Help with pressure sensor.

[DavyJones]

Well I have to be about the dumbest person and why this eluding me on how to correct this is beyond me. Anyway I am working with a 16f877 and I've got a pressure sensor thats range is .5vdc to 4.5vdc and I am trying to get the full range of the AD but I am stuck. It works perfectly when I hook it up my AD output reads .47v which is perfect since there is some degree of percission. I don't know why they don't make these to read 0v to whatever but they don't So my quesiton is how to programatically drop that first .5v and still take advantage of the entire 1-1024 range of the AD converter or is there a better way?


Thank you.
David

[mackrackit]

Maybe this will help?
http://www.picbasic.co.uk/forum/show...16&postcount=2

[DavyJones]

Maybe this will help?
http://www.picbasic.co.uk/forum/show...16&postcount=2

Mack, That does help allot thank you. I am an amateur at best and not a good one, but I try. I am starting to understand the math. I do read the datasheets, I don't always understand them but I read them and try none the less. The part I am having difficulty is with the settings of ADCON0 and ADCON1 and setting of VREF. I am getting readings and it seems that proportinatly it goes up as the PSI goes up. I've incorporated your suggested math into my program thanks. The problem I think I am having is the VREF is and one or more of the bits I have set in ADCON? is not correct. At a dead stop with 0 psi I get a reading and that most likely is proportinate to the
.47 vdc the voltmeter is telling me the sensor is putting out and being presented to RA0. Ok now the dumb questions: Do I need to actually present a +V and -V on RA3 and RA2 or as I understand it setting ADCON1.0 through .3 to 0000 sets it up to use the Vdd and Vss the pic is using itself?
I thought the range would be 0 to 1024 not 0 to 256 why did I think that?
I thank you for your help.
David

[mackrackit]

Lets go with small steps and we will come back to the 10 bit resolution~1024.

And lets not worry about the VREF for now. That will take extra hardware to setup the reference voltage. When we get there all you will need for your application is V+.

Read through these two links, Bruce makes very nice tutorials.
This one is using your chip but does not use the ADCIN command. I like this way and it will help you to understand how the ADC works.
http://rentron.com/PICX2.htm

This one is using ADCIN and a different PIC, but will help with your understanding.
http://rentron.com/serial.htm

Get some of the above working and post the code you have. Then we will go for the 10 bit stuff.

[DavyJones]

Mack, I acutally started my project with this example from Bruce good call. I did not completely understand it but I did in fact hook up a 5k pot to my board to see how the ADC worked.

This time I pretty much did this line for line with my pressure sensor attached to see how it would react. I am getting a reading. My volt meter is reading .47vdc from the sensor. The serout is bouncing back and fourth between 24 and 25. I would assume that the way Bruce's program is set its 0-255 analog output I would then assume that if my sensor was .5 to 4.5 then the full range of the output would be 4/255 or .0156 volts per increment of the 255 this 24*.0156 = .3744. I think my math is correct there from your previous example but since I can see/know that my sensor is putting out .47 I either did something slighlty wrong or... I've got a mistake in my program.

P.S. Thanks for the help.
David

INCLUDE "modedefs.bas"
' Hardware specific settings
@ DEVICE pic16F877, XT_OSC ' System Clock Options
@ DEVICE pic16F877, WDT_ON ' Watchdog Timer
@ DEVICE pic16F877, PWRT_ON ' Power-On Timer
@ DEVICE pic16F877, BOD_ON ' Brown-Out Detect
@ DEVICE pic16F877, LVP_ON ' Low-Voltage Programming
@ DEVICE pic16F877, CPD_OFF ' Data Memory Code Protect
@ DEVICE pic16F877, WRT_ON ' Flash Memory Word Enable
TRISA = %00000001
' Allocate variables
x var byte
SO VAR PORTC.6
BAUD CON 32
DEFINE OSC 10
ADCON1 = 0 ' Set PortA 0 to A/D inputs
Pause 100 ' Wait for LCD to start
SerOut2 SO,baud,[12,"Hello World AD Test",10,13]
pause 1000
Serout2 so,baud,[12,27,"[1;1H","Reading: "]
Goto mainloop ' Skip subroutines
' Subroutine to read a/d converter
getad:
Pauseus 50 ' Wait for channel to setup
ADCON0.2 = 1 ' Start conversion
Pauseus 50 ' Wait for conversion
Return
' Subroutine to get pot x value
getx:
ADCON0 = $41 ' Set A/D to Fosc/8, Channel 0, On
Gosub getad
x = ADRESH
Return
mainloop:
Gosub getx ' Get x value
Serout2 so,baud,[27,"[1;11H", dec3 x]
Pause 100 ' Do it about 10 times a second
Goto mainloop ' Do it forever
End

[Melanie]

Your reading is 24 or 25...

Try calculating 24/255*5v = 0.470v...

That's equivallent to ADCVoltage=ADCCount/ADCMaxScale*VRef

Close enough for you?

BTW... you can't use the above math in PBP as you have to scale it up a little to maintain INTEGER integrity, but I'm sure you can figure how to do that.

And in answer to a previous point you made right at the start, the reason 0-5v Sensors don't actually go 0-5v, is that by going say 0.5v-4.5v you can tell if they go open or short circuit or otherwise go defective as their output will be out of range by being too low or too high. If it actually started at 0v and somebody pulled the plug on the Sensor on your Hospitals Life Support machine, you'd never know - and just might wreck your day.

[DavyJones]

Your reading is 24 or 25...

Try calculating 24/255*5v = 0.470v...

Close enough for you?

BTW... you can't use the above math in PBP as you have to scale it up a little to maintain INTEGER integrity, but I'm sure you can figure how to do that.

Yes, Not sure why I am getting this math all assbackwards. I think I've spent to many years as a mumps programmer. I think I get this part so far. My range of .5 to 4.5 isn't relivant... yet. the 24/25 reading is going off a REF of 5v which as you pointed out makes complete sense to me. I keep frogetting to take that into account but I am learining and willing to learn. I jsut can't seem to get past that part I think I need to programatically figure out how to factor in my range is only going to be .5 to 4.5 or 0 to 4 or I must have to give the pic a VREF. If you can point me in the right direction I would appreciate it.

With in mind and I believe I get it so far I will post the program that I have been working on with some success. I think I am close.
Here is my program and my logic/math

So if my scale is 0-1024 with an internal ref of 5v and using an earlier example form Mack I would conclude give my sensor range of 0-150psi with an proportinate output voltage of .5 to 4.5v then...

(150-0) / (4.5-.5) = 37.5 psi per volt
150/1024 at a 5v ref range is .1464 psi per reading

That part I understand completely. The next part is were I am lost.
Converting the .5 -> 4.5v with an offset somehow so that works out to
0 -> 4 then calculating out that voltage to PSI which would be easy would it not be my vdc * 37.5?

0-1024 with a 5v ref gives me .0048 volts per incrment or 204.8 increments per volt. If I leave the 5v ref as it is then I think my offset would be -102.4 so to get the PSI reading I would take:

Given ADVal = reading from the ad result

( ADVal - 102.4 ) * .1464

Let's say I know the PSI is 75 then the calc should work... Half of 1024 would be 512 so ( 512 - 102.4 ) * .1464 = 59.96
but... my reading would not be 512 for 75 psi it would be 512+102.4 right? arghhhhhhhh I'l be back I have to sweep up the last of my hair!

Thanks for your help. I am trying.
P.S. It's a beautiful day here in Pennsylvania

David

INCLUDE "modedefs.bas"
' Hardware specific settings
@ DEVICE pic16F877, XT_OSC ' System Clock Options
@ DEVICE pic16F877, WDT_ON ' Watchdog Timer
@ DEVICE pic16F877, PWRT_ON ' Power-On Timer
@ DEVICE pic16F877, BOD_ON ' Brown-Out Detect
@ DEVICE pic16F877, LVP_ON ' Low-Voltage Programming
@ DEVICE pic16F877, CPD_OFF ' Data Memory Code Protect
@ DEVICE pic16F877, WRT_ON ' Flash Memory Word Enable
' ADC definitions 16F877 - (8) 10bit channels
'*****************************
DEFINE ADC_BITS 10 ' (8) 10 Bit channels
DEFINE ADC_CLOCK 3 ' tell the us per osc clock tick?
DEFINE ADC_SAMPLEUS 50 ' how long to sample the pin
DEFINE OSC 10
DEFINE LOADER_USED 1
'INIT ADC
ADCON1 = %10001110 ' Just one analog input needed see page 112 of 16f877 spec
ADCON0 = %11000001 ' Analog converter Fosc/32, ADC module is ON
'ADCON0.7 = 1 I DID THAT ABOVE RIGHT?

