Resistors and ESP32 ADC

[Lonestar78]

Dear @nliaudat,

quick one....i use v54, rev 1.34....there i find 1,07 Ohm resistors between BEMF and Ground (R3).
Measuring BEMF on the open header gives me ~10mv if a valve opens/closes without load, ~30mV on endstop (open and klack klack klack...).
Given that the ESP32 adc reads 0 for 0-0,1v how do you get ADC readings?
i always get 0....damn, can it be, that the 1,06 Ohm Resistor is to small? Why did you change it from 12 Ohm?
Attenuation is set to 2.5db and 12bit resolution.

Any hint is welcome.
May it be, that i have to re-solder bigger (= higher ohm) to R3?
also could use 10kOhm there and another 10kOhm between header for BEMF and VCC.
What do you think?

Thanks!
Christian

[nstadigs]

Greetings!

1.07 ohm resistors don't work for me either. See #9.
I resoldered 12 ohm resistors to my boards and that works well for me. I don't know why @nliaudat settled for 1.07 after that discussion, but since 1.07 seems to work for others, I have a working solution myself and the fact that I don't understand half of these things I just dropped it. 🤷‍♂️

[Lonestar78]

It might depend on the esp32. The S2 version reads from 0 volt, not only from 0,1... Holen Sie sich Outlook für Android<https://aka.ms/AAb9ysg>

…

________________________________ From: Niklas Stadigs ***@***.***> Sent: Monday, September 19, 2022 8:34:09 AM To: nliaudat/esp32_8ch_motor_shield ***@***.***> Cc: Christian Winkler ***@***.***>; Author ***@***.***> Subject: Re: [nliaudat/esp32_8ch_motor_shield] Resistors and ESP32 ADC (Discussion #19) Greetings! 1.07 ohm resistors don't work for me either. See #9<#9>. 12 ohm resistors works well for me. I don't know why @nliaudat<https://github.com/nliaudat> settled for 1.07 after that discussion, but since 1.07 seems to work for others, I have a working solution myself and the fact that I don't understand half of these things I just dropped it. 🤷‍♂️ — Reply to this email directly, view it on GitHub<#19 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AFRV6WWDHCZTNJM2TCC3N6DV7ACODANCNFSM6AAAAAAQPR6L6Y>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[nliaudat]

Hi,
L9110s chips have a buildin resistance between gnd and bemf which is undocumented. 12 ohms Did not work for some actuators, that's why I change it to the smallest available 1.07ohm.

You can add a resistor in open headers between gnd and bemf to fit your needs. I do not remember if it setup serial or parallel.

[nliaudat]

Any news ? Have you tested with external power supply ?

[Tecnologic]

No had no time. Hopefully today evening

[flexmatics]

OK, i removed on CH1 + CH2 the two 1,07Ohm resistors and replaced with 2x 10Ohm on the openheader between the two BEMF pins and GND, this seems to work.

Confirm, that this fix has also worked for me.

[nethawk2]

I use v54, rev 1.34 and replaced the eight 1,07 Ohm resistors by 10 Ohm resistors too. Now BEMF works, if i set BEMF trigger to 0.20.
My first SMD soldering ...

[nliaudat]

I use v54, rev 1.34 and replaced the eight 1,07 Ohm resistors by 10 Ohm resistors too. Now BEMF works, if i set BEMF trigger to 0.20. My first SMD soldering ...

No need to remove the 1,07o resistors, you can add some in the free headers labeled BEMF to GND

[Tecnologic]

I now hat time to post my findings. My HmIP-VdMot motors are rated on the Label with 120mA max. and 3V3 supply voltage. During running the draw 20mA while opening and 30mA while closing. When they run to block while opening they draw roughly 40mA before the click click noise (gear teeth slipping) is coming. When closing they draw 120mA before the drive pops up from the valve.

So i switched to dual ads1115 as adc for all channels and used analog_threshold at 40mA for opening and 80mA for closing and they work like a charm.

