It converts an 8Hz square wave into a sine, and produces two anti-phase waves that each drive a resistive load.
It appears to work as intended, but, as a learning process for me, I was wondering if anyone here could please critique the design and suggest any improvements.
The other way of producing a respectable approximation to sine wave is use a faster clock - say 128Hz, divide it down to 8Hz and clock the 8Hz square wave through it through a 16-stage shift register at 128Hz.
You get a stair-case approximation to sine wave by hanging suitably weighed resistors on each of the 16 outputs of the shift register and summing them - with an op amp adder into a single output (Finite Impulse Response filte r). The 128Hz triangular wave error on the output is a lot easier to filter out than the 24Hz, the 40Hz and 56Hz the components in your square wave.
Longer shift registers are perfectly practical, but getting the resistors p recise enough to see any benefit from it gets tricky.
Getting a quadrature waveform means having second shift register with the s ame value resistors shifted along a bit. Less filtering means less phase er ror between the in-phase and quadrature outputs.
Since it's running from a single supply, it might be best to use a single divider to generate a (buffered) bias voltage for the driver op-amps, one that has some capacitive filtering, at least. Depending on the op amps used, some won't function correctly if they're sinking a lot of bias current into those resistors (though FET-input types will probably work okay.)
There are ways to make two buffered antiphase sine waves from a single-ended input using fewer components.
It'll only make a nice-looking output sine from a square wave significantly below the cutoff freq of the filter made from U1.
That's tricky. Given a digital source, a three level approximation a can el iminate the third harmonic component - 0V for 60 degrees, 1 V from 60 to 18
0 degrees, 0V from 180 to 240 degrees, and -1V from 240 to 360 degrees.
Don Lancaster's "magic sinewaves" list a whole range of messy looking two l evel square waves which have very low harmonic content, for low order harmo nics. Higher order harmonics are horrible and have to be filtered out, but that doesn't take much in the way of filter delay.
I've looked at them from time to time, and while it's a cute idea - particu larly at low frequencies - it hasn't been competitive at frequencies I've b een interested in synthesising.
As I understand, you mean to use an op amp buffered voltage divider to generate a virtual earth between +6V and GND.
formatting link
Thank you for the suggestion. I will keep it in mind if the built circuit has this problem.
I have another version that uses a single transistor phase inverter, but I found the biasing to be a bit tricky and, still requiring buffers, it did not reduce the part count by much.
I've fiddled around and those seem to be the best RC values for the present setup.
it's description of the step response of the filter.
it means it has a extreme gain peak at that frequency.
Attached is a modified version of your schematic I've probably gone a bit overboard on this one getting it to work this well in real life, with real parts, could be hard, perhaps reduce R21 to 860 and reduce r22 to compensate for that.
--
begin-base64 644 Moving Field Amp Single-4.asc
VmVyc2lvbiA0DQpTSEVFVCAxIDE4NDQgNjgwDQpXSVJFIC0yNDAgLTIwOCAt
use a faster clock - say 128Hz, divide it down to 8Hz and clock the 8Hz squ are wave through it through a 16-stage shift register at 128Hz.
ghed resistors on each of the 16 outputs of the shift register and summing them - with an op amp adder into a single output (Finite Impulse Response f ilter). The 128Hz triangular wave error on the output is a lot easier to fi lter out than the 24Hz, the 40Hz and 56Hz the components in your square wav e.
rs precise enough to see any benefit from it gets tricky.
he same value resistors shifted along a bit. Less filtering means less phas e error between the in-phase and quadrature outputs.
HP had the kind of engineers that could make anything work. Showing that so mething could work and showing that it's now the cheapest way of solving Ke vin Foster's problem are rather different targets.
ElectronDepot 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.