Boost converter wisdom

Hi, all,

I'm doing a miniaturized version of my low noise diode laser controller for a big biomed customer. It's supposed to replace an existing unit for retrofitting, as well as being designed in to their new products.

The wrinkle is that it has to run off a single +5V supply instead of our usual +24V. Sooooo, it needs a boost converter.

To reduce the dissipation in the TEC driver, I'm going Class H there, as I mentioned in another thread, which means I also need a buck converter. Bucks and inverting buck-boost converters I've done lots of, but I've never done a positive boost before.

There's also a possibility that +5V won't be enough headroom on the Class H stage. Some TECs such as CUI's use much finer fingers and so get equivalent delta-T and heating capacity at higher voltages and lower currents. Thus I'd like to have a population option to run the buck off the output of the boost. This is of course fairly fraught, given the stability problems of boost converters running in CCM and the negative input resistance of all switchers. Also, space is at a huge premium on this board.

Advice welcome!

Cheers

Phil Hobbs

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

Stability isn't a problem in general. Only specifically for peak current mode controllers. This mode is pretty popular, but it isn't the only kind that's good. CCM isn't required either -- for their market -- so they aren't very concerned about that...

If you must operate in CCM, consider an average current mode controller. Easy enough to do discrete, or there's probably a few you can find integrated?

Tim

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

Sometimes I buy a cheap isolated dc/dc converter brick and stack its output on top of the incoming supply. Like if I have 12 and want 15, just add 3.3 volts to the 12.

dc/dc things are so cheap that designing something only makes sense for high volumes.

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

Downside, they're noisy as sin.

Tim

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

This one is probably 10k-20k/year. Mostly I need it to be quiet and to fit in a matchbox without taking the matches out. ;)

Cheers

Phil Hobbs

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

Nice thing about CCM is that it avoids the free ringing of the suddenly-unloaded inductor, which reduces EMI by a lot. That unstable zero is a pain, though. Since most of the rest of the circuit is basically a linear regulator, I might just slow the loop way way down and use a big output cap to accommodate the buck regulator's transients.

Thanks. I'll revisit the math and see what comes out.

Cheers

Phil Hobbs

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

Some are. The upside is that, managed properly, there is no ground loop on the main board.

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

Is it really that bad? The peak energy of that ringing is in the MF to HF band. Easy to filter. The hard switching transients, which are there regardless, carry the pernicious VHF content. And it's a double-whammy, because DCM has a shorter falling edge (hard switch-on), while CCM has reverse recovery (if applicable) plus full height edge.

Buck has the same transients, of course, so if you can do that in your design already, boost is no problem at all.

Have you considered doing the class H by modulating the SMPS directly -- in other words, a post-reg configuration? Doesn't even need to be an LDO or pass device in that case; it could be your favorite C-mult if you like (given suitable adjustments for loop speed, that is). :-)

Tim

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

Anyone tried these recom ones?

formatting link

George H.

Reply to
George Herold

Phil Hobbs wrote

..

Cannot be answered without current, ripple, specs. ebay is full of boost converters. Made in China, 5$ a piece or so for a few amps.

Reply to
<698839253X6D445TD

I don't understand "Class-H" in this context. Why not a four switch buck-boost?

I do them all the time. No big deal, other than there is a DC path from the input to the output to remember. Startup can be a real bitch!

Ah, do you mean an "H bridge" or "full bridge"? "Class-H" means something completely different to me.

No problem as long as you have enough input capacitance on the second converter. People often forget this little detail. A boost after a buck is easier than the other way around, too, because the boost's inductor is in the right place.

Your efficiency is going to suffer, though. A four-switch buck-boost might be a better plan, or even a SEPIC? A SEPIC buys some isolation, as well.

This tool is rather useful for anyone playing with switchers.

formatting link

Reply to
krw

Not those specifically, but they look worth a try. RECOM is generally higher quality than others. Sheesh, and you pay for those oddball voltages too, don't you...

Hmm, no EMC certs, that's maybe worrying?

