Designing a ripple low PS for DDS function generator

[matthieu1973]

TL;DR Summary

help with designing a MC34063 based ripple free PS.

Hello to all.

I am trying to design a ripple free (read as ripple free as possible) power supply (PS) for my DIY DDS function generator.
I am (was) in the possession of the hyland 5v to 12v PS which wrecked due to a stupid action on my side, my bad.
so i was going to repair it, but i found that this PS has not really a clean output.

So I decided to go the other way and try to design a better one with all the info that is available on this digital highway.
I have a background in physics and some experience with tinkering on electronics but this is a first time for me designing anything so I would appreciate some nudges into the right direction thought wise.

As basis of my design I am currently thinking of using the MC34063 instead of the XL6008 the hyland PS uses, as this IC is duscussed in EEVblogs tutorial "EEVblog #110 - Let's Design a DC to DC Switchmode Converter" .
But am still open to change on this.

The design will need the following specs:
1: As DC input I have a 100W variable PS with 12 to 24V output in 2V steps.
2: Designed maximum load capability of 1A.
3: As DC output I will need +12V, -12V and +5V
4: the 5V is for powering the DDS, the +/-12V is just as the 5V used for adjusting the offset and amplitude, clean input on all three is there for important.

As basic setup I have "designed" the following specs.
1: Complete separate rail for all voltages.
2: DC input either of 14 or 24V (i am not sure which would be most benefitial, if there was a difference)
3: MC34063 setup as either step-down or voltage inverter as desired for that rail, all according to the datasheet
(calculations are being done but depend on some factors I hope to discuss here)
4: MC34063 output to be designed to be slightly higher that the intended voltage.
5: MC34063 output will go thrue a output filter (to be designed but thinking of capacitance multipliing) to filter as much of the ripple out as possible
6: from output filter to a LM7812, LM7912 and LM7805 to set the right voltage.
an other option here would be to use the LM7812 output as input for the LM7805.
7: Output into the DDS generator.

For now I am going with these specs and will continue the calculations.

And need to deside the switching frequency.
For this I have the following data from the data-sheet between 100Hz to 100KHz

and I am not sure what would be best, personally I feel as low as a frequency would create less amount of ripples, but in practise I find most designs going for the highest working frequencies, which kinda baffals me.

So going from this. anyone have any constructive criticism?

 

[berkeman]

Welcome to PF.

For voltage regulation on RF transceiver circuits that I've designed and worked on, I would use a switching DC-DC front end for efficiency, but follow that with low drop-out linear regulators. That has worked well for me, to keep the full range of sensitivity of the RF receiver sections.

 

[matthieu1973]

@berkeman: thanks for your reply!
the MC34063 would be a switching DC-DC front end here right?
or is that a wrong judgement? and as low dropout liniear regulator a LM350A?

Kind greats

Matthieu

 

[berkeman]

I haven't used the LM360A, but it looks like it could work. Be sure to check the datasheet for the LDO -- often there will be a minimum and/or maximum output capacitance requirement for stability reasons.

 

[matthieu1973]

ps: you speak of LM360 while i mentioned the LM350 is it an error or do you actuallyy refer to the 360?
Both exsist but the 350 seems more appropriate to me for this cause but i could be wrong ofcourse.

 

[berkeman]

ps: you speak of LM360 while i mentioned the LM350 is it an error or do you actuallyy refer to the 360?
Both exsist but the 350 seems more appropriate to me for this cause but i could be wrong ofcourse.

No, my apologies. That's a typo on my part that I didn't notice.

 

[DaveE]

Summary:: help with designing a MC34063 based ripple free PS.

feel as low as a frequency would create less amount of ripples

Nope. The energy is delivered to the output in pulses at the switching frequency. Smaller pulses more often will reduce the ripple. That also allows you to have an output filter that is either simpler or works better. 200KHz is (roughly) the sweet spot for this sort of low noise hard switched PS design.

If you are going to have linear regulators to clean up the outputs, then you could have one primary side SMPS that drives a transformer with 3 output windings. Be mindful of minimum load issues if you do this, so the outputs will track reasonably well.

Analog.com (was Linear Technologies) has the best low noise SMPS control ICs as I recall. They and TI.com have a lot of support information available about different topologies and controllers. You might check out Maxim too (also bought by Analog).