http://przyrbwn.icm.edu.pl/APP/PDF/119/a119z4p10.pdf -
Above link summarizes LED noise well.
Defects that occur with aging determine noise.
Edit - not to forget about defects right from the manufacturer !!
"Non-radiating recombination" = a cheap chinese blue that drops
in output after 5K hours. 😀
I saw tempco figures posted between 1.8mv/K - 2.2mv/K
http://www.onsemi.com/pub_link/Collateral/TND328-D.PDF
3+ mv for the reds ...
But , there a three different blends for red ! (cool link ahead)
Technical LED Color Chart
So between InGaAs and GaAsP red's , one could have 1.6 - over 3mv/K.
PS - the resulting red LED CCS would range between NTC and PTC ,
depending on the sourcing.
I have red led CCS's that are NTC by design (a specific red is used).
If you know the EXACT blend of your particular LED (UV to IR), the
Tc and Vf are predictable enough to design around.
A "junkbox" or very old ,used might be harder to design around because
of aging or other unknowns.
OS
Above link summarizes LED noise well.
Defects that occur with aging determine noise.
Edit - not to forget about defects right from the manufacturer !!
"Non-radiating recombination" = a cheap chinese blue that drops
in output after 5K hours. 😀
I saw tempco figures posted between 1.8mv/K - 2.2mv/K
http://www.onsemi.com/pub_link/Collateral/TND328-D.PDF
3+ mv for the reds ...
But , there a three different blends for red ! (cool link ahead)
Technical LED Color Chart
So between InGaAs and GaAsP red's , one could have 1.6 - over 3mv/K.
PS - the resulting red LED CCS would range between NTC and PTC ,
depending on the sourcing.
I have red led CCS's that are NTC by design (a specific red is used).
If you know the EXACT blend of your particular LED (UV to IR), the
Tc and Vf are predictable enough to design around.
A "junkbox" or very old ,used might be harder to design around because
of aging or other unknowns.
OS
Last edited:
Feh. Try 3 stroppy teenage girls, the youngest of whom is capable of throwing an armchair. Soon they will be unleashed on an unsuspecting world. (cue maniacal laughter).
So far the mancub just squeals loudly, but that will change.
Die of the LST676 looks about that
http://www.leds-and-more.de/images/SMD/PLCC2/SMD-3528-PLCC2-S.jpg
I'd assume a miniature LED as the TLSU1008 (25mA If) need not have a die that's even smaller.
Good references. Thanks!
I got to the HLMP-6000 by starting with the HP "standard red" ones in radial 3mm packages, which were the ones alleged to match the silicon Vbe tempcos in the PMI app note. I believe the material was then listed as GaAsP. I'm pleased that they seem to be lower noise than some other examples given.
One additional thing to consider: the tempcos of the transistor and probably the LED will vary somewhat with operating currents.
As far as competition with other current generators, one can derive a figure of merit in terms of reference current to noise ratio. As my number crunch some posts back shows, for the 1mA example, the emitter resistor strongly dominates the current noise at the collector for reasonable assumptions about the LED and the transistor. As we attempt to get higher currents the resistor current noise will go up reciprocally as the square root of the resistance, and the current go up just reciprocally. However, the importance of the base bias noise (from the LED), the base current noise, and the transistor voltage noise from rbb' will all get more important. One really would like more delta voltage to play with. But it remains a very useful circuit, as well owing to having rather high voltage compliance. I use them a lot to supply currents to the input devices of a diamond complementary output stage, usually embedded in an overall loop.
Compared to the "ring-of-two", despite the transistors readily obtainable in opaque packaging, thus alleviating any residual fears of photonic interference, the tempco is fairly large and one only has a Vbe to play with. Thus the resistor's contribution will be proportionally greater, all other things equal. In addition the base current of the output device and collector current of the second device must be supplied by something reasonably quiet and stable. But then that's true for the LED bias as well.
Nothing has been said about the output impedances of these circuits either. And of course there are many modifications to improve them if needed. A bit of cascoding has the additional advantage of greatly reducing the self-heating effects with voltage swing, as well as mitigating Early effects.
I believe Walt Jung has done a lot of good work with current sources and has published about this on his site.
Yes, although the LEDs measured were fair thumpers.
Figure 2 suggests, for a fresh LED of that type at reasonable current, both optical and electrical noise @1kHz all in is about 0.1nV/sqrt(Hz) or so..........
