External Vref+ on nucleo-h723zg

[moivet]

I'm trying to create an ADC project using nucleo-h723zg. I want to measure AC voltage and current (simultaneously). I created electronics with isolation amplifiers, which have, as an output, a differential signal with a common DC voltage. The common DC voltage is set by the manufacturer to be 1.45V.

I followed the STM32H723 manual and noticed that the input common voltage (for differential inputs) on the ADC should be around Vref+/2. So I need Vref+ to be 2.9V. So I created that reference voltage regulator on my separate electronics board. I also created a small power-up delay (~1ms) for this Vref+ voltage. 2.9V LDO's ground is connected to Agnd on the nucleo board.

I removed R36 from NUCLEO board and connected my new Vref+ source.

ADC signal is within 2.9V span (Agnd - Vref+), together with a DC common voltage. Also, I created a start delay for ADC signal as I did for Vref+ voltage (~1ms).

In software, I set VREFBUF mode to "external voltage reference".

When I start the system (nucleo board + my electronics board), STlink reports that the target can't be found. Probably, the microcontroller is burned. On board, everything looks good. LDO's are ok (3.3V, and 2.9V), and cold.

Am I doing something wrong? Any suggestions? Is there any manual explaining external Vref+ connection on these nucleo boards?

[TDK]

What you're doing should work.

Does STM32CubeProgrammer report the correct target voltage? Does the chip recover if you remove the external LDO?

Maybe attach a picture of your setup showing the connections.

FWIW, in my experience you don't need common mode to be VREF+/2. It will work just fine with VREF+ at 3.3 V.

[moivet]

Hi...

Well, STM32CubeProgrammer is reporting a core voltage of 3.29V. I also measured everything, from voltages to the jumpers. I also measured if any used pin is burned to GND or VDD. Everything is ok.

I removed the external LDO and soldered R36 bypass again, to reverse modifications, but it was still the same. Think the microcontroller is done. I'll get the new board. But I want to be sure not to destroy the next one.

When I started with this project, I just connected the voltage and current signals to the nucleo board. Just to see if it is going to work. I got samples, and I calculated RMS values. But they were off. I used cubeMonitor to visualise data. And I saw that voltage samples in some random places have zero value or some noise going from zero to the voltage level.

I've read that this issue can be from using Ethernet, as I do. It draws too much current, and adc is showing that in the data.

Wanted a pure signal read, so I created two LDO's. Also, when I'm building extra electronics, I've decided to go with proper voltage levels. One for the Vdda (3.3V, which I'm not using, for future use if possible), and Vref+ 2.9V.

First, I started to integrate Vref+, but this happened. Maybe with only vref+ supply signal will get clean, but if not, I will be forced to add Vdda supply also. And I don't want to break any more boards.

I see there are connectors for both Vref+ and Vdda on nucleao board. So there should be a way to use them.

Is there some software issue? Do I need to change something manually in the software?
For hardware modifications, I used schematics from 2019 (mb1364-h723zg). Maybe some newer hardware modifications need to be considered?

Attached connection picture...

[AScha.3]

>Any suggestions?

Restore prev. "standard" ; R36 , remove your vref ext. connection;

Try to get st-link connection to cpu ....use CubeProgrammer . (show, what its telling)

+

Let vref as it is, all you "loose" is some percent of max. input value for ADC. But no danger of damaging anything.

[moivet]

Hi AScha.3...

Everything is restored, but still, I can't access the microcontroller... screen picture attached

AScha.3...What do you suggest to do with the lost samples? I get a signal like in the attached picture.
And on the oscilloscope signal is ok.
And sample rate is around 5kHz (triggered by a timer). Dual regular mode. Double buffer dma.

[AScha.3]

>but still, I can't access the microcontroller

- power cycle , unplug USB...reconnect it.

Try with mode: hardware reset;

if nothing...last chance: set boot0 hi, try connect to bootloader (user port USB).

+

>What do you suggest to do with the lost samples?

(If cpu still working.... :) )

Calibrate ADC in diff.mode done ?

Try sampling without DMA , just as a test; if all samples fine, DMA is problem (cache mangement ?).