Author Topic: A niche chip (LT4320) for mains powered AC -> low volt DC "linear" based psu ?  (Read 38579 times)

0 Members and 1 Guest are viewing this topic.

Offline BravoVTopic starter

  • Super Contributor
  • ***
  • Posts: 7549
  • Country: 00
  • +++ ATH1
Not sure if this is old news, just found out that this little gem should be one of the nice component to replace those scorching hot silicon rectifier.

For sure if I'm going to build another linear based power supply (one off project), this will be in my shopping list.

Also I'm curious, if the input is pure DC, isn't this chip will act like an overkill reverse polarity protection circuit ?  ???

Any second thought ?

Linear Technology LT4320 -> http://www.linear.com/product/LT4320



« Last Edit: September 24, 2013, 03:32:11 pm by BravoV »
 

Offline Bored@Work

  • Super Contributor
  • ***
  • Posts: 3932
  • Country: 00
Re: A niche chip for mains powered AC -> DC "linear" based psu ?
« Reply #1 on: September 24, 2013, 06:41:43 am »
A niche chip for mains powered AC -> DC "linear" based psu ?

Mains powered? Not directly. Max. input voltage is 72 V.
I delete PMs unread. If you have something to say, say it in public.
For all else: Profile->[Modify Profile]Buddies/Ignore List->Edit Ignore List
 

Offline BravoVTopic starter

  • Super Contributor
  • ***
  • Posts: 7549
  • Country: 00
  • +++ ATH1
Re: A niche chip for mains powered AC -> DC "linear" based psu ?
« Reply #2 on: September 24, 2013, 06:44:45 am »
A niche chip for mains powered AC -> DC "linear" based psu ?

Mains powered? Not directly. Max. input voltage is 72 V.

My mistake, I mean through step down transformer as traditional AC->DC low voltage linear psu.

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16385
  • Country: za
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #3 on: September 24, 2013, 09:04:00 am »
Not a bad device, solves the problem in a very elegant way.
 

Offline kizzap

  • Supporter
  • ****
  • Posts: 477
  • Country: au
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #4 on: September 24, 2013, 01:40:07 pm »
I saw this one a few weeks ago and thought it was a brilliant idea. I mean who doesn't like the thought of extremely low losses for rectification?

The only thing that made me sad is the fact that you need to use external MOSFETs with it. Give me a drop in replacement option for current SIL bridge rectifiers, and I will LOVE it.

-kizzap
<MatCat> The thing with aircraft is murphy loves to hang out with them
<Baljem> hey, you're the one who apparently pronounces FPGA 'fuhpugger'
 

Offline BravoVTopic starter

  • Super Contributor
  • ***
  • Posts: 7549
  • Country: 00
  • +++ ATH1
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #5 on: September 24, 2013, 03:25:11 pm »
The only thing that made me sad is the fact that you need to use external MOSFETs with it. Give me a drop in replacement option for current SIL bridge rectifiers, and I will LOVE it.

I'm assuming that your view is from pro or commercial perspective ?
As an hobbyist, those 4 mosfets are insignificant factor to me, either in cost, board space or effort in building the circuit.  >:D


I wonder how it works internally. Must be some simple logic but seems to be powered from the rectified output over a very wide range.

Looks like it, especially they emphasizing in the datasheet the importance of placing at the DC output with "minimum" caps 1uF ceramic + 10uF electrolytic type, these must be just for supplying it self right ? cmmiw, ee wannabe here.  :P

Besides who the heck will use that total low cap value at the output like that anyway ? Unless they're trying to generate high current and "high ripple" signal.    :P

Btw, its very unlikely that its powered from the AC side right ?

Offline digsys

  • Supporter
  • ****
  • Posts: 2209
  • Country: au
    • DIGSYS
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #6 on: September 24, 2013, 03:37:56 pm »
There's been many iterations of FET bridges to minimise power loss, but usually require a heck of a lot more components.
I'll definitely whip a proto pcb for our 20-30A P/S units. Hopefully there are no "limitations" or "side effects".
Not having direct on/off control of the FETs may be a dis-advantage, but the IC looks very interesting.
Thanks for the heads-up.
Hello <tap> <tap> .. is this thing on?
 

Offline BravoVTopic starter

  • Super Contributor
  • ***
  • Posts: 7549
  • Country: 00
  • +++ ATH1
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #7 on: September 24, 2013, 03:41:58 pm »
I'll definitely whip a proto pcb for our 20-30A P/S units. Hopefully there are no "limitations" or "side effects".
Please, update us once you used it.  :-+

Hopefully there are no "limitations" or "side effects".

Noob question, regarding the huge inrush current at power up, especially at the output with the large value capacitor, is this better and/or more reliable compared to plain old silicon diodes ?
« Last Edit: September 24, 2013, 03:43:36 pm by BravoV »
 

Offline Bored@Work

  • Super Contributor
  • ***
  • Posts: 3932
  • Country: 00
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #8 on: September 24, 2013, 03:57:24 pm »
Hopefully there are no "limitations" or "side effects".