This is surely not going to frighten the horses............🙄
No. But actually the dependence of electrical noise on current density is complicated according to their research. Granted that the LEDs they tested are big suckers, too little current can be noisy.
And I'm not sure how much the Al and the In and their compounding change things for the red ones.
But a very nice paper in any case, particularly the determination of correlations.
True by the way all my horses are battle hardened not gun shy any way . Interesting to see 10 hz to 20khz used for noise bandwidth.
What is the reason to use LEd in place of zener diode? The LED doesnt have a steep V-I curve. As a regulator or constant voltage source, the LED is not very good.
THx-RNMarsh
THx-RNMarsh
Last edited:
For some red ones the temperature coefficient of forward voltage is a good match to a silicon transistor's Vbe, hence making simple and stable current generators possible.
And, based on several measurements, they have very low noise in that forward-biased voltage.
The LED seems like a cheaper alternative only. But with reduced performance as a regulator. Noise of zener can be very low... dynamic Z of 10 ohms and if suitably bypassed isnt noisy.
The TempCo of +2mv/K and low noise can be found in zener around 5.6 - 5.9v.
file:///C:/Users/15-P006/Desktop/The%20Circuit%20Designer's%20Companion%20-%20Tim%20Williams%20-%20Google%20Books..html
THx-RNMarsh
The TempCo of +2mv/K and low noise can be found in zener around 5.6 - 5.9v.
file:///C:/Users/15-P006/Desktop/The%20Circuit%20Designer's%20Companion%20-%20Tim%20Williams%20-%20Google%20Books..html
THx-RNMarsh
You want minus rather than plus 2.2mV/K for the particular circuits, so a lower voltage indicated. But although not bad, and helped in the current generator app due to higher voltage across the emitter resistor, the noise is still much higher than a red LED. And that's without any capacitor's bypassing.
I regulate with a zener first , power my red LED CCS's after.
They are cheap , visually indicate operation .... one can pick
the right Tc in the LED to counteract another PTC/NTC elsewhere.
I've achieved the same performance using either a 2Q or LED.
A 2Q will achieve higher PSRR , but 2 - 2Q's (like in a blameless VFA) -
will be more device dependent for the circuits total Tc.
I can tweak the LTP Tc on my "blameless" .... (negate the VAS CCS)
Edit , I agree ... very bad voltage regulator.
OS
Last edited:
Plus and neg TC can be had with zener, of course. The zero TC is around the 5v level. It doesnt matter that an LED may be quieter un-bypassed. Bypass the zener. And, have a higher performing volt regulating function.
Or is noise the main or most important issue? For what app, topology or use?
THx-RNMarsh
Last edited:
Not at forward voltage range where zeners and LEDS might otherwise be considered similar, as used in bias apps for example. Often the exact convenient forward voltage drop is the deciding factor in choice of diode, subject to correct tcoeff IME.
At 10mA current, an HLMP6000 red LED has about a 3R impedance. Cheap surplus red LEDs clocked in at about 4R5-5R. I measured about 10R slope resistance from the 6.2V Zeners I have on hand, so normalizing for voltage, the slope resistance is a wash.
What's the size of a cap that would have a significantly lower impedance, which is the point of bypassing? Quick calculation- to get a decade lower impedance from the Zener, you need 1R in the bypass. Putting aside subsonics and ignoring ESR, at 20Hz that's about 8000uF. Is that a practical solution?
The most cost effective is the LED for a regulator. And, they can be quiet. The example (6.2v) zener at 10 Ohms Z is fine IMO. Esp for line level or PA circuits. I would use an LED with a CCS as others do. But then costs are higher. Which brings you back to the single part - zener. Its all a wash Except the zener regulates better than an LED.
If you want both better volt regulation and lowest noise.... use a CCS+zener followed by a Cap-multiplier.
THx-RNMarsh
If you want both better volt regulation and lowest noise.... use a CCS+zener followed by a Cap-multiplier.
THx-RNMarsh
Last edited:
Steven, I wanted to make sure that you will rethink over the universal applicability of that “music is bringing us all together”
I remembered your words today (me listening to the old cassette, everyone driven out of the house)
Prepare thyself to deal with a miracle
https://www.youtube.com/watch?v=17JgEYTmE-o
🙂
George
Last edited:
That's wacky. Good old Rahsaan. And recorded from LP, one I had never seen. I'm going to post it on my FB page, perhaps with a warning.
Hard to believe he's been gone since 1977. When he first came to my attention I had to try playing multiple horns as he did. I stopped at two.
- Status
- Not open for further replies.