Voltage reference level - how to choose?

[Magnethicc]

Hello everyone,
I have a question on how to select the voltage of the reference in SMPS feedback loop.
In the compensator you connect your reference voltage to the non-inverting input of the Error Amplifier.

How do you choose the reference voltage? why 2.5V and not 5V? why 5V and not 1.24V? what are the design criteria when choosing the reference voltage?


Thank you.

 

[KlausST]

Hi,

* Often used a bandgap reference voltage. About 1.25V, simple, cheap and good quality.
* if you use a higher voltage. like 5V, then you can not externally (simple) build a regulator for low output voltage.
Example: With a 1.25V reference you can build 1.25V regulators, 2.5V, 3.3V, 5V regulators with just adding a voltage divider in the feedback path. But when the reference is 5V it´s much more difficult to design a 3.3V regulator.

Klaus

 

[cupoftea]

Yes thats it......and the TLV431 comes with 1.24V reference so that you can do , say 2 vout SMPS's......but as you know, many SMPS chips have an 0.8V reference......and indeed, some PFC chips have a 7V reference (LT1248)......with the LT1248, its a PFC chip, and vout is 390V....so your feedback signal is less drowned by noise if you only divide it down to 7V......so in that case, its a SNR thing.......but the cheap TL431 is the cheapest reference on the market, and its 2.48V(?).....its used for many other things other than SMPS...and because its so cheap, its gotten adopted into the world of SMPS.

 

[Easy peasy]

We often use 10V for noise rejection ...

 

[cupoftea]

We often use 10V for noise rejection ...

This is a good comment...

...and i believe it harps back to what this thread was on anout from #10 onwards....

https://www.edaboard.com/threads/high-voltage-output-smps-shouldnt-use-basic-tl431-opto-feedback.402396/

...and that the added transistor there would mean not having to use such a high reference voltage?

It does appear that with HV output SMPS, the amount by which you have to divide down the vout is excessive and unfortunate....and results in poorer SNR.....hence the transistor circuit in #10 of the above.