[herrfrei]

@Tecnologic: What is the update interval you reach with the two boards on i2c? Could you please provide your yaml config?

[Tecnologic]

Hi,
This is my yaml https://github.com/Tecnologic/floor-heating-controller

[herrfrei]

Hi, it looks that you are not using the ads1115 anymore and developed a new board. May I ask you, why?

[Tecnologic]

I have problems with the reliability of the values endpositons. Some valves do not completely close occasionally. Some times they completely block end I need to adjust the threshold and so on. It is running right now and working ok. But I need to monitor and maintain it every month.

From my point of view it's the simple current peak endstop mechanism which is flawed. I do industrial motor control for a living and can't sustain such a solution. But with esphome it's difficult to get this better with pure yaml.

[nliaudat]

ads1115 smt chip costs 5.42$ at JLCPCB ==> +11$ price increase for the board.

[Tecnologic]

The adafruit breakout boards are 3€ a piece. But for a PCB redisign I would use the opamp circuit I posted above with an gs8631 opamp (lcsc stock) and a cd74hc4067 analog multiplexer.

The cd74hc4067 is supported by esphome and with filter settings for the sensors one can do the offset and inversion of the opamp signal to get a valid current reading.

By the way I tried the easy eda links to join the design but I have no rights to make changes. Need some guidance there. Other wise I would use kicad.

[nliaudat]

Many thanks, I've set you as developper in easyeda (default is observer)

[Tecnologic]

Thanks, don't expect anything next week, hole lot of work that week.

[nstadigs]

Hi! Has there been any progress on this? Or should I get me some ads1115's?

[Tecnologic]

For now I run two units with ads1115 without flaws. It's just a bit ugly

[Tecnologic]

For now I run two units with ads1115 without flaws. It's just a bit ugly

[nliaudat]

For information, the 1R resistor for BEMF works great with ESP32-C3.
I've done some tests with https://github.com/nliaudat/floor-heating-proportional-valve and I can read from 0.00v (not 0.08V) with steps of 0.01v correctly.
I mean that the problem of reading BEMF is related with some different quality of ESP32-WROOM-32D.

[nstadigs]

Yes, I understand that. I was thinking that instead of adding components to the circuit for reading BEMF as Tecnologic suggested we could modify the board to fit the pinout of the S2 devkit. This would at least require less components.

[nliaudat]

changing the devkit is not so easy as pinout :

• The S2 & C3 compile only with esp-idf framework and that means the web server is not available anymore.
• S2 have ADC1 10 channels: GPIO1 - GPIO10 free (ADC2 is always used by wifi) = ok but the board is expensive (~10$)
• C3 have ADC1 5 channels: GPIO0 - GPIO4 free = ok but expensive (ESP32-C3-DevKitM-1 = ~ 12$)

There are differences between espressif documentation and devkit. As example, the wroom-32D has 8 ADC1 channels, but only 6 are available (ACD1_CH0 and ACD1_CH3 to ACD1_CH7) while ADC1_CH1 and ADC1_CH2 are not available (even the pins are not exposed in the ESP32 Development Board).

Changing of devkit must be carefully studied and the price raise from 3.5$ to 10-12$ for the devkit
Unfortunately, I have just changed jobs and I no longer have any free time.

[nstadigs]

I see. Thanks for the explanation. Maybe we should just go the opamp route proposed by Tecnologic then?

Good luck with your new job!

[nliaudat]

Thats a good solution with minimal effort. I'm currently looking in to running the actuators current controlled because they are quite unreliable when running a pid_climate. But this would be possible with the ads1115 or the s2 esp2.

@Tecnologic
I have tested pid climate but the very high update frequency was bad for valve life.
I've used the PID_mode: kp: 30 ki: 0.005 kd: -24000 for my low temperature underfloor heating.

Have you find good parameters ?

