I have been thinking about building a Sulzer-based regulator, using a led as a reference instead of a zener diode. It is cheaper, and I have read that it is quite low-noise too. I believe that with a classic filtering it would be excellent, biased at around 1mA.
What do you think? And a bonus question: do regulators work better when they use as reference a small or big part of the output voltage, or is this irrelevant?
EDIT: Is there any reason why certain colors are superior to others, in terms of noise/stability?
What do you think? And a bonus question: do regulators work better when they use as reference a small or big part of the output voltage, or is this irrelevant?
EDIT: Is there any reason why certain colors are superior to others, in terms of noise/stability?
Last edited:
@audiostrat
Interesting reading: LED noise masurement
Yes, if you need low noise low voltage reference, LED is your best choice. If you need 15V or more, it is easier to use Zener than stack of LEDs.
Bias your LED with low current. Yes, dynamic impadance will be slightly greater if biasing with low current, but as you will use additional RC filtering, dynamic impedance doesn't matter.
With LED, be careful of PSRR and use CCS to privide constant current to your LED and to increase PSRR.
Your noise will be lower if you use reference closer to output voltage. You can use 10V reference and divide to 5V with resistor network. Never use 1V reference and crank up the gain (5x) to reach 5V, if you need low noise.
Green, red, yellow LEDs have lower noise than blue and white (generally speaking).
Interesting reading: LED noise masurement
Yes, if you need low noise low voltage reference, LED is your best choice. If you need 15V or more, it is easier to use Zener than stack of LEDs.
Bias your LED with low current. Yes, dynamic impadance will be slightly greater if biasing with low current, but as you will use additional RC filtering, dynamic impedance doesn't matter.
With LED, be careful of PSRR and use CCS to privide constant current to your LED and to increase PSRR.
Your noise will be lower if you use reference closer to output voltage. You can use 10V reference and divide to 5V with resistor network. Never use 1V reference and crank up the gain (5x) to reach 5V, if you need low noise.
Green, red, yellow LEDs have lower noise than blue and white (generally speaking).
Last edited:
You Might want to look here. http://www.diyaudio.com/forums/parts/35821-some-noise-measurements-leds-zener-diodes.html
LM329 buried Zener are still in stock even though EOL
lots of industrial ADC/DAC Vrefs have been made over the years, some innovations like the AD XFET promise even lower noise
but for audio heavy RC low pass can give fine dynamic noise - the kind you can hear - while TC, drift, accuracy really don't matter as much for many audio circuits
lots of industrial ADC/DAC Vrefs have been made over the years, some innovations like the AD XFET promise even lower noise
but for audio heavy RC low pass can give fine dynamic noise - the kind you can hear - while TC, drift, accuracy really don't matter as much for many audio circuits
Zeners are horribly noisy as all reverse biased PN junctions. Forward biased junctions (like LEDs) are way more quiet.
LEDs are practical as they show you that current runs. Temperature wise they are not the most stable but for a lot of circuits that is not so critical.
best,
A filtered Zener is usually low enough noise for most audio circuits.
A stack of red LEDs will have more noise than a single red LED. Filtering will reduce that noise.
300mW to 500mW Zeners are cheaper than red LEDS. expect 1p:2p when bought in bulk.
A stack of 6 red LEDs comes to about 12p compared to 1p for a 12V Zener.
A stack of red LEDs will have more noise than a single red LED. Filtering will reduce that noise.
300mW to 500mW Zeners are cheaper than red LEDS. expect 1p:2p when bought in bulk.
A stack of 6 red LEDs comes to about 12p compared to 1p for a 12V Zener.
I like TL431 as well, a very handy device. Way back when I was making voltage references I concluded that HF noise is of little consequence because you can always filter it out. What was much harder to deal with was LF noise on the order of Hz or less. It wasn't practical to filter it out and it didn't always average to zero over a short time period. Not good for a calibration standard. For audio circuits I'd think only the HF noise is an issue, and this is easier to deal with, meaning there are really no horribly bad choices. PPM DC stability just isn't an issue.
A Zener diode is a diode which allows current to flow in the forward direction in the same manner as an ideal diode, but also permits it to flow in the reverse direction when the voltage is above a certain value known as the breakdown voltage, "zener knee voltage", "zener voltage" or "avalanche point" from Wiki.
Thanks Author
Thanks Author
LM329 reference is an excellent choice for a 12V or 15V regulator.
It will a) center (more or less) the common-mode voltage on the opamp input pins at half-supply (== 1/2 output voltage) and b) allow the reg to be run at low gain with a high feedback factor. And c) it will allow for enough volts accross a constant current source or resistor to feed the reference. All this will help perfomance compared to references that are significantly lower or higher voltage than half the output voltage.
It will a) center (more or less) the common-mode voltage on the opamp input pins at half-supply (== 1/2 output voltage) and b) allow the reg to be run at low gain with a high feedback factor. And c) it will allow for enough volts accross a constant current source or resistor to feed the reference. All this will help perfomance compared to references that are significantly lower or higher voltage than half the output voltage.
Last edited:
If you have a shared design (common GND for 9V and 16V and common input) you could just use a single reference and two opamps all running from the 16V line, one at slightly lesser gain to give 9V instead of 16V.
You could also use TL431's with a bit of filtering instead of LM329. Many options to choose from, depending on what level of circuit perfomance and complexity/parts count you're after.
You could also use TL431's with a bit of filtering instead of LM329. Many options to choose from, depending on what level of circuit perfomance and complexity/parts count you're after.
Each supply sill be separate.
I will implement Sulzer regulator, which already has very good performance for my needs. What is a fact though, as I stated above, is room. I need physically small references. LM329 is TO-92 which is good, a zener is also as small as it gets. TL431 is a whole IC as I see, so is ruled out.
As I said before a zener should suffice (the original Sulzer uses that), but why not try something better that fits my tight layout. LM329 is good as far as both performance and size is concerned. But if it is not going to suit me for the 9V supply, I won't use it after all and it's ok. But I think it will, won't it?
I will implement Sulzer regulator, which already has very good performance for my needs. What is a fact though, as I stated above, is room. I need physically small references. LM329 is TO-92 which is good, a zener is also as small as it gets. TL431 is a whole IC as I see, so is ruled out.
As I said before a zener should suffice (the original Sulzer uses that), but why not try something better that fits my tight layout. LM329 is good as far as both performance and size is concerned. But if it is not going to suit me for the 9V supply, I won't use it after all and it's ok. But I think it will, won't it?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.