'Varibles

SO VAR PORTC.6
ADval VAR word ' Storage for A/D result
PRESSURE VAR word
PRESS VAR word
ADavg VAR WORD[2]
READING var byte
Value VAR WORD

'CONSTANTS
AvgCount CON 64 ' = Number of samples to average
FAspread CON 25 ' = Fast Average threshold +/-
BAUD CON 32
'**************************************
'****START OF PROGRAM*****
'**************************************
BEGIN:
SerOut2 SO,baud,[12,"Hello World AD Test",10,13]
pause 2000
Serout2 so,baud,[12,27,"[1;1H","Pressure: "]
MAINLOOP:
ADCIN 0,adval ' Read A/D channel 0 to adval variable
Serout2 so,baud,[27,"[3;11H",DEC5 adval]
'average out press first before calcuating true pressure
READING = 1
value = adval
gosub average
press = value
Serout2 so,baud,[27,"[2;11H",DEC5 PRESS]

'PRESSURE = ( 58 * PRESS - 58 ) / 100
pressure = ( 1464 * press - 1464 ) / 10
value = pressure
READING = 2
GOSUB AVERAGE
PRESSURE = VALUE
Serout2 so,baud,[27,"[1;11H",DEC5 PRESSURE]
Pause 200 ' Wait period between updates
GOTO MAINLOOP
END

' -=-=-=-=-=-= Average Analog values -=-=-=-=-=-=-=-=-=-=
' compliments of Darryl Taylor
AVERAGE:
IF Value = ADavg[READING] Then NoChange
IF ABS (Value - ADavg[reading]) > FAspread OR Value < AvgCount Then FastAvg
IF ABS (Value - ADavg[reading]) < AvgCount Then RealClose
ADavg[reading] = ADavg[reading] - (ADavg[reading]/AvgCount)
ADavg[reading] = ADavg[reading] + (Value/AvgCount)
GoTo AVGok
FastAvg:
ADavg[reading] = Value
GoTo AVGok
RealClose:
ADavg[reading] = ADavg[reading] - (ADavg[reading]/(AvgCount/4))
ADavg[reading] = ADavg[reading] + (Value/(AvgCount/4))
AVGok:
Value = ADavg[reading] ' Put Average back into Value
NoChange:
Return

[mackrackit]

Do not mix up the ADCIN defines and the bit banging from Bruce's example.
Some more reading..
http://www.picbasic.co.uk/forum/showthread.php?t=2188

And a snippet for 10 bit banging..

ADCON1.7 = 1

	gosub getAD



	press.highbyte = ADRESH

	press.lowbyte = ADRESL

	return

Do a search for 10 bit ADC also..
http://www.picbasic.co.uk/forum/showthread.php?t=4751

[DavyJones]

Do not mix up the ADCIN defines and the bit banging from Bruce's example.
Some more reading..
http://www.picbasic.co.uk/forum/showthread.php?t=2188

And a snippet for 10 bit banging..

ADCON1.7 = 1

	gosub getAD



	press.highbyte = ADRESH

	press.lowbyte = ADRESL

	return

Do a search for 10 bit ADC also..
http://www.picbasic.co.uk/forum/showthread.php?t=4751

Mack, Malanie Ok so let me step backwards and make this simple. Let's go back to Bruce's example first and get that working first as you suggested I got ahead of myself.

Per Melanie and Bruce's example the pic is right on. My reading of 24
24/255*5v = 0.470v

So let me work on that. That would mean that for every bascially .5 volt I would see my count add 24 (I realize it's .47 but for the sake of argument we'll say .5) So for 1 volt it would be 48 then 1.5 volt 72 etc to 4.5 volt top out on my sensor at 216 reading.

If I got from 24 = 0 PSI to 216 = 150 PSI that would be 216-24 for 192 increments at a max of 150psi = .78125 PSI per incrment. So the calculation would be
( adval - 24 ) * .78125
( 24 - 24 ) * .78125 = 0
(120 - 24 ) * .78125 = 75
( 216 - 24 ) * .78125 = 150

Right half way between 216 and 24 would be 96 + the offset of 24 I would get a reading of 120 so 120 - 24 = 96 * .78125 = 75

I think I got this so far.

