3
\$\begingroup\$

For a linear power supply after the bridge rectifier stage, what is the best way to filter the output power?

  • Is it possible to use typical capacitance multiplier circuit with a power BJT or MOSFET, or should I use normal high capacitance capacitors in parallel?
  • What are the benefits and use case for either of these?
  • Why should I choose one over the other?
\$\endgroup\$
10
  • \$\begingroup\$ Exactly how low noise is ultralow noise? Give us some target figures. 10 nV/√Hz? 1 nV/√Hz? 100 nV/√Hz? \$\endgroup\$
    – Hearth
    Commented Aug 15, 2021 at 4:47
  • \$\begingroup\$ Best way, for what use case? How much current? And do you really mean noise, or do you just mean ripple? \$\endgroup\$
    – Justme
    Commented Aug 15, 2021 at 7:19
  • \$\begingroup\$ @Justme I mean for both ripple and noise. 3A continuous current. And maybe 1uV RMS ripple? \$\endgroup\$ Commented Aug 15, 2021 at 7:31
  • \$\begingroup\$ capacitance multipliers do not reduce noise, they only reduce ripple. \$\endgroup\$
    – tobalt
    Commented Aug 15, 2021 at 10:18
  • \$\begingroup\$ @tobalt Capacitance multipliers are great for reducing noise. With a BJT, you can get down to the emitter noise level, usually very low. \$\endgroup\$
    – John Doty
    Commented Aug 15, 2021 at 12:25

2 Answers 2

7
\$\begingroup\$

You need a reservoir capacitor to store energy between pulses from the rectifier.

A capacitance multiplier is a type of series regulator it does not store (much) energy.

There's no reason why you can't use both, a reservoir capacitor to store energy followed by a capacitance multiplier to chop the peaks off the ripple, allowing a low-ripple output without sacrificing too much power factor by using an oversized reservoir capacitor.

\$\endgroup\$
8
  • \$\begingroup\$ I like the way this is couched, despite being short. Says a lot, tersely. +1. \$\endgroup\$
    – jonk
    Commented Aug 15, 2021 at 6:09
  • \$\begingroup\$ Good post that emphasizes the right things, namely that it is sort of a series regulator and doesn't store energy and thus works against ripple (turning it into heat) but does not reduce noise. It buffers the potential (incl. noise) of the input RC. To reduce this noise, only a larger C is effective. \$\endgroup\$
    – tobalt
    Commented Aug 15, 2021 at 11:41
  • \$\begingroup\$ But then why would you use a capacitance multiplier instead of an actual regulator, which has effectively infinite capacitance? \$\endgroup\$ Commented Aug 15, 2021 at 14:34
  • \$\begingroup\$ @tobalt what's the difference between ripple and noise? \$\endgroup\$ Commented Aug 15, 2021 at 20:38
  • \$\begingroup\$ @user253751 capacitive multiplier has a reduced headroom requirement versus LM7805, and even LDO regulators. etc \$\endgroup\$ Commented Aug 15, 2021 at 20:40
0
\$\begingroup\$

Based on your comment about a bridge rectifier, I’m assuming you’re talking about AC to DC conversion.

what is the best way to filter the output power?

This depends on the application and requirements. There’s usually a trade off between physical space, efficiency, and output ripple. The type of load may affect how much capacitance you can have on the output as well.

Is it possible to use typical capacitance multiplier circuit with a power BJT or MosFET?

Not only is this possible but it’s fairly common for low noise applications. It’s more common to use a dual capacitance multiplier however which allows you to control the capacitance without relying heavily on the gain of the transistor which is device dependent.

or should I use normal high capacitance capacitors in parallel?

This depends on your requirements

What are the benefits and use case for either of these? And why should I choose one over the other?

The capacitance multiplier is typically used in very low noise situations where the amount of capacitance needed is impractical to accomplish directly. Some reasons not to use it however are

  1. If you don’t need really low noise, you may end up using more space with a capacitance multiplier than just using parallel capacitors.

  2. The transistors are going to dissipate some of the power, so you will lose efficiency.

  3. The capacitance multiplier is more complicated which introduces some risk that may be unnecessary if you can meet your requirements with just capacitors.

\$\endgroup\$
5
  • \$\begingroup\$ Hey Ryan, The use case is for a power supply. So, after the transformer to the bridge rectifier stage and then comes the DC filter stage. I am aiming to achieve very low noise and ripple for the psu. Is it possible for you to explain the number 2 reason further more? Like what type of power loss will I have? And also any formula to calculate the power loss through the pass transistor? Also, do you have suggestions for the pass transistor to use for 24V@7A? Also, for the number 3 point, what are the other risks you think it has and also why is it complicated if I am using a power BJT/FET, \$\endgroup\$ Commented Aug 15, 2021 at 4:07
  • \$\begingroup\$ Hi Roger, this sounds like a fun project. If you need help with a specific design, we will need to know as much specific information about the design as you can give us. Do you have a noise requirement, or can you tell us what the power supply will be used for? The power loss in the transistor is essentially the diode drop times the collector current (V_CE * I_C). A capacitance multiplier is a more complex circuit than a few capacitors. It’s easier to get it wrong, hence more risk. \$\endgroup\$
    – Ryan
    Commented Aug 15, 2021 at 6:42
  • \$\begingroup\$ the overall specifications are a little vague because I am trying to build a bench precision low noise power supply to test all of my future projects. I will be working with Radio equipment and also normal microcontroller based automation stuff. I am going forward through the design step by step starting from the mains input and transformer and then slowly going towards the regulated output. Trying to figure out each module individually. \$\endgroup\$ Commented Aug 15, 2021 at 7:03
  • \$\begingroup\$ Right now I am trying to figure out the bridge rectifier and secondary filtering stage after the transformer. Also, I was wondering if I should use a single or double fuse after the transformer's secondary winding and then build up the rectifier and filter section. But then for the fuse, I have to take account of the filter capacitors inrush current. Also. I am planning to build 3 galvanic isolated channels. So, I should start the isolation from the bridge rectifier stage isn't it? Like right after the transformer secondary AC output? Also, the ripple I am trying to go for is 1uV RMS. \$\endgroup\$ Commented Aug 15, 2021 at 7:06
  • \$\begingroup\$ Just want to implement all the best possible design practices and requirements while building this project and the lowest possible ripple I can get at a reasonable size and also reasonable for the use case. Low voltage and current Radio testing and also low level microcontroller stuff. \$\endgroup\$ Commented Aug 15, 2021 at 7:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.