Archive of the former Yahoo!Groups mailing list: MOTM

previous by date index next by date
  topic list  

Subject: Circuit design question...

From: "Brousseau, Paul E (Paul)" <noise@...>
Date: 2001-06-22

OK, I know this isn't exactly on-topic for MOTM, but I'm hoping you'll
forgive me and not banish me to SDIY. Those guys are way over my head. :)

I want to know how to design a circuit which I will endever to now explain.
If it doesn't make any sence (which I'm sure it won't), let me know!

There are three inputs, all in the 0-10V range. The first two are upper and
lower limits, and the last is a "value" (I can't come up with a better
word!). The output range is 0-10V. I want the circuit to output a voltage
that represents the location of the value voltage between the upper and
lower limits. If the value is greater than or equal to the upper, output
0V. If its less than or equal to the lower, output 10V. If its somewhere
between, output the, er, division?

OK, this is a good time for an example!
Upper limit = 8V.
Lower limit = 3V.
Value = 4V.
Output = 10 - 10 ∗ ( (4 - 3) / (8 - 3) ) = 10 - 10 ∗ (1 / 5) = 10 - 2= 8V

Another example:
Upper limit = 10V
Lower limit = 5V
Value = 7.5V
Output = 10 - 10 ∗ ( (7.5 - 5) / (10 - 5) ) = 10 - 10 ∗ (2.5 / 5) = 10 - 5 =
5V

One more example:
Upper limit = 9V
Lower limit = 0.5V
Value = 1V
Output = 10 - 10 ∗ ( (1 - 0.5) / (9 - 0.5) = 10 - 10 ∗ (0.5 / 8.5) = 10 -
~0.59 - ~9.41V

A last example:
Upper limit = 9V
Lower limit = 0.5V
Value = 6V
Output = 10 - 10 ∗ ( (6 - 0.5) / (9 - 0.5) ) = 10 = 10 ∗ (5.5 / 8.5) = 10 -
~6.47 = ~3.53V

It looks like I bias off the upper and value voltages by the inverse of the
lower voltage. But how to do division and scale to 10V? After that, it's
inversion and bias.

To put this all in context, I have been thinking about a filter. Initially,
just low-pass. You set the cutoff frequency (lower limit). And you set
the, er, spread (upper limit). (Perhaps this is better describer as the
zero-attenutation freq. and the full-attenuation freq.?) The filter input's
goes through a frequency detection, the result of which is the "value"
above. After going through this scaling circuit described above, which will
tell an op-amp how much to attenuate. If the frequency is less than the
cutoff, no attenuation. If it's greater than the cutoff plus spread, full
attenuation. If it's somewhere between, attenuate accordingly. No
resonance.

Of course, this should be done in the analog domain. C'mon, you don't think
I'd settle for some D/A converters and micorcode? :)

Thanks for any help!

--PBr