# reverse current of 1N914 diode

#### Xufyan

Joined Aug 3, 2010
114
What is the reverse saturated current of 1N914 diode,
according to data sheets it is constant , however according to caculations it also depends upon the value of 'R'

like,

Vo = kt/q * ln (vi/R*Is)

if we now arrange this equation for 'Is' the final equation is so weird :/ i want know what exactly the value of Is should i use for my calculations ???

i am designing simple Natural Logarithm Opamp in which diode is a feedback factor

#### steveb

Joined Jul 3, 2008
2,436
What is the reverse saturated current of 1N914 diode,
according to data sheets it is constant , however according to caculations it also depends upon the value of 'R'

like,

Vo = kt/q * ln (vi/R*Is)

if we now arrange this equation for 'Is' the final equation is so weird :/ i want know what exactly the value of Is should i use for my calculations ???

i am designing simple Natural Logarithm Opamp in which diode is a feedback factor
As we've tried to caution in the past, reverse saturation current (Is) is FAR from constant. It is temperature dependent and device dependent. There is no R for the diode, hence "Is" does not depend on R. The value R is a component in the log-amp circuit.

Joined Dec 26, 2010
2,148
In the absence of better model data, the value of Is may be better determined from a curve-fitting exercise working from forward conduction data.

For a number of reasons, reverse currents as given in data-sheet or even measured values give poor estimations for Is. In particular, leakage paths other than through the working junction may be significant, and data-sheet figures are often worst-case rather than typical.

A better estimate for Is would be given by taking a value from a simulation model. The copy of the LTSpice model which I have gives Is=2.52nA, and an ideality factor N of 1.752. Other models may differ somewhat, but the Is value is typically far less than the quoted leakage.

N.B Do not ignore the non-ideality factor N. In contrast to some transistors, these diodes require a substantial correction.

#### Wendy

Joined Mar 24, 2008
22,332
The nearest thing I've heard of is PIV (Peak Inverse Voltage), and it is the point at which the diode becomes at risk of breaking down from excessive voltage in reverse bias. Cross that line at your own risk.

When I design I treat the current as 0, even nanoamps sounds a bit high for some of what I've seen.

#### steveb

Joined Jul 3, 2008
2,436
A better estimate for Is would be given by taking a value from a simulation model. The copy of the LTSpice model which I have gives Is=2.52nA, and an ideality factor N of 1.752.
This might be a better estimate, but I fear you are giving the OP false hope that he can achieve something practical from a simple log-amp circuit that is so sensitive to reverse saturation current. The device to device variation and the strong temperature dependence mean that every circuit must be tuned to the given device and every circuit must have the diode mounted on a thermoelectric cooler with an active temperature control feedback system keeping the temperature constant.

If someone really wants an accurate value for Is, then a measurement using a thermal chamber is the way to go. The Is can be characterized for a particular device versus temperature. (note, this can be a tricky procedure in order to isolate the effects of saturation current from that of leakage current)

If this is just a modeling task, then one needs to know the device model and the parameters that are used in the simulations (as you pointed out), but I don't really see the value of modeling such an impractical circuit.

Last edited:

Joined Dec 26, 2010
2,148
The equation using Is is orthodox and quite practical for modelling forward current. Such equations are behind many simulations of diodes and transistors.

The relationship between Is and real-world reverse currents is not simple. I would not normally expect a model value of Is to be a reliable indicator of leakage current.

Joined Dec 26, 2010
2,148
This might be a better estimate, but I fear you are giving the OP false hope that he can achieve something practical from a simple log-amp circuit that is so sensitive to reverse saturation current. The device to device variation and the strong temperature dependence mean that every circuit must be tuned to the given device and every circuit must have the diode mounted on a thermoelectric cooler with an active temperature control feedback system keeping the temperature constant.

If someone really wants an accurate value for Is, then a measurement using a thermal chamber is the way to go. The Is can be characterized for a particular device versus temperature.

If this is just a modeling task, then one needs to know the device model and the parameters that are used in the simulations (as you pointed out), but I don't really see the value of modeling such an impractical circuit.
Real log amplifiers often use multiple junctions integrated into a common chip so that some of the effects of Is variation can be offset, well as (possibly) a temperature controlled environment. This was recently discussed with another student.

