Hello! I have a problem with ADC - reading analog values specifically for this project. I don't understand why analog values change by themselves, within small limits, but they change and this is unacceptable. The problem is not in the electrical circuit (hardware) solution, but the analog value reading configuration is incorrect I think. Attached is a video where you can see that the values on the display jump around within large limits, but the battery voltage is actually measured, which is stable, and there are no disturbances (high frequency, etc.). GCB helps contents various solutions, but I have tried them, it does not work. There are still solutions out there, but I don't know which one to use. For example, #define LowSpeed 0 or Highspeed does not make any difference. Here in the text is a part of the program code for defining and reading the analog input. I used #chip 16f18446, 32 and analog resolution 8bit. I hope someone has had a similar experience and has an idea how to solve it.
The full program code is attached.
''==== ANALOG IN
#define volts_AN4 an4
dir porta.4 in
#define ACS758 an5
dir porta.5 in
#define LowSpeed 0
''==================== A ============================ 50A ACS758-50U (unidirectional)====
Lim_amps_AN2=ReadAD(ACS758)
If Lim_amps_AN2<31 then Lim_amps_AN2=31'' 0,6V quiscient voltage ACS758
amps_AN2=scale(Lim_amps_AN2, 31, 184, 0, 500)
if amps_AN2<4 then amps_AN2=0
Amp_disp_str=str(amps_AN2/10)
Amp_disp_str_mod=str(amps_AN2%10)
Here is a revised program. Documented and with #option explicit added. I had to reformat and document so I could get to this far, then I added #option explicit, then I resolved those issues.
But, this analysis ( so far ) will show you why you should always use #option explicit.
I have not looked at the ADC issue as I think changing the register.bit PREV3 may have an impact on the ADC operations. As PREV3 is an ADC management register.bit
So, compiling the ERV version
Warnings have been generated:
50A AKKU CHARGER ACS758_16F18446.gcb (93): Warning: Variable name PREV3 is used for a system bit on the current chip
50A AKKU CHARGER ACS758_16F18446.gcb (358): Warning: Comparison will fail if NOW3 is any value other than 0 or 1
PREV3 .. is an ADC register.bit and a 0 or 1.
The warning was present without #option explicit and you must have ignored.
So, resolve these issues first. Then, we can see look at the ADC issue.
Thank you very much for the correction! Thanks for the suggestions! I will delve into the correction. But yesterday after posting the post I found a problem!!! The problem was in the switching power supply. The output voltage of which is regulated by this circuit (from 8.6-16V). I noticed that when the charging process is underway, the voltage value is stable. I tried to power my circuit from another power source, and also only from the battery. The value became stable. Increasing the capacitance of the filter capacitor did not help, and a large capacitor does not work for me, because the goal of my project is to create a 20A (30A peak) miniature charger. Then I introduced changes to the circuit (see the circuit in the appendix) - diode D6, which helps power my circuit from the battery. And it helps! The values became stable! Maybe this program is useful to someone, as well as some idea from the circuit.
You can use the internal voltage reference (in the PIC) for its ADC, then it will not be sensitive to noise from the power supplying as if using VDD for reference (which is default).
Also if you put a resistor in series with the "supply" then maybe high frequency noise and transients may have a better change of being smoothed out by C4 (if that is the problem).
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
For the project I used the MW LOP series compact power supply. It is really a successful product from Mean Well. In my case 20A current (about 350W) and the power supply unit did not heat up, but only warm.
https://www.meanwell.eu/newsInfo.aspx?c=1&i=4144
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thanks Roger! For filtration, you need to use a choke, that would be the best solution. I don't want to try anymore. With this diode I solved my problem, and now I can measure the voltage of the battery without connecting the power supply! So the charger has an additional bonus :)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thank you Anobium for your great work. You didn't just fix the program, you wrote all the comments exactly according to the program's algorithm. Great. Fantastic.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Do you have some to write up a short overview? All the URL etc.
I will help you publish.
And, you should use INTOFF and INTON around the ReadAD()s. Why? You are accessing variables in the ISR that could change when the event happens. Example is amps_AN2 = scale(Lim_amps_AN2, 31, 184, 0, 500) if the ISR happened in between amps_AN2_H and amps_AN2 being set by the scale() function... you WILL have the incorrect value in the ISR for amps_AN2.
So, examine the code very carefully for places where variables ( specifically non Byte variables) accessed in the ISR could be partially updated.
Do share you end state code.
