Buck regulator chips

Modern buck regulators seem to have way too many features that I can't turn off or customize.

The otherwise very nice AOZ1282-1 snookered me by having a 600-us fixed short-circuit protection interval--it wouldn't start up into a 220 uF capacitor, even with a very light load. Fortunately, switching to 22 uF fixed it, and I had some with the right footprint.

And then many go into stutter mode at light loads, e.g. the MCP16311. Its brother, the MCP16312, is claimed to stay in vanilla PWM mode throughout the load range, but the data sheet lies. That bit me once when I was trying to make an isolated +24 -> +-5V converter for an industrial control system, using two coupled inductors. Simulated great, but the undocumented stutter mode of the MCP16312 screwed it up.

Chips these days.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs
Loading thread data ...

Grin, On the glass half full front, at least there are these IC's. What if you had to design it all with discretes?

I'm beating my head on some unknown bit of stray capacitance. The first circuit worked, the second doesn't. The second pcb now looks like a battle field of blue jumper wires. I've got it isolated to an inch of trace, but I can't see how it couples.... oh a new idea... I think I'm going to redo the pcb, but I'd like to understand a bit more what I did wrong.

George H.

Reply to
George Herold

Quit being so offensively reasonable. ;)

This one would probably work using an astable driving a PMOS at fixed duty cycle, followed by an LC lowpass (which is really all a buck is) and an LM1117 (or zener plus metal-can 2N2222) for regulation.

Yeah, you learn a lot from those post-mortems. Using SiGe:C BJTs is an education in layout all by itself.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

My latest fave is the TPS54302. It's a 6-pin SOT23, nice and dumb, and behaves very well.

There is no usable model (Web-bench never works for me) so I had to breadboard the various compensations, but that's done now.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Looks like a nice part, but distributor stock appears to be very thin.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

I'd better check on that. It's very similar to the AOZ part but, of course, different pinout.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

The fancier synchronous controllers allow you to run in full-wave mode, and some of the regular-switch kinds too. They're usually the upscale ones that feel like giving you those extra control pins to play with. So, they may break whatever budget or layout size restrictions you had.

That short protection mode is good to know about, thanks. Seems like I see that on chips with poor (or no) current limiting, or fixed timing designs (like a peak current mode controller, that turns on /every/ cycle and for at least a minimum pulse width, hence not actually limiting current into a short). I mean, even my discrete designs deal with that problem, but alas, we're stuck with ICs in practical designs.

What's so bad about burp-ripple in an "industrial control system"? In your low noise opto gadgets, I can see that, but..?

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

Yeah, there's a big run on quite a few chips at the moment, it seems. The AOZ1282-1 is in "full production" but nobody has any more till at least January. It's the 400 mA, 1-MHz version of the vanilla 1282, which is a 1A, 500 kHz part. Faster is nice when you're coming off a

24V rail.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

I was using it to make an isolated supply, so I need it to keep on PWMing under all circumstances. (It was a 'flybuck' design.) The main output didn't need much current.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Ah, kind of thought so. I've done that before, put in a SimpleSwitcher when I meant to use a synchronous one -- I just put a dummy load on the main rail, since efficiency wasn't a problem there.

I recall playing with TPS54233 in a coupled buck/neg flyback circuit and seeing it work just fine at most combinations of loads. Load on the negative output (and insufficient load on the positive buck output) draws negative current through the sync rect, which it seems to handle just fine.

Or I'm misreporting this because I tested it years ago. I can dig up the proto board if needed.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

...I'm greatly misreporting that, as it's not even a synchronous regulator. :^)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

Huh? There's a TINA model AND a PSpice model. The PSpice model should run just fine in LTspice.

Oooops! It's encrypted. Sorry about that... I can run it, you can't. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

[snip]

Those of you who build systems with parts from multiple sources should opine, "No _generic_ Spice model, no appearance on the BOM."

For me... models are no big deal.

When one of my customers wants a custom chip to work along-side an OTS device, but no generic Spice models exists for the OTS device, I simply make creating a Spice model a part of my quote... with the proviso that it can't be published/used elsewhere.

My customer gets a system-development advantage... and I get another "One more nose rubbed in the dirt" notch added to my ammo-belt >:-} ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

g

in output

er when

Well, you kind of thought wrong then. ;)

The Microchip part is a sync buck, and so would have worked fine if it had really been as the datasheet described.

Cheers

Phil Hobbs

Reply to
pcdhobbs

We just nabbed 500 of the TI parts from Arrow.

The 54302 works fine making 24 into 5 or 3.3. It gets inefficient making +1 (FPGA core voltage) from 24. For that, I went from 24 to 5, then 5 to 1.

The TI claims to run up to 3 amps, and it does, but it gets pretty warm. 2.5 seems more prudent. Not bad for a SOT-23.

It's not entirely clear how the heat gets out of this chip, so we hung some copper on everything.

formatting link

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

The breadboard worked fine. I hacked an eval board and used the actual caps that we'll use in production. Load current stepping gave me the transient response to tweak the loop compensation. The suggested TI values weren't very close, possibly because of cap ESR issues.

The breadboard gave me thermal information that a sim wouldn't. And I enjoy soldering and scoping stuff now and then. I wouldn't really trust the sim anyhow.

I wonder if the sim includes the spread-spectrum stuff. I was concerned if the spread-spectrum would show up as noise on the final DC output. Now I know.

The SS is pretty radical:

formatting link

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

But then you live in a very sheltered world, where foundries live and die by the quality of their libraries, so that a sufficiently skilled and careful worker can be pretty sure that the silicon will adequately match the model.

The board-level and (especially) the mixed-technology worlds aren't like that at all.

Cheers

Phil Hobbs

Reply to
pcdhobbs

Which was about 40% of the total world supply at that point. Weird--you don't look at all like Bunker Hunt. ;)

Yup, vias are free, pours are cheap, but field fails are _expensive_.

BTW we took your advice and standardized on the highest commonly-available wall wart voltage (+24). We use a pretty strict UVLO setting (19V) so that if somebody connects the wrong supply, nothing happens.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Maybe they will be the next bitcoin.

Yeah, somebody might accidentally plug a 12v wart into a 6v box, but it's unlikely anyone will have anything above 24, so no transzorbs needed. Maybe reverse protection, though.

This one is nice, rated 36 watts with good margin.

formatting link

Comes with all the international adapters.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

I went looking for wall warts and found that 48V seemed to be the next 'standard' value. I built a gizmo with one. (Power plug gets pitted by arcs when it is disconnected 'hot'.)

George H.

Reply to
George Herold

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.