What kind of SMD LED is this?

[swgertsch]

LED (green) is on the main board of a LinkSys SD2008 network switch (manufactured around 2005).



The reason I ask is because the LED does not illuminate when using the diode test mode on my DMM (121GW), but I can get the LED to illuminate if probed with a PS set to 2V drawing 10mA.

The diode test mode of the GW121 measures 3.2V, so why doesn't the LED illuminate? 

[golden_labels]

Multimeters rarely make LEDs emit light.

With GW121 the maximum current delivered seems to be (15V - 3.2V) / 2200Ω ≈ 5mA, based on specs.

[swgertsch]

I tried the 15V range on the 121GW and the LED does just begin to illuminate.

In my original post I should have included the Vf and Vr measurements: on the 3V range Vf=.80, Vr=.66. On the 15V range: Vf=2.0, Vr=.31.
These readings seem to be at odds with my understanding/experience with LEDs. This LED is one of many (13) on the PCB and all the LEDs measure the same, so I am confident it isn't just a "wacked" LED.

One thing that is apparent is this LED needs a lot more current to stimulate photon emission.

I checked some SMD "status" green LEDs on another "newer" board and the readings where in line with what I would expect. 3V range: Vf=2.2, Vr=OFL, 15V range: Vf=2.6, Vr=OFL.

Visually the circa 2005 LED appears to have a different construction and electrical characteristics than a modern LED. So is this some sort of LED variant?       

[golden_labels]

Even a single LED model may have variance of a volt across different bins. And then Vf of different LEDs varies greatly. 3.2V is not unusual for a green LED; see LiteON’s green C230TGKT for an example.

In 3V testing mode of 121GW the current supplied is too low to reach Vf of the LED. Obviously, considering that 3V < 3.2V . Therefore you are receiving an invalid measurement result. Keep in mind diodes are non-linear and that forward voltage is not a constant. A LED has a range of currents it should be operated at and its stated Vf is simply the voltage measured at those currents.

I guess your confusion arises from various Vf tables one may find online, grouping LEDs by color. They may be very roughly true for low-power THT LEDs from eBay or old devices. But not in general.

[cortex_m0]

Visually the circa 2005 LED appears to have a different construction and electrical characteristics than a modern LED. So is this some sort of LED variant?       

The LEDs of today are much more efficient than those from 20 years ago. I would imagine anything that would have been inside a consumer grade Linksys switch would not have been state of the art even when the switch was new, so consider the below example illustrative and not necessarily related to the LEDs you've measured.

A 1206-case Gallium Phosphide (GaP) green LED I used on a design 10 years ago (P/N LTST-C150GKT) and was recently declared obsolete had a Vf of 2.1V, If of 10mA and luminance of 6mcd typ, or 0.26cd/W

A modern Wuerth InGaN Green LED, P/N 150120VS75000 also in 1206 case, has a Vf of 3.2V, If of 20mA, and luminance of 450mcd typ. Yes, this is more power, but the light output and efficiency are way higher at 7.0cd/W.

The old part, I would probably drive at the full 10mA. The new one, possibly less than 0.5mA.

[tooki]

Multimeters rarely make LEDs emit light.

With GW121 the maximum current delivered seems to be (15V - 3.2V) / 2200Ω ≈ 5mA, based on specs.

As a blanket statement, that’s complete nonsense. Not every LED will illuminate on every multimeter, but a great many will.

As cortex_m0 already said, many modern LEDs are insanely efficient. On my current project, I have various diagnostic LEDs, and to keep power consumption low, I am using white LEDs (which are the most efficient ones I have) and run them at around 50 μA IIRC. More than bright enough to be seen indoors.

[SteveyG]

Multimeters rarely make LEDs emit light.

With GW121 the maximum current delivered seems to be (15V - 3.2V) / 2200Ω ≈ 5mA, based on specs.

As a blanket statement, that’s complete nonsense. Not every LED will illuminate on every multimeter, but a great many will.

As cortex_m0 already said, many modern LEDs are insanely efficient. On my current project, I have various diagnostic LEDs, and to keep power consumption low, I am using white LEDs (which are the most efficient ones I have) and run them at around 50 μA IIRC. More than bright enough to be seen indoors.

Probably more correctly, InGaN LEDs are usually easily lit by multimeters with an output voltage above 2.5V, but AlGaInP, GaInP LEDs are still widely used for indicators which will not always be illuminated in diode test mode.

[gabiz_ro]

You tested in circuit.
Desolder one and check again.

[tooki]

Multimeters rarely make LEDs emit light.

With GW121 the maximum current delivered seems to be (15V - 3.2V) / 2200Ω ≈ 5mA, based on specs.

As a blanket statement, that’s complete nonsense. Not every LED will illuminate on every multimeter, but a great many will.

As cortex_m0 already said, many modern LEDs are insanely efficient. On my current project, I have various diagnostic LEDs, and to keep power consumption low, I am using white LEDs (which are the most efficient ones I have) and run them at around 50 μA IIRC. More than bright enough to be seen indoors.

Probably more correctly, InGaN LEDs are usually easily lit by multimeters with an output voltage above 2.5V, but AlGaInP, GaInP LEDs are still widely used for indicators which will not always be illuminated in diode test mode.

