Understanding diode (LED) resistance at V=0

[phillip_at_work]

I'm working with some bare chip diodes for an application where I need extreme luminosity. These are the diodes here:

https://www.irtronix.com/leuv-v625a6

I notice that some of the diodes have very significant resistance (on the order of megaohms) from anode to cathode and some have very small resistance (tens of ohms to a few kOhms). A large resistance seems to make more sense to me. However, I don't notice any difference in the performance of the diodes when I evaluate radiometric flux with an integrating sphere. How can I explain this?

 

[berkeman]

Welcome to PF.

How are you making these measurements? It looks like at their operating point, the effective impedance is pretty low, as expected...

 

[phillip_at_work]

By operating point, are you referring to the range of voltages sufficient to forward bias the diode? Why then do I see large resistance for some and small for others? I'm simply measuring with an ohmmeter from anode to cathode with no power source connected.

 

[berkeman]

By operating point, are you referring to the range of voltages sufficient to forward bias the diode?

I'm referring to the $V_f$ numbers listed in the chart for 500mA operating current...

Does your DVM have a "diode" check function? Although it may not be strong enough to do much with these big LEDs...

 

[phillip_at_work]

Using the diode check function, I see about 2.1V across 14 of these diodes wired in parallel. This is consistent with my expectations, but I'm not sure it tells me much about the resistance when no power is applied to the diode. I thought that measuring the diode resistance with no power across the junction should return near infinite resistance always. Is this not the case?

 

[tech99]

Remeber that LEDs are light sensitive so room lighting will affect the results.

 

[DaveE]

The resistance at 0V is, of course, the slope of the V-I curve at that operating point. For diodes that is essentially a leakage current measurement, since the junction isn't forward biased. Leakage current is small compared to the normal forward biased operating region and, at the device level, isn't a parameter that's well controlled. Note that there is no characterization at 0V in the datasheet. This may be particularly true for an array of parallel diodes. Things like illumination or the actual junction temperature may cause significant differences.

Still, "tens of Ohms" does sound surprisingly low. Suppose your array was operated at 3.5V & 0.5A, then if there really was a 20Ω resistor in parallel with it, it would divert 175mA away from the junction, which would have a noticeable effect on the intensity. Diodes are very nonlinear, so you may want to think about why you need to measure the 0V parameters and what they mean for your application.

This is a case where you really shouldn't just assume that your multimeter will give you a meaningful answer, unless you understand how the meter is making it's measurements. For example, the bias current applied in most ohmmeters is dependent on the range, so the meter may be selecting different operating points and exaggerating differences.

Regardless of this particular problem, it is a good idea in any EE lab to get to know how your meter makes these measurements. This is easily done with a couple of DMMs measuring each other. No, they aren't all the same. I would use different DMMs in different circumstances depending on their individual quirks. Don't assume your instruments are "fool proof", they only appear that way when used in common applications. As soon as you see confusing results, like this one, you should look more carefully at how the measurements are actually being done. It wouldn't be too hard to build your own with a power supply, resistor, voltmeter, ammeter, and a calculator; then you'll know better how the device is actually being excited. A curve tracer would be great if you have access to one, or you could plot your own DC V-I curve by hand.

 

[DaveE]

Remeber that LEDs are light sensitive so room lighting will affect the results.

Yes! This is a great point. All diodes are photodiodes too. This especially matters if they have clear packaging.

Most DMMs won't measure the true slope when there is a DC bias involved.

 

[tech99]

An interesting way to investigate a diode at low Vf is to apply a known current, in microamps or nanoamps, by using a large series resistor with a high voltage, and then measure the voltage developed across the diode.

 

[DaveE]

Also, if you want a simple slope measurement when DC doesn't work, you can buy (or build) an AC ohmmeter. Note that this isn't the "AC" range on a normal DC ohmmeter, it applies a low frequency sinusoid as excitation and rejects all other frequencies in the measurement. Also easy to set up with a lock-in amplifier, if you have those laying around. Commonly use in TEC screening, for example.

 

[phillip_at_work]

Many thanks for the insightful replies. We are actually already conducting IV curve testing and simultaneously measuring flux.

We were concerned about the resistance of the diode when unpowered for tangential reasons. I think it's safe to say, though, that manufacturers don't characterize diode behavior in this case because it has no bearing on the operation of the diode while forward biased. Please correct me if I'm wrong, but it also seems that the "diode test" method available for most multi-meters (or simply creating our own IV curve, as we're already doing), is the gold standard way to evaluate the health of the LED.

I am still somewhat curious about what DaveE says here:

The resistance at 0V is, of course, the slope of the V-I curve at that operating point. For diodes that is essentially a leakage current measurement, since the junction isn't forward biased.

Is there an intuitive way (or no intuitive, if you like) to understand why this is the case?

Many thanks,

 

[DaveE]

Many thanks for the insightful replies. We are actually already conducting IV curve testing and simultaneously measuring flux.

We were concerned about the resistance of the diode when unpowered for tangential reasons. I think it's safe to say, though, that manufacturers don't characterize diode behavior in this case because it has no bearing on the operation of the diode while forward biased. Please correct me if I'm wrong, but it also seems that the "diode test" method available for most multi-meters (or simply creating our own IV curve, as we're already doing), is the gold standard way to evaluate the health of the LED.

I am still somewhat curious about what DaveE says here:Is there an intuitive way (or no intuitive, if you like) to understand why this is the case?

Many thanks,

It's just the Shockley equation, I think. There are innumerable web resources to learn about PN junctions, at all different levels of complexity. I think you'll get a better description than I can give by picking your own from a google search.

Step 1 is to assume initially that all PN junctions are the same; whether they are power rectifiers, RF diodes, PV cells, photodiodes, etc. The distinctions between them is more about optimization than basic physics.

BTW, I said "leakage current", which isn't really right. That would be current that doesn't flow through the junction (like around the edges). It's really saturation current in the junction plus, at the device level, leakage current around the junction. In normal devices (meaning not nm scale ICs or HV reverse biased rectifiers), the leakage current should be negligible. I don't have a clue about a simple description of leakage current, there are several sources and it is very process/design dependent.