new (to me) 90 degree all-pass

I invented this! Well, after maybe 15,000 other people invented it first.

formatting link

If I use the AD835 multiplier in my IQ modulator, with differential inputs, the E1 thing is free.

--
John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
 Click to see the full signature
Reply to
John Larkin
Loading thread data ...

Only at one frequency. The phase shift over the spectrum ranges from 0 to 180 degrees. It does have the advantage that the input impedance is constant, at least if the pick-offs are of infinite impedance.

--
Tim Wescott 
Wescott Design Services 
 Click to see the full signature
Reply to
Tim Wescott

Fortunately, I only need to work at one frequency, 14.7456 MHz. Turns out that's a common baud rate clock, so I can buy oscillators cheap.

One of the resistors might be a pot, to tweak the phase shift.

--
John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
 Click to see the full signature
Reply to
John Larkin

Yes, that's an old one. I used it back in the 1960's for some reason. Works well.

Reply to
John S

You can do almost a whole octave if you add an inductor to each leg. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

Huh, I learned about that when a potential hire mentioned it in a job interview. playing around, I put two in a loop, (with one the other way around.) added some gain and made a quadrature oscillator*. (it was analog so it depended on how well things matched)

George H.

*there's another quadrature oscillator based on the all-pass, so it's just a riff on that.
Reply to
George Herold

It seems to me that you're re-inventing the SSB tecnology. Please get a good book first, it will save plenty of unnecessary work.

The RC phasing network was already used in the first phasing type SSB generators for carrier telephony. I dug out my old Radio Handbook by Editors and Engineers, of year 1963, with plenty of phasing-type contructions.

Another popular way of generating single-frequency 90 degree phase difference was to have two loosely coupled tuned circuits, one tuned 45 degrees above the frequency and the other 45 degrees below.

Using the current components, it is probably easiest to usee two flipflops connected as a twisted ring and fed with 4 times the desired carrier frequency. If only the target frequency is available, it should be easy to generate the fourfold frequency with a PLL, and the ring counter can be used in the loop.

--

-Tauno
Reply to
Tauno Voipio

The four-section RC phase-sequence network in AoE III, Figure 7.48, page 456, works well over two decades, with about 1-deg of ripple. Only needs 16 Rs and Cs.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

Due to some amazing oversight, I don't have a copy of AoE up here in the cabin, and I bet the local bookstore doesn't either.

The usual SSB phasing method network has two outputs that stay 90 degrees apart, but both squirm around relative to the input, over frequency. I want a real 90 degree phase shift, relative to the input, in my Q path. Of course, it's easier to make 90 degrees when the frequency never changes.

I played around with opamp integrators, but it looks like one needs a huge opamp GBW to make a good 90 deg integrator at 14.5 MHz. And current-mode opamps don't make good integrators.

A series RLC is an interesting 90 deg shifter, too.

--
John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
 Click to see the full signature
Reply to
John Larkin

I'm a circuit designer. I invent things.

I know all about the SSB phasers. But SSB modulation only needs two signals that are 90 degrees relative to one another. I want 0 and 90 degree shifts that are referenced to the input. Ham radio operators don't care much about absolute phase and amplitude control.

The zero degree phase shift path is pretty easy to design. My little RC-CR works fine at a fixed frequency.

Thanks for the suggestions, but none would work in my application. I need to precisely simulate an inductive transducer, gain and phase and in-out impedances. It's easier for the customer if he can program the actual I and Q gains, 0 and 90 degrees.

--
John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
 Click to see the full signature
Reply to
John Larkin

You mean a differential transformer?

What hinders you to design the PLL so that it gives the 0 and 90 degrees RF outputs from the twisted ring dicvider?

You have to admit at least a quarter-cycle delay from the input to make your phase shifter causal. Of course the time delay can be hidden into full cycles of delay.

--

-TV
Reply to
Tauno Voipio

It's a special thing that does behave sort of like an LVDT, but it runs at 14 MHz.

There is no PLL. Two Rs and two Cs are sure easy. And I need amplitude linearity from the excitation to the output of my box.

OK, I'll concede that my box will be causal.

--
John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
 Click to see the full signature
Reply to
John Larkin

Many years ago I designed a circuit that produced a 90 degree phase shifted signal using a serial analog delay line (SAD1024) and a phase locked loop to generate a clock signal at a multiple of the signal frequency such that it had a delay corresponding to 90 degrees. This was for a test set using power line frequencies, and would only work at audio frequencies.

Another idea is to use an EPROM with sine values and a D/A to create the waveform, but again this would be practical only for audio frequencies, and not the 14.7456 MHz signal you are working with. BTW, I use that frequency for the clock of numerous PIC designs, both for accurate baud rate generation, and for ADC sampling rates that are multiples of 60 and 50 Hz power line signals for true RMS measurement. It's 25 * 65536 * 9.

Paul

Reply to
P E Schoen

You can use a shift register for a similar effect. This example...

was devised for a GenRad portable instrument where the floppy drive had speed variation issues, but I've used a similar approach for quadrature satellite receiver VCO's... up to 5.2GHz ;-) ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

Yes it's not relative to the input phase, but RC phase-sequence networks give you the needed 0-deg differential signal along with the 90-degree one.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

Add a diffamp, like I did, and you're back to true 90 degrees. I think.

--
John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
 Click to see the full signature
Reply to
John Larkin

allpass_1.jpg

A slightly more general version is given in AOE 2. If you had ever read the book, you might have come across it. It has been mentioned here before.

AOE 2 and 3 both talk about phase sequence phase shifters for SSB.

The phase shifts don't "squirm around". As shown in fig 7.49 in AOE 3, both phase shifts monotonically increase with frequency, from -180 degrees to +

540 degrees with increasing frequency. The phase difference does squirm aro und a bit, but not much, in the region of interest.

Much easier.

Two 45 degree phase shifts in sequence are rather less demanding.

Not all that interesting.

--
Bill Sloman, Sydney
Reply to
bill.sloman

Or make part of one of the capacitors a varactor and have it servoed by an xor/diode mixer/whatever so an accurate 90 deg is maintained over age and temp?

piglet

Reply to
piglet

I've seen controllable phase shifters done (at audio frequencies) with a CMOS logic gate output as a variable resistance.

Posted on s.e.d. about a decade back:

formatting link

Regards, Allan

Reply to
Allan Herriman

Maybe I'm starting to understand.

You have a reference sine at 14.756 MHz, and you'd like to make a copy with adjustable phase and amplitude - right?

How about matching the input signal to coax, feed that to a 0 degree

3 dB power splitter and taking one output from the splitter and another through a quarter-wave coax from the other output. You'll lose the cable attenuation in the delayed signal, but it is small.

You then need two Gilbert cells fed with the I (straight) and Q (delayed). For precision work, the Gilberts need calibration, but it is doable. Also small phase errors in the delay cable can be compensated by feeding some of the baseband I to the Q channel or the other way round, adjusting the gains to compentate for the additional signal.

--

-TV
Reply to
Tauno Voipio

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.