Tim

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

To some given level. As I like to say, "there's SMPS quiet, and then there's instrument quiet."

That's one reason I really like the TI Simple Switchers. No sharp edges. Schottky diodes are a lot better than synchronous rectifiers there.

It's not boost vs. buck AFAICT, it's CCM vs DCM.

That's how Class H works--the idea is to maintain just enough headroom for the amp to work properly.

The linear output stage helps a lot.

Cheers

Phil Hobbs

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

I'm not looking for someone to design it for me--I'm interested in advice on what potholes to look out for.

I have several. Too big, too noisy, and the actual output current is about a third of the rating.

Cheers

Phil Hobbs

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

Noise. Putting a Class D driver on a TEC is a recipe for all sorts of birdies on the laser bias, because the capacitance from the TEC to the cold plate is surprisingly large (nanofarads).

No, it's Class H--that is, you servo the SMPS output to get the minimum supply headroom for the (linear) output stage to work properly. That preserves the low noise of the linear stage while avoiding excess dissipation. The output stage is an asymmetric BJT bridge--one side is a current conveyor and the other side is a normal emitter follower type. Because I have a higher supply voltage available for the bridge driver, I avoid the V_BE drop on the emitter follower side, at least in the cooling direction, which is what I mostly care about.

Takes more board space, though, doesn't it? I really only have space for an integrated solution. Keeping the dissipation down is also mainly a board space issue, since the thing is mains powered.

Thanks

Phil Hobbs

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

You still need to specify current and ripple.

The one I have (for bit higher voltage performs very well. Same for the bucks, and I have a boost/buck too.

It s not clear to me, you have a 24 V unit, have only 5 V, add a box, put nice Chinese boost converter in it, maybe if you ask them nicely they will supply it in a nice box with 'made in USA'? (not sure these days, and rightfully so), pay them the 10 $, add a sticker on it 'Made By The Great Dr Hobbs' and charge a hundred, and have a pizza.

Does not the great JL do the same and use Meanwell power supplies, wallwarts made in (whisper) C H I N A', to evade all those mains power related requirements for his extremely high tech boxes? I do the same, I use those Chinese (n-th time I mention China. oh well) to power all sort of stuff, some have adjustable voltage and current limit with 10 turn pots, use it to charge batteries. It is good stuff, I cannot make it for that, not even the board only, and no failures so far in ? 4 years?

Or reverse engineer, I have a circuit diagram from at least one somewhere.

formatting link
starts at 0.76 $ with free shipping for 2A.

And now you do not have to re-design anything.... These seem to be the latest housing trend, not sure about the 5 V in.. more expensive, but 'box included' so to speak:

formatting link

Anyways, plenty on the market.

Reply to
<698839253X6D445TD

Of course, your filled-matchbook constraint adds some limits.

One nice topology is a resonant tank with a saturating switch that's on only at low voltage. No spikes.

--
John Larkin         Highland Technology, Inc 

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

I've tested some CUI surface-mount dc/dc converters, the PDS1 types. They seem happy at 2x rated load. That footprint is multi-sourced.

--
John Larkin         Highland Technology, Inc 

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

If the switcher keeps the amp's supply rail just above the load voltage, doesn't the amp degenerate into a filter, specifically a c-multiplier?

Why not let the switcher be the control element? You don't need bandwidth.

A buck or buck-boost switcher followed by a DPDT switch would be simple. Merge the c-multiplier into that somehow.

--
John Larkin         Highland Technology, Inc 

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

It needs to be bipolar, so the output amp is an asymmetric bridge: class-B current conveyor on one side, emitter followers on the other. I have a higher supply voltage available for the drivers, so the emitter follower doesn't lose me a V_BE drop in the direction I mostly care about, namely cooling.

Yup, that would work OK with a digital control loop, which this version will have. The little TEC driver board is designed already, and there's no NRE budget for doing it over. It doesn't make too much difference in board area since class H doesn't need significant thermal pours.

Cheers

Phil Hobbs

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

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.