Clean +5V supply for ADC circuit - Buck - Boost - LDO solution

Hi Shmuel,

I've pinged a colleague to help answer this question, but I wanted to add my thoughts first.
The TPS7A20 is coming out later this year and if your schedule allows for it, it might be a very good candidate in a very small device package.
The PSRR plots from the advanced datasheet are shown below.  By choosing the appropriate output capacitor you can improve the PSRR at the switching frequency of your boost converter.  It is expected to come in a 0.603mm x 0.603mm package. 

You can gain PSRR by lowering the switching frequency of the boost converter, which will impact the physical size of the boost converter magnetic.  If you need more PSRR than what is shown in any TI linear regulator drawing, you can place two copies of LDO's in series which will cause the PSRR values to add for increased noise rejection.  So if an LDO has 50 dB of attenuation at Fsw, then 2 copies will have 100 dB of attenuation.  The TLV751 is a dual LDO which might be a good choice if two LDO's in series are needed, and the TPS7A20 will not be ready in time for your schedule.  My colleague will respond with additional feedback soon.

Thanks,

- Stephen

Dear Stephen,

Thank you for your quick response.

I looked at the TPS7A20 and even though the package is a great size - the design of the footprint (X2SON) might be too hard for our PCB Production to make (we had a lot of trouble with a device that had a pitch 0.4mm and the X2SON 0.22mm clearance from PAD to middle pin is probably too small !). Do you have anything else ?

What about the Buck - Boost ?

Best regards

Shmuel 

Hi Shmuel,

Here are the packages the TPS7A20 is expected to come in.
Will either of the other packages for the TPS7A20 work for you?

X2SON: 1mm x 1mm

DSBGA: 0.603mm x 0.603mm

SOT23-5: 2.90mm x 1.6mm

My colleague also suggests the TPS7A87 for a duel channel device (again, wired in series to increase PSRR by 2x). 
If you think a single LDO device will provide sufficient PSRR (which again will reduce space), the TPS7A90 or LP5912 may work well here.

If these devices will work for you, we can engage the low voltage switching team for an optimum boost converter.

Thanks,

- Stephen

Dear Stephen,

Thank you so much for responses and suggestions. 

The TPS7A20 packages are not really going to work for us and also we would prefer to take a device that has already been released and been on the market already

We have very little room so a dual channel device is not really an option (we also have a Buck Boost feeding the LDO)

The LP5912 looks good but I am leaning more towards the smaller package of the MIC5219-5.0YMT-TR which appears to give similar ratings !

For the Buck-Boost I am currently considering the MAX77827 set to +5.3V !

Best Regards

Hi Shmuel,

Upon review it really looks like the LP5912 is better suited for you here than the competitor component.
We can also look at the switching converter if you would like me to engage someone from the switching team.

Let's work through the design challenge. 
The switching converter frequency is 2.5 MHz typical, and across temperature it can be from 2.2 MHz to 2.8 MHz.
Let's assume this is the switching frequency of the chosen device regardless of manufacturer for the moment.

If you are designing a switching converter with a boost derived topology, it typically has low bandwidth due to the RHP zero and high ripple content.  Low bandwidth results in larger output voltage deviations due to transient responses.  You are space constrained so you cannot add a secondary low pass LC filter after the power stage of the boost converter.  You need high PSRR at 2-3MHz.

Worst Case Analysis - Dropout Voltage
At 5.3V input you will want to accommodate tolerance (temperature, steady state variation, transient variation, IR drop).  Let's assume 3% tolerance, so 5.141V to 5.459V, but the ripple content of the boost derived converter may make this 3% goal challenging in a space constrained environment.  If the LDO also has 3% tolerance, then 5V * 1.03 = 5.15V.  So a worst case analysis will not work here as the LDO output = 5.15V while the LDO input = 5.141V.  And 3% tolerance including transients as well as steady state tolerances can be hard to achieve.  The MIC5219 accuracy is +/-2% but that does not include transient loads or IR drop.  In addition you will want to account for the worst case dropout voltage. As discussed below, the MIC5219 has 150mV of dropout at 100mA of load, versus the LP5912 dropout of 25mV.

PSRR
PSRR is sensitive to the headroom voltage, which is Vin - Vout.  Operating at the dropout voltage (the minimum headroom which will still allow proper regulation on the output) will give much less PSRR than, say 1V of headroom.  Above 1V of headroom and you don't see any significant gains in modern LDO's.  If you want to reduce power loss as much as possible, you want a device with the lowest dropout voltage.

The LP5912 looks much better in this application than the MIC5219.  The MIC5219 uses a BJT pass element which will have higher dropout than the LP5912 which uses a MOSFET.  For a MOSFET the dropout is simply the load current multiplied by Rds_on.  At 100mA the MIC5219 requires 150mV of dropout vs just 25mV for the LP5912. 

If you need PSRR at 2-3MHz, the LP5912 can equal or exceed the MIC5219.  The output capacitor will dominate at these frequencies and we can use low loss ceramics to dial in the PSRR at these frequencies.  The datasheet uses 1uF on the output but that can be increased to 2.2uF or 4.7uF to lower the PSRR improvement you see at higher frequencies.  An EVM would be very helpful here to test your options.

There is a statement in the MIC5219 that would give me pause and want to test my solution before incorporating it into production.  Regarding the output capacitor selection: "Ultra-low-ESR capacitors could cause oscillation and/or underdamped transient response."  Ceramic capacitors will likely meet this statement as the datasheet speaks about tantalum and electrolytic type capacitors being used on the output.  You need to make sure the MIC5219 does not oscillate if using an "ultra-low ESR" capacitor.  The LP5912 was designed for low ESR ceramic output capacitors and will remain stable if used in accordance with the datasheet.  I'm not sure about the MIC5219.

If you would like to discuss further, please let me know.

Thanks,

- Stephen

Dear Stephen,

Thank you for your very clear explanation and clarification. As I continued my research I actually moved closer to choosing the LP5912. I am now in the middle of finalizing whether to choose this part or the or the MAX38902. It appears that both will give similar noise reduction results but the LP5912 will in fact give me a smaller and simpler design solution. What do you think ?

If you don't mind please see the following thread that I have on this topic(maybe we can close this one and continue there ?) :-

 https://e2e.ti.com/support/power-management/f/196/p/923257/3420102#pi320995=2

Best regards

Shmuel 

Dear Stephen,

Thanks to John's and your great responses and very accurate comparisons to the competitors part, I have more or less decided to move over to TI LP5912 ! Thanks for all your help.

Best regard

Shmuel