Last edit: Anobium 2025-03-30
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I inserted INTOFF and INTON. I hope I understood correctly. But this program code is so large that I can't write any more instructions because the chip memory is full. I wanted to create some more functions with the display, but that's no longer possible.
Thanks Anobium. Yes, I already did that, as you can see, my program code is divided into subroutines. Otherwise, the chip would be full. But the number of subroutines is also limited. "I will help you publish." Where do you think this could be published? My colleague will beautifully assemble the PCB board and PSU in the next few days in a box and then the device will be tested with real loads. I hope there will be no errors or problems.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anobium, good idea. I'd be happy to share the entire project, as there's still a gerber file, etc. But first, you need to put everything in a box and test it in real life, as there are still many nuances that need to be worked on - transistor/diode cooling radiators, ventilation, power supply operation at max load, etc. After the test period, I'll let you know when everything works perfectly, okay?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello! However, disturbances appear when the battery starts to be charged with a higher current - over 10A, that is, at the load of the switching power supply unit. Again, the voltage measurement ADC does not work as it should, the values jump. Then maybe these noises come from the transformer of the switching power supply unit? Roger suggested the idea: You can use the internal voltage reference (in the PIC) for its ADC, then it will not be sensitive to noise from the power supplying as if using VDD for reference (which is default). How to implement it? Maybe you can explain please?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
What if you run the PIC from a separate battery supply, during the high current measurements?
If it doesn't make a difference, I'm thinking that with the high currents involved combined with PWM It could be RF crap spreading in the grounds etc. Maybe better filtering is needed for the sampled voltage. It could be transients in the measured voltage showing its ugly face, corrupting the sampling (that you are catching some unwanted peaks, while looking for slower, mean values). PWM can be very problematic in this regard. Maybe you even need screening the sensitive parts.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Yess. Thanks! Everything is fine. I fixed the problem. I added C5 = 0.1uF , see the attached electrical diagram. There is C2 = 47uF, I left it. I first tried to change C2 to 0.1uF , then I don't get the result as good as this solution. So far, everything seems to work fine. There will be a test period with max loads.
Good.
Yes, electrolytics are often lossy at high frequencies due to series self induction. So often a polyester capacitor is used in parallel. Having it in a slightly separated stage as you have (as long as the potentiometer is not at its end) yields even better crap-rejection.
👍
1
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello! I have a problem with ADC - reading analog values specifically for this project. I don't understand why analog values change by themselves, within small limits, but they change and this is unacceptable. The problem is not in the electrical circuit (hardware) solution, but the analog value reading configuration is incorrect I think. Attached is a video where you can see that the values on the display jump around within large limits, but the battery voltage is actually measured, which is stable, and there are no disturbances (high frequency, etc.). GCB helps contents various solutions, but I have tried them, it does not work. There are still solutions out there, but I don't know which one to use. For example, #define LowSpeed 0 or Highspeed does not make any difference. Here in the text is a part of the program code for defining and reading the analog input. I used #chip 16f18446, 32 and analog resolution 8bit. I hope someone has had a similar experience and has an idea how to solve it.
The full program code is attached.
Here is the video, attached is the diagram.
Here is a revised program. Documented and with #option explicit added. I had to reformat and document so I could get to this far, then I added #option explicit, then I resolved those issues.
But, this analysis ( so far ) will show you why you should always use #option explicit.
I have not looked at the ADC issue as I think changing the register.bit PREV3 may have an impact on the ADC operations. As PREV3 is an ADC management register.bit
So, compiling the ERV version
PREV3 .. is an ADC register.bit and a 0 or 1.
The warning was present without #option explicit and you must have ignored.
So, resolve these issues first. Then, we can see look at the ADC issue.
Unused pins may introduce noise to the circuitry. Try pullups for all unused input/pins (if you haven't already).
Could be the unused pins - good point.
I just did a code review on my new version. I think this is caused by the interrupt firing during the ADC read.
Thank you very much for the correction! Thanks for the suggestions! I will delve into the correction. But yesterday after posting the post I found a problem!!! The problem was in the switching power supply. The output voltage of which is regulated by this circuit (from 8.6-16V). I noticed that when the charging process is underway, the voltage value is stable. I tried to power my circuit from another power source, and also only from the battery. The value became stable. Increasing the capacitance of the filter capacitor did not help, and a large capacitor does not work for me, because the goal of my project is to create a 20A (30A peak) miniature charger. Then I introduced changes to the circuit (see the circuit in the appendix) - diode D6, which helps power my circuit from the battery. And it helps! The values became stable! Maybe this program is useful to someone, as well as some idea from the circuit.
