Freq. Independent Phase Shifter

The square wave jitter is always one clock p-p, namely 8 ns. I didn't believe this, but one of my guys, who's a lot smarter than I am, convinced me. Making a square wave does violate the Nyquist criterion, which is why the waveform reproduction is imperfect. Sine waves are theoretically perfect even at high/oddball frequencies, imperfect only because of quantization and using imperfect filters, and the dac sinc thing. That Nyquist/Shannon stuff is amazing.

John

Email me for manuals or more info. jjlarkin atsign highlandtechnology dotcom.

Oh, if you mathematically bandlimit the samples before you put them into the lookup table, you can make a square wave (granted, with finite rise and fall times) that has way under one-clock jitter, limited mainly by dac quantization and lowpass filter perfection. Shaping the transition of the samples also allows pulse widths, or any other waveform, to be created with way under one-clock resolution. It makes Nyquist and Shannon happy again.

John

If you square up the sinewave output, the jitter is much less than one clock. It may seem a bit odd to filter the output down to a sine, only to square it up again afterwards, but it *does* get you a much lower jitter square.

Jeroen Belleman

Right; if you don't violate the Sampling Theorem, waveform reproduction is theoretically perfect. Jitter becomes limited by dac quantization and filter perfection, so can be a tiny fraction of the clock period.

John

This is standard for a DDS with an accumulator, and not good, if you generate high frequency signals. E.g. for 31.7 MHz output you have a period of 31 ns, so this means 25% jitter for the worst case (of course, if you can devide the 128 MHz without rest by the selected output frequency, jitter would be near zero).

But I like the idea to square up the sinewave output. Is this possible for a wide frequency range output?

I was just curious, but I guess it is a 4 digit price, too much for me for using it for my electronics hobby :-) Any reason why you don't offer it in a shop, like I can buy and compare e.g. HP signal generators?

Yes. I addressed that issue in another post. As long as you don't violate the Sampling Theorem, jitter will be low.

The niche we go for is lots of synchronized channels, for machinery simulation and such. There aren't many many-channel generators around. We can connect these boxes and generate unlimited numbers of synchronized/phase shifted/swept/ratioed waveforms. One of our customers has several 64-channel setups, for simulating jet engines into engine control computers.

John

That's pretty much what I did for a CCD shutter motor controller for a high def color camera a couple of years ago. The accumulator drove three lookups to drive the three-phase motor outputs. In normal mode the thing ran open loop with the 'F' register setting the RPM. An optical encoder was used to sense position (and RPM). During start-up and if the motor wasn't where it should be (acceleration needed), the 'F' register was set to the encoder value (position) plus a constant until the RPM hit a threshold. Didn't take much of a Virtex-4. Timing wasn't hard to meet either. ;-)

This is fun stuff. The 3-phase dac outputs (or sine/cos) can drive stepper motor windings and get beautifully smooth motion. Come to think of it, 3 phase will probably microstep nicer than quadrature, since it won't have flat spots like many 2-phase stepper motors do.

All the classroom signals-and-systems stuff gets very real when you roll your own DDS.

John

It was a three-phase motor. This part of the system wasn't my responsibility but the engineer assigned to it was totally clueless. I was a contractor, so no problem. I volunteered to help him (ended up doing the work) since I was working by the hour. ;-) You're right, it was fun stuff.