General > General Technical Chat
LEDs as a low noise shunt regulator?
RoGeorge:
--- Quote from: mawyatt on December 27, 2021, 02:52:15 pm ---He sent me a note long ago (think he was still at Analog Devices) and told me my Current Source had been published in Current Sources & Voltage References by Harrison (I had no idea it was published outside EDN), also later asked if he could use the Equal Valued 3rd Order Butterworth active filter he had seen in EDN in a book he was writing.
--- End quote ---
You mean this EDN article, about Peaking Current Source?
https://www.edn.com/peaking-current-source-has-high-rejection/
Interesting circuit I didn't know before, thank you.
Wow, and better stabilization than the Widlar Current Source?! :-+
Wyatt Current Source vs. Widlar Current Source :box:
First round score: 1 - 0 ;D
mawyatt:
Thanks, yes that's it :)
The story behind this current source is in late 80s to early 90s we were developing the first fully integrated MicroWave Receiver in a new silicon Bell Labs process (CBIC V2) and based upon a unique "Synchronous Homodyne" architecture, which is a variation of what was later called Direct Downconversion or Zero IF and culminated in patent 5603111.
During the course of the development a requirement arose which demanded operation from a single +5VDC source which had significant noise, both low and high frequency content (from a noisy switcher) and a size requirement which wouldn't allow any decoupling capacitors or other external components other than 1 resistor and capacitor (used for offset correction)!! With this in mind we set out to develop on-chip regulators which had good LF and HF rejection, and also a curvarture-corrected bandgap reference. The regulators and bandgap reference required various current sources and we needed very high overall PSRR, so we came up with this Peaking Current Source Reference. The bandgap design with this current source got some interest and was published within a Bell Labs Journal/Paper before later we got permission to release to EDN.
Classic example of a stiff requirement being the "Mother of Invention" :)
Best,
TimFox:
The Morgan Jones references to using "cheap" LEDs were originally for cathode bias voltage in vacuum-tube preamplifiers, to avoid large bypass capacitors. In that application, Merlin Blencowe (in "Designing High-Fidelity Tube Preamps, 2016, pp 210-211) points out that in forward-biased diodes (rectifiers or LEDs), the noise current riding on the DC value is given by shot noise calculations, but the noise voltage depends on that noise current and the slope resistance, which is why IR or red LEDs (with lower forward voltage and slope) are often preferred.
David Hess:
Using an older LED with -2mV/C temperature coefficient to bias a single transistor current source/sink makes a low noise temperature compensated current source/sink.
The transistor Walt Jung recommended is a Zetex "super e-line" type with very low base spreading resistance which results in low noise. Others make these now for their high gain and low saturation voltage but they are not always easy to identify.
I now wonder if Zetex named those transistors "super e-line" because they were exposed using the 546 nanometer e-line. The way I remember it, they bought obsolete photo-lithography equipment originally used for ICs and made transistors with it, so they could support a much finer feature size.
Navigation
[0] Message Index
[*] Previous page
Go to full version