1. I never said “always”, I was very clear on this.
2. The old LED chemistries tend to be lower voltage, so have little trouble illuminating on even rudimentary diode test modes. The newer, higher voltage chemistries may require a higher test voltage, but many multimeters are perfectly capable of supplying this, including many of the most popular models (like the Fluke 87 V).

I own a number of different multimeters, some cheap, some expensive, which I think are very representative of the world of digital multimeters. I just tested numerous LEDs (some vintage, some modern) with six different meters:

• Fluke 87 V ($$$)
• Keysight U1252B ($$$)
• Keithley 2015 ($$$$)
• Aneng AN8008 ($)
• Mastech MY64 ($)
• No-name VC830L (¢)

All of them could light and measure the least efficient of all the LEDs, a very old red LED that lit very dimly on all the meters. (Even at 20mA it’s a very dim LED.)

All of them could light and measure a typical vintage red LED.

All of the meters except the Mastech could light the most challenging of all the LEDs, a modern UV LED (those have a Vf approaching 4V at 20mA), enough to see it in a lit room. Only the Fluke, Keithley, and Aneng could measure it, but the no-name and the Keysight could light it without having a reading.

A modern blue LED, two different modern green LEDs, and a white LED were lit by all meters, but again, only the Fluke, Keithley, and Aneng could measure them.

A modern red LED, a modern yellow LED, a vintage yellow LED, and a vintage green LED lit on all meters, and could be measured on all except the no-name.


Remember that the statement I was refuting is this one:

Multimeters rarely make LEDs emit light.

These quick tests not only disprove that nonsense claim, but disprove it decisively, with literally only a single LED, the challenging UV LED, refusing to illuminate visibly on one single meter out of six. I tested 11 different LEDs, meaning a total of 66 tests. Of those 66 combinations of meters and LEDs, only one single one failed to light.
 
So I stand by my statement that most LEDs will light on most multimeters.

Had the question been “which meters can measure high-Vf LEDs?”, the discussion would be rather different. But the question was whether they emit light.

[Edited repeatedly to fix the mathml, which isn’t parsed in the preview, misconstruing the dollar signs.]

[tooki]

P.S. The three expensive meters all use a test current of 1mA, while the three cheap meters use 1.42-1.66mA.

5mA, as golden_labels suggested as being insufficient, is entirely unnecessary to test LEDs.

The maximum test voltages at 1.0-1.66mA varied from 2.63V (Mastech) to 7.26V (Fluke 87 V)*, with all except the Mastech being above 3V, explaining why it alone was unable to light the UV LED, which drops about 2.95V at 1mA, 2.85V at 100μA, and 2.75V at 10μA.


*The Keithley actually uses 12.9V and 12.7V test voltages for 100μA and 10μA diode test currents, respectively.

[swgertsch]

There appears to be significant differences between how a DMM "diode" test mode does its magic; specifically what test voltages and test current limits are used.

I couldn't find a "theory of operation" description for a DMM diode test function. The 121GW user manual states the test voltage and test current of its two diode ranges (3V/1.4mA and 15V/7mA) and states "A nominal 2.2K resister (including a PTC) is in series with the voltage source and is what limits the test current." These are good clues as to how the 121GW diode does its magic, but I am not confident I have an accurate understanding of the magic with respect to determining the diode reading. Does anyone have inside knowledge of how the magic is preformed?

Note: The measured voltage at the DMM probes for each range: 3V and 15V ranges measured 3.2V and 16.3V respectively. I interpret this to be the diode test voltage supply is unregulated.

[tooki]

There appears to be significant differences between how a DMM "diode" test mode does its magic; specifically what test voltages and test current limits are used.

I couldn't find a "theory of operation" description for a DMM diode test function. The 121GW user manual states the test voltage and test current of its two diode ranges (3V/1.4mA and 15V/7mA) and states "A nominal 2.2K resister (including a PTC) is in series with the voltage source and is what limits the test current." These are good clues as to how the 121GW diode does its magic, but I am not confident I have an accurate understanding of the magic with respect to determining the diode reading. Does anyone have inside knowledge of how the magic is preformed?

Note: The measured voltage at the DMM probes for each range: 3V and 15V ranges measured 3.2V and 16.3V respectively. I interpret this to be the diode test voltage supply is unregulated.

No magic: it’s a constant-current source typically between 1-2mA (except in high end meters with selectable test currents), and it measures the voltage drop across the diode.

Using a series resistor to create a current-limited voltage source is close enough to a current source for simple diode testing purposes, where precision is neither expected nor required. If you subtract the typical 0.7V of a silicon diode from the 3.2V and 16.3V, respectively, and then divide that by 2200 ohms, you get very close to 1mA and 7mA. The specced 3V and 15V, divided by 2200 ohms, gives roughly the 1.4mA and 7mA short circuit current given in the manual.

In my testing given in an earlier reply, I mention that the expensive meters all used 1mA test currents, while on the cheaper ones it varied. This probably means the expensive meters use an actual regulated constant-current source, while the cheaper ones use simpler current-limited voltage sources like the 121GW. The true CC source is probably superior if you need to truly characterize the diode (at exactly that current), but for ordinary testing, really doesn’t matter. 

See also: https://www.eevblog.com/forum/beginners/diode-test-principle/

Full diode characterization (creating a V-I curve) is basically testing at many different currents to show how the forward voltage drop varies with current flow.