[Tecnologic]

OK. Set KD to 0. Thats a floor heating that is just a big pool sink for energy. There is nothing the Controller needs to compensate quickly. Then determine the rate of ur Temperature Update. My is 60s. Then assume the time constant of ur room. Typical Something like 8h. Then U choose how much the valve should react for a 1K Error in Temperature, i would start with 0.03 (1=100%) for 1K. So Kp would be 0.03. With the time constant of 8h Set as TN that gives a ki = Kp / (TN / UpdateRate) = 0.03/(8h*3600s/60s) = 0.0000625

[nliaudat]

For information, esphome support now the 8 channels adc128s102.
It cost ~4.2$ (much more than the ESP32 devkit)

[nliaudat]

@Tecnologic Do you mind adding a LM358 opamp chip could fix the ADC reading ?

[Tecnologic]

Yes. U need one amp per Channel so a Quad amp or an an analog multiplexer Like i suggested

[Karl-opec]

A Quad Op-Amp LM324 cost less then 10cent at JLCPCB some Resitors and capacitors to add, would be a cheap solution to fix the ADC-Readings.

[nliaudat]

[Karl-opec]

First, for my understanding, what is the reason to have the 1.07 Ohm Resistor two times per channel block(CH1,Ch2)? These resistors are in parallel, which halves the resulting value (0.5 ohm). So a 1ohm resistor would be sufficient. (Power 625 µW)

The Voltage over the resistor is a measure for the motor current. The blocking Motor current according to this Graph Vd-mot Motor current is 25mA. Over a 1Ohm resistor this lead to 25mV to Ground as CH(1,2)-BEMF. so the Gain of the OP-Amps should be greater than 5 to have a voltage over 100mV for the ESP32 ADC.

Am I right or do i miss something?
The capacitor in the output of the op-amp makes BEMF to ESP32 ADC AC-coupled. I think we need it DC-coupled.

These are my very general comments. My time for having done such design are long time away, Perhaps @Tecnologic can make comments more valid.

[nliaudat]

Many thanks for you support !

I am not an expert on electronics, but the L9100s motor driver is a bit uncommon ... If you remove the 1R resistor, it do not works well.
Some people need to cut them off and replace with 12Ω resistors in series to get it works. (but not everybody, it seems to be random, adding 12Ω resistors works without cutting 1R)
My own board works great since 3 years with 1R resistors and ADC is very good. I mean there is factory ADC calibration problem with some cheap boards.

My homematic motors run between 0.01A to 0.024A and endstop is at 0.08A-0.1A. It differs from motor ...

I have no experiences with opAmp, that's why I ask.
You are right for the capacitors. They are useless.

Easyeda link

[Karl-opec]

The Shunt Resitor from L9110s GND pin to electric GND of the circuit is definitly needed. You need it to generate a Voltage which is a measure for the motor current. The Motor Current is what you want to measure . The unknown value of an internal resistor is of no interest.

[nliaudat]

@Tecnologic @Lonestar78 @Karl-opec : Can you validate the schematic for LM324 opamp ?

calculation done at http://mustcalculate.com/electronics/noninvertingopamp.php?vin=0.01&r1=10k&r2=50k

[nliaudat]

Many thanks.
Feel free to commit your improvements in Easyeda V56 Karl-opec improvements
I have no time actually to do - new job. sorry

[xdubx]

did the v56 needs a new label? On the pcb stands still "PCB v55 rev 1.45"

[nliaudat]

v56 is the same as v55. No time to do the changes

[nstadigs]

I have JLCPCB let build me the PCB v55 rev 1.45. It works very good so far. I have replaced the 1.07 R with 2 ohms. so in parallel it is 1 ohm. With the gain of the op-amp (6) this gives blocking voltages at 25mA / 1 Ohm * 6 = 0.150V. setting these values as bemf_trigger it works great. I propose you add also the low pass-filter which @Tecnologic mentioned and it would be a additional improvement. The resistor issue would be solved.

You could save in addition 4 Resitors, if you leave off one of the shunts per channel. 1 Ohm is sufficiant. There are not two needed.

@Karl-opec You don't happen to have a copy of your BOM and pick-and-place files do you? :)

[Karl-opec]

@nstadigs : you find them under the above link.

