Analog neagtive delay generator - please help

What kind of signal do you want to delay? If it's a periodic clock, all sorts of phase shifters or delay lines will work. A 0.75T delay line looks like a -0.25T delay. Or invert phase and delay 0.25, same result.

If it's a general signal, it's impossible.

John

Well, so much for the stock market investment machine I was going to build!

Or a PLL.

Ayup.

Firstly, what is 'delay locked loop'?

When one needs a signal and a delayed copy of that signal, one uses a delay line (simplest is just a long cable). Then, the events on the signal cable are occurring prior to those on the delay line, and you can refer timings to the delay-line-output and call the direct signal 'negative delay'. That's how an oscilloscope can trigger on a pulse and still show the lead-up to the pulse on the display.

It is a delay line with a variable tap, or some sort of variable RC delay, that is controlled in a similar manner to a PLL to get controlled delay.

They're way popular in FPGAs these days, to de-skew clocks, double clocks (by XORing a clock with a delayed version), have controlled phase differences between clocks, and other useful things that you need if you want to build really, really, really fast logic on a chip that's only really, really fast.

Even a small negative delay could be put into a feedback loop to get arbitrary long negative delays, enough to see the Drudge Report a week in advance.

John

It's a delay line in the feed back path of an oscillator. A "digital PLL", if you will.

The problem I am trying to solve is : Consider an analog PLL and the reference signal is pure sinusoidal. In a "locked in" mode, the VCO output will be a sinusoidal signal of exactly same frequency as the reference signal, but has an added phase (basically delay), because of the way the VCO works. So is there a way to remove this phase ??

That depends on your phase detector. If you're talking about a classic mixer-style phase detector that returns zero error when the VCO is 90 degrees off from the reference -- yes.

So you can solve that, either by making a network that phase shifts the VCO by 90 degrees before applying it to the phase detector, or -- if your reference is also a pure tone -- by using a digital reference detector that returns zero error when the phases match.

John Larkin a écrit :

Cool. So with a small TLine put in a feedback loop will give looong delay line? And maybe varying the loop gain will make the delay variable?

Better patent that before others do.

Thanks for the insight. I do remember the phase detector output vs. phase graph fom Paul Gray's book. However, I wish to verify a few more issues:

1. A reference signal and 90 degree phase shifted VCO output, when fed into the phase detector produces a zero error - so what signal is being fed into the loop filter ??
2. Suppose I split the raw VCO output (just before adding the 90 degree phase) and try to use it for something - what is the phase difference between this and the reference input the the phase detector ??

Thanks for the helpful > That depends on your phase detector. If you're talking about a classic

A very common instance of this is 'advanced sync' used in television to make a video source create a signal early so by the time it goes through processing it's at the desired time. As Larkin said, it only works with a periodic signal. AND, that 'advanced' signal is only delayed 'almost' 1 time period back so it seems advanced. It's not - so Tim Westcott can't have his investment machine.

G=B2

Bummer!