Phase shifter

"anwar"

** This question has been beaten to death on SED and SEB many times.

There simply is no such filter circuit as a broad band 90 degree phase shifter.

The closest thing is to build a pair of phase shift networks ( using all-pass filters) such that the DIFFERENCE in phase shift at the two outputs is close to 90 degrees over some range of frequencies. This is sometimes called a Hilbert Transformer or an "out phaser".

.... Phil

--
View using a fixed font:


                 +--------------->I 0°
                 |           
      +----------|----------+---->Q 90°
      | +-----+  | +-----+  |
      +-|D   Q|--+-|D   Q|  |
        |    _|    |    _|  |
CLK---+-|>   Q| +--|>   Q|O-+
      | +-----+ |  +-----+
      +---------+

The devices are "D" type flip-flops.

--- Oops...

. +--------------->I 0° . | . +----------|----------+ . | +-----+ | +-----+ | . +-|D Q|--+-|D Q|--|---->Q 90° . | _| | _| | .CLK---+-|> Q| +--|> Q|O-+ . | +-----+ | +-----+ . +---------+

Pudding:

Version 4 SHEET 1 880 680 WIRE 560 16 16 16 WIRE 272 64 144 64 WIRE 432 64 272 64 WIRE 144 112 144 64 WIRE 432 112 432 64 WIRE 16 160 16 16 WIRE 64 160 16 160 WIRE 352 160 224 160 WIRE 608 160 512 160 WIRE 16 208 -80 208 WIRE 64 208 16 208 WIRE 352 208 320 208 WIRE 560 208 560 16 WIRE 560 208 528 208 WIRE -80 240 -80 208 WIRE 16 288 16 208 WIRE 320 288 320 208 WIRE 320 288 16 288 WIRE -80 352 -80 320 WIRE 144 352 144 256 WIRE 144 352 -80 352 WIRE 272 352 272 64 WIRE 272 352 144 352 WIRE 432 352 432 256 WIRE 432 352 272 352 WIRE -80 416 -80 352 FLAG -80 416 0 SYMBOL Digital\\dflop 144 112 R0 WINDOW 3 8 12 Invisible 0 SYMATTR InstName A1 SYMATTR Value trise 100n tfall 100n vhigh 5V SYMBOL voltage -80 224 R0 WINDOW 3 24 104 Invisible 0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V1 SYMATTR Value PULSE(0 5 0 10n 10n 500e-6 1e-3) SYMBOL Digital\\dflop 432 112 R0 WINDOW 3 8 12 Invisible 0 SYMATTR InstName A2 SYMATTR Value trise 100n tfall 100n vhigh 5V TEXT -64 384 Left 0 !.tran .01

-- JF

e

this

se

r.

Very nice, but you had to divide the original frequency by two. That may or may not be a problem for the OP.

George H.

--
Thanks. :-) 

By four, actually.

Are you trying to get two square waves that are shifted 90 degrees from one another? Then see the circuit that John Fields posted.

Are you trying to get a circuit that shifts _all_ off a square wave by

90 degrees? Then see Phil Allison's post.

Do you just want to know what happens when you phase shift a square wave by 90 degrees, but don't need to actually generate the signal? Then do the work in Mathemagic land, using the Fourier transform. You'll find that the edges of the square wave turn into spikes with infinite amplitude, though.

Do you have a need to take an existing square wave and derive another square wave that's shifted by 90 degrees? If you know the frequency exactly, you can delay the thing. If you don't, you can either lock onto it with a phase locked-loop that uses an XOR phase detector (see the venerable old 4046 and its derivatives), or you can lock onto it with a little synthesizer that generates a 4x wave with zero phase shift, and divide down by four (using John's circuit).

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

Well, he could use one of these:

____ ----+-------------------------------\\ \ | ||xor >----- | +----//____/ | | +--[inv>o--[inv>o--[inv>0--+

to give him one spike at each transition.

That would at least bring the 1:4 down to 1:2. ;-)

Cheers! Rich

is

A circuit can be invented for 90 degrees, but please specify tolerances and limitations. Is it 90 degrees plus or minus 0.001 degree or 5 degrees? Can the frequency be limited below 50 megacycles per second ? With wide tolerance, a circuit can be simulated using a counter running with fast sampling. For a 50 megahertz square wave, sample at 2 gigahertz and use logic.

There is something called a delay-locked-loop (DLL) which can phase shift a square wave by 90 degrees; but the output has jitter and it takes time to lock. Xilinx FPGAs have DLLs in blocks called Digital Clock Managers (DCMs).

wave

this

phase

shifter.

Thanks guys for all of your replies. It is very helpful. I've tried the circuit by JF and it working very good. By using 2 D-flip flop, I can do 90 degree phase shift for any frequency.

Yes there is some issues regarding the original frequency but it not really a problem for my simulation.

Thanks again especially to JF:)

Problem solved:)

--------------------------------------- Posted through

wave

this

phase

shifter.

Thanks guys for all of your replies. It is very helpful. I've tried the circuit by JF and it working very good. By using 2 D-flip flop, I can do 90 degree phase shift for any frequency.

Yes there is some issues regarding the original frequency but it not really a problem for my simulation.

Thanks again especially to JF:)

Problem solved:)

--------------------------------------- Posted through

"Tim Wescott"

** Other ambiguities the OP needs to clear up include:

1. Is your "square wave" signal symmetrical in time ?
2. Is it symmetrical about zero value ?
3. What frequency is it and how does it vary ?
4. What amplitude is it and does that vary ?

As the reality if usenet is that posters all believe they never need reveal what they are actually up to - it is all a mad guessing game for anyone attempting to reply.

.... Phil

"Globemaker"

Can the frequency be limited below 50 megacycles per second ?

** Must be an old ham radio looney.

.... Phil

Not all, perhaps, but, sadly, most.

Hmmm. There may be an extra comma in that.

--
Rich Webb     Norfolk, VA

90

onto

reveal

Hi Phil,

for your question this is my answer:

--------------------------------------- Posted through

90

onto

reveal

Hi Phil, for your question this is my answer:

1. Is your "square wave" signal symmetrical in time ? yes.the square wave is a symmetrical in time. Actually the square wave come out from a timer and d-flipflop. So the signal will have 50% duty cycle. So like JF's suggestion another d-flipflop can make it 90 degrees phase shift.

I'm not very sure if square wave is same with clock signal? But if the signal coming out from the flipflop is a square wave, then the signal do not have zero value(zero dc bias).

The frequency varies with the timer. So it can be from 1kHz up to 12kHz.

The amplitude stay the same.

Sorry again if my post is not very clear.

--------------------------------------- Posted through

How about an integrator followed by a zero-crossing detector? The integrator output will be a triangle wave with peaks aligned with the square wave edges. The zero crossings of the triangle will be at 90 degrees. (That assumes the input square was symmetrical about zero. If not, you will need to set the threshold of the comparator to an appropriate level.)

It wouldn't have a huge frequency range, since on the low end the integrator would saturate and on the high end its output would get low enough to make noise and DC offset an issue. A little hysteresis in the comparator could reduce the noise issue, at the expense of exact 90 degree phase shift.

But might be fine over a limited range, depending on what it's needed for.

Best regards,

Bob Masta DAQARTA v6.01 Data AcQuisition And Real-Time Analysis

Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, FREE Signal Generator Pitch Track, Pitch-to-MIDI Science with your sound card!

--
My pleasure! :-)

--
Aha!

555 strikes again? ;)