In the BOM I have replaced the 1,07 Resistor against 2.0 ohm resistors . NR. @ JLCPCB C17942 Manufacturer Nr. 1206W4F200KT5E

[xdubx]

I have one from azdelivery and the problem with the 0.08v BEMF to.
Try the next days my other ESP32 i hope one of them works fine.

The c und S2 version working without problems? Can someone recommend a shop where you are sure not to get a buggy esp.

[xdubx]

OK i double checked the electrical voltages. One of my valves vom homematic have weird value.

[nliaudat]

New version v57 rev 1.47 :

Replace the 1.07R with 2 ohms

[nliaudat]

The opamp do not work in latest version.
I do not understand why ...

[nliaudat]

I have received 5 last week, but the BEMF values reported by the opamp are the same as without. It should have a gain of 6x. It works, but without amplification.

[nliaudat]

I tested to inject 0.2V directly in the BEMF pin and I get 1.2V as output. The opamp works, but if I connect the esp32, the AC/DC converter and the Vmot motors, I get only the previous BEMF values near 0.3V.

[Karl-opec]

@nliaudat getting close to 0.3 V ist absolutliy fine. U(bemf) = R * I(mot) * V(opamp) -> 0,3V = 1ohm* 0,05A *6 .

R(shunt) CH1 || R(shunt) CH2. 2ohm parallel 2Ohm -> 1ohm resulting.

50 mA Motor Current ist Blocking motor. I have setup in the software as Motor Endstop Voltage 0,180 V. since then it work . No more Pop up's from the valve etc. Look at this chart to see the Motorcurrent over valveopenening.

If you want higher values of U(bemf) you can increase V. But this not recommend becaus noise is also increased.

following find my definition of sensor_adc.yaml:

sensor:
### in use the motor use 15-20 mV. At endstop, it raise to 50 mV
  - platform: adc
    pin: GPIO36
    name: "BEMF 1" #CH1 + CH2
    update_interval: 100ms
    attenuation : 0db
    id : BEMF_1_2_sensor_ADC
    accuracy_decimals: 3
    filters:
    - sliding_window_moving_average:
        window_size: 9
        send_every: 9

This averages the voltage measurment of 900ms und gives a value every 900ms.

this works for me since 4 Month now

[nliaudat]

@Karl-opec many thanks for your sensor_adc.yaml sharing. I've tested it for fast an hour with a small modification to get 0.000V instead of 0.075

    filters:
    - offset: -0.075

My BEMF trigger is set to :

ch1_bemf_trigger: "0.105"
(0.180-0.075) instead of 0.180

It seems to works.

BUT : I do not understant why my previous BEMF trigger value (0.200 to 0.350 depending the valve) without opamp is the same or less with opamp.

[nliaudat]

@Karl-opec : do you want I add you as developper ? You have more skills than me in electronic and could improve the project.

[Karl-opec]

Thank you for you offer, yes it would be nice.

your previous BEMF trigger can be reached. look at the Motor Current chart. While Blocking the motor can gain higher currents and this is leading to high BEMF Trigger Values. But remember you want to measure the Motor Current. and more than 30-35 mA are not needed to trigger.

because of different blocking currents for openning and closing i have split the trigger levels for open and close.

the use of the filter is not realy of any use. as the trigger level is the switch pont.

[nliaudat]

What is your easyeda member name ?
You are invited in github as collaborator

[Karl-opec]

my Username is Karl23

[nliaudat]

Everybody who's want to contribute must say.
I have closed the public easyeda sharing cause nobody commited an improvement since 3 years and it was a rant

The main goal is to make an open source heating controller for everybody. The only contributors are in that discussion and you are welcome to improve or get more involved.

Regards and many thanks for your improvements

[nliaudat]

Here is the easyeda link to view the project : https://oshwlab.com/project/join/b2642362910a768d05dcb075a611a4d2
I'll add everybody to the project as developper again if you want to contribute.