I don't know why I am struggling with this so much but I think this is right so far. This looks correct to me so far let me hook this up to my compressor and take a look

David

[DavyJones]

I know that math is right but I can't get calculation. This is a stupid question I know and I really should know but I don't. What do I have to do to get this to multipl the decimal.

( 216 - 24 ) * .78125 = 150

[DavyJones]

I know that math is right but I can't get calculation. This is a stupid question I know and I really should know but I don't. What do I have to do to get this to multipl the decimal.

( 216 - 24 ) * .78125 = 150

woohoo I think I got it. Were X is my value read from the AD. I get a nice looking number. What do you think?

x = x - 24
res = x * 7812
res = div32 1000

[mackrackit]

By George I think he has it

[DavyJones]

By George I think he has it

Mack,
Thank you for your help today. I can't believe I had so much trouble with that. I wrote a program in mumps a few years ago that converted a number to base 36 for this wacky interface checksum I had to write and I don't think I struggled that hard.

I was able to use my knew found logic and incoporate that same philosphy into my 10 bit program I originally started to work on but stepped back from to get the 8bit to work and understand this a little better. I know you pointed out a couple things about the high and low bits. I won't try to kid you I am not sure what you were trying to tell me there. I put this program together from bits and pieces by looking at others and their examples.

I can run this side by side with the 8 bit program and I get the same results. Is there any advatage of reading the analog value using ADCIN vs. Bruces example? Any disadvatages?

Thanks again today.
David

INCLUDE "modedefs.bas"
' Hardware specific settings
@ DEVICE pic16F877, XT_OSC ' System Clock Options
@ DEVICE pic16F877, WDT_ON ' Watchdog Timer
@ DEVICE pic16F877, PWRT_ON ' Power-On Timer
@ DEVICE pic16F877, BOD_ON ' Brown-Out Detect
@ DEVICE pic16F877, LVP_ON ' Low-Voltage Programming
@ DEVICE pic16F877, CPD_OFF ' Data Memory Code Protect
@ DEVICE pic16F877, WRT_ON ' Flash Memory Word Enable
' ADC definitions 16F877 - (8) 10bit channels
'*****************************
DEFINE ADC_BITS 10 ' (8) 10 Bit channels
DEFINE ADC_CLOCK 3 ' tell the us per osc clock tick?
DEFINE ADC_SAMPLEUS 50 ' how long to sample the pin
DEFINE OSC 10
DEFINE LOADER_USED 1
'INIT ADC
ADCON1 = %10001110 ' Just one analog input needed see page 112 of 16f877 spec
ADCON0 = %11000001 ' Analog converter Fosc/32, ADC module is ON
ADCON0.7 = 1 'I DID THAT ABOVE RIGHT DO I NEED TO RESET THIS OR
'DID MISS SOMETHING ABOUT SETTING VALUES THIS WAY?

'Varibles

SO VAR PORTC.6
ADval VAR word ' Storage for A/D result
PRESSURE VAR word
PRESS VAR word
ADavg VAR WORD[2]
READING var byte
Value VAR WORD
RES var word

'CONSTANTS
AvgCount CON 64 ' = Number of samples to average
FAspread CON 25 ' = Fast Average threshold +/-
BAUD CON 32
'**************************************
'****START OF PROGRAM*****
'**************************************
BEGIN:
SerOut2 SO,baud,[12,"Hello World 10 bit AD Test",10,13]
pause 2000
Serout2 so,baud,[27,"[2;1H","Pressure: "]
MAINLOOP:
ADCIN 0,adval ' Read A/D channel 0 to adval variable
adval = adval - 96
res = adval * 1953
adval = div32 1000
'average out press first before calcuating true pressure
READING = 1
value = adval
gosub average
press = value
Serout2 so,baud,[27,"[2;11H",DEC5 PRESS]

'PRESSURE = ( 58 * PRESS - 58 ) / 100
'pressure = ( 1464 * press - 1464 ) / 10
'value = pressure
'READING = 2
'GOSUB AVERAGE
'PRESSURE = VALUE
'Serout2 so,baud,[27,"[1;11H",DEC5 PRESSURE]
Pause 200 ' Wait period between updates
GOTO MAINLOOP
END

