# computing the sine function

• posted
I'm sure there is some way to make an analog circuit that takes as an
input a conituously changing voltage or current and outputs the sine in
the form of a continuously changing voltage or current. Can anyone tell
me how difficult such a circuit is to build? How much error do they
have? Are they expensive to build?
SM
• posted
there is not quite enough info here as to your needs.
perhaps an oscillator followed by a VCA is what you want. the input voltage drived the VCA. the output would be an amplitude modulated sine wave.
(if the sine wave is an RF frequency we call this an AM transmitter) :)
• posted
How accurate? Over what temperature span? How many volts/amps do you want to handle? One,two, or four quadrants? How much bandwidth do you need? How much power can you afford to run it? How much money do you want to spend? How many pieces do you need to make?
There's got to be a dozen ways of doing this, or more. Ways that pop into my mind: - Discrete analog function generator, pots and diodes - Use a PIC with analog input and output, and do the sine math digitally - Electric motor driving the wiper on a special potentiometer wound with a sine-shaped resistance element. - A-D converter/ EPROM with sine table/ D-A converter - A synchro transformer could probably do it. - I'm sure Analog Devices makes or made some custom chip that will do the job with 0.1% precision over a range of 10 volts at \$65 a piece in quantities of 10,000.
And I'm sure the list goes on from there.
Bill
• posted
It would need to operate around room temperature. I have not done a thorough analysis of my needs, but I suspect I only need to have 1% precision, however 5% might prove to be acceptable. The voltage will be between -C*pi and +C*pi where C is arbitrary (The voltage range will be less than -+1 volt, but it is easily adjusted). The range of \$25 to \$35 US is exceptable, however I would like to spend less than \$10 US (This price figure assumes I'm buying in bulk). I'm not sure about bandwidth, I've forgotten what the term means. As far as power goes the standard power that comes out of the wall socket will be available. This power may also be converted to another form first. Another important parameter is that digital electronics can not be used.
Thanks, Steve
• posted
"bandwidth" i.e., how *quickly* do you need to see sin(x) for a new value of x?
Why *not* digital electronics? Religious prohibition?
How many 6SN7s am I allowed to use?
Can I transform, rectify, filter and regulate AC power or must I use it as 120 VAC ? How many kilowatts can I dissipate to do this?
Bill
bandwidth,
• posted
This is pretty obviously a class project. I'm not sure how I feel about this--how far do you go to help someone? Even without live help the internet is still an enormous advantage compared to when I was in school.
dave y.
• posted
Some latency is permisable, but I'd like to keep it down to 0.005 seconds.
I do not want a quantitized signal. If I went digital I would need a sampling rate of 150+KHz. I also think that I can minimized noise with an analog circuit.
6SN7??? I don't know very much about electronic components.
You can pretty much do what ever you want, however ir must be able to operate continueously, so use of baterries (maybe capacitors must be limited)
• posted
Hmmm...if I help, the sine function generator will have 6SN7s in it. And just this morning I figured out why a diode function generator can't work for the given parameters.
Bill
• posted
To me, "latency" implies a uniform delay between the input changing state and the output changing state...which I suppose is what I'd asked before. Perhaps I'd better instead ask how many times per second must you produce a new value of sin(x) for a new x?
What engineering reason do you have for not wanting a "quantized" signal? If a 16-bit standard is good enough for compact disc audio, then your 5% allowable tolerance could be met easily with only 5 bit conversion. The sampling rate of 150 kHz is common as dirt and would not be a serious constraint - 150 GHz would be past the current state of the art, though.
But 5 ms does rule out some of the more Rube Goldbergian electromechanical converters.
All the really easy analog methods that I know would only convert for +/- 90 degrees - you asked for +/- 180 degrees which becomes non-monotonic.
Have you checked the Burr-Brown and Analog Devices web sites for any combination of chips that might solve this problem? Have you read any of the literature on analog computers to see if there are techniques that are applicable?
And the 5 people who still know what a 6SN7 is are enjoying the idea of using them in this application. I don't think I can get very many for \$25, though.
You sure you don't need a cosine, too? I'm trying to think what the application might be...some kind of circular-scan (S scan? A scan? Can't recall) radar display is my first idea. But you'd need cos(x), too.
Bill
• posted
I need to find a new value of sin(x) 150,000 times/second. The reason for not using digital circuitry is so that I can have amazing resolution. I do not want to aproximate a value produced by analog circuitry (even with its 5% error) with a value produced by digital cicruitry.
Is there a place where I can get A/D and D/A that have a sampling rate of 150kHz and 32-bit precision?
I will have to look into the references you've given.
Thanks, Steve M.
Bill Shymanski wrote:
• posted
I'm baffled. You can accept 5% error but you ask for 32 bits digital conversion? This makes no sense to me. 5% error corresponds to slightly less than 5 bits resolution, an 8-bit A/D -D/A combination would be quite low cost.
Bill
cos(x),

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.