Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Help:DC-DC converter design

Status
Not open for further replies.

weeyndha

Advanced Member level 4
Joined
Dec 12, 2002
Messages
103
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,296
Activity points
607
Hello All,

I'm about to design a DC-DC converter with Vin = 7.2V and multiple output VO1=24V (Io=100mA), VO2=+15V Io=300mA, VO3=-15V Io=300mA, VO4=5V Io=2A. Which topology should I choose?

Rgds
 

You can use a flyback topology to get multiple outputs very easy.

Bastos
 

The flyback topology is the best for multiple outputs.

Look at national.com for the LM2587 switching regulator.
 

this might not be the best job for flyback - look at all combinations of vin/vout, you will need weird transformers to put it all on one block - many of national's products are built for one customer, not so good for general use.. (they have nice tutorials though!)

i would reccommend boost, boost, charge pump, buck in your order.. this takes more components but efficiency is high and these components are easy to find, compared to the weird multi-coil transformer that a flyback will require..

Have you built many DC-DC? You are familiar with each topology? In buck topology (Vout lower than Vin), you put a switcher between Vin and your inductor. Switcher allows little packets of power into inductor and output - if you need 2A!! on your 5v rail you MUST separate this system from your other supply, else they will ripple a lot due to how hard 5v rail is working.

In boost topology, the switcher comes after the inductor.. Switcher pulls inductor to 0v for a little while, then lets it spring up to high voltage - a diode makes sure the higher voltage is directed to the output. Both boost and buck are easy to work with and very high efficiency (>90%). They can also provide LOTS of power - good workhorses.

If you only need 300mA on -15v rail, I say build a 600mA +15v boost, then take half of that power and invert to -15v with a charge pump. A charge pump charges a capacitor to +15v, then flips it over to instantly give -15v. Also high efficiency, but only 80-90%.

A flyback will put all of these together into one transformer. You switch one side with 7.2v, and many other secondary coils will provide your other voltages. But they all need to be perfectly synchronized - hard to do if you need to make your own transformer.. Say 5v lead is a little too high - then flyback switcher will go a little lighter than it should and all other voltages will droop. With multiple regulators they are all independant! You can buy 3 of the same regulator chip, and adjust them all to different voltages with only resistor divider.

Let me know if you get this, I can refer you to some data sheets.
 

Flyback is a low cost solution for multi output power supplies; the cross regulation of its outputs is higher than forward mode SMPS, so if you use one of the outputs as master output(ususlly high current/low voltage one) to closing the loop, other outputs will track it good BUT not very good! (typically +-5%)
In fact the coupling of secondary windings determine the quality of cross regulation, thus the transformer must be cunstructed very carefully.
for more info to power integrations app. notes:
**broken link removed**
AN-16,17,18
also you can use more outputs to closing the loop and increasing the cross regulation, but for a limited range of load; also the design precedure is a time consuming try and error process!
you can find such ckts in other powerpoint app.notes.
I think if your application is not so economically critical use the method electronrancher discussed, i.e. using several independent regulator, its plus is easy design precedure of independent regulators!

BEST!
 

very good analysis electronrancher and goodboy_pl. Can you guys (or anybody) supply me with a link/website/App-notes where it gives description of how to build a ~400W PC Computer power supply (12V, 5V*~30A, 3.3V*~40A, -12V and -5V I think)? schematic/parts and descritpion of how it works will be a very good help...
I have been trying to find this for a while, but with no luck :( . Can somebody help me.

Thanks in advance...
 

Are you trying to build one professionally? If you just need a power supply for a home project, you will save time and money by simply buying a pre-made "brick" DC-DC converter. If you are starting your own company or something, I can find some datasheets and application notes that may interest you.
 

electronrancher, yes, I am trying to 'help' in co-founding a company, though this is not my field (I am an RF engineer), but I think I can help in co-founding such a company (small scale), so where can I find such info/app notes etc.? Thanks for sharing the info, and waiting for ur reply....

Cheers...
 

Many switch = many noise ?

Electronrancher,
Compare to flyback which is only use 1 switch, combination boost, buck and charge pump means more than one switch, would it be increase noise?

Rgds
 

Hi,
Maybe, you should think about such topology: buck 7.2V->5V and to get the others outputs by non-regulated push-pull? I'm afraide, you cannot get good cross-regulation in Flyback...

Regards
 

one flyback plus linear regulator help you

one flyback plus linear regulators fit for you
 

i would stay away from LDO's since they dissipate a lot of power in the pass transistor - you guys all know how hot 7805's get!

and no, the reason that seperate switchers give LESS noise is that in a flyback topology, all outputs come off the same transformer. When one output works hard (5v, 2A) and other outputs don't (12v, 300mA) the other outputs are still coupled to the hardworking 5v so they pick up switching noise.

Seperate converters will be very quiet since 5v can bang away into a big inductor and big capacitor, and the other outputs can use smaller inductors and caps for their light duty. Everyone will be very silent since they are happy with their own components.

National semi has a little downloadable evaluation program that helps you choose the best components quickly - you should start here once you decide on a topology. ps-Their chips are overpriced though! ;)
**broken link removed**

If you need help with anything, let me know.
 

be carefull when combining several independant switched power supplys!! Especially when each one is running with it's own frequency you can easiely increase output noise. No matter how good the filtering on the ouput of a smps is, you will always be able to destinguish between a switched and a conventional power supply. When having several independant switches the mixture of all noise frequencies can be found on every ouput.
So I would recomment coupling the phases of all these power switches (same sawtooth/triangle generator, for each feedbackloop one comparator).

The effect of noise is really worse for audio applications or high performance measurement.
 

So, which one is the best solution?

Dear All,

Since I'm building handheld data acquisition, which topology should I choose? Are there any references about measuring dc-dc converter performance?

Rgds
 

ok i agree with aoxamox (oa?) ;) running several 500kHz switchers in parallel is more noisy than 1 500khz, multi-outuput flyback, but!!!
1) having high and low power outputs on the flyback will be much noisier
2) noise will appear ABOVE the switcher for the most part, any good switcher will not couple other switchers' noise downstream into the other supplies - this is guaranteed for current-controlled PWM switcher and only slightly likely for PFM switcher (if it's crappy)
3) syncing them all together WILL cause more noise than you are trying to fix - all switchers turning on at one instant? that will suck 10A out of your supply even if you have 3 300mA switchers - the BEST thing to do would be (if your switcher had external clock, which most dont) run a single 500kHz oscillator (xtal divided down) then generate pseudonoise with XOR, shift register, etc like a cellphone. run all the switchers off different levels of the shift register, and they will all pseudonoise themselves into one very quiet system. please, this technique is not even on the market yet - it is the next generation ONLY for very very high-performance switched supplies, like cellular tower repeaters. spread spectrum brings many problems if implemented carelessly.

But my main point is...
4) If you want to build a PC power supply with very noisy components (processor, drives) the noise of your switchers will be tiny compared to these - please don't worry unless you need noise rejection of 60-80dB (1MEG-100MEG) which only interferes with audio if it is less than 20kHz. Most switchers are in the 300-500kHz range specifically to avoid harming audio quality.
 

Hello all,

I think I'm going to build using boost and buck topology. Thank you all for your time.

Rgds
 

Don't forget the best paper design can fall apart if the transformer construction and pcb/circuit layout is sloppy!
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top