DIY Digital Caliper

[MitkoDyakov]

Hey!

I am trying to design a digital caliper. So far I have got my hands on a couple of cheap calipers and a few papers and patents. I have create footprints for both the scale and the sensor board and seem to be working quite well. You can follow the build log (link in the bottom) for more details, also the footprints are available if you want to check them out too. 

https://hackaday.io/project/194778-diy-digital-caliper

The difficulties I am having now are connected with conditioning of the signal I am receiving. 

I used this schema to get a reading with my scope



The result is as follows:



Now from what I gather I need to amplify or at least buffer this signal, then transform it to a sine wave and then detect the zero crossing. Compare the phase shift to a base sine frequency to determine the direction and displacement.

1. Are my assumption of how the thing works correct at the first place?
2. What should I be looking for when designing the amplification circuit?
3. How do I choose R/C values for creating my sine? 
4. How do I detect zero crossing when my sine wave does not go negative?

Cheers,
M

[ali_asadzadeh]

Just following the project, I think you can either choose hardware zero cross detection, say you would make an op-amp comparator with the one side of op-amp set to mid VCC scale, which does not works perfectly or you could choose the second option which is to apply DSP algorithms to Detect Zero crossing, which needs reading the wave form with ADC and apply some math formulas to it, and extract zero crossing.

[tszaboo]

For a quadrature encoder, which things are based on, you need two signals. So I'm going to assume you are only showing one of the signals. I think some of the issue comes from the 22 MOhm value that you have on the board. A smaller value would probably make the signal cleaner.
Then about how to amplify the signal: Comparators. You feed the output of the opamp into a comparator with hysteresis (feedback resistor from  the output to the DC input).

[MitkoDyakov]

For a quadrature encoder, which things are based on, you need two signals. So I'm going to assume you are only showing one of the signals. I think some of the issue comes from the 22 MOhm value that you have on the board. A smaller value would probably make the signal cleaner.
Then about how to amplify the signal: Comparators. You feed the output of the opamp into a comparator with hysteresis (feedback resistor from  the output to the DC input).

The dotted line on the first row is the resultant sine wave if sense pad is located on top of all excitation pads. I call this the base frequency, mind you that it does not exists physically (you can't measure it with a scope). 

And depending which way you shift the caliper head the resulting sine wave you measure on the sense pad get phase shifted to the right or to the left.

So I don't have two signal in the sense that a quadrature encoder has, but rather one signal to measure to determine the phase shift.

[moffy]

You are talking about a sine wave but it looks like you are using a square wave for excitation is that the case?

[MitkoDyakov]

You are talking about a sine wave but it looks like you are using a square wave for excitation is that the case?

This is correct. The signals bellow are one of four pair of signal generated. Each pear is shifter 90 deg from the one before it. And inside the pair each signal is 180 degree shifted.



You can see that it is PWM with variable duty cycle. In my mind this is like digitizing the sine wave and encode the duty cycle corresponding to each sample point.


*in my case it does not go negative just goes to 0

[moffy]

That makes some sense, I think personally, that I would filter the PWM signal to sinusoidal before applying it to the sensor, that way you are only adding sine waves of a single frequency and you have a direct comparison between excitation and output without the need of phase shifting filters. You might need a  buffer amp, that's a small disadvantage but it would make problem solving much easier to visualise, even if it is only for development purposes.

[ali_asadzadeh]

MitkoDyakov would you please share kicad project files including schematic and PCB into the project repo, also for easier access please mention the GitHub repo in here too.

This is a working Zero crossing code that I have implemented before on some industrial equipment, it uses 64 samples per sine wave cycle to calculate the Zero crossing

Code: [Select]

#define ADC_samples 64

//we need two cycles to calculate the Frequency, ok!
short oldSamples[128];

const int BLOCK_SIZE =ADC_samples;
const int NUM_TAPS =32;
arm_fir_instance_q15 S;
q15_t firStateF32[BLOCK_SIZE + NUM_TAPS - 1];

q15_t firCoeffs32[NUM_TAPS] = {
       -1,     -2,      8,     39,     83,     97,     14,   -213,   -535,
     -762,   -596,    232,   1778,   3773,   5659,   6810,   6810,   5659,
     3773,   1778,    232,   -596,   -762,   -535,   -213,     14,     97,
       83,     39,      8,     -2,     -1
};


void init_myFilter(void){
/* Call FIR init function to initialize the instance structure. */
  arm_fir_init_q15(&S, NUM_TAPS, firCoeffs32, firStateF32, BLOCK_SIZE);
}

void dsp_zero_cross(short* input,int size,float* zero_1st,float* zero_2st, float* zero_3st)
{
int i;
float first_zero = 0,second_zero = 0,third_zero = 0;
    int index[4] = {0};
    int index_index = 0;
int x1 = 0;
int x2 = 0;
int x3 = 0;
short temp[ADC_samples];
//first copy from the last portion to the first portion
for(i=64;i<128;i++)
oldSamples[i-64] = oldSamples[i];
//now copy the input to the last portion
for(i=64;i<128;i++)
oldSamples[i]= input[i-64];


arm_fir_fast_q15(&S,oldSamples,temp,BLOCK_SIZE);
arm_fir_fast_q15(&S,oldSamples+64,temp,BLOCK_SIZE);


for (i = 1; i < size; i++)
  {
if((temp[i]<=0 && temp[i-1]>0) || (temp[i]>=0 && temp[i-1]<0))
    {
      index[index_index] = i;
      index_index++;
if(index_index == 4)
{
break;
}
    }
  }

  if(index_index == 0)
{
*zero_2st=-1;
    *zero_1st=-1;
*zero_3st=-1;
return;
}
  else if(index_index == 1)
{
*zero_2st=-1;
*zero_1st=-1;
*zero_3st=-1;
return;
}
  else if (index_index == 2)
  {
x1 = index[0];
x2 = index[1];
if((temp[x1 - 1] - temp[x1]) == 0 || (temp[x1 + 1] - temp[x1]) == 0 || (temp[x2 - 1] - temp[x2]) == 0  || (temp[x2 + 1] - temp[x2]) == 0 || x2 == (size - 1) || x1 == (size - 1))
{
*zero_2st=-1;
*zero_1st=-1;
*zero_3st=-1;
            return;
}
else
{
first_zero = ((float)(temp[x1] * temp[x1 + 1])/((float)(temp[x1 - 1] - temp[x1])*(float)(temp[x1 - 1] - temp[x1 + 1]))) * (x1 - 1);
first_zero += ((float)(temp[x1 - 1] * temp[x1 + 1])/((float)(temp[x1] - temp[x1 - 1])*(float)(temp[x1] - temp[x1+ 1]))) * (x1);
first_zero += ((float)(temp[x1 - 1] * temp[x1])/((float)(temp[x1 + 1] - temp[x1 - 1])*(float)(temp[x1 + 1] - temp[x1]))) * (x1 + 1);
////////////////////
second_zero = ((float)(temp[x2] * temp[x2 + 1])/((float)(temp[x2 - 1] - temp[x2])*(float)(temp[x2 - 1] - temp[x2 + 1]))) * (x2 - 1);
second_zero += ((float)(temp[x2 - 1] * temp[x2 + 1])/((float)(temp[x2] - temp[x2- 1])*(float)(temp[x2] - temp[x2 + 1]))) * (x2);
second_zero += ((float)(temp[x2 - 1] * temp[x2])/((float)(temp[x2 + 1] - temp[x2 - 1])*(float)(temp[x2 + 1] - temp[x2]))) * (x2 + 1);
*zero_1st = first_zero;
*zero_2st = second_zero;
*zero_3st=-1;
return;
}
  }
else if (index_index == 3)
{
x1 = index[0];
x2 = index[1];
x3 = index[2];
if((temp[x1 - 1] - temp[x1]) == 0 || (temp[x1 + 1] - temp[x1]) == 0 || (temp[x2 - 1] - temp[x2]) == 0  || (temp[x2 + 1] - temp[x2]) == 0|| (temp[x3 - 1] - temp[x3]) == 0  || (temp[x3 + 1] - temp[x3]) == 0 || x3 == (size - 1) || x2 == (size - 1) || x1 == (size - 1))
{
*zero_2st=-1;
*zero_1st=-1;
*zero_3st=-1;
            return;
}
else
{
first_zero = ((float)(temp[x1] * temp[x1 + 1])/((float)(temp[x1 - 1] - temp[x1])*(float)(temp[x1 - 1] - temp[x1 + 1]))) * (x1 - 1);
first_zero += ((float)(temp[x1 - 1] * temp[x1 + 1])/((float)(temp[x1] - temp[x1 - 1])*(float)(temp[x1] - temp[x1+ 1]))) * (x1);
first_zero += ((float)(temp[x1 - 1] * temp[x1])/((float)(temp[x1 + 1] - temp[x1 - 1])*(float)(temp[x1 + 1] - temp[x1]))) * (x1 + 1);
////////////////////
second_zero = ((float)(temp[x2] * temp[x2 + 1])/((float)(temp[x2 - 1] - temp[x2])*(float)(temp[x2 - 1] - temp[x2 + 1]))) * (x2 - 1);
second_zero += ((float)(temp[x2 - 1] * temp[x2 + 1])/((float)(temp[x2] - temp[x2- 1])*(float)(temp[x2] - temp[x2 + 1]))) * (x2);
second_zero += ((float)(temp[x2 - 1] * temp[x2])/((float)(temp[x2 + 1] - temp[x2 - 1])*(float)(temp[x2 + 1] - temp[x2]))) * (x2 + 1);
///////////////////
third_zero = ((float)(temp[x3] * temp[x3 + 1])/((float)(temp[x3 - 1] - temp[x3])*(float)(temp[x3 - 1] - temp[x3 + 1]))) * (x3 - 1);
third_zero += ((float)(temp[x3 - 1] * temp[x3 + 1])/((float)(temp[x3] - temp[x3 - 1])*(float)(temp[x3] - temp[x3+ 1]))) * (x3);
third_zero += ((float)(temp[x3 - 1] * temp[x3])/((float)(temp[x3 + 1] - temp[x3 - 1])*(float)(temp[x3 + 1] - temp[x3]))) * (x3 + 1);
*zero_1st = first_zero;
*zero_2st = second_zero;
*zero_3st= third_zero;
}
}

