Author Topic: Would this source selector circuit work?  (Read 1012 times)

0 Members and 1 Guest are viewing this topic.

Offline Nominal AnimalTopic starter

  • Super Contributor
  • ***
  • Posts: 7198
  • Country: fi
    • My home page and email address
Would this source selector circuit work?
« on: February 23, 2020, 08:13:52 pm »
I have two 5V sources that can power my circuit.  If only one of them is connected, I would like to avoid the (Schottky) diode drop.  I came up with this circuit:

This is essentially steady-state operation, as switching occurs at the rate of a human plugging or unplugging the device.
The resistors ensure the VIN lines are pulled to ground when disconnected.  I'm not sure of their resistance, but I'd prefer them high rather than low.  Any suggestions?

When VIN1 is connected but VIN2 is not, the upper P-Enh MOSFET conducts, passing the load current.
When VIN2 is connected but VIN1 is not, the lower P-Enh MOSFET conducts, passing the load current.
When both VIN1 and VIN2 are connected, neither MOSFET conducts, and the Schottky diode with the higher potential on its anode passes the load current (assuming the diode drop is lower than the intrinsic body diode in the MOSFET).

If this works, it would be rather optimal for my needs (given suitable P-channel enhancement mode MOSFETs and Schottky diodes).  Is there something I'm overlooking?  Problems in real life scenarios that I didn't see in simulation?  Better suggestions for this use case?
« Last Edit: February 23, 2020, 08:16:09 pm by Nominal Animal »
 

Offline atmfjstc

  • Regular Contributor
  • *
  • Posts: 121
  • Country: ro
Re: Would this source selector circuit work?
« Reply #1 on: February 24, 2020, 07:48:56 am »
Don't the MOSFETs also have a R_DS_on ? You might get a voltage drop across them too if the current is high enough...
 

Offline Nominal AnimalTopic starter

  • Super Contributor
  • ***
  • Posts: 7198
  • Country: fi
    • My home page and email address
Re: Would this source selector circuit work?
« Reply #2 on: February 24, 2020, 11:56:59 am »
Don't the MOSFETs also have a R_DS_on ? You might get a voltage drop across them too if the current is high enough...
If the current were so high that the voltage drop over the MOSFET was higher than the forward voltage of the Schottky diode, it would be the Schottky diode that passes the current.

Practical parts:

A typical part used for this would be something like DMG2305UX-7, with RDS(ON) less than 50mOhm with VGS(TH) = -4.5V.  Even at 4A current, the voltage drop would be 0.2V.  My current needs are modest, typically in the 100mA-500mA range, corresponding to a 0.005V to 0.025V voltage drop.

The DMG2305UX-7's (SOT23-3) cost 0.333€ (and 3A Schottky diodes like PMEG3030EP,115 0.369€) in singles at Mouser; making this configuration just sixty cents more expensive than just the two Schottky diodes that most circuits use.  If one were to just chuck an LDO to drop the voltage to 3.3V or whatever one needs, this would not make any sense, but I intend to use something better, like TPS82084 (2.45€ in singles at Mouser; 90%+ efficient at 1mA+ from 5V to 3.3V; 94%+ at 100mA+, if TI Webench design report is accurate).

Background:

Using a Schottky diode with a 0.3V forward voltage turns 6% of the input power to waste heat. At 500mA input current, the DMG2305UX-7 drops only 0.025V and turns just 0.5% of the input power to waste heat.  At 2A, that increases to 0.1V and 2%, correspondingly.  In a hobbyist project like the ones I bumble with, it is really more about that waste heat in small plastic enclosures with horrible thermal characteristics, than being super efficient just for efficiency's sake.

Does that 6% matter?  Well, if I can avoid it by dropping two SOT23-3 jellybean parts that cost less than a third of an euro apiece, it would be worth it for me.  But if there are problems in using such circuits in real life, then no.  This is why I'm asking.
 

Offline mariush

  • Super Contributor
  • ***
  • Posts: 5171
  • Country: ro
  • .
Re: Would this source selector circuit work?
« Reply #3 on: February 24, 2020, 12:35:18 pm »
Did you look at ideal diodes like let's say MAX40200 : https://datasheets.maximintegrated.com/en/ds/MAX40200.pdf  ?

There's also OR switches and various ICs that do this switching, did you check such alternatives?
 

Offline Nominal AnimalTopic starter

  • Super Contributor
  • ***
  • Posts: 7198
  • Country: fi
    • My home page and email address
Re: Would this source selector circuit work?
« Reply #4 on: February 24, 2020, 01:12:07 pm »
Did you look at ideal diodes like let's say MAX40200?
I did look at MAX40200.  I'm scared of trying to solder BGA or the WLP-4 package without a reflow oven, and the SOT23 package has a much higher RDS(ON) (or Vf), 0.1V typical at 500mA (0.175V max)...  DMG2305UX-7 has only about a quarter of that voltage drop at the same currents, maximum.  Cost-wise, two MAX40200 would cost me the same as two DMG2305UX's and two PMEG3030EP,115, in singles at Mouser.

That said, you pointing it out made me realize even MAX40200 well outperforms Schottky diodes, which means (two or more of them in SOT23-5) are definitely a reasonable alternative to the circuit in my first post.  Thanks!

There's also OR switches and various ICs that do this switching, did you check such alternatives?
I tried to look for something like that, but finding them without knowing what they are is surprisingly hard :).

Do you have any suggestions?  I'm dealing with 3.0-5.5V at 0-500mA typical, but would prefer a bit of headroom for the current, just in case.
 

Offline langwadt

  • Super Contributor
  • ***
  • Posts: 4857
  • Country: dk
Re: Would this source selector circuit work?
« Reply #5 on: February 24, 2020, 01:32:55 pm »
something like this was (is?) used on the raspberrypi,
 

Offline Nominal AnimalTopic starter

  • Super Contributor
  • ***
  • Posts: 7198
  • Country: fi
    • My home page and email address
Re: Would this source selector circuit work?
« Reply #6 on: February 24, 2020, 01:41:52 pm »
something like this was (is?) used on the raspberrypi
Yes, it is an ideal diode configuration; I discussed it in this message.  It is used in case the right side is powered, but the left side is not.  In my opinion, the MAX40200 is better suited for this.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf