Buck regulator of the month....

[mikeselectricstuff]

Whenever I design in a buck regulator seems like there's a different 'cheapest' one...

This is the nicest one I've found for a long time....
http://www.digikey.co.uk/product-detail/en/AOZ1051PI/785-1279-1-ND/2769839

0.8 to 5V out, 3A (slightly cheaper AOZ1050 for 2A)
Synchronous rectifier = efficient (>90% typ)  and no external diode
Minimal external components, ceramic caps
Small inductor
Sensible SO-8 thermal package
Only minor gripes are the thermal pad is output not ground,  and input only up to 18v.
And the best part.... Under GBP0.40 100x, making it a good choice even for much lower
power situations where a 7805 runs a bit too warm.

[T4P]

Too bad it's only on Arrow and Digikey, those 2 companies cost a bomb to deliver to me

[Bored@Work]

This is the nicest one I've found for a long time....

And a nice datasheet. I knew they are an American company without looking it up, when I saw they were arguing with poles/zeros in the datasheet. Some American EEs are really obsessed with them. Must have to do with their typical curriculum.

0.8 to 5V out,

The datasheet says one can go up to 0.85 * Vin. For a 12V input that would be more than 10V out.

Only minor gripes are the thermal pad is output not ground,  and input only up to 18v.

And that they didn't make use of the NC pin for something useful, like providing the internal 5V LDO output, the reference voltage or adjusting the current protection.

[vxp036000]

And when your circuit self destructs from parasitic oscillation, we all know why.  Another typical "the math is meaningless" engineer (I'm using the term engineer loosely here).  The designers I knew that did not bother with bode plots found themselves walked out the door after their circuits consistently failed to perform.

And a nice datasheet. I knew they are an American company without looking it up, when I saw they were arguing with poles/zeros in the datasheet. Some American EEs are really obsessed with them. Must have to do with their typical curriculum.

[Kremmen]

[..]
And a nice datasheet. I knew they are an American company without looking it up, when I saw they were arguing with poles/zeros in the datasheet. Some American EEs are really obsessed with them. Must have to do with their typical curriculum.
[...]

Obsession? We used to consider it one element of competence. In fact we still do.

[vxp036000]

I hope we still consider feedback theory an element of competence.  I certainly won't let any EE near a design unless he/she demonstrates a strong understanding of feedback.  Without it, low noise regulators break into oscillation, oscillators won't oscillate, RF amplifiers self destruct from parisitic resonance, and the list goes on.

Obsession? We used to consider it one element of competence. In fact we still do.

[mariush]

The minimum 4.5v in is a bit of a downside, depending on what you need it for.

Maybe you want to use 4 rechargeable AA batteries to power something... 4x1.2v is kind of close to 4.5v

Also, maybe you want to make a usb powered device and user plugs the phone charger instead of your power adapter and it turns out that charger outputs 6v at no load, 4v or less at 1A... like that cheap chinese usb hub power supply Mike made video of.

[T4P]

The minimum 4.5v in is a bit of a downside, depending on what you need it for.

Maybe you want to use 4 rechargeable AA batteries to power something... 4x1.2v is kind of close to 4.5v

Also, maybe you want to make a usb powered device and user plugs the phone charger instead of your power adapter and it turns out that charger outputs 6v at no load, 4v or less at 1A... like that cheap chinese usb hub power supply Mike made video of.

4x1.2v ... only applies to Ni-MH/Ni-Cad's
But isn't it adjustable ?

[mikeselectricstuff]

Re. all the bode plots etc. I can't say I really understand it - I rely on the Bob Pease method of testing PSU stability -- "Bang on the output and see what happens" - by applying load transients you can pretty easily tell how close it is to stability or otherwise.
 

[mariush]

4x1.2v ... only applies to Ni-MH/Ni-Cad's
But isn't it adjustable ?

The datasheet says it takes minimum 4.5v, maximum 18v, so it won't work with various scenarios like generating 1.8v from 2 AA batteries and like I said, it might be bad to use it with USB power... sometimes even the computers output 4.8v+... if you further plan to use a diode for extra safety it gets messy.

But then, maybe for stuff running on batteries something like Intersil's ISL9111 might be a better deal : http://search.digikey.com/scripts/DkSearch/dksus.dll?WT.z_header=search_go&lang=en&site=us&keywords=ISL9111&x=0&y=0

It starts from as low as 0.6-0.8v and can reach 97% efficiency (with 3.2-4v input), output up to 5v... nice chip.

[vxp036000]

I'll try to make it simple.  Plot the magnitude of the transfer function vs. frequency.  Now plot the phase of the transfer function vs. frequency.  Make sure that the gain (i.e., magnitude of the transfer function) drops below unity where the phase crosses 180 degrees.  This assumes an inverting amplifier.  When we design regulated power supplies, we always check the bode plot to make sure the feedback is negative up to and including the unity gain frequency.  Otherwise, who knows what sort of crazy response the circuit will end up with.

Re. all the bode plots etc. I can't say I really understand it - I rely on the Bob Pease method of testing PSU stability -- "Bang on the output and see what happens" - by applying load transients you can pretty easily tell how close it is to stability or otherwise.

[Bored@Work]

Yes, obsession. There are different ways to describe the required loop compensation, just that Americans obsessively prefer arguing poles and zeros.

The datasheet is such an example. Arguing with poles and zeros over almost one page but beating around the bush and not giving the transfer function, only fragments. Complete with the typical American phrases about "setting the pole-zero". And at the end of one page of arguing the datasheet just gives two simple equations to calculate Rc and Cc, derived not the least from using rules of thumb (e.g. 1/10 of the switching function). But great that we talked about poles and zeros.

And this is typical for US literature. Somehow they prefer to come up with poles and zeros. And the second US specialty is arguing with them and arguing about moving poles and zeros around in the s-plane. Instead of giving the transfer function they like to argue with artifacts of the transfer function. They just can't help it.

BTW if you look up my previous posts you'll find that I am one of those engineers disagreeing with Dave's "no one needs math" approach. But I am also one of those engineers who think there is no point arguing poles and zeros just because you are obsessed with them.

[vxp036000]

I apologize; I probably should have looked at the datasheet.  Yes, there is no excuse for not simply giving the transfer function and showing magnitude / phase plots.

Yes, obsession. There are different ways to describe the required loop compensation, just that Americans obsessively prefer arguing poles and zeros.

The datasheet is such an example. Arguing with poles and zeros over almost one page but beating around the bush and not giving the transfer function, only fragments. Complete with the typical American phrases about "setting the pole-zero". And at the end of one page of arguing the datasheet just gives two simple equations to calculate Rc and Cc, derived not the least from using rules of thumb (e.g. 1/10 of the switching function). But great that we talked about poles and zeros.

And this is typical for US literature. Somehow they prefer to come up with poles and zeros. And the second US specialty is arguing with them and arguing about moving poles and zeros around in the s-plane. Instead of giving the transfer function they like to argue with artifacts of the transfer function. They just can't help it.

BTW if you look up my previous posts you'll find that I am one of those engineers disagreeing with Dave's "no one needs math" approach. But I am also one of those engineers who think there is no point arguing poles and zeros just because you are obsessed with them.