PDA

View Full Version : rctime problem

les nesman
08-17-2005, 07:34 PM
DetailedProblem=I'm trying to set up the stamp to be an ohmeter.
I will be measuring resistance in the range of 100-200 ohms

As a test I use a 1000 micro farad cap and a 220 ohm resistor.

When i run the program the rctime = 48000

Using the equation from the Basic Stamp reference manual.
Rctime units = 635 x R (inKOhms)xC (mF)
so R = rctime/635*C
R = 48000/635*1000 = .075kohms =75 Ohms
That's pretty far off. What am I missing here?

A picture of the circuit and a screenshot of my program/debug window are attached.

Thanks -Keith

Jon Williams
08-17-2005, 07:48 PM
The capacitor is the culprit -- those big electrolytic caps have a big tolerance range. Use the results from a known R to adjust for the capacitor value.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jon Williams
Applications Engineer, Parallax

les nesman
08-17-2005, 08:47 PM
The resistor that is there measures 216 with a fluke ohmmeter.

So C=rctime/(635*.216) = 356mF

Now I replace the 216 ohm resistor with a coil that measures 157ohms on my digital ohmeter.
The rctime is 19834. I plug it into the equation.
R = rctime/(635*C) = 19834/(635*356) = .087 kohms

Something is still funky??
Thanks -keith

allanlane5
08-17-2005, 09:40 PM
A "coil"? You're trying to calibrate an ohm-meter with an inductor?

And electrolytics are known for lots of leakage current.· You probably want to use a ceramic capacitor.

Post Edited (allanlane5) : 8/17/2005 2:43:36 PM GMT

les nesman
08-17-2005, 09:52 PM
The calibration was actually done with the 216 ohm resistor.

The only other resistor I have is about 420 ohms and that drives the rctime

over 65000 so it defaults to zero.

So I figured since I had coils that measure around 157 ohms. I would use those as a test to

see how the ohmmeter works. Sounds like that wasn't such a good idea. Please explain why an inductor would make the rctime be so far off?? Thanks.

les nesman
08-17-2005, 11:04 PM
Now I have replaced the 220 ohm resistor with two 220 ohm resistors in paralell. Rctime now reads 00001.

Chris Savage
08-17-2005, 11:06 PM
Les,

·· Could you try a known configuration?· Please try a 10K resistor with a .1uF cap.· This will go to the I/O pin through a 220 ohm resistor.· PAUSE value will be 1.· See what you get then.·

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Chris Savage
Parallax Tech Support
csavage@parallax.com (mailto:csavage@parallax.com)

Chris Savage
08-17-2005, 11:08 PM
Oh, I can't tell by your pic, but is the negative side of the capacitor toward Vss?· Electrolytic Caps like you are using are polarized.· Ceramic Disc caps are not.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Chris Savage
Parallax Tech Support
csavage@parallax.com (mailto:csavage@parallax.com)

les nesman
08-17-2005, 11:21 PM
Yes, the negative side of the cap is connected·to VSS.

Beau Schwabe (Parallax)
08-17-2005, 11:36 PM
les nesman said...
Please explain why an inductor would make the rctime be so far off??
http://www.greenandwhite.net/~chbut/lc_oscillator.htm

The LC circuit resembles a mass-spring system. Initially, the electrical energy from the capacitor is transferred into the magnetic energy of the inductor. When the electrical energy of the capacitor becomes zero, the process is reversed.· The magnetic energy from the inductor is transferred into the electrical energy of the capacitor. Electromagnetic oscillation occurs when energy is transferring between the capacitor and the inductor.

Because of the nature of the RCTIME command of charging/discharging a capacitor via a resistor, substituting the resistor with an inductor will produce a behavior
similar to what is described above.

Perhaps the question should be asked.· Do you need to measure the resistance of an inductor?· If so, then the attached schematic will produce aproximately 0.01 Volts
per Ohm with a resistor or inductor.· To measure the voltage you will need to use an ADC of some kind.

·
In contrast, see· http://www.greenandwhite.net/~chbut/rc_cir.htm· for an RC comparison.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe (mailto:bschwabe@parallax.com)

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 8/17/2005 4:43:48 PM GMT

les nesman
08-18-2005, 12:32 AM
In response to Chris.

I tried using the .1mF cap and 10 k resistor. rctime = 688.

So that works well.

In response to Beau.

Yes, I was building the application to measure the resistance of solenoid coils. Will setting up the attached circuit with an analog to digital converter be pretty straightforward or do you have to be pretty experienced to make that fly? Obviously I'm a·very electronically challenged if I can be stumped by rctimehttp://forums.parallax.com/images/smilies/smile.gif· Thanks very much for your reply.

Beau Schwabe (Parallax)
08-18-2005, 12:51 AM
One question we also need to know.... What kind of resolution are you looking for? In other words how accurate do your results need to be?

There are several ways that your goal can be accomplished some are better than others. The circuit that I provided simply converts resistance
into voltage ( I thought that might be easier to deal with ) using as easy to find components I could think of. In theory you can directly use an
ADC without my circuit by placing the resistor in question across a voltage divider, but with the relatively low resistance values you are looking
at, the range will be limited.

There are several "how to's" here in the forum as well as in the Stamp manual on how to use an ADC.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe (mailto:bschwabe@parallax.com)

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 8/17/2005 5:54:20 PM GMT

les nesman
08-18-2005, 12:55 AM
I will be looking for a min. resolution of about + or - 5 ohms. The range only needs to be from 30 ohms to 500 ohms. I looked at the datasheet for the LTC1298 and was a bit intimidated.· Does it look like using a voltage divider and an adc would work for this application?

Tracy Allen
08-18-2005, 01:01 AM
The formula you need to calculate the resistance of the device under test is a little more complicated than what I gave in the earlier email.

R_unknown = RCT_value / constant - 1000

220 1k DUT
P0 --/\/\--o----/\/\---/\/\----.
| |
'-----||-------------o- Vss
1 uF

That does not account for the output resistance of the Stamp (around 30 ohms for an SX chip, higher for a PIC), nor for the 220 ohm resistor in series with the pin, which is there to protect the pin from the large discharge current of the 1uf capacitor as well as to protect it from ESD from outside that might come in from outside in a test

The formula is,

1.3 = 5*exp(-tm/(R*C)) ' on a calculator (not stamp math!)

Where 1.3 volts is the Stamp threshold, 5 is the Stamp Vdd supply, tm is the time it takes for the capacitor to discharge from 5 volts down to 1.3 volts, and R and C are the parameters. However, that assumes that the capacitor is discharging from 5 volts, which is only a good approximation when the resistance R of the Device Under Test is much much larger than the output resistance of the Stamp. Otherwise, it starts from,

V = 5 * (R / (R + 220 + Rout)
where Rout is the output resistance of the Stamp and 220 is the protection resistor. Substituting 50 ohms for Rout, the starting voltage is,

V = 5* (1000/(1270) = 3.93 volts.

So, the RCtime value will be considerably shorter than you would expect from the generic formula that is valid when the resistance is high. When R is variable, that has to be accounted for in the formula, which complicates it greatly:

1.3 = (5 * (R / (R+220+Rout)) * exp(-tm/(R*C)) ' on a calculator (not stamp math!)

------------------------------------------

It is really a lot better for calculations if you can use a larger resistor and a smaller capacitor. Say, a 10k ohm minimum and a 100 ohm protection resistor and a 0.1uf capacitor. That will give times around 1000 microseconds (rctime values of 500 on the BS2, higher on the other stamps).

100 10k DUT
P0 --/\/\--o----/\/\---/\/\----.
| |
'-----||-------------o- Vss
0.1 uF

The error from the output and protection resistance is less than 1%, and a much smaller fraction of the total.

Here is a URL with more RCTIME tutorials:
emesys.com/BS2rct.htm (http://emesys.com/BS2rct.htm)

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com (http://www.emesystems.com)

Post Edited (Tracy Allen) : 8/20/2005 6:55:48 PM GMT

Beau Schwabe (Parallax)
08-18-2005, 01:02 AM
30 Ohms to 500 Ohms ?

What happened to 100 Ohms to 200 Ohms?· http://forums.parallax.com/images/smilies/smilewinkgrin.gif

You might be able to implement something like this and have two seperate scales.

0-300 Ohm and 200-500Ohm (small overlap)

Note:

Calibrate by setting the scale and placing the LOWEST resistance value for that scale in the appropriate place.

Adjust the Rcal for that scale until you read 1 Volt.· done!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe (mailto:bschwabe@parallax.com)

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 8/17/2005 11:19:00 PM GMT

les nesman
08-18-2005, 08:47 PM
Tracy,
Thanks for the great explanation of why rctime can't measure a small resistance. I am going to try the 10k resistor idea. -Keith

Tracy Allen
08-21-2005, 02:04 AM
The attached circuit uses an op-amp to help with measuring small resistances, still with RCTIME, not an adc. The circuit converts the small resistance to a proportional small current to charge the capacitor. The voltage divider from Vdd creates a 100mV reference at the noninverting input of the op-amp. Feedback puts the same voltage across the resistor under test.

http://forums.parallax.com/attachment.php?attachmentid=38604

i = Vi / Rdut enforced by feedback op-amp

The transistor can be either a logic tye mosfet or a pnp (high beta preferable). The current through the resistor equals the current through the capacitor, which charges at a linear rate,

i = C dV/dt => dV = i dt / C charging C

Intagrated, the voltage at the Stamp pin is,

Vout = Vo - (i * t / C) Vo is initial condition
= Vo - Vi/(Rdut*C) * t

The voltage would start at 5 volts and go down to the Stamp threshold at 1.3 volts, a total of 3.7 volts change, in RCTIME 0,1,result.

Then with Vi=0.1 volt and choosing C=0.1µf,

Vout = Vo - t/Rdut t in microseconds,

substituting Vo=5 volts, Vout=1.3 volt threshold

Rdut = t / 3.7

The op-amp has to be a single supply type, and preferably one with a small offset voltage. The LM358 would work, but more modern precision op amps like the LT1078 would be better. Best if it can swing rail to rail at the output.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com (http://www.emesystems.com)

Post Edited (Tracy Allen) : 8/20/2005 7:04:42 PM GMT