3
\$\begingroup\$

I'm looking for schottky diodes to make 2 power rails (powering from charger vs powering from battery) meet together while minimizing voltage drop. Currents are expected to be up to 2-2.2A peak for near-empty battery and for, well, effectively indefinite time (long enough not to be considered a pulse or anything, it's for a device that can be on for, say, an hour).

I'm desperately trying to limit my voltage drop to 0.2V (I can tolerate up to 0.3V, but more losses would be undesirable, unless unavoidable).

I've been looking into Schottkies at mouser and digikey and whatever datasheet I open, there is always an instantaneous I-V curve that shows voltage drop depending on current and junction temperature.

Examples: datasheet datasheet

I'm confused by the word "instantaneous". But since there is no other I-V curve, I assume my course of action is the following:

  1. Determine peak constant current I will have
  2. Look up power dissipation on the graph (amps -> power loss in watts)
  3. Estimate junction temperature (power loss watts * thermal resistance degrees per watts -> junction temperature degrees)
  4. Lookup expected voltage drop on the instantaneous I-V curve taking junction temp into account.

Do I understand it right that the instantenous I-V characteristic also applies to constant current?

Is this the right course of action?

Also, this power rail goes into buck-boost, so the less battery juice there is, the higher the current will be to boost it, the higher the current the higher the temp and the smaller is the voltage drop, which is a desired behavior if I want to get as much as possible from the battery before the buck-boost goes into undervolt shutdown. (maybe I should just place a heater next to the diode to keep it hot to minimize the voltage drop; joke)

To make everyone's life a little easier:

schematic

The output of the buck-boost is 3.3V with peak consumption estimated at 1.5A. The Q thing is just a beefy p-mosfet to be controlled by physical switch, irrelevant, imagine it's not there (it's only development stage anyway, everything is subject to change).

While looking for the schottky with small voltage drop at desired currents, I discovered that close to none diodes actually fit my requirements, a few barely come close at best (mainly high-amp like 10A+, like in the datasheets I provided). Others have voltage drops I would expect from regular silicon diodes - often 0.6V and even more at 2A currents.

Is there a way to make power rails with such current meet with less power loss than I'm looking at now? (if yes, I still would like to know about diodes, knowledge is knowledge after all)

\$\endgroup\$
2
\$\begingroup\$

Is there a way to make power rails with such current meet with less power loss than I'm looking at now?

Consider looking at PowerPath, Ideal Diodes, and Load Switches from such vendors as ADI. For example this one, the LTC4376: -

enter image description here

enter image description here

Basically it's a MOSFET used as a near-ideal diode. You can even get two of them in one package I believe thus forming the diode-OR functionality it appears that you need. The LTC4353 has two in one package: -

enter image description here

Do I understand it right that the instantaneous I-V characteristic also applies to constant current?

Yes it does.

\$\endgroup\$
10
  • 1
    \$\begingroup\$ appreciate your reply; I tend to avoid linear technology ICs because only Bezos and Musk can afford them, but I will look at these devices \$\endgroup\$
    – Ilya
    Aug 31 '21 at 12:57
  • 2
    \$\begingroup\$ Roll your own from standard MOSFETs then. Not much needed in terms of gate control to make ideal diodes from MOSFETs. Here's a TI document on the subject. (just a reverse protection circuit). \$\endgroup\$
    – Andy aka
    Aug 31 '21 at 12:59
  • 1
    \$\begingroup\$ Because they need some minimal voltage above about 2 volts for them to start cooking. \$\endgroup\$
    – Andy aka
    Aug 31 '21 at 13:02
  • 1
    \$\begingroup\$ See also this and this. \$\endgroup\$
    – Andy aka
    Aug 31 '21 at 13:05
  • 1
    \$\begingroup\$ MOSFETs will balance providing there is enough gate-source voltage. Below the point where this happens they may not balance at all well and smoke may be produced. \$\endgroup\$
    – Andy aka
    Aug 31 '21 at 14:14
0
\$\begingroup\$

You have a battery that may need charging and a BMS and a AC/DC power source that you want to share for a Buck-Boost 3.3V 1.5A output.

The ideal choice for this may be the DW01-P chip and 2 power FETs. This includes the BMS and control signals for the dual FETS , the latter which you can get in one IC or one PCB.

I will list the datasheet features below.

 Reduction in Board Size due to Miniature
Package SOT-23-6.
 Ultra-Low Quiescent Current at 3μA
(Vcc=3.9V).
 Ultra-Low Power-Down Current at 0.1μA (Vcc=2.0V).
 Precision Overcharge Protection Voltage 4.25V ± 50mV
 Load Detection Function during Overcharge Mode.
 Two Detection Levels for Overcurrent Protection.
 Delay times are generated by internal circuits. No caps needed 

Also the parameter to consider is not the max average load current but rather the max peak current under any condition including surge startup. then apply Ohm's Law and Watt's Law. This ripple voltage will affect the performance of your boost regulator.

A rule of thumb is RdsOn <= 1% of Vmin/Ipk which is usually limited by DCR and ESR of the boost regulator using logic level FET's .

I could recommend a better Regulator but that's not encouraged here to think outside the box and answer unasked questions or allow too many questions.

https://cdn.sparkfun.com/assets/learn_tutorials/2/5/1/DW01-P_DataSheet_V10.pdf

  • similar chips are DW01A offshore and BQ97100 from TI

Next question?

\$\endgroup\$
3
  • \$\begingroup\$ I'm sorry I tried really hard, but I failed to understand how most of your reply has anything to do with my question. I already have a charging circuit with all the necessary protections. \$\endgroup\$
    – Ilya
    Aug 31 '21 at 13:34
  • \$\begingroup\$ you're very helpful \$\endgroup\$
    – Ilya
    Aug 31 '21 at 13:35
  • \$\begingroup\$ Now you almost have an integrated solution in a tiny package. \$\endgroup\$ Aug 31 '21 at 13:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.