Forums

LDO regulators whose outputs can be ORed together

Started by Peter November 11, 2021
Sometimes one needs to power a circuit from one source or another.

Most LDOs, or indeed most normal regs, feed current back up to the
source. LDOs tend to use a PMOS pass transistor which has a parasitic
diode.

I am doing a design where I am using the Ricoh R1191 for this

https://eu.mouser.com/ProductDetail/Ricoh-Electronic-Devices-Company/R1191N033B-TR-FE?qs=%2Fha2pyFaduhEV6ZG3xOqbaXpStP%2FIzlm74g8V5lGNcwdefxpMkR8XA%3D%3D

which has a series diode, so the dropout voltage is about 0.7V.

It's not dirt cheap but not crazy-priced either.

I am wondering why this is rare. Is it not possible to make a PMOS
device without the parasitic diode? Or have some other series element
which gets turned off when there is no input? It reminds me of an
active rectifier in switching power supplies, to avoid the Vf of the
diode(s). There is even a circuit for a bridge rectifier, although
that was commercially implemented with a complicated chip to drive the
four gates, IIRC.

One obvious solution is to use a normal LDO and have a diode in series
with the input, so long as you can be sure nothing funny will be
hapenning inside with the ground lead which could still pass negative
current.
On 11/11/2021 5:01 PM, Peter wrote:
> Sometimes one needs to power a circuit from one source or another. > > Most LDOs, or indeed most normal regs, feed current back up to the > source. LDOs tend to use a PMOS pass transistor which has a parasitic > diode. > > I am doing a design where I am using the Ricoh R1191 for this > > https://eu.mouser.com/ProductDetail/Ricoh-Electronic-Devices-Company/R1191N033B-TR-FE?qs=%2Fha2pyFaduhEV6ZG3xOqbaXpStP%2FIzlm74g8V5lGNcwdefxpMkR8XA%3D%3D > > which has a series diode, so the dropout voltage is about 0.7V. > > It's not dirt cheap but not crazy-priced either. > > I am wondering why this is rare. Is it not possible to make a PMOS > device without the parasitic diode? Or have some other series element > which gets turned off when there is no input? It reminds me of an > active rectifier in switching power supplies, to avoid the Vf of the > diode(s). There is even a circuit for a bridge rectifier, although > that was commercially implemented with a complicated chip to drive the > four gates, IIRC. > > One obvious solution is to use a normal LDO and have a diode in series > with the input, so long as you can be sure nothing funny will be > hapenning inside with the ground lead which could still pass negative > current. >
There's a family of chips called an "ORing controller" that does this on the secondary side, I'm remembering one in particular that IIRC, just had two inputs and one output and acted like an ideal diode for both, simply picking whichever input was higher and preferentially passing it on to the output. It won some award like one of the best new chips of 2013 from one of the magazines that gives out awards like that. Sorry can't be more specific, maybe someone else will recall which product it was in particular. I think there wouldn't be a family of devices to perform this specific function if it were simple to do efficiently and reliably in a general way.
On Thu, 11 Nov 2021 22:01:33 +0000, Peter <nospam@nospam9876.com>
wrote:

>Sometimes one needs to power a circuit from one source or another. > >Most LDOs, or indeed most normal regs, feed current back up to the >source. LDOs tend to use a PMOS pass transistor which has a parasitic >diode. > >I am doing a design where I am using the Ricoh R1191 for this > >https://eu.mouser.com/ProductDetail/Ricoh-Electronic-Devices-Company/R1191N033B-TR-FE?qs=%2Fha2pyFaduhEV6ZG3xOqbaXpStP%2FIzlm74g8V5lGNcwdefxpMkR8XA%3D%3D > >which has a series diode, so the dropout voltage is about 0.7V. > >It's not dirt cheap but not crazy-priced either. > >I am wondering why this is rare. Is it not possible to make a PMOS >device without the parasitic diode? Or have some other series element >which gets turned off when there is no input? It reminds me of an >active rectifier in switching power supplies, to avoid the Vf of the >diode(s). There is even a circuit for a bridge rectifier, although >that was commercially implemented with a complicated chip to drive the >four gates, IIRC. > >One obvious solution is to use a normal LDO and have a diode in series >with the input, so long as you can be sure nothing funny will be >hapenning inside with the ground lead which could still pass negative >current.
Or put a schottky diode on the output but take the feedback from after the diode. -- Father Brown's figure remained quite dark and still; but in that instant he had lost his head. His head was always most valuable when he had lost it.
jlarkin@highlandsniptechnology.com wrote

>>One obvious solution is to use a normal LDO and have a diode in series >>with the input, so long as you can be sure nothing funny will be >>hapenning inside with the ground lead which could still pass negative >>current. > >Or put a schottky diode on the output but take the feedback from after >the diode.
Sure, but then you are building your own regulator.
On a sunny day (Thu, 11 Nov 2021 22:01:33 +0000) it happened Peter
<nospam@nospam9876.com> wrote in <smk3rt$72m$1@dont-email.me>:

>Sometimes one needs to power a circuit from one source or another. > >Most LDOs, or indeed most normal regs, feed current back up to the >source. LDOs tend to use a PMOS pass transistor which has a parasitic >diode. > >I am doing a design where I am using the Ricoh R1191 for this > >https://eu.mouser.com/ProductDetail/Ricoh-Electronic-Devices-Company/R1191N033B-TR-FE?qs=%2Fha2pyFaduhEV6ZG3xOqbaXpStP%2FIzlm74g8 >V5lGNcwdefxpMkR8XA%3D%3D > >which has a series diode, so the dropout voltage is about 0.7V. > >It's not dirt cheap but not crazy-priced either. > >I am wondering why this is rare. Is it not possible to make a PMOS >device without the parasitic diode? Or have some other series element >which gets turned off when there is no input? It reminds me of an >active rectifier in switching power supplies, to avoid the Vf of the >diode(s). There is even a circuit for a bridge rectifier, although >that was commercially implemented with a complicated chip to drive the >four gates, IIRC. > >One obvious solution is to use a normal LDO and have a diode in series >with the input, so long as you can be sure nothing funny will be >hapenning inside with the ground lead which could still pass negative >current.
IIRC the correct way to parallel 2 voltage regulators is have each one sense its output current and if too high drive the current reference of the other one higher until both deliver the same current. If you just parallel voltage controlled ones then one is likely to do all the work due to minuscule output voltage differences. For example one could go into current limit at 100% current and the other will then do say 10%. Much simpler to get or design one bigger one?
On Fri, 12 Nov 2021 07:07:56 +0000, Peter <nospam@nospam9876.com>
wrote:

> >jlarkin@highlandsniptechnology.com wrote > >>>One obvious solution is to use a normal LDO and have a diode in series >>>with the input, so long as you can be sure nothing funny will be >>>hapenning inside with the ground lead which could still pass negative >>>current. >> >>Or put a schottky diode on the output but take the feedback from after >>the diode. > >Sure, but then you are building your own regulator.
No, just using a chip that has a feedback pin. That's common in LDOs. What regs are you considering? Note: an LM317 is not an LDO. A lot of people have taken to using "LDO" for most any linear reg. -- Father Brown's figure remained quite dark and still; but in that instant he had lost his head. His head was always most valuable when he had lost it.
On Fri, 12 Nov 2021 08:09:35 GMT, Jan Panteltje
<pNaOnStPeAlMtje@yahoo.com> wrote:

>On a sunny day (Thu, 11 Nov 2021 22:01:33 +0000) it happened Peter ><nospam@nospam9876.com> wrote in <smk3rt$72m$1@dont-email.me>: > >>Sometimes one needs to power a circuit from one source or another. >> >>Most LDOs, or indeed most normal regs, feed current back up to the >>source. LDOs tend to use a PMOS pass transistor which has a parasitic >>diode. >> >>I am doing a design where I am using the Ricoh R1191 for this >> >>https://eu.mouser.com/ProductDetail/Ricoh-Electronic-Devices-Company/R1191N033B-TR-FE?qs=%2Fha2pyFaduhEV6ZG3xOqbaXpStP%2FIzlm74g8 >>V5lGNcwdefxpMkR8XA%3D%3D >> >>which has a series diode, so the dropout voltage is about 0.7V. >> >>It's not dirt cheap but not crazy-priced either. >> >>I am wondering why this is rare. Is it not possible to make a PMOS >>device without the parasitic diode? Or have some other series element >>which gets turned off when there is no input? It reminds me of an >>active rectifier in switching power supplies, to avoid the Vf of the >>diode(s). There is even a circuit for a bridge rectifier, although >>that was commercially implemented with a complicated chip to drive the >>four gates, IIRC. >> >>One obvious solution is to use a normal LDO and have a diode in series >>with the input, so long as you can be sure nothing funny will be >>hapenning inside with the ground lead which could still pass negative >>current. > >IIRC the correct way to parallel 2 voltage regulators is have each one sense its output current >and if too high drive the current reference of the other one higher until both deliver the same current. >If you just parallel voltage controlled ones then one is likely to do all the work >due to minuscule output voltage differences. >For example one could go into current limit at 100% current and the other will then do say 10%. >Much simpler to get or design one bigger one? > > >
You can certainly diode OR the input of a single reg, from two sources. -- If a man will begin with certainties, he shall end with doubts, but if he will be content to begin with doubts he shall end in certainties. Francis Bacon
On a sunny day (Fri, 12 Nov 2021 10:08:39 -0800) it happened John Larkin
<jlarkin@highland_atwork_technology.com> wrote in
<3bbtogh6jstka4039uimng3q8baguk94i3@4ax.com>:

>On Fri, 12 Nov 2021 08:09:35 GMT, Jan Panteltje ><pNaOnStPeAlMtje@yahoo.com> wrote: > >>On a sunny day (Thu, 11 Nov 2021 22:01:33 +0000) it happened Peter >><nospam@nospam9876.com> wrote in <smk3rt$72m$1@dont-email.me>: >> >>>Sometimes one needs to power a circuit from one source or another. >>> >>>Most LDOs, or indeed most normal regs, feed current back up to the >>>source. LDOs tend to use a PMOS pass transistor which has a parasitic >>>diode. >>> >>>I am doing a design where I am using the Ricoh R1191 for this >>> >>>https://eu.mouser.com/ProductDetail/Ricoh-Electronic-Devices-Company/R1191N033B-TR-FE?qs=%2Fha2pyFaduhEV6ZG3xOqbaXpStP%2FIzlm74 >>>g8 >>>V5lGNcwdefxpMkR8XA%3D%3D >>> >>>which has a series diode, so the dropout voltage is about 0.7V. >>> >>>It's not dirt cheap but not crazy-priced either. >>> >>>I am wondering why this is rare. Is it not possible to make a PMOS >>>device without the parasitic diode? Or have some other series element >>>which gets turned off when there is no input? It reminds me of an >>>active rectifier in switching power supplies, to avoid the Vf of the >>>diode(s). There is even a circuit for a bridge rectifier, although >>>that was commercially implemented with a complicated chip to drive the >>>four gates, IIRC. >>> >>>One obvious solution is to use a normal LDO and have a diode in series >>>with the input, so long as you can be sure nothing funny will be >>>hapenning inside with the ground lead which could still pass negative >>>current. >> >>IIRC the correct way to parallel 2 voltage regulators is have each one sense its output current >>and if too high drive the current reference of the other one higher until both deliver the same current. >>If you just parallel voltage controlled ones then one is likely to do all the work >>due to minuscule output voltage differences. >>For example one could go into current limit at 100% current and the other will then do say 10%. >>Much simpler to get or design one bigger one? >> >> >> > >You can certainly diode OR the input of a single reg, from two >sources.
Of course, but that is different, and even then the highest input will deliver everything, although having high internal resistance sources / 'feeble' diodes would share ..
On 11/12/2021 9:24 AM, jlarkin@highlandsniptechnology.com wrote:
> On Fri, 12 Nov 2021 07:07:56 +0000, Peter <nospam@nospam9876.com> > wrote: > >> >> jlarkin@highlandsniptechnology.com wrote >> >>>> One obvious solution is to use a normal LDO and have a diode in series >>>> with the input, so long as you can be sure nothing funny will be >>>> hapenning inside with the ground lead which could still pass negative >>>> current. >>> >>> Or put a schottky diode on the output but take the feedback from after >>> the diode. >> >> Sure, but then you are building your own regulator. > > No, just using a chip that has a feedback pin. That's common in LDOs. > > What regs are you considering? > > Note: an LM317 is not an LDO. A lot of people have taken to using > "LDO" for most any linear reg. >
Some LDOs tie the bulk of the FET to ground so they don't conduct in reverse, as so: <https://www.ti.com/product/TPS7A37>
bitrex <user@example.net> wrote

>Some LDOs tie the bulk of the FET to ground so they don't conduct in >reverse, as so: > ><https://www.ti.com/product/TPS7A37>
Nice chip but not cheap. The Ricoh ones are still a good option.