Diode leakage reading with very different results between meters.

[Avelino Sampaio]

I recently adapted my analog multimeter to expand the ohmic scale to x1M. The project can be found in the link below. The schematic is also in the images below.

https://www.eevblog.com/forum/beginners/attempt-to-implement-an-x100k-ohmic-scale-in-an-analog-multimeter/

With this X1M scale I was able to go beyond the 200M of my digital multimeter unit UT89XE. I can capture the leakage of some diodes and transistors. And it is in this issue of diodes that I am intrigued and do not fully understand what is happening. When I measure the leakage of the 1n4148 diode with the UT89XE I get 116M, but when I measure it with the analog multimeter the reading is almost 500M. So what could be happening between the meters for such a different result and which is probably correct?

[CaptDon]

My first guess is that the test voltage applied is different between the two instruments which would indeed give a different leakage result. Also, to say that the diode leakage measures 'XXX' ohms is totally inaccurate. What you are really seeing is 'leakage current' usually in microamperes. It would be more accurate to say 'xx' microamperes of leakage registers as 'XXX' megaohms on your meters. The old military microwave diode testers for 1N21 and 1N23 mixer diodes registered in microamperes and not ohms since measurement of ohms is meaningless. It is also worth noting that many of the old V.O.M. meters used 22 volts or more on the high resistance ranges and were notorious for destroying the germanium diodes being tested and would also destroy most unprotected current production MOSFET's as well as older unprotected CMOS devices. I have a Ballentine DMM that has only 7.2 volts of battery but generates nearly 15 volts for testing on the high resistance ranges which could be enough to destroy Logic Level Fet's. One nice unknown feature of many DMM's is the use of a precise 1ma. of current to measure ohms on the low ranges. When you see something like 620 ohms forward you will immediately know the diode is a standard silicone with a forward voltage drop of .62 volts @ 1ma. This is a quick way to determine silicone, schottky, germanium or multicell special diodes used for bias networks. Cheers mate!!

[iMo]

The leakage current of the 1N4148 will vary based on the voltage applied (talking reverse currents and voltages) and temperature. Typically it is in single nA digits at low voltages - see below.

In your above schematics the voltage at the diode (and the rev current) will be almost the same S1 switch setting regardless. What you need to do is to measure the reverse diode current instead of the voltage drop at the diode (as you do in your above schematics, imho).

[iMo]

For Example (simplified):

[Kleinstein]

The leakage of the diode is very temperature dependent (e.g. can go up by a factor of 2 for 5-10 K more). Another point is that light can cause extra leakage current and with a different test current the relevance of that extra current varies. There is also a chance that the protection part at the input of the meter can contribute. Finally the hum pickup can get rectified and cause errors in the reading, when large.

[Avelino Sampaio]

I don't know if this is correct, but I took a reading of the voltages applied to the ut89xe scales. On the 200R scale it applies 3v, on the scales up to 2M it applies 1v, and on the 20M and 200M scales I'm reading 0.5376v. Considering that the reading is correct, 0.5376v/116M ::> 4.63nA. Very consistent with the graph presented. As for the result of the analog meter, we have approximately 5v/500M ::> 10nA, which is also very consistent with the graph presented. I deserve a scolding for not paying attention to the datasheet.

[Kleinstein]

When reading the voltage of the ohms mode one gets different results depending on the range.
For the low ranges the voltage is essentially the open circuit voltage and a 200 ohm resistor would produce less voltage, likely some 200 mV.
For the high range (20 M) the input resistance of the meter reading the voltage is relevant. So the 0.54 V would correspond to some 54 nA of test current. from the meter causing that voltage at the typical 10 M input resistance of the meter in volt mode. With a higher restance the voltage can usually go higher, possibly to 2 or 3 V.

The diode leakage in the reverse direction is usually not very dependent on the voltage, except for very low voltages below some 50 mV.
Analog meters may have the test current in ohms mode with the different polarity compared to a voltage reading - this could also make a difference.

[CaptDon]

All of the graphs show diode leakage in reverse bias is very dependent on voltage.

[Avelino Sampaio]

Based on what I'm observing here, I decided to put six of these diodes in series. Since there will be a voltage division between them, I deduce that the leakage current would be lower in each of them but as a final result it would not be too far from the measurement of just one diode. And here are the results, very interesting.

[iMo]

All of the graphs show diode leakage in reverse bias is very dependent on voltage.

People say that at low reverse voltages the reverse current is "relatively" constant and determined by thermally generated carriers and surface leakages around the contacts (on the chip). As the reverse voltage starts to approach the diode's breakdown voltage the reverse current begins to rise due to avalanche or Zener effects..
PS: perhaps this..

[Avelino Sampaio]

My first guess is that the test voltage applied is different between the two instruments which would indeed give a different leakage result. Also, to say that the diode leakage measures 'XXX' ohms is totally inaccurate. What you are really seeing is 'leakage current' usually in microamperes. It would be more accurate to say 'xx' microamperes of leakage registers as 'XXX' megaohms on your meters. The old military microwave diode testers for 1N21 and 1N23 mixer diodes registered in microamperes and not ohms since measurement of ohms is meaningless. It is also worth noting that many of the old V.O.M. meters used 22 volts or more on the high resistance ranges and were notorious for destroying the germanium diodes being tested and would also destroy most unprotected current production MOSFET's as well as older unprotected CMOS devices. I have a Ballentine DMM that has only 7.2 volts of battery but generates nearly 15 volts for testing on the high resistance ranges which could be enough to destroy Logic Level Fet's. One nice unknown feature of many DMM's is the use of a precise 1ma. of current to measure ohms on the low ranges. When you see something like 620 ohms forward you will immediately know the diode is a standard silicone with a forward voltage drop of .62 volts @ 1ma. This is a quick way to determine silicone, schottky, germanium or multicell special diodes used for bias networks. Cheers mate!!

You made me realize that I can read the current directly instead of resistance. The full scale is 250nA (5v/20M), which coincides with the full scale of the 250v DC voltage. So I can perfectly use the DC scale to read the current. Since there are 50 divisions, I have 5nA per division (250nA/50). On this x1M scale I can then read from 1nA to 250nA. As an example, I took the reading of a 1n5711 SCHOTTKY diode and the reading, using the DC voltage scale, was 10nA. Another test was with the 1n4007, which had a result of approximately 2nA. Very consistent with the graph.

Thank you

[BennoG]

Why do you have the meter in AC voltage.
and de jacks in the current inputs.

In this situation the measuring makes no sense at all.

Benno

[CaptDon]

I agree with Benno, W.T.F. are you measuring with the black meter?? The red lead in the 'OUTPUT' jack??? That is normally a capacitor coupled jack and should only pass A.C. voltage or current. Second, the 1000 volt A.C. scale??? Again, what the hell, how would that measure leakage current? As stated by Benno, the lash up of the black meter in the pictures makes no sense.

[Avelino Sampaio]

I agree with Benno, W.T.F. are you measuring with the black meter?? The red lead in the 'OUTPUT' jack??? That is normally a capacitor coupled jack and should only pass A.C. voltage or current. Second, the 1000 volt A.C. scale??? Again, what the hell, how would that measure leakage current? As stated by Benno, the lash up of the black meter in the pictures makes no sense.

I apologize for the confusion, but I thought it was clear at the beginning of my post that the meter underwent an adaptation. I even attached a link to the topic with the details of the adaptation. One of the adaptations was the removal of the capacitor from the OUTPUT connector, so that I could use it exclusively. When I use the OUTPUT connector, the main selector has no effect and can be in any position. I adapted another selector to the meter, so that I can select the x10k, x100k and x1M scales, when I am using the OUTPUT connector.

[Avelino Sampaio]

The leakage of the diode is very temperature dependent (e.g. can go up by a factor of 2 for 5-10 K more). Another point is that light can cause extra leakage current and with a different test current the relevance of that extra current varies. There is also a chance that the protection part at the input of the meter can contribute. Finally the hum pickup can get rectified and cause errors in the reading, when large.

Kleinstein, I was curious to check the issue of the incidence of light on the behavior of diodes. In the 1n4148 the variation is very small, but I have here the 1n5061 which is crazy. It is so sensitive to light that even my movement around the lab causes its leakage current to vary. I could even use it as a sensor.

image 1: with the lab light.
image 2: cutting the light with my hand.
image 3: direct light on the diode.

[CaptDon]

Sincere apologies, I didn't realize you internally modified the black meter. When I looked over your earlier post I thought you had measured the apparent behaviour of the ohms scale. Now it is more clear what you have done. Best wishes in your diode investigations. I have been stripping some old audio amplifiers that are damaged beyond worthwhile repair mainly for the output devices that are still good, a supply of high wattage low resistance emitter resistors and the bias diodes which internally may have as many as four diodes in series. I keep these items in stock for repairing amplifiers that I use daily and for amplifiers I repair for other musicians in the local area. If I put a used part in their amplifier I generally don't charge for the part or only charge 25% of its new value. And yes I do tell them "I can fix it today if you don't mind the replacement part being previously used.

[David Hess]

What you are really seeing is 'leakage current' usually in microamperes. It would be more accurate to say 'xx' microamperes of leakage registers as 'XXX' megaohms on your meters.

Ohms measurement applies a calibrated current and reads the voltage.  So what we are seeing is the voltage returned by applying a specific low current.  On a 20 megohm range, that is likely 100 nanoamps to make 2 volts full scale which is well within the leakage range of a gold dopped switching diode like the 1N4148.

Different meters could produce different results because they use different currents and diode leakage is not linear, or the diode could literally detect the charge pumping from the meter's ADC.

When reading the voltage of the ohms mode one gets different results depending on the range.
For the low ranges the voltage is essentially the open circuit voltage and a 200 ohm resistor would produce less voltage, likely some 200 mV.
For the high range (20 M) the input resistance of the meter reading the voltage is relevant. So the 0.54 V would correspond to some 54 nA of test current. from the meter causing that voltage at the typical 10 M input resistance of the meter in volt mode. With a higher restance the voltage can usually go higher, possibly to 2 or 3 V.

The meter's divider which produces its 10 megohm shunt resistance is disconnected from the output when in ohms mode.  Older meters used the 10 megohm decade divider as part of the ohms converter to produce the calibrated and scaled current, but I do not know what newer automatic ranging meters do since their decade divider is arranged differently.