Been snowed in for a couple of days and playing around with odd ideas. Broke open a "new style" LED buld to investigate the series LED "filament" used in them.
It provide a contant voltage drop of exactly 82.7V from 8ma to 12ma. "Constant" in that I could not measure the slope with a 3-digit voltmeter, a better instrument would be needed.
Placed in the cathode circuit of a 12AU7 cathode follower replacing a 27K load, diode followed by an 11K load instead, it also made a very low distortion level-shifter, provided that the current was around 10ma quiestent.
I've looked around for any information on these kinds of uses, but haven't found any. Has anyone else experimented with these?
Thanks...
It provide a contant voltage drop of exactly 82.7V from 8ma to 12ma. "Constant" in that I could not measure the slope with a 3-digit voltmeter, a better instrument would be needed.
Placed in the cathode circuit of a 12AU7 cathode follower replacing a 27K load, diode followed by an 11K load instead, it also made a very low distortion level-shifter, provided that the current was around 10ma quiestent.
I've looked around for any information on these kinds of uses, but haven't found any. Has anyone else experimented with these?
Thanks...
Isn't it made from lots of tiny LEDs in series? So you should get the same result as that.
I think he is talking about the bulbs that have those leds that are supposed to imitate filaments.
If it were just a bunch of little leds in series, then I would expect the voltage to change enough to read it on a basic meter as you vary the current.
If it were just a bunch of little leds in series, then I would expect the voltage to change enough to read it on a basic meter as you vary the current.
I don't see what else it could be than a string of LEDs, but I would also expect a measurable voltage change. If it were a string of 27 LEDs (just over 3 V per LED) and if each LED would behave as an ideal diode, you should still get 0.2766 V of voltage change between 8 mA and 12 mA at room temperature - assuming the effect of self-heating is negligible, that is. With non-ideal diodes with an emission coefficient between 1 and 2, it can be up to twice as much.
Last edited:
Those are wide band lamp strings, 85V to 277V...I expect some SMPS type control in it.
There are also 90-260V, and 100-250V ranges available.
There are also 90-260V, and 100-250V ranges available.
Right. That's why I was asking what bulb the OP was using. If its a standard stock item, then I could measure it with better equipment and figure out what is going on. The chances that it is actually useful is pretty small, but hey, I need replacement lights anyways and I am stuck inside my house with nothing better to do until the cold weather passes.
I'am loving the idea of replacing zeners with LED strings. Could you use some of those fake LED vintage lights.
Common tungsten light bulbs have an EI characteristic opposite that of LEDs & other SS diodes.
On a log-log graph SS diodes have a slope of 2 since they are square law. A resister has a slope of One.
On a log-log graph common tungsten bulbs have slopes approximating 0.75. The Edison Bulb has a slope of One.
The Tungsten bulbs are a slow reacting constant current device, perfect for use in William Hewlett's HP 200,
Wien Bridge Audio Oscillator. They also work very well in a bridged 'T' oscillator. 👍
On a log-log graph SS diodes have a slope of 2 since they are square law. A resister has a slope of One.
On a log-log graph common tungsten bulbs have slopes approximating 0.75. The Edison Bulb has a slope of One.
The Tungsten bulbs are a slow reacting constant current device, perfect for use in William Hewlett's HP 200,
Wien Bridge Audio Oscillator. They also work very well in a bridged 'T' oscillator. 👍
Semiconductor diodes ideally have an exponential relation between voltage and current, not a quadratic one.
A few points:
1, I smashed the bulb without checking the brand, Shame on me, I'm going to buy a known common brand and test again
2. Re the cathode follower, the LED diode string (that's what it is) is followed by a resistor, the voltage shifted signal is taken from the point where the resistor and diode string meet. If there was no resistor there would obviously be no signal, just a really weirdly biased tube.
3. The "slope" I refer to is the slope of the upper, most vertical part, of the lines in the graph Baoudouin0 posted (thanks).
Had it ended with nothing more interesting I'd not have bothered to post, What surprised me is that over a certain range of current the "verticalness" of the LED string's response was so good I won't actually believe it without testing it better. Maybe I "got lucky", maybe I botched the test, or maybe the LED string has an engineered property that's actually pretty remarkable.
Thanks for the replies,
1, I smashed the bulb without checking the brand, Shame on me, I'm going to buy a known common brand and test again
2. Re the cathode follower, the LED diode string (that's what it is) is followed by a resistor, the voltage shifted signal is taken from the point where the resistor and diode string meet. If there was no resistor there would obviously be no signal, just a really weirdly biased tube.
3. The "slope" I refer to is the slope of the upper, most vertical part, of the lines in the graph Baoudouin0 posted (thanks).
Had it ended with nothing more interesting I'd not have bothered to post, What surprised me is that over a certain range of current the "verticalness" of the LED string's response was so good I won't actually believe it without testing it better. Maybe I "got lucky", maybe I botched the test, or maybe the LED string has an engineered property that's actually pretty remarkable.
Thanks for the replies,
Some time ago I measured the current-voltage relation of a red and a white LED, both salvaged from bicycle lights found on the street.
Voltage-to-current characteristic of the white LED:
The dark curve shows the measurements, the orange curve is an exponential curve: 45 uA at 2.5 V, a decade per 110 mV (emission coefficient just over 1.9).
The reverse current at 2.9 V was immeasurable with my set-up (< 100 pA).
Voltage-to-current characteristic of the white LED:
The dark curve shows the measurements, the orange curve is an exponential curve: 45 uA at 2.5 V, a decade per 110 mV (emission coefficient just over 1.9).
The reverse current at 2.9 V was immeasurable with my set-up (< 100 pA).
This curve is for a bright red LED from a bicycle light. I took a 9 V battery, some crocodile clip wires, a bunch of resistors and a cheap TDM600 digital multimeter, which has a 1 Mohm input resistance when measuring DC voltages. With this set-up and correcting for the current through the meter, I measured the LED forward voltage for currents from 732 nA to 1 mA and got this result for the forward current in A versus the forward voltage in V (dark blue curve):
It fits nicely with the red exponential curve, which is 25 uA at 1.65 V with a decade/93 mV slope (emission factor about 1.6).
Using two resistors to make a voltage divider and using the meter's 200 mV range as a 200 nA range (as 200 mV/1 Mohm = 200 nA), I measured the LED's reverse current to be 0.3 nA at 1.52 V and 0.5 nA at 2.05 V. These leakage current measurements are quite inaccurate.
It fits nicely with the red exponential curve, which is 25 uA at 1.65 V with a decade/93 mV slope (emission factor about 1.6).
Using two resistors to make a voltage divider and using the meter's 200 mV range as a 200 nA range (as 200 mV/1 Mohm = 200 nA), I measured the LED's reverse current to be 0.3 nA at 1.52 V and 0.5 nA at 2.05 V. These leakage current measurements are quite inaccurate.
I did. Again and again. I just don't get what he meant with »followed by a 11 k load«. Where is this load? In the following stage?
Best regards!
