Post by thread:[EE] Step-up/down DC-DC help, please?

I am trying to design a DC-DC converter (which is going to be a part of a bigger design) with the following specs: Vin = 12V Vout = 5 - 20V (variable) Load: 100uA-5A (well, maybe 3A max) Very relaxed requirements to line and load regulation (it need not be too accurate (+/- 0.5V) neither fast (0.1 sec)). Ripple of 0.5 - 1V is also OK. As you can see the load is highly variable with maximum power quite high. I can think of at least 3 approaches to this: 1. Step-up (dedicated chip as 34063, 3843...) to fixed Vmax=20V, followed by step-down as needed(another control chip). Seems to be complicated and costly. 2. Up/down at once like SEPIC (3843?) Not sure if it will be possible get it work for all loads with acceptable efficiency. (Efficiency itself is not of much concern in my design, but heat dissipation still matters....) 3. Since I am going to have a PIC in my design anyway, use its PWM outputs to control two FETs to form step-up/ step-down converter "on demand". This could be the simplest and cheapest solution, but wouldn't it be too complicated to form stable control loop in software. I will appreciate any thoughts on the above. Which approach looks better to you? Which control chips (if any) to use? May be I have completely overlooked something. If you can point me to any existing design - that's even better. - Arkady

Do you need to maintain a common ground? If not you can use a buck-boost topology which is similar to the boost but the return is made to the positive supply rail. The disadvantage is that the negatives are different for the input and the output. Essentially you are generating a boost voltage and then subtracting the supply voltage. As far as a suitable boost converter is concerned I think the topic has been pretty well covered in the past (archives). An alternative may be a simple converter driving a transformer. Your spec is pretty wide so it shouldn't be too hard to get something working You may need to add an additional load in order to be able to control the voltage at the lowest current level. Richard P On 09/11/05, Arkady <spam_OUTark1TakeThisOuTmyrealbox.com> wrote: {Quote hidden}> I am trying to design a DC-DC converter (which is going to be a part of a > bigger design) with the following specs: > > Vin = 12V > > Vout = 5 - 20V (variable) > > Load: 100uA-5A (well, maybe 3A max) > > Very relaxed requirements to line and load regulation (it need not be too > accurate (+/- 0.5V) neither fast (0.1 sec)). Ripple of 0.5 - 1V is also OK. > > > > As you can see the load is highly variable with maximum power quite high. > > I can think of at least 3 approaches to this: > > 1. Step-up (dedicated chip as 34063, 3843...) to fixed Vmax=20V, followed by > step-down as needed(another control chip). > > Seems to be complicated and costly. > > 2. Up/down at once like SEPIC (3843?) > > Not sure if it will be possible get it work for all loads with acceptable > efficiency. (Efficiency itself is not of much concern in my design, but heat > dissipation still matters....) > > 3. Since I am going to have a PIC in my design anyway, use its PWM outputs > to control two FETs to form step-up/ step-down converter "on demand". > > This could be the simplest and cheapest solution, but wouldn't it be too > complicated to form stable control loop in software. > > > > I will appreciate any thoughts on the above. Which approach looks better to > you? Which control chips (if any) to use? May be I have completely > overlooked something. If you can point me to any existing design - that's > even better. > > > > - Arkady > > -

> Do you need to maintain a common ground? > If not you can use a buck-boost topology which is similar to the boost > but the return is made to the positive supply rail. > The disadvantage is that the negatives are different for the input and > the output. Essentially you are generating a boost voltage and then > subtracting the supply voltage. > As far as a suitable boost converter is concerned I think the topic > has been pretty well covered in the past (archives). Yes, I didn't mention that but I do need common ground.