ICs for Sine wave signal generation (Any PDIP left?)

Question

I've been looking into ICs for sine wave signal generation and find that most ICs only generate square waves, and are sold as clock generators. I notice that the difference between square wave Clock Generators and true sine wave function generators is poorly delimited on RS Electronics. Both on RS and digikey, it seems Analog Devices is the only true supplier of something that can generate a sine wave, with something like the AD5932. Other companies like Silicon Labs or Linear Technology are just offering up oscillators and square wave clock generators.

Wondering if the market is truly this small. Are there other manufacturers of these chips? Any that still create PDIP chips? All we could find was the Exar XR2206 but is obsolete so my only options seems to be buying it from ebay or alibaba, which I'd like to avoid if possible. I looked a little further; it seems that all other function generator PDIP chips (MAX038, ICL8038) are going obsolete. Guess I may just have to up my soldering game!

I realize it's better to move to this SMD ICs but for basic prototyping and my beginner skill set, it's a lot easier to look for PDIP components. I'm trying to get into some Visual Light Communication and Li-Fi stuff, and learn how to transmit and receive a signal through an LED photodiode pair. I'd prefer to work with sine waves for now instead of an on-off keying system so that I have less noise, harmonics, and interference to worry about.

Any advice people can lend is appreciated. Any alternate websites or manufacturers that you could suggest?

Thanks!

Jordan

jc2048 · Accepted Answer

This would be easier if you explained a little more what you are trying to achieve. My cheap function generator gets its sine wave (about 1% distortion) by shaping the triangular wave [using the non-linear characteristics of diodes]. The chips you mention above do something similar, only better because of the closer control they have over the semiconductor characteristics. If you try and do it by low-pass filtering of a square wave, the filter needs a steep cut-off to get the third harmonic down to an acceptable level [try it with a simulator and different orders of filter]. That will also have implications if the phase of the sine wave coming out relative to the square wave is important to you. If the frequency is variable, you'd need to accurately track it with the filter. You're getting into quite a lot of involved analogue design. If you need low harmonic content, then generating it with a wein bridge oscillator might be one way to go, but you can't move the frequency quickly and it takes time for the amplitude to settle. If you want it as simple as possible and aren't too concerned about frequency stability, traditional oscillators like the Colpitts and Hartley are possibles. At high frequencies you can move the frequency around using varactor diodes. At low frequencies, you might be able to do it crudely by switching capacitors into circuit with processor IO pins (just a thought I'm throwing in - don't ask me how to actually do it). A crystal oscillator should be capable of producing a good sinewave - the crystal is a good filter - but you won't be able to move the frequency by very much. If the frequencies are low, you'd probably be better off doing what Clem suggests and generating it from a microcontroller. If you massively oversample, the reconstruction filter on the output can be quite simple. If the frequencies are higher, you could do the same with programmable logic of some sort. The great advantage of this approach is that you might find you don't even want a sine wave - perhaps adjusting the waveform might improve the received waveform at the far end (preemphasis). Is it a high power LED you're driving? If so, by working linearly like this, you'll need a power amplifier. It's a lot more involved than just modulating the LED with PWM, which a simple LED driver will do for you as long as the frequencies aren't too high. Perhaps you should look at an old book on rf design for ideas. At one time, radio transmitters were basically a sine oscillator, a modulator of some sort, and then an amplifier stage.

michaelkellett · Answer

Give us an idea of frequency range, acceptable distortion and any modulation requirements.

The AD DDS chips are pretty good but I'm puzzled as to why you want to drive LEDs with sine waves - it's terribly inefficient since you need to bias the LED and modulate the bias (because it can't emit negative light ).

Have a look at some IRDA device and remote controller app notes - you'll get a good idea of low cost industry practice for low speed systems.

MK

michaelkellett · Answer

Your link for IRDA is where I meant.

I think the best bet for your system might be to use time division multiplexing - it will get round so many problems.

You could drive the transmitter LEDs with any convenient frequency and you will only need a single bandpass filter on the receiver. If the transmitter is part of a lighting system the lights will be on most of the time but could have short periods where they were modulated - or rather where they were all off except one which was being pulsed.

If the modulation window was 100uS wide you could get 25 cycles of 250kHz in, if you want to keep below 100kHz you'll only get 10 cycles in which feels like too few (because a reasonable narrow bandwidth filter will have the effect of reducing the amplitude of the first few cycles (amount depends on the filter).

You could easily encode the source of the signal into the pulses.

MK

ICs for Sine wave signal generation (Any PDIP left?)

Top Replies