good enough experiment. If it bugs you, change your simulation time to say 1mS - you hardly notice it at all....

Bill

Junction capacitance would make the voltage rise slightly if anything, not droop. Zeners over 6v go up in voltage with a rise in temperature so it can't be temperature either. It must be a bug in LTSpice.

[jcx]

you may need to adjust/inform your expectations of simulation, modeling, real world device tolerance/production spread

Spice DC solvers try to find the "steady state" solution before running the .tran simulation

these 2 different stages of simulation use different algorithms, can come to different conclusions - sometimes at very gross levels - much less the 10ppm you seem concerned about


a big weakness of Spice for matching "real world" results is that there is no "live" modeling of device thermal effects - the standard Spice transisitor, diode equations can be assigned a junction temperature -but is is fixed for the duration of the sim

[eranrund]

I see.

I was hoping to use LTSpice to experiment with basic stuff (such as the Zener regulation) I am trying to learn.
Guess I'm gonna have to just take its result with a grain of salt

Thanks again

[ingenieus]

LTspice is pretty good, especially when you consider the price being charged for it.

It is not perfect. Better models do give better results. A good start in this regard is Bob Cordell's website, which has some improved models (including Zeners if I remember correctly).

You can also have a look at this thread.

Things you should know about LTSpice