The value of such a task may lie in helping the student to better appreciate the significance of of the exponential diode behaviour, as well as other circuit techniques.

Actually, such a circuit does have at least one practical application. As a temperature sensor.

#### steveb

Joined Jul 3, 2008
2,436
Actually, such a circuit does have at least one practical application. As a temperature sensor.
As you know, a simple log amp is not a practical temperature sensor unless you do more compensation tricks at the circuit level. The circuit the OP is discussing would need to be characterized for every different 1N914 device that was installed. That's hardly practical because the calibration is too complex to just be a pot-adjustment and would need complex nonlinear formulas decoded by a microprocessors. Compare that to the available circuits and ICs that offer linear and stable temperature measurements to high accuracy.

It's not clear to me that the OP is fully appreciating the issues here, so I just want to stress these concepts. The OP says,

"i want know what exactly the value of Is should i use for my calculations"

Unless we are clear, we will be sending him on a fool's errand.

Joined Dec 26, 2010
2,148
My remark about temperature sensing was essentially humorous, although it is true that diode forward voltage is used to sense temperature, particularly in the relative sense of temperature rise detectors.

We have no information on the degree of accuracy required by this application, so I think it may be an exaggeration to say that complex nonlinear compensation would necessarily be required. A linear approximation with two constants might suffice to meet a modest requirement over a limited range of current.

The variability of device parameters is however a useful lesson. The other student with such a circuit (in the same class?) appeared to be surprised that a device model gave a result a couple of tens of mV from his measured result. I hope that he now realises that the actual variation could be considerably more. It is true that viewed in terms of apparent input current shift such variations are pretty dismal, for a diode having maybe a decade of conduction variation per 120mV change of Vf.

In the end, it would be interesting to hear more from the OP about what his real requirements are.

#### crutschow

Joined Mar 14, 2008
26,823
My understanding is that a small, diode-connected (collector connected to base) bipolar transistor generally has a more ideal diode characteristic than a standard diode. You might try using those instead of a diode.

Joined Dec 26, 2010
2,148
The ideality factor for a small transistor is often closer to one, but this has the result that its logarithmic slope is smaller. This may increase the relative effects of amplifier offsets. (More of a problem for students who are asked to use mouldy old 741s.)

That said, it is also true that transistors may be more easily be obtained in IC arrays, if the student wants to make a somewhat more serious attempt at a temperature compensated amplifier.

#### steveb

Joined Jul 3, 2008
2,436
i want know what exactly the value of Is should i use for my calculations ???
So let me address this question directly, because this is the issue that is important, and the question itself reveals a misconception.

You should NOT want to know the exact value of Is to use in calculations. To stress why, lets look at a similar issue that arises with bipolar transistor circuit design. With bipolar transistors we often consider the current gain (beta) which is the ratio of collector current to base current. The value of beta is very sensitive to manufacturing tolerances and somewhat sensitive to temperature and current as well, hence we must design circuits to be insensitive to variation in beta. So, instead of considering the EXACT value, we consider a typical value and we consider the possible range of variations. If the circuit is insensitive to changes in beta, then it is not critical to use, or even know, an exact value.

Similarly, the reverse saturation current has the same concerns, only it may be worse because Is varies by orders of magnitude, while beta may change well within an order of magnitude. So, you need a typical value and tolerance range for the room temperature variation of Is, and you need a general understanding about how Is changes with temperature. Then you design circuits that are tolerant to wide variations in Is.

Of course, all rules have exceptions, and designing sensors (like Adjuster's temperature sensor example) is an exception. Personally, I once designed a high speed transistor amplifier that required one compensation capacitor to be hand selected to match a transistor beta. This was necessary because I was pushing the speed limits of available devices and a sensitive topology gave me the best speed. But, these situations are very rare in design work, and the end result is sometimes less than convenient.

#### steveb

Joined Jul 3, 2008
2,436
Another thought comes to mind. Isn't the 1N914 in a glass case? I can't remember for sure, but if it is, the device will be sensitive to light, and this might effect a log-amp performance.