}



[MitkoDyakov]

You can find the all I have here: https://github.com/MitkoDyakov/BluePillCaliper/

Documents - are all written I could find on the subject, patents, paper, work done by others.
Hardware - the footprints for the t-scale and sensor and also gerbers and schematics
Software - I need to update that but, for the software I have written just the excitation of the pads and the display.

   

[ali_asadzadeh]

Hardware - the footprints for the t-scale and sensor and also gerbers and schematics

Thanks for sharing, But it would be nicer if you could share the Kickad files for schematics and PCB, since We can modify it easier.

[MitkoDyakov]

Hardware - the footprints for the t-scale and sensor and also gerbers and schematics

Thanks for sharing, But it would be nicer if you could share the Kickad files for schematics and PCB, since We can modify it easier.

You can find it in BluePillCaliper/Hardware /Footprints - KICA

[ali_asadzadeh]

You can find it in BluePillCaliper/Hardware /Footprints - KICA

There is only footprints, Kicad project files extensions is like this .kicad_pcb .kicad_pro .sch unless, I'm missing something :'(

[MitkoDyakov]

here there are

https://github.com/MitkoDyakov/BluePillCaliper/tree/main/Hardware/Footprints%20-%20KICAD/CaliperPCB
https://github.com/MitkoDyakov/BluePillCaliper/tree/main/Hardware/Footprints%20-%20KICAD/CaliperBlade

[ali_asadzadeh]

here there are

https://github.com/MitkoDyakov/BluePillCaliper/tree/main/Hardware/Footprints%20-%20KICAD/CaliperPCB
https://github.com/MitkoDyakov/BluePillCaliper/tree/main/Hardware/Footprints%20-%20KICAD/CaliperBlade

Thanks for sharing, is there any updates regarding zero crossing?

