Comparator circuit thoughts....

[kevlar129bp]

Hello all,

It has been a long time since posting on this forum. I hope everyone is well. I am currently working (or thinking about working) on a circuit for one of my cnc machines. It currently exists in analog discrete form, but I thought a pic version would be well suited for an upgrade...but my thoughts are scattered with it! Hopefully one of you fine folks can lead me down the right path.

Description of machine: 3 axis cnc plasma cutting gantry

Description of circuit: torch height control

Circuit operation:

Torch voltage is scaled from 120vdc to 1.2vdc.
Torch voltage is then fed into a window comparator.
Window comparator threshold voltage is variable from .06vdc to 1.85vdc. (representing 60-185vdc)
Comparator has hysteresis of .01vdc + and - of the set threshold voltage. (representing a 2vdc "lock range")
As the torch voltage swings out of the lock range, either comparator goes high, causing the cnc software
to "servo" the torch back into range.
The circuit also has an enable input, provided by the plasma cutter itself.
It is simply a dry contact, "ok to move" signal which indicates the torch is cutting.


My thoughts for the pic are:

The cnc software allows direct commands to be sent from the serial port of the PC, via VB scripting.
Therefore:

1. Include all of the above features.
2. Add the option of changing the comparator threshold voltage through serial with say...

SERIN2 SerPin,84,["V",dec3 TargetVDC]
3. Add the option of changing the comparator hysteresis through serial with say...

SERIN2 SerPin,84,["H",dec1 TargetHyst]

Are these thoughts doable, and if so, can you fine folks lead me in the right direction?

Thanks to you all,
Chris

[mackrackit]

Here is a half baked thought.

Look at using an ADC input.
60 to 185. Scale this by 60 giving 1 to ~3 for the input. Set VREF for 4 volts. Now you have a 1 volt buffer either way for when the plate suddenly warps or something else happens.

Now the optimal voltage can be set as VAR SetPoint and the hysteresis can now be SetPoint +- what ever. Beyond the +- all stop or torch out or...

Have the analog at 10 bit giving 1024 steps in the 4 volt range I would think would allow for a nice smooth operation.

[kevlar129bp]

Thanks mackrackit,

That sounds like a great idea, albeit leads me to a ton more questions.
If I understand correctly, is this how it would go?....

If I divide the TorchVdcMax by 1024 ----> ( 200/1024=.195)

So, if I do something like this:


SERIN2 SerPin,84,["V",dec3 TargetVDC]
TargetADC=TargetVDC*.195 'EX:TargetADC=120*.195 ----> 615.3846

The "615.3846" correlates to the ADC target count? ----> (200/1024)*615 = 120.117

I'll stop right there for the moment...I'm way confused at this point!
I think for the moment, I'de much rather focus on commanding the torch voltage rather than both the torch voltage and hysteresis. Maybe it will calm my nerves a bit:o.
Can you correct my mindset to this thing, or shoot me an easier way to comprehend it? Thanks a ton so far!

Chris

[mackrackit]

Normally a PIC will only handle a maximum voltage of 5 volts. When you said


# Torch voltage is scaled from 120vdc to 1.2vdc.
I thought you were doing this with a voltage divider to bring the 120 down to 1.2, stating that the divider you planned on using was 100 to 1.
So...

I was thinking that being the expected voltages from the torch are 60 to 185 a voltage divider that would bring the 60 volts down to 1 volt for the low end and the 185 high end volts would then be ~3 volts. That is where the divide (scale) by 60 came in.

Many PICs have a 10 bit ADC. Even with a 10bit if you want you only have to use the first 8.
The difference is the resolution. There is also a voltage reference available when using the ADC. VDD can be used or some other voltage less than VDD. Some have a low end VREF but I figure on this one zero volts will be fine.

This all means that if an 8 bit (256 steps)ADC setup is used with VREF set for 4 volts, each step represents ~0.015 volts. In real life with the torch voltage being divided by 60, 0.015 would equal 0.94 volts. The torch voltage would have to change 0.94 volts before the PIC sees the change.