You can use the internal voltage reference (in the PIC) for its ADC, then it will not be sensitive to noise from the power supplying as if using VDD for reference (which is default).
Also if you put a resistor in series with the "supply" then maybe high frequency noise and transients may have a better change of being smoothed out by C4 (if that is the problem).
For the project I used the MW LOP series compact power supply. It is really a successful product from Mean Well. In my case 20A current (about 350W) and the power supply unit did not heat up, but only warm.
https://www.meanwell.eu/newsInfo.aspx?c=1&i=4144Thanks Roger! For filtration, you need to use a choke, that would be the best solution. I don't want to try anymore. With this diode I solved my problem, and now I can measure the voltage of the battery without connecting the power supply! So the charger has an additional bonus :)
Thank you Anobium for your great work. You didn't just fix the program, you wrote all the comments exactly according to the program's algorithm. Great. Fantastic.
Do you have some to write up a short overview? All the URL etc.
I will help you publish.
And, you should use INTOFF and INTON around the ReadAD()s. Why? You are accessing variables in the ISR that could change when the event happens. Example is
amps_AN2 = scale(Lim_amps_AN2, 31, 184, 0, 500)if the ISR happened in betweenamps_AN2_Handamps_AN2being set by the scale() function... you WILL have the incorrect value in the ISR foramps_AN2.So, examine the code very carefully for places where variables ( specifically non Byte variables) accessed in the ISR could be partially updated.
Do share you end state code.
Last edit: Anobium 2025-03-30
I inserted INTOFF and INTON. I hope I understood correctly. But this program code is so large that I can't write any more instructions because the chip memory is full. I wanted to create some more functions with the display, but that's no longer possible.
'chip memory is full' you still have two pages left and plenty of RAM left.
What are you try to add? Got an example of a memory full issue?
Try cutting up the code in the main loop, and the code prior to the main loop.
Add the following then call it, as an example. Most of that main loop can be in a sub.
Thanks Anobium. Yes, I already did that, as you can see, my program code is divided into subroutines. Otherwise, the chip would be full. But the number of subroutines is also limited. "I will help you publish." Where do you think this could be published? My colleague will beautifully assemble the PCB board and PSU in the next few days in a box and then the device will be tested with real loads. I hope there will be no errors or problems.
I moved the splash screen. That should work.
You could move all the main loop to a sub and call that sub. This will permit the relocation of this new sub to another place in the page(s).
Elektormagazine are publishing many GCBASIC projects.
Anobium, good idea. I'd be happy to share the entire project, as there's still a gerber file, etc. But first, you need to put everything in a box and test it in real life, as there are still many nuances that need to be worked on - transistor/diode cooling radiators, ventilation, power supply operation at max load, etc. After the test period, I'll let you know when everything works perfectly, okay?
Hello! However, disturbances appear when the battery starts to be charged with a higher current - over 10A, that is, at the load of the switching power supply unit. Again, the voltage measurement ADC does not work as it should, the values jump. Then maybe these noises come from the transformer of the switching power supply unit? Roger suggested the idea: You can use the internal voltage reference (in the PIC) for its ADC, then it will not be sensitive to noise from the power supplying as if using VDD for reference (which is default). How to implement it? Maybe you can explain please?
I think I tested this a while ago, but I can't find the code I used.
Maybe you find this useful:
"To use a voltage reference source for ADC operation sett the AD_REF_SOURCE constant to your chosen source."
https://gcbasic.sourceforge.io/help/_analog_digital_conversion_overview.html
Example here:
https://gcbasic.sourceforge.io/help/_readad10.html
What if you run the PIC from a separate battery supply, during the high current measurements?
If it doesn't make a difference, I'm thinking that with the high currents involved combined with PWM It could be RF crap spreading in the grounds etc. Maybe better filtering is needed for the sampled voltage. It could be transients in the measured voltage showing its ugly face, corrupting the sampling (that you are catching some unwanted peaks, while looking for slower, mean values). PWM can be very problematic in this regard. Maybe you even need screening the sensitive parts.
Yess. Thanks! Everything is fine. I fixed the problem. I added C5 = 0.1uF , see the attached electrical diagram. There is C2 = 47uF, I left it. I first tried to change C2 to 0.1uF , then I don't get the result as good as this solution. So far, everything seems to work fine. There will be a test period with max loads.
Good.
Yes, electrolytics are often lossy at high frequencies due to series self induction. So often a polyester capacitor is used in parallel. Having it in a slightly separated stage as you have (as long as the potentiometer is not at its end) yields even better crap-rejection.