The first trap is in the first picture of the datasheet. Already postd above



It says DC to 600Hz, but that requires the LT4320-1, not the LT4320 as shown in the picture.
I delete PMs unread. If you have something to say, say it in public.
For all else: Profile->[Modify Profile]Buddies/Ignore List->Edit Ignore List
 

Offline digsys

  • Supporter
  • ****
  • Posts: 2209
  • Country: au
    • DIGSYS
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #9 on: September 24, 2013, 04:19:11 pm »
Quote from: BravoV
  Please, update us once you used it.  :-+ 
I have a new run of PCBs coming up soon, hopefully have something in 2-3 weeks. We're always looking at new ideas
at replacing bridges / rectifiers, so this is timely.
Quote from: BravoV

Noob question, regarding the huge inrush current at power up, especially at the output with the large value capacitor, is this better and/or more reliable compared to plain old silicon diodes ?
All silicon can withstand quite severe inrush currents. It's the RDSon / pulse-time / energy > heat that kills them.
FETs can have extremely low RDSon figures and are "better" by an order of magnitude +
Hello <tap> <tap> .. is this thing on?
 

Offline BravoVTopic starter

  • Super Contributor
  • ***
  • Posts: 7549
  • Country: 00
  • +++ ATH1
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #10 on: September 24, 2013, 04:19:45 pm »
The first trap is in the first picture of the datasheet. Already postd above

It says DC to 600Hz, but that requires the LT4320-1, not the LT4320 as shown in the picture.

C'mon, its just a typo.

Assuming it will be used on mains stepped down using transformer at 50 or 60Hz as most of the cases, whats wrong with LT4320 even though it is designed for DC up to only at 60Hz ?
« Last Edit: September 24, 2013, 04:31:05 pm by BravoV »
 

Offline BravoVTopic starter

  • Super Contributor
  • ***
  • Posts: 7549
  • Country: 00
  • +++ ATH1
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #11 on: September 24, 2013, 04:21:18 pm »
digsys, eagerly to hear the results from you, thanks !  :-+

Offline johansen

  • Super Contributor
  • ***
  • Posts: 1181
a close look at the datasheet suggests that they charge those mosfets up with some kind of onboard silicon capacitive voltage doubler.

at a maximum current of 800uA, dropping to 300uA once the fet gets to 5 volts on the gate
and discharge of 1.x mA
bottom gate pull up is 6mA, pull down is 12

you will need a pnp assisted turn off at a minimum.
 

Offline Bored@Work

  • Super Contributor
  • ***
  • Posts: 3932
  • Country: 00
Re: A niche chip for mains powered AC -> low volt DC "linear" based psu ?
« Reply #13 on: September 24, 2013, 04:33:32 pm »
Assuming it will be used on mains stepped down using transformer at 50 or 60Hz as most of the cases, whats wrong with LT4320 even though it is designed for DC up to only at 60Hz ?

Nothing, if you only use it up to 60 Hz. But if you happen to e.g. work in a 400 Hz environment it can cause you some grieve.

Linear seems to get a bit sloppy with their datasheets.
I delete PMs unread. If you have something to say, say it in public.
For all else: Profile->[Modify Profile]Buddies/Ignore List->Edit Ignore List
 

Offline digsys

  • Supporter
  • ****
  • Posts: 2209
  • Country: au
    • DIGSYS
Quote from: johansen
a close look at the datasheet suggests that they charge those mosfets up with some kind of onboard silicon capacitive voltage doubler.
At 50/60Hz (even 600Hz) that's heaps of time ! I'm more interested on how they work out Vin-Vout / current direction sense / control.
That's always the "tricky" part !!
Quote from: johansen
you will need a pnp assisted turn off at a minimum. 
not sure what you mean ???
Hello <tap> <tap> .. is this thing on?
 

Offline johansen

  • Super Contributor
  • ***
  • Posts: 1181
 

Offline digsys

  • Supporter
  • ****
  • Posts: 2209
  • Country: au
    • DIGSYS
Still not sure what you mean?
I can't see LT making a custom chip and "forgetting" to have enough drive dynamics to the FETs, after all, most will
only ever operate at 50/60Hz. Plus we have no idea of internal topography and if we are even able to reference
extra drive devices? Did you find more app circuits? Well, I'll find out in a couple weeks or so anyway.
Hello <tap> <tap> .. is this thing on?
 

Offline johansen

  • Super Contributor
  • ***
  • Posts: 1181
Still not sure what you mean?
I can't see LT making a custom chip and "forgetting" to have enough drive dynamics to the FETs, after all, most will
only ever operate at 50/60Hz. Plus we have no idea of internal topography and if we are even able to reference
extra drive devices? Did you find more app circuits? Well, I'll find out in a couple weeks or so anyway.
i don't consider 6mA of gate pull down current sufficient for anything.

i suppose in theory it works.
i mean i've seen 60 volt 60 amp fets driven directly from an LM358 without a gate drive resistor.
i've also seen the same fets driven from something 1000 times "stiffer" but then they used a 2.2K ohm gate resistor.
both of those examples are from 300-500 watt push-pull APC UPS
 
but that is still on the order of 10 times as much current as that chip can deliver.
 