A 10 bit setup works the same but with 1024 steps. Now the PIC can see a 0.23 volt change on the torch.

This could be tightened up by lowering the VREF and/or scaling the 60 to maybe 0.5 volts, but I figure a little cushion would be good on both ends. I may have more than is needed...

Now using all of this. Say you want the torch voltage to be 120. With the 4 volt VREF and the voltage dividing, 4 volts at the ADC is equal to 240 at the torch, so we want to see 2 at the ADC to be good. The ADC will give a value of 512. If the ADC is + or - 512 the torch height will need adjusted.

Let me know how much I confused things :)

[kevlar129bp]

mackrackit,

No more confusion than before :). Actually, the way you described the voltage scaling makes more sense to me now. As far as the torch voltage is concerned, maybe I added confusion from the beginning by not stating what would be a good baseline hysteresis to go with. After trying to get my head around the scaling, I started looking at the numbers more, and in doing that, I think a resolution of 1 volt change at the torch would be more than ok. That being said, that's where I figured using a torch value ranging from 0 - 256 Vdc...

8 bit ADC
4 Vdc Vref
Vdc per ADC count : 4 / 256 = .015625
Torch Vdc scale : 1:64
Torch Vdc per ADC count : .015625 * 64 = 1 Vdc

Ex: Torch Vdc = 115


ADC counts * Vdc per ADC count = ADC Vdc Now



115 * .015625 = 1.796875 Vdc


1.796875 * Torch Vdc scale = Torch Vdc



1.796875 * 64 = 115 Torch Vdc

So... for every 1 Vdc change at the torch, there will be a 1 count change at the ADC... correct?

If this is the case, coding for this arrangment should be way easier?
Am I making this easier or harder on myself? I think it's sinking in slowly but surely. Thanks for assisting me on my brain fryer!

Chris

[mackrackit]

I think you got it.

[kevlar129bp]

Thanks mackrackit,

I'll hack away at a bit of the code and see what I get... Will you be around for some assistance in that arena, when I get to that point?

Thank you again... TONS!

Chris

[mackrackit]

We are always here....

Start the testing with a POT to simulated the torch.

Which PIC are you going to use?

[kevlar129bp]

mackrackit,

In looking at my options, I think the 16F88 is going to do the trick. I'de like to plan ahead to do something like this on the front of the cnc pc. I dunno, let me know what you think... And again, thank you so much for your help. I'm working through the code as we speak :confused:

Chris

[mackrackit]

At this point I would say whatever you have laying around. The 88 is good for starters. Might run out of memory or pins later but might not.

But how are you writing code if you have not picked a PIC?

[kevlar129bp]

I have opted to write the code for the '88. Hopefully I don't run out of space :). I'll keep you posted on that one.

[kevlar129bp]

Major stupid question...
If I declare something like Number var BYTE...
Then I do: Number = 1.01567
Does that get truncated to 1, or something else, or rejected altogether?

Wow, I feel like a moron for asking! Sorry.:(

Chris

[mackrackit]

No problem with the questions.....
Unless you get into floating point routines we are stuck with integer math.
But if You need to be more accurate there are ways to do so. Basically mutiply by a power of ten and ...... Well you get the idea.

So it would become 1.

The PBP manual has a good section about the math.