' -=-=-=-=-=-= Average Analog values -=-=-=-=-=-=-=-=-=-=
' compliments of Darryl Taylor
AVERAGE:
IF Value = ADavg[READING] Then NoChange
IF ABS (Value - ADavg[reading]) > FAspread OR Value < AvgCount Then FastAvg
IF ABS (Value - ADavg[reading]) < AvgCount Then RealClose
ADavg[reading] = ADavg[reading] - (ADavg[reading]/AvgCount)
ADavg[reading] = ADavg[reading] + (Value/AvgCount)
GoTo AVGok
FastAvg:
ADavg[reading] = Value
GoTo AVGok
RealClose:
ADavg[reading] = ADavg[reading] - (ADavg[reading]/(AvgCount/4))
ADavg[reading] = ADavg[reading] + (Value/(AvgCount/4))
AVGok:
Value = ADavg[reading] ' Put Average back into Value
NoChange:
Return

[mackrackit]

Is there any advatage of reading the analog value using ADCIN vs. Bruces example? Any disadvatages?

I like Bruce's example because I feel it allows more control.
Example:
I can when reading multiple channels, or one channel for that matter, switch from 8 or 10 bit with
ADCON1.7 = 1
or
ADCON1.7 = 0

I have not checked to see if one way is faster or uses less code space.

[DavyJones]

I like Bruce's example because I feel it allows more control.
Example:
I can when reading multiple channels, or one channel for that matter, switch from 8 or 10 bit with
ADCON1.7 = 1
or
ADCON1.7 = 0

I have not checked to see if one way is faster or uses less code space.

Mack,
Gotcha that would make sense. If you don't mind can we discuss the 8 bit vs. 10bit and let me rant a little about some observations. With regards to the ADC isn't just the obvious difference between reading 8 bit vs. 10 bit just precision? There is nothing else to consider is there? I can either get a return value from 0 -> 255 or 0 -> 1024. One thing I noticed given the 8 bit response it tends to not bounce around as much as the 10 bit. The 10 bit response raw data when I am at 0psi bounces all over the place from 92 to 100. I have to run that through Darryl's average routine to get that stablized. That helps but doesn't stop it completely which brings me to a question is there any way to get that to settle down? The other question is whats the proper term for a signle interval on the ADC read? Is it tick or interval, increment?

My other observation/question based on your statment about switching back and fourth between 10 bit and 8 bit. From my observations and looking at many examples of code there is always more then one way to accomplish something in the coding. I didn't miss anything here right I mean:

ADCON1 = %10000000
'ADCON1.7 = 1 'I DID THAT ABOVE RIGHT?

Setting ADCON1.7 = 1 is essentially the same thing as my example correct? You're just setting the individual bit right?

After this experience and reviewing what you taught me and looking at Bruce's example more closely that you could accomplish the 8 bit using the ADCIN by just setting ADC_BITS=8 in the beginning and the read on ADCIN basically puts both the ADRESH and ADRESL together for you in your variable. If you choose to use Bruce's method and wanted to do the 10 bit you've got to put those two result registers together right? Setting ADC_BITS is a one time thing at the beginning?

It's tough even with the datasheet until someone get's you pointed in the right direction. I think I am intellegent and I should be able to figure this stuff out but thats not always so. Now that you've given me direction it's much clearer I appreciate the help I get from this forum.

Good morning

David

[mackrackit]

Mack,
Gotcha that would make sense. If you don't mind can we discuss the 8 bit vs. 10bit and let me rant a little about some observations. With regards to the ADC isn't just the obvious difference between reading 8 bit vs. 10 bit just precision? There is nothing else to consider is there? I can either get a return value from 0 -> 255 or 0 -> 1024. One thing I noticed given the 8 bit response it tends to not bounce around as much as the 10 bit. The 10 bit response raw data when I am at 0psi bounces all over the place from 92 to 100. I have to run that through Darryl's average routine to get that stablized. That helps but doesn't stop it completely which brings me to a question is there any way to get that to settle down? The other question is whats the proper term for a signle interval on the ADC read? Is it tick or interval, increment?

Just a precision thing. With more precision you can can expect more "bouncing" sometimes noise will cause this or the sensor it's self. A lot of sensors a pretty sensitive Sometimes a capacitor from the ADC pin to zero will help smooth thing out, This may even be needed with some 8 bit applications.

