Yeah, like wanting precise control of the min and max voltages.
And using modern parts that we have in stock.
But he does hold a commanding lead
Sure, just keep adding parts.
But ideally, don't you want to dynamically change the RC value of the integrator when you change the frequency a lot, when it's variable ?
How do the modern function generators make triangle waves ? Not the old ICL8038 method I don't suppose ?
I used to use the old two op-amp integrator and comparator method a lot. Always worked.
boB
The problem at hand wasn't variable frequency, but you'd probably make variable current sources/sinks.
YEARS ago I did just that, but added components to clean up the distortion, etc. If I can find the drawing, I'll post it.
...Jim Thompson
[snip] ...Jim Thompson
Already posted (designed 1986), see...
Function_Generator_OmniComp-GenRad.pdf
on the S.E.D/Schematics Page of my website.
...Jim Thompson
It should be possible to match against any distribution.
That's obviously not true but even if it were, why work around a goofy sorta-triangle wave when a good one is easy to make?
It might not be obviously true, but it certainly isn't obviously not true. The 'goofy' isn't obvious to me, at all; that 'jump' only occurs at the triangle apices, and it's 30 mV on a 3.3V triangle wave. The reason to use a triangle wave is the long straight sweep, not the pointy-top shape. Similarly, the reason to use a square wave is the abrupt rise/fall not the kinda-flat top. I think we all agree on that, because we all used a flip-flop for the square (faster risetime than an op amp used as Schmitt trigger).
Because it works. And the work-around is a no-op (to exercise an ADC, the sweep should exceed the input range, or it doesn't hit all codes - the excess just needs to be greater than 30 mV-before-offset-and-scaling).
It's more useful to improve the triangle by symmetry control (because that allows for a high duty cycle of the upsweep, with short 'reset' delay). I've also done jobs where different measurements on a slow upsweep and faster downsweep improved my throughput.
You can, of course, eliminate the 'jumps' without adding a third chip, by using a (CMOS) '555 and dual op amp, integrating with the first section and inverting-with-attenuation with the second section (and take feedback from that inverted second section). Oddly, though, that would be slightly limiting, because the second section has to be a compensated op amp. That can get you a +/- 5V square and +/- 4V triangle, and benefits from the triangle being amplitude-regulated (to the '555 power rails).
With a small change (triangle wave was +/- 1.33, now +/- 4V) here's some component values.
And, a valid LTspice netlist, enough to do a simulation; note the triangle wave crest 'anomaly', and there's also some opamp-induced ripple after each peak.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required