Small footprint 5V voltage regulator

Started by Leslie Rhorer October 12, 2018
I need a 5V DC 3A maximum / 400mA continuous power supply with a 24 - 30 VAC, 60 Hz
input.  The kicker is it must fit in a very small footprint.  I have worked up a
number of different designs, but in every case at least one component refuses to
fit.  Ideally, the design should fit in no more than 1.4 square inches on both sides
of the board.  If I make some compromises, I can stretch that to 3 square inches on
the top side of the board and 1.4 square inches on the bottom.  The bottom side has
very limited headroom - less than 1/4 in.  The top side has no height limits within
reason.  It needs to have a decent efficiency, because the AC input has a somewhat
limited current capacity.  Cost is an important factor.

Any ideas?
On 10/12/2018 10:27 AM, Leslie Rhorer wrote:
> I need a 5V DC 3A maximum / 400mA continuous power supply with a 24 - > 30 VAC, 60 Hz input. The kicker is it must fit in a very small > footprint. I have worked up a number of different designs, but in > every case at least one component refuses to fit. Ideally, the > design should fit in no more than 1..4 square inches on both sides of > the board. If I make some compromises, I can stretch that to 3 > square inches on the top side of the board and 1.4 square inches on > the bottom. The bottom side has very limited headroom - less than > 1/4 in. The top side has no height limits within reason. It needs > to have a decent efficiency, because the AC input has a somewhat > limited current capacity. Cost is an important factor. > > Any ideas? >
LMR23630 is good, but produces a very fast rising edge (650 ps) and so may need a shield. I have a board with one of those and a TPS61175 boost, plus input and output LCs, that all fit under a single 1 x 1.5 x 0.24 inch shield. Dunno how it works yet--stuffed boards will be here in a week. Sure is nice and small though. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net https://hobbs-eo.com
On Fri, 12 Oct 2018 07:27:03 -0700 (PDT), Leslie Rhorer
<rhorerles@gmail.com> wrote:

>I need a 5V DC 3A maximum / 400mA continuous power supply with a 24 - 30 VAC, 60 Hz
input. The kicker is it must fit in a very small footprint. I have worked up a number of different designs, but in every case at least one component refuses to fit. Ideally, the design should fit in no more than 1.4 square inches on both sides of the board. If I make some compromises, I can stretch that to 3 square inches on the top side of the board and 1.4 square inches on the bottom. The bottom side has very limited headroom - less than 1/4 in. The top side has no height limits within reason. It needs to have a decent efficiency, because the AC input has a somewhat limited current capacity. Cost is an important factor.
> >Any ideas?
Some little (non-isolated) dc/dc synchronous switcher chip, an inductor, and a few passives should fit into under 1 square inch. The problem becomes how to rectify the AC. And that devolves to finding a filter capaitor. So you need about 1000 uF 50 volts worth of cap(s) on the tall side of your board, and a schottky bridge. Sounds like you have lots of room. -- John Larkin Highland Technology, Inc lunatic fringe electronics
On Friday, October 12, 2018 at 8:10:48 AM UTC-7, John Larkin wrote:
> On Fri, 12 Oct 2018 07:27:03 -0700 (PDT), Leslie Rhorer > <rhorerles@gmail.com> wrote: > > >I need a 5V DC 3A maximum / 400mA continuous power supply with a 24 - 30 VAC, 60
Hz input.
> Some little (non-isolated) dc/dc synchronous switcher chip, an > inductor, and a few passives should fit ...about 1000 uF 50 volts worth of cap(s)
on
> the tall side of your board, and a schottky bridge.
Or, with a voltage-doubler rectifier, two caps and two diodes would do the DC input part; needn't be Schottky, at these voltages the diode drop is low. UVR2A331 capacitor (330 uF, 100V) is cheap, two for under a buck in 1k qty. Not sure what a Schottky bridge costs, but don't need a 1000 uF for this kind of ripple current nowadays, even with the bridge.
The problem is, the LMR23630, like most of the other switching chips I have
investigated, can't handle more than 36V.  Rectified 30V AC will produce at least 
42V, and with line spikes - which are exceedingly likely - the input can go well
beyond that.  Line transformers are huge, and after adding some large filter
capacitors, none that I found would allow me to populate the board successfully. 
Obviously, if I can get the input voltage to something under 20VDC, I can easily
find off-the-shelf buck converters for about $1 - $2.  I thought of using a
controlled full wave rectifier, but I wasn't able to get it, the filter caps, and
the buck converter on the board together.  Perhaps one of you can come up with a
better design for something like a partially or fully controlled bridge rectifier?
On Fri, 12 Oct 2018 17:54:08 -0700 (PDT), Leslie Rhorer
<rhorerles@gmail.com> wrote:

>The problem is, the LMR23630, like most of the other switching chips I have
investigated, can't handle more than 36V. Rectified 30V AC will produce at least
>42V, and with line spikes - which are exceedingly likely - the input can go well
beyond that. Line transformers are huge, and after adding some large filter capacitors, none that I found would allow me to populate the board successfully. Obviously, if I can get the input voltage to something under 20VDC, I can easily find off-the-shelf buck converters for about $1 - $2. I thought of using a controlled full wave rectifier, but I wasn't able to get it, the filter caps, and the buck converter on the board together. Perhaps one of you can come up with a better design for something like a partially or fully controlled bridge rectifier? A flyback converter could use a small chip and an outboard mosfet. But that needs a transformer, which is harder to find than an inductor. You could do a SEPIC with a stock dual-winding inductor. Do you need isolation? That suggests a flyback maybe. -- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
https://lcsc.com/product-detail/DC-DC-Converters_TI_TPS54360DDAR_TPS54360DDAR_C44377.html

?
On Friday, October 12, 2018 at 8:16:27 PM UTC-5, Lasse Langwadt Christensen wrote:
>
https://lcsc.com/product-detail/DC-DC-Converters_TI_TPS54360DDAR_TPS54360DDAR_C44377.html
> > ?
Hey, it looks like that will work. Thanks! With this I can get by with only about 16 small components, including an inductor of less than 0.1mH, which should be pretty small. Since I am still worried about transients (OK, so I am paranoid), and since the 24VAC input is optional for this device, I think I will implement it as a plug-in module. The main device has optional Ethernet / WiFi operation. If the user chooses Ethernet, they can go with PoE, the optional 24VAC inverter or an external 5V supply. If they choose WiFi, then they still have the option of 24VAC or 5V external.
Leslie Rhorer wrote:
> > > The problem is, the LMR23630, like most of the other switching chips > I have investigated, can't handle more than 36V. > Rectified 30V AC will produce at least 42V, and with line spikes - which are >
exceedingly likely - the input can go well beyond that.
> >
** Filter electros are *extremely* effective at supressing line spikes. .... Phil
On 2018-10-13, Leslie Rhorer <rhorerles@gmail.com> wrote:
> On Friday, October 12, 2018 at 8:16:27 PM UTC-5, Lasse Langwadt Christensen
wrote:
>>
https://lcsc.com/product-detail/DC-DC-Converters_TI_TPS54360DDAR_TPS54360DDAR_C44377.html
>> >> ? > > Hey, it looks like that will work. Thanks! With this I can get by > with only about 16 small components, including an inductor of less > than 0.1mH, which should be pretty small. Since I am still worried > about transients (OK, so I am paranoid), and since the 24VAC input is > optional for this device, I think I will implement it as a plug-in > module. The main device has optional Ethernet / WiFi operation. If > the user chooses Ethernet, they can go with PoE, the optional 24VAC > inverter or an external 5V supply. If they choose WiFi, then they > still have the option of 24VAC or 5V external.
POE is 48VDC. can you re-use the POE buck module for your 24VAC input that'd save space. you'd need bigger capacitors for 24VAC operation and some way to switch between POE and AC but all else should be good. -- Notsodium is mined on the banks of denial.