My other observation/question based on your statment about switching back and fourth between 10 bit and 8 bit. From my observations and looking at many examples of code there is always more then one way to accomplish something in the coding. I didn't miss anything here right I mean:

Yep there is always more than one way do do anything. I coding we call this a style.

ADCON1 = %10000000
'ADCON1.7 = 1 'I DID THAT ABOVE RIGHT?

Setting ADCON1.7 = 1 is essentially the same thing as my example correct? You're just setting the individual bit right?

For your chip
ADCON1 = %10000000
makes all of th ADC pins analog, left justified, VREF+ at VDD and VREF- at VSS.
If you want some of the ADC pins digital or change the VREFs then you will run into a problem.
ADCON1.7 = 1 or 0 is just changing that particular bit.

After this experience and reviewing what you taught me and looking at Bruce's example more closely that you could accomplish the 8 bit using the ADCIN by just setting ADC_BITS=8 in the beginning and the read on ADCIN basically puts both the ADRESH and ADRESL together for you in your variable. If you choose to use Bruce's method and wanted to do the 10 bit you've got to put those two result registers together right? Setting ADC_BITS is a one time thing at the beginning?

Not quite... In the data sheet I have for this chip take a look at section 11.4.1.
The register containing ADRESH and ADRESL is really 16 bits wide. ADRESH and ADRESL are 8 bits each.

It's tough even with the datasheet until someone get's you pointed in the right direction. I think I am intellegent and I should be able to figure this stuff out but thats not always so. Now that you've given me direction it's much clearer I appreciate the help I get from this forum.

This all can be quite confusing, if this forum was not here I would not have a clue either.

[erice1984]

I have a similar problem, I got everything to work, except translating the math.

My lowest ADVAL is 0 and it's also the HIGHEST pressure.

I saw how the math worked out for him, but I am lost.

[erice1984]

AD Res 1024

Lowest voltage = .085v = 100psi = ADVal 18

Highest voltage = .999v = 10psi = ADVal 205

In order to switch everything around all I did differently was subtract my highest voltage (1volt) from lowest voltage (.085volt) and went from there!

[erice1984]

So then I thought it out a little more, and instead of converting to voltage etc.. ADVal represents that value...

SO

highest ADVal = 205

205 - whatever the adval is at for the input from the sender = x

x * .5347 = pressure

on my 187 resolution scale here.

[erice1984]

didnt work very well... lost again

[mackrackit]

AD Res 1024

Lowest voltage = .085v = 100psi = ADVal 18

Highest voltage = .999v = 10psi = ADVal 205

In order to switch everything around all I did differently was subtract my highest voltage (1volt) from lowest voltage (.085volt) and went from there!

Can you give the make and model or data sheet for the sensor?

Maybe show your code too.

If that voltage range is correct I would go with 8 bit resolution and a reference voltage. Are you really getting a lower reading with a higher pressure?

[erice1984]

yes, its an autometer 2242 pressure sender unit. its really annoying.

all I have is
15 ohms = 100psi
175 ohms = 10psi


What I am doing right now is using regulated current. ~5mA

[erice1984]

reason why I am not using the 8bit adc is because I would have to use a lot more current to get it up out of the 19mV range, and not really wanting to do that since it seems to vary more, its probably my hardware tho, LM317T current reg.

[mackrackit]

I think I see the problem.

The ADC is for measuing a voltage.

Use the sensor for R1 of a voltage divider and a fixed resistor for R2. The feed that output to the PIC.

http://en.wikipedia.org/wiki/Voltage_divider

[Acetronics2]

woohoo I think I got it. Were X is my value read from the AD. I get a nice looking number. What do you think?

x = x - 24
res = x * 7812
res = div32 1000

Hi, Davy

I think you could have got it better !!!

.78125 = ... exactly 25/32 !!!

so, res = ( x * 25 ) >> 5


Alain

[erice1984]

What if I switched the VRef+ and Vref- to

Vref+ = 0v
Vref- = 5v

Would that work?

[mackrackit]

What if I switched the VRef+ and Vref- to

Vref+ = 0v
Vref- = 5v

Would that work?

Not sure.
From what I found out about that sensor it has a variable resistance output. So to use it with an ADC I think you are going to need to change it to a variable voltage with a voltage divider.