[MitkoDyakov]

I was looking at this schema on electronics stack exchange. Which seems to be what I need given the fact that I would power it from 3V



And I build a prototype... a few prototypes



The ones I designed were tragic. I have questions for the design from stack exchange and they are the following:

1. How do I choose values for C1 and C2
2. I understand why R1 and R2 are there, but I don't get why some people choose Mohm values for them but here 10k is fine
3. I did not had MCP6021, but I did had MCP6V11. Is there anything special about MCP6021?

And for people who will ask me about pictures from the scope. The "thing" had nothing resembling the original signal, looks random noise at 1.5V 50mV peak to peak.

[tooki]

I was looking at this schema on electronics stack exchange. Which seems to be what I need given the fact that I would power it from 3V



And I build a prototype... a few prototypes



The ones I designed were tragic. I have questions for the design from stack exchange and they are the following:

1. How do I choose values for C1 and C2
2. I understand why R1 and R2 are there, but I don't get why some people choose Mohm values for them but here 10k is fine
3. I did not had MCP6021, but I did had MCP6V11. Is there anything special about MCP6021?

And for people who will ask me about pictures from the scope. The "thing" had nothing resembling the original signal, looks random noise at 1.5V 50mV peak to peak.

Please attach your images as attachments here, rather than embedding external URLs that don’t work.

[notsob]

How about these analogue ones

https://ceeshop.com.au/products/limted-banana-calipers

[MitkoDyakov]

How about these analogue ones

https://ceeshop.com.au/products/limted-banana-calipers

the hell with my project I am jumping ship to the banana calipers

[MitkoDyakov]

I have updated the schematics. You can find my solution on the git hub repo and build logs are on the hackaday page - link in my first post.

[ali_asadzadeh]

I have updated the schematics. You can find my solution on the git hub repo and build logs are on the hackaday page - link in my first post.

what are the updates? tell us the good news
There is no software in the repo, is there any plan to add them too?

[MitkoDyakov]

Hey Ali,

I have added MCP6S21 and a fixed ref voltage using ADR510 and AD8603 wombo combo. This way I can amplify the signal and offset it to the middle of the ADC range. And also the amp gain is configurable with SPI which is great.

There is no code because I just got the signal conditioning acceptable. I have send some boards to contributors that will help with this part.

Stay tuned, to get the update you can like the hackaday page. I will post here from time to time also.

Cheers!

[ali_asadzadeh]

Stay tuned, to get the update you can like the hackaday page. I will post here from time to time also.

Thanks for the update, I think you should definitely add code and kickad project files to the repo (we have only Gerber files for now ), so we can build and test and give feedback on every level, this way the project moves faster.

[MitkoDyakov]

I can't figure something. Lets say we have this setup. Three plates the red one is connected to a MCU output and it is driven HIGH or LOW. The grey plate spans bellow the red and the blue plates but it is not connected to anything and finally the blue plate is connected to the ADC of the MCU.



When we set the output of the MCU to HIGH we get this on the scope (which is kind of what we expect, but why does it go down):



But when we go and set the output to LOW we get this:



Now why does it dip bellow 0 volts?

To take the screenshots I also connected the scope ground to the MCU ground. The measurements are taken from the "input" lead.

[moffy]

It is a low value capacitor coupling the output of the MCU to the input of the ADC, which has resistance to 0V/gnd. The capacitor removes the DC and the resistor discharges the cap, a C-R circuit hence what you see.

[jwet]

I've done some playing with this kind of stuff over the years- I developed the DRO system for Sherline after looking at a bunch of low cost solutions.  Here are a couple of inputs that I haven't seen in the discussion so far.

Book- "Capacitive Sensors" by Baxter, IEEE press.  Only text I've seen that covers these in detail.  A small part of the book- engineering library?

There was a project in "Home Shop Machinist"- Village Press many years ago on this topic.  The guy ended up CNC engraving the spars himself.  He couldn't get 1 meter length and the PCB accuracy wasn't great.  The builder made the spars but used off the shelf readout heads that he could buy cheap that had something like SPI outputs.  I've never seen the asics in these things for sale anywhere.  I googled this article and it didn't pop right up- try Village Press.

Good luck.