Power Path Management for: 4.2V to 3.3V, 100mA peak and circuit separation

Hello Den,

This is a very good question.

Fundamentally the designer needs to know the current profile to determine whether a LDO or Switcher is best suited for a low power application.

LDO's can have very low Iq as compared to a switcher.
However LDO's  efficiency under load can be less efficient than a switcher.

In cases where the average current is very low, or zero, an LDO will provide better total system efficiency.
As the average power, peak power, increases you will find that a switcher is overall more efficient.

But you must know the power profile.

See this application note that provides a treatment of LDO vs Switcher for an RF application.

4061.Powering LPW products swra173b.pdf

If you don't know, or can not measure, the power profile you may just have to test in the lab.
Select a LDO>=100mA with low Iq, and select a switcher >=100mA with low Iq and try them both.

For short duration pulses with long standby light load periods, an LDO will likely provide the best overall average efficiency.

Hello Ed,

thank you for your reply.

After discussing it with our Embedded team (they basically tell me what they want to have on the PCB), I think the best choice will be for sure SW. There is a small change in my design now, we are going to add a part which can draw up to 0.3-0.4A over a short period of time. So overall current requirement is around 0.5A in the worst case and in the "waiting mode" the system will draw about 7mA - 10mA (don't know yet about the "sleep mode"), so SW is a more efficient choice.

The problem is that with all the "time to market" issue I am not given much (actually any) time for prototyping i.e. I don't have time for the lab test. It is a revision of the old product which I have developed and tested before. However, now we have a few more components one of which is a Doppler which has fast transient power consumption and another element which draws up to 0.3A as I've mentioned before. This other element is regulated through a PWM of 10kHz.

Thus, I have to decide if I should use one uniform Power Supply, in my case 3V3 and just add a few Caps to the Inputs of the ICs which draw a lot of current, or separate it with two different ICs? I actually favor one uniform PWR supply since it is better for the PCB layout and keeping overall plane impedance to minimum. What would you suggest in this situation? Again I dont have much time for testing, thus I ask people with more experience they I have.

Kind Regards,
Den

Hello Den,

Splitting the 3.3V would be a good idea.

The transient from the 400mA  will put some deviation on the 3.3 rail.

Digital needs a clean 3.3V.

You may be able to use a LDO for digital if the current isn't too high.

Efficiency of an LDO is always Vout/Vin.   78%.

But total power loss at low loads is the real question.

>What is the digital load?

Sometimes digital draws only 20mA or so, where an LDO makes a good choice.

Use a switcher for the transient.

I did a WEBENCH design that I will share:
You can experiment with it if you care to.

https://webench.ti.com/appinfo/webench/scripts/SDP.cgi?ID=84E4CC4EB9D6F641

TLV62568A 1Amp DC DC is used.

Efficiency is 95% at 500mA.

Efficiency drops as Iout drops. As any switcher would.

This has Power Save Mode where it lowers the switching frequency when the inductor goes discontinuous.

2630.4.2Vin 3.3Vout 500mA TLV62568A.pdf

Hello Ed,

I've never used WEBENCH before but it looks like a really useful tool if you frequently use TI components which I do.

This second element I was talking about which can draw up to 400mA/3.3V is a Peltier element which I regulate with DRV8837. I've tested it on TI's Dev. Board and it works really good, especially for such a small package of the DRV8837 IC.

So, right now I kind of have the whole picture:

I have my main digital circuitry which probably will work with 10-15 mA, a Doppler which sometimes can draw for a short period of time ca. 50-60mA and then a Peltier which can (but probably never will) draw up to 400mA. I still don't want to use LDO for my main circuity since I think it wastes too much power and an efficient SW can do a better job, like TPS62742. Hence, for my main 3.3V digital power supply I want to use something similar to TPS62742. The question is what to do with the Peltier element. The problem is that it should be able to work from as low as 0.2mA up to 400mA... (gradual heating and cooling) we regulate it through DRV8837 with PWM and we might want to be able to apply very little current to Peltier. Do you have an efficient solution to this problem? (again I don't have current profile :/ )

Also in case you have something more efficient than TPS62742 with 3.3V and less overall current since I don't need 400mA, that would be great.

Regards,

Den

Thank you for your help, Ed! 

I have next to zero knowledge about Peltiers so any input from your side is much appreciated. As a matter of fact, connecting Peltier directly to the battery was my first thought, but very annoyingly Peltier which we ordered (not my decision) is rated at Umax 4.1! And since LiPo at the maximum charge is 4.2V, that is not going to work.

Nevertheless, I wrote to the company from which we've bought these Peltiers and asked them if it might still be ok to drive this part with 4.2V even if it is rated at 4.1V. They are going to test it and thus I am waiting for their reply. 

But for now, as it seems, I will have to get another component, i.e. not TPS6274x, but one which will deliver more power. I really prefer to have one PWR supply. I can add a few caps and a Ferrite Bead (Pi filter) next to the noisy component, but this way I will have a very uniform PWR plane. Also as you've said it yourself, the Peltier element is very slow and this is the only component that draws a relatively large amount of current. The rest of the circuitry does not draw that much. Or is there still some other variable I do not know/understand?

Kind Regards,
Den

Hi Ed,

so as it seems, I should be using 4.1V power supply for the Pletier, or lower - period. Also, R&D team wants to be able to change the max. Vm on DRV8837 during assembly. Hence, I will have to separate Vm with a power supply which can go from 1.5V to 3.6V (new limits). Do you have a SW power supply with good efficiency for the output Voltage from 1.5 to 3.6V for my Peltier element? The voltage level should be set through resistor divider.

for the rest of the circuitry I will take TPS62740.

Kind Regards,

Den

Hi Den,
The original design I presented would still be the best approach for your Peltier.

https://webench.ti.com/appinfo/webench/scripts/SDP.cgi?ID=84E4CC4EB9D6F641

TLV62568A 1Amp DC DC is used.

Efficiency is 95% at 500mA.
Design report: 4.2Vin 3.3Vout 500mA TLV62568A.pdf

Using a 1Amp switcher will place the "sweet spot" of efficiency right at your maximum power point.

Of concern is still the architecture.
BQ51050b is designed to charge a LiIon battery.
So we must assume that your system is connected to a battery.
And that battery will range from 4.2V fully charged down to about 3.0V discharged, 3.6V is the plateau.
If your desire is to get 3.6V from the battery, when it is almost discharged then you will need to boost, 3.0 to 3.6V.

TLV62568A has a 100% duty cycle mode that will allow Vout to follow Vin as Vin drops below the Vout set point.
Example: If Vout is set for 3.6V, you will get 3.6Vout from Vbat=4.2(full) to 3.6V.
Then Vout = Vin as Vin drops below 3.6V.

To maintain Vout when Vin<Vout and Vin>Vout you will need a buck boost solution.
Here is a link to the WEBENCH design I just ran for this.
https://webench.ti.com/appinfo/webench/scripts/SDP.cgi?ID=43D8A9AF4BF8D999

Vin 3.0-4.2; Vout 3.6V 500mA; TPS63024 3Amp buck boost.
A 3Amp part is needed because of higher peak currents boosting from 3V to 3.6V at 500mA output.
Vout is adjustable as you need for the Peltier.