Several questions about hardware

[Ampchamp]

Hello all,
I have some questions about the internal analog interfaces (I’m referring to the mainboard version 3 resp. „mainboardv1.1.pdf“)

1) The current measurements are realized with the MLX91205 hall sensors, these are supplied with 5V and have an output voltage of 2.5V at 0A loadcurrent. A first resistor divider (1k2-3k3) is placed on the sense board.
On the mainboard the current signals connect to JP7.6 and JP7.9, followed by an additional resistor divider (3k3-4k7), afterwards they’re routed to the µC (ADC5 / ADC8).

If I calculate / simulate the voltage levels with the combined resistor dividers, I see at the output of the sense board / input of the mainboard a 0A-voltage level of 1.65V. This is almost exactly 3.3V/2 and I would expect this 0A-voltage level also at the ADC inputs of the µC.

But, this voltage is divided once more by the second resistor divider on the mainboard, which results in a 0A-voltage of 970mV. This is rather surprising, particularly because it limits the complete measuring range to 1.94V.

Is the concept really the way it is described? Or do I misunderstand something?

Thanks for your replies,
Fabian.

[johu]

Basically you wouldn't populate R50 and R51 in that use case. (Bestückoption)

[Ampchamp]

Ok, understood.

But without R50, you don’t have a parallel-resistance to R2 (3k3 on the sense board), which leads to an overall division factor of 0,733 and a 0A-voltage of 1,83V at the µC.
From a theoretical / calculative standpoint, you need a parallel resistance of 8kΩ to R2 to reach a division factor of 0,66 and thus 1,65V at 0A.
So you need to populate R48 with 0Ω and R50 with 8kΩ. But then, you don't have an additional R-C filter at the mainboard.

Or is the 0A voltage at the ADC input of the µC measured during initialization and this value is furthermore used as an offset correction?

Is there somewhere a BOM in which it is described, what components are populated or have which value in which version?

This also leads me to my next question:

2) As written in another thread, I don’t like the implemented internal temperature sensors so much but prefer other types such as PT1000 or screwable NTCs.
As the temperature sensor is supplied with 5V, but the ADC input is limited to 3,3V, the sensor voltage should stay within this range.
With a PT1000 sensor, it is no problem with the 1k2 pullup resistor. But, the voltage at the ADC input only vary between 2V and 2,84V going from -50°C to +150°C, so you only use a small part of the possible voltage range at the ADC.
On the advantage side: the curve is quite linear.

But if I want to use for example a K45 NTC with 2k2 @25°C, it is needed to place a parallel resistor of about 2k to the NTC for limiting the max voltage and having a good voltage range for measuring (3,11V @ -55°C; 0,3V @ +125°C).
Without parallel resistor, the voltage at the ADC input will be over 3,3V for temperatures below 0°C even with the K45-1k.
Unfortunately, this parallel resistor isn’t included in the schematic and the layout.

Johannes: Do you think it is a problem to implement a temperature curve in the software, which is only valid, when an additional resistor is mounted? My concern is, that someone uses the NTC and gets wrong temperature values, because the parallel resistor isn’t considered.
Or (again), is there a BOM, in which this additional component could be listed (with a remark, when to populate them)?

CU, Fabian.

[johu]

Yes the firmware does an offset correction in stop mode.
There is no BOM, just the instructions in the schematic. I will open source the the design files once I've sold off the first batch. Then you would be very welcome to become maintainer for the BOM in wiki.
For PT1000 there would still be 1042 digits or 5.2 digits/°C. I think thats plenty for temp measurement.

K45: There is not way to determine if you're seeing a real temperature or one that is generated because of a missing resistor. Very low temperatures are not so relevant for operation, so you could just let the signal go out of range below 0°C. The controller won't mind as there is a 10k series resistor.