Stm32u5 mixed power domains - 1.8V & 3.3V ?

[sb_st]

Hi! I'm trying to muddle my way through a project in which I'd like to use an stm32u5 MCU. 

I would like to extend the RAM/FLASH of the MCU by using PSRAM and eMMC; it seems that higher densities of PSRAM require being run at 1.8V, so I am imagining powering at least the digital side of the MCU with a 1.8V rail. 

My application also involves motor control; I'll be using the STSPIN233, which can also be powered by a 1.8V rail. I would like to perform current sensing, and here's where my question comes in. I'd like to use a current sense amplifier, e.g. TI's INA181 (or similar). These devices need a 3.3V rail for power. I am curious whether it is better to:

• Power the analog domain of the stm32u5 at 3.3V. The current sense amplifiers + shunt resistors are scaled such that I produce a total rail-to-rail voltage near 3.3V, biased at 1.65V.
• Power the analog domain of the stm32u5 at 1.8V. The current sense amplifiers + shunt resistors are scaled such that I produce a total rail-to-rail voltage near 1.8V, biased at 0.9V.

I've not done a mixed-power-domain setup before, so I'm unsure if there are nuances or gotchas there to be aware of. Conversely, if I choose the second option above, I'm unsure if it's "better" or "worse" for the ADC to ultimately 'see' a more narrow voltage range, in terms of accuracy or performance? 

Any advice?

Thank you!

[AScha.3]

Hi,

as i see it : 

you want use > INA181 (or similar). These devices need a 3.3V rail for power.

So output is in 0...3v3 range -> "natural choice" is 3v3 range for analog range - right ?

+

>I've not done a mixed-power-domain setup before

Here i see no reason to do it.

+

>I'm unsure if it's "better" or "worse" for the ADC to ultimately 'see' a more narrow voltage range, in terms of accuracy or performance?

Easy - if your signal is in 0 +/- 0.1V range , its best to amplify it , to match the ADC full scale input, better is not possible.

But if you dont want to spend some cents for an opamp...ok, then need to match as good as possible.

But never forget: with some switching bridges around, there will be some electro-/ magnetic "noise" , spikes, so the smaller the signal, the more noise is in the measured result. So try to get big signal levels...and big ADC range, if possible.

[sb_st]

Hi! Thanks for responding!

---

you want use > INA181 (or similar). These devices need a 3.3V rail for power.

So output is in 0...3v3 range -> "natural choice" is 3v3 range for analog range - right ?

---

Well, not as I understand it - the output is based on 1) the gain of the current amplifier, and 2) the choice of shunt resistor. So, I can run the INA181 at 3.3V, but scale my shunt resistor so that the output ranges from 0-1.8V. 

>I've not done a mixed-power-domain setup before

---

Hm, this seems to contradict what you just wrote above. As I mentioned in my initial post, I need to run PSRAM/eMMC at 1.8V. So, my VDD domain for the MCU will be at 1.8V. If I run VDDA at 3.3V as you suggest, that is two different power domains I am using...isn't it?

---

Easy - if your signal is in 0 +/- 0.1V range , its best to amplify it , to match the ADC full scale input, better is not possible.

But if you dont want to spend some cents for an opamp...ok, then need to match as good as possible.

But never forget: with some switching bridges around, there will be some electro-/ magnetic "noise" , spikes, so the smaller the signal, the more noise is in the measured result. So try to get big signal levels...and big ADC range, if possible.

---

Yes, I understand all of this - I have some experience with current sensing and motor control, and have used INA181's to prototype my design so far. My question is: are there any relative advantages, drawbacks, or 'gotchas/things to look out for' when using ADC+VDDA at 1.8V vs. ADC+VDDA at 3.3V? Particularly when the digital domain of the MCU (VDD) is at 1.8V? Or are both options effectively equivalent, conceptually?

Thanks again!

[AScha.3]

Hi,

It seems I misunderstood you.

If you anyway need/want 1v8 VDD , i would choose also the VDDA at 1v8 , just to keep it as simple as possible.

From ds U59xx :  VDDA min. 1.8 V in general op. conditions , so you get full/all analog performance .

Just dont forget : 2. The voltage booster on the ADC switches must be used when VDDA < 2.4 V (embedded I/O switches).

• When VDD is below 1 V, other power supplies (VDDA, VDDIO2, VDDUSB) must remain below VDD + 300 mV.

• When VDD is above 1 V, all power supplies are independent.

So just look at ds of the cpu you choosing, to check: 1v8 is in range for full analog performance.

And read the footnotes...  (booster has to be switched on, etc.)

[MHoll.2]

As the STM32U5 has a separte VDDA domain, there is no better or worse, dipends on what Your prefer, I whould probably use 3.3V for the ADC, because this gives me less problems with noise / disturbence on the Analog signals.