[kevlar129bp]

Cool... I guess I'll put this mess up for you to take a look at. Mark it up at will. Thanks a ton....and here goes :eek:


' Name : THC.BAS
' Author : Me
' Notice : Copyright (c) 2009 None
' : All Rights Reserved
' Date : 10/20/2009
' Version : 1.0
' Notes :
'
'
include "MODEDEFS.BAS"
'
'\\\\ADC DEFINES////
DEFINE ADC_BITS 8 'Number of bits in ADCIN result
DEFINE ADC_CLOCK 3 'ADC clock source (rc = 3)
DEFINE ADC_SAMPLEUS 50 'ADC sampling time in microseconds
CVRCON = 143 'Reference on, not tied to pin, set to 3.59vdc
ANSEL = 64 'Enable analog select, select channel 6
CMCON = 7
'
'\\\\SERIAL PORT SETUP////
DEFINE HSER_RCSTA 90h 'Enable serial port
define HSER_TXSTA 24h 'Enable transmit
define HSER_SPBRG 25 'Set baudrate to 9600
DEFINE HSER_CLOERR 1 'Automatic clear overrun error
INTCON = 64 'Enable peripheral interrupts
PIE1 = 32 'Enable AUSART receive interrupt
'
'\\\\PINS SETUP////
TRISB.0 = 0 'Up LED Output
TRISB.1 = 1 'Enable THC Input
TRISB.3 = 0 'LCD Pin
TRISB.4 = 0 'Dn LED Output
TRISB.6 = 0 'THC Enabled LED Output
TRISB.7 = 1 'Scaled Torch Voltage Analog Input
'
'\\\\VARIABLE DECLARATIONS////
TargetVDC var byte '000-256 Allowed
TargetVDCDiv var byte
TorchPin var PORTB.0
TorchVoltsNow var byte
TorchDown var PORTB.4
TorchUp var PORTB.0
EnableIn var PORTB.1
TorchVoltsFull var byte
'
'
'
'\\\\POWER UP STARTS HERE////
Pause 500 'Wait for LCD to startup
TargetVDC = 120 'Start up target volt value
'
'
'\\\\LOOK FOR SERIAL DATA ON AUSART RX PIN////
on interrupt goto NewSerialData
'
NewSerialData:
hserin [wait ("V"),DEC3 TargetVDC]
TargetVDCDiv = TargetVDC / 256
goto mainloop
'
'\\\\START MAIN PROGRAM LOOP HERE////
Mainloop:
TargetVDCDiv = TargetVDC / 256
IF Enablein = 1 then
ADCIN 6,TorchVoltsNow
if torchvoltsnow > TargetVDCDiv then
high torchup 'Turn off TorchUp Pin
low torchdown 'Turn on TorchDown Pin
else
high Torchdown 'Turn off TorchDown Pin
Low torchup 'Turn on TorchUp Pin
endif
TorchVoltsFull = TorchVoltsNow * 256
'\\\\SEND STATUS OUT TO LCD////
LCDOUT $FE, 1
LCDOUT $FE, 2
LCDOUT "PLASMA VDC : "TorchVoltsfull
PAUSE 50
LCDOUT $FE, $C0
LCDOUT "TARGET VDC : "TARGETVDC
PAUSE 50
HSEROUT TorchVoltsFull
else
LCDOUT $FE, 1
LCDOUT $FE, 2
LCDOUT "---CONTROL IS---"
PAUSE 50
LCDOUT $FE, $C0
LCDOUT "----DISABLED----"
PAUSE 50
ENDIF
Goto mainloop

Don't be cruel....newbie with ambition!

Thanks a ton!

Chris

[kevlar129bp]

Typo in:
TorchPin var PORTB.0
Change to
TorchPin var PORTB.7
;)

[mackrackit]

This is for comparators


CVRCON = 143

you want


ADCON1 = %00100000 'Pin #2 is voltage reference


The LCD lines and HSEROUT line should be something like


LCDOUT "TARGET VDC ", #TARGETVDC
HSEROUT [#TorchVoltsFull]


The divide by multiply by 256 should not be there. The 8bit ADC goes from 0 to 255. As we talked before withe the voltage divider feeding the ADC and using using the reference voltage. Well it worked pretty neat for you. The torch voltage will be the same "number" as the ADC value.

Other than that it should do something :)

Now I will toss something else at you.
Think how fast the torch move up and down. Those few line of code before the ADC is read could mean trouble. You may want to increment the movement.