Offline johansen

  • Super Contributor
  • ***
  • Posts: 1181
it seems to me this chip is aimed at very low power consumption devices.
and at 4$ per chip in a 10 lead MSop device it seems to be targeted for very small devices as well.
1mA of gate drive sounds like its enough but in practice i think we're looking at something that might be able to handle a 10 amp diode bridge but not much more without additional gimmicky to built a more robust gate drive.
i think you could make a bootstrap cap, diode, zenar, npn/pnp driver to assist this device that could drive an amp or five through a big Fet from just another mA or three of power dissipation, but my brain is mush right now.

assuming reasonable expenditure on diodes, you can get 100mv per 10 volts PIV rating reasonably cheap.
so for a 35 volt dc output from 24vac, we need 40 volt mosfets (i would run them that high, but many won't, so they will go for 50 volt fets)
at 10 amps dc current we go for 10mOhm fets at 1$ each. plus 4$ for the part and two square inches of board space.
that's 8$ to get from 10 watt of loss from standard silicon at 1 volt drop to 2 watts of loss.

or i can get some .35 volt drop schottky diodes and let them radiate the 7 watts of heat away at 100C, for 6$.
2$ increase to save 5 watts is tempting, but that's 40 cents per watt that could be spent elsewhere.

not a bad trade off really, but they could have integrated a bootstrap capacitor and made a larger, higher voltage device.
« Last Edit: September 24, 2013, 10:07:53 pm by johansen »
 

Offline BravoVTopic starter

  • Super Contributor
  • ***
  • Posts: 7549
  • Country: 00
  • +++ ATH1
Well, I'll find out in a couple weeks or so anyway.

Which mosfet do you have in mind ?


Questions on the mosfet selection criteria quoting the datasheet :

Quote
Other Considerations in MOSFET Selection

Choose the lowest available total gate charge, QG, and correspondingly the lowest CISS, COSS, and CRSS. Choosing a lower gate capacitance while meeting RDS(ON) speeds up the response time for full enhancement, regulation, turn-off, and input shorting events.

VGS(th) must be a minimum of 2V or higher. A gate threshold voltage lower than 2V is not recommended since too much time is needed to discharge the gate below the threshold and halt current conduction during a hot plug or input short event.

Say we working at common 50/60Hz, although it stated to use lowest gate charge, but how low is low enough ? <= 1000 pF ? <=2000 pF ? I can't find it in the datasheet.

Am I correct at that VGS threshold requirement, basically it rules out those logic level mosfet ?
« Last Edit: September 25, 2013, 03:15:57 am by BravoV »
 

Offline digsys

  • Supporter
  • ****
  • Posts: 2209
  • Country: au
    • DIGSYS
Quote from: BravoV
  Which mosfet do you have in mind ? 
Parts are ordered from USA. Hopefully 7-10 days max. I have dozens of high power Mosfets in stock.
I'll just plug various buggers in and see what happens :-)  In one app paper they specify up to 30A parts, so I'm sure pretty
much anything will work. Now to do a PCB bridge replacement.
Hello <tap> <tap> .. is this thing on?
 

Offline BravoVTopic starter

  • Super Contributor
  • ***
  • Posts: 7549
  • Country: 00
  • +++ ATH1
I'll just plug various buggers in and see what happens :-)

 :clap:

Offline BravoVTopic starter

  • Super Contributor
  • ***
  • Posts: 7549
  • Country: 00
  • +++ ATH1
Interesting brief insight about this chip found in -> Linear Tech. Press Release : Ideal Diode Bridge Controller Minimizes Rectifier Heat & Voltage Loss (PDF) and this chip is just released recently in June 2013, pretty new.

Quoting few interesting topics raised by johansen

Quote
...<snip>...

The LT4320 switch control smoothly turns on the appropriate two MOSFETs, while keeping the other two off to prevent reverse currents.

An integrated charge pump provides the gate drive for the external low on-resistance N-channel MOSFETs without requiring external capacitors.

The choice of MOSFETs offers the greatest flexibility in power levels ranging from one to thousands of watts.

Offline tszaboo

  • Super Contributor
  • ***
  • Posts: 8217
  • Country: nl
  • Current job: ATEX product design
I wrote about this chip some time ago, also some considerations, etc... http://nandblog.com/chip-review-ltc4320/
The bottom line is, if you use some linear regulator on the output, it might be not worth it, since you will dissipate the voltage anyway. Sometimes it is useful.
 

Offline digsys

  • Supporter
  • ****
  • Posts: 2209
  • Country: au
    • DIGSYS
Update: ICs are here, PCB layout done (based on a H-Bridge layout I did in the past, it should handle 30A - bolted to some Al naturally)
It will be 2-3 weeks though until I do a PCB run. Itching to see how it performs.
Hello <tap> <tap> .. is this thing on?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf