Strange display power consumption issue

[radiolistener]

I just got powerful M6-PD100w power bank, I'm planning to use it as a power supply for my computer during power outage. Played with it a little and found strange power consumption issue of my display.

There is Samsung 971P display connected through DVI-HDMI adapter to Raspberry Pi 4B computer.
When I connect display to the power bank, and RPI is powered from mains, the power bank shows power consumption 12 V x 2 Amps. DMM also shows 11.6 V x 2 Amps. That is expected behavior, because the same 2 Amps is consumed when I power display from a mains power supply.

But very strange thing happens when I power both  display and RPI from the power bank. In such case power bank shows RPI power consumption 5V x 0.6-1.0Amps (which is expected value), but display power consumption drops down to 12V x 1.5 Amps... 

My first thought was is that there is error on the power bank indication, but DMM shows that display consumption really drops down to 1.5..1.6 Amps. The voltage is almost the same 11.5..11.6 V.

So, I have no idea why display consume 0.5 Amps less current when I power RPI from the same power bank? I well understand Ohms law and checked all things several times, and it shows that the power consumption is really drops down.

I tried to connect other 1 Amps load instead of RPI to the power bank, but it don't reduce display power consumption. The RPI power consumption is really 0.6-1.0 Amps and don't depends if I power it from power bank or mains power supply.

This looks crazy, because display works the same with the same intensity, but it's power consumption is reduced for 0.5 Amps.

I don't see how it's possible . Any idea?   

[Faringdon]

Hi, It would be good to see an actual current sense resistor shunt in the wire leading to the display, and then see with a scope what is the voltage on that sense resistor in each case.
...And you can cleverly use capacitance(s)  to make sure that the current in the sense resistor is flat DC, (in each case) so as to make the comparison easier.

I mean, i am not sure exactly where you are measuring current?, or how...remember that mains power supplies have this thing where, if not power factor corrected, they draw quite a bit of reactive current aswell....which can sometimes make things look misleading. Also, if the current you are measuring is not flat DC, then some meters wont be able to measure it properly.

Also, if the current waveshape in the sense resistor is not flat DC....then you can RC filter the voltage of the sense resistor, to get it flat,  in order to get the average value....do this in each case, and compare....to see if the current really is different in each case that you describe.

[radiolistener]

I mean, i am not sure exactly where you are measuring current?, or how...

Measured with DC clamp meter UNIT UT210E. The power is DC voltage.

I will check it with oscilloscope later

[PlainName]

Off the wall idea:

The HDMI is providing RPi power. When the Pi is mains powered the 5V rail is lower than the 5V supplied by the monitory, so the HDMI 5V is used. Hence 1.5A for the display and 0.5-1A for the Rpi. When the Pi is powered from the powerbank, the Pi 5V is at least as high as the HDMI 5V (same source) so the Pi is powered from the powerbank instead of the display.

[Faringdon]

This comes up for UT210E
https://www.farnell.com/datasheets/2724182.pdf

...likely very low bandwidth, datasheet doesnt even say

[radiolistener]

Off the wall idea:

The HDMI is providing RPi power. When the Pi is mains powered the 5V rail is lower than the 5V supplied by the monitory, so the HDMI 5V is used. Hence 1.5A for the display and 0.5-1A for the Rpi. When the Pi is powered from the powerbank, the Pi 5V is at least as high as the HDMI 5V (same source) so the Pi is powered from the powerbank instead of the display.

I also think there is probably something related with HDMI, but RPI is powered from genuine PSU which has 5.2V, it is higher than 5V provided from powerbank.

[radiolistener]

This comes up for UT210E
https://www.farnell.com/datasheets/2724182.pdf

...likely very low bandwidth, datasheet doesnt even say

As I remember, according to my measurements it has about 2 kHz bandwidth, but since I measure DC current, bandwidth is not relevant here...

[radiolistener]

Just tested with oscilloscope. Nothing unusual some a little spikes, but they almost the same.

First two pictures when power bank connected to display only.
The last two pictures when power bank connected to display and RPI.

Also I found that there is DC current unbalance 0.3 Amps through HDMI cable.
The same 0.3 Amps unbalance through USB cable which provides power for RPI.

So it seems that the display consumes 5V x 0.3 Amps through path: power_bank->USB->RPI->HDMI->display... 
Any idea how it happens?

[Faringdon]

It seems to me that the display could not possibly "know" if the RPI is powered from the mains , or the power bank(?)

The display current looks very pulsey indeed....i think you need to use a  sense resistor and multi  stage RC to filter that signal flat in each case...then maybe you will find that in actual fact, the average current is the same in each case. This is all i can think of for the moment.

[radiolistener]

No, the current is not the same.
As I mention before the missing 0.3 Amps current flows through USB->RPI->HDMI->display. I see it with DC clamps on HDMI cable.
But I don't understand why it happens.

[Wolfram]

Are you measuring current through the positive or negative terminal of the power input? If you connect devices together with interfaces that include a ground, you can create multiple ground paths between devices, and the current from the power supply negative terminal can split between these paths depending on their relative resistance.