Frequency dependent phase shifts... Need help!

[SmD]

I'm currently trying to build an analog circuit that will phase shift sine waves of various frequency (from 900Hz to 1100Hz) in a very specific way. What I'm trying to achieve is a constant time delay at the output (in respect to the input) of the circuit without modifying the amplitude of the sine waves. More specifically, here's what I want:

Input/Output:
- Waveform: Sine Wave
- Voltage: 0.1V - 5V
- Frequency: 900Hz - 1100Hz

Output:
- Time Delay: 0.25ms

The time delay of 0.25ms suggests a phase shift of:
- 81 deg for 900Hz
- ...
- 90 deg for 1000Hz
- ...
- 99 deg for 1100Hz

I've talked with many profs in my department about this, but unfortunatly since I'm a physics student, none of them has a lot of experience in electronics. Fortunatly, I found this forum and I know at least one of you is smart enough to help me with what I want...

Any design ideas or schematics would be greatly appreciated...

Thanks

SmD

 

[Nigel Goodwin]

I'm rather confused as to what you want?, what EXACTLY are you trying to do? - not how you think it should be done (like above), but exactly what and why.

 

[zachtheterrible]

A single common-emitter amplifier causes a phase shift of 180 degrees I believe.
I'm sure that it could be tuned to produce the phase shift that you desire.

 

[audioguru]

An Allpass filter will provide the phase shift required, without changing the amplitude.

 

[mstechca]

I'm currently trying to build an analog circuit that will phase shift sine waves of various frequency (from 900Hz to 1100Hz) in a very specific way.

and you are doing this because...

Input/Output:
- Waveform: Sine Wave
- Voltage: 0.1V - 5V
- Frequency: 900Hz - 1100Hz

All I can say is avoid using logic gates for your oscillator.
Voltage and amplitude are similar. Power (voltage *current) and amplitude are the same.

Output:
- Time Delay: 0.25ms

Use this equation as an approximation: 0.000025 = Capacitance (farads) * resistance (ohms)
Make the resistance high enough or your circuit will blow.

The time delay of 0.25ms suggests a phase shift of:
- 81 deg for 900Hz
- ...
- 90 deg for 1000Hz
- ...
- 99 deg for 1100Hz

I don't deal with degrees much, so I can't help you here.

Any design ideas or schematics would be greatly appreciated...

what is it you want out of all this?

 

[audioguru]

MStechca,
Your lowpass filter will reduce the amplitude at 1100Hz more than at 900Hz. The requirement is: "without modifying the amplitude" which can be done by the allpass filter I recommended.

 

[SmD]

Thanks audioguru... at first glance, I think this is exactly what I need.

Unfortunately, I can't really elaborate on what I'm doing due to the complexity and the nature of my research. What I can tell you though is that the system consists of a waveform generator which will generate a signal of 0.1v to 5v at a frequency of 900Hz to 1100Hz (remotely controlled) at the source of the circuit, and we need a constant time delay (0.25ms) in respect to the generated signal at the output without distorting the signal. This means that I need a circuit which has a flat response, so the simple RC low pass filter will not work. I’m also aware that I can use two remotely controlled generators to achieve the same results. Even though we have the resources to do so, for a specific reason, this setup cannot be used.

I also apologize for my lack of experience in electronics. Even though I have a hard time designing what I need, I can fairly easily (with lots of equations) read and understand circuits. After all, I’m only studying physics…

For those who are curious… when I’m done with this research I will be able to tell you guys exactly what I’m doing …

Thanks for your help…

SmD

 

[Roff]

As Audioguru said, you can use an allpass network. You could possibly do this more easily digitally (A/D, delay, D/A) but I'm basically an analog guy, so I did a little research...
I found a program called **broken link removed**. I downloaded it, and by playing around a little, I got it to give me seven different implementations of the same transfer function, which does what you want... except for one thing. For some reason, some of the implementations have inverted outputs.
Filter free will give you frequency response and schematics. To satisfy my own curiosity, I have redrawn and simulated two of the implementations (below), with a plot of their respective phase plots. I got rid of the inverting output stage that the program had added to the upper circuit.
The upper circuit (out2) has slightly flatter amplitude response. Out1 has about 0.2dB dip around 1kHz. This may be because I rounded the component values to 2 places past the decimal.
If you have any design or analysis experience, I encourage you to play with Filter Free. It's pretty cool.
BTW, these circuits will give you 250usec delay from DC 1.1kHz, and out to a couple of KHz, depending on how much delay variation you can tolerate.
PS I want to make it clear that the schematic below contains two separate circuits that do the same thing.

 

[Nigel Goodwin]

Unfortunately, I can't really elaborate on what I'm doing due to the complexity and the nature of my research.

The reason I asked is because we get this all the time!, people ask vague questions like this without really knowing what they want - there are loads of answers given, none of which are really satisfactory, because we don't know what the real requirements are. Very often, it turns out that the original question bore no resemblance to the actual requirement, and the objective could be achieved far simpler in a different way.

If you want to simply delay the signals? (I see no reason whatsoever to mention phase shifting?), you need a delay line of some kind. In the past 'bucket brigade' devices were available, and used for this purpose, now they seem to have been discontinued, and it's more commonly done digitally - digitising the signal, passing through memory (to give the delay) and converting it back to analogue again.

In both cases, it's simple to adjust the exact delay time.

 

[motion]

I’m also aware that I can use two remotely controlled generators to achieve the same results. Even though we have the resources to do so, for a specific reason, this setup cannot be used.

It's really hard to venture a suggestion if you keep things to yourself. I have a few ideas but I'd rather keep them to myself if you are unwilling to give some more details. For example, why can't you use two sinewave sources delayed by a fixed amount? What is that specific reason?

 

[SmD]

Thanks for the amazing link Ron H... this will surely help!

Unfortunatly I'm at work right now, so I can't try the program out, but I'll definitely play with it as soon as I get home...

SmD