Designing a frequncy synthesizer

[Shahil]

Hi!

I want to design a simple, frequency synthesizer that synths from 1KHz to 16KHz is 1Khz steos. The required frequency should ideally be chosen by a "switch" of such.

Does anyone have, at least, some ideas or uselful links. I've googled this but I've been presented with rather complex circuitry. Ideally, the cct. should be a simple cct consisting of basic components (transistors, caps, resitors, diode...) rather than an programmable ICs.

Help...anyone...someone...please??:shy:

 

[Averagesupernova]

Impossible to do without programmable ICs or MANY transistors or even logic gates configured as counters. However, there are phase locked loop ICs designed with parallel inputs. So they are technically programmable, but are more easily controlled with just switches than serial input ICs are. It has been over 10 years since I had a look at a data sheet for one of them. Do you know the basics of phase locked loop? How accurate does it need to be? There are other ways around this if you don't need quartz crystal accuracy.

 

[berkeman]

Hi!

I want to design a simple, frequency synthesizer that synths from 1KHz to 16KHz is 1Khz steos. The required frequency should ideally be chosen by a "switch" of such.

What kind of distortion specs do you want on the output? Can it be a square wave (probably not)? If it can be a square wave, just make a configurable divider circuit down from some higher single frequency. How accurate do the frequencies have to be? Do they need to be 1.000kHz, 2.000kHz, etc.? Or can you have some be slightly off to help with the divide ratios down from the higher frequency.

If you need sine waves and can tolerate slight (<1%) frequency errors, the easiest way to do this is to program a PROM with the waveforms, use a switch to select which of the 16 areas of the PROM to cycle through, and connect an R-2R ladder to the 8 data lines to make a simple DAC. You need to choose the PROM size and the overall clock rate to give you the 16 waveforms with reasonable frequency accuracy. You also need to have the "end address" of each waveform stored, so that you know when to reset your address counter to the beginning...

 

[Shahil]

Thanks for the help...

The specs are bascially very open...I do want the step is 1KHz, 2Khz, #Khz...up to 16KHz.

The actaul distortion amount isn't really paramount. The project is basically a simple assignement given to us by a lecturer. It'sa rather advanced electronics course (final year level) but we have been concentrating on the workings of only basic components to solve problems.

Hence the use of ICs and even PROM might be a bit over the top.

I did have the idea of using a 555 timer to control the actual frequency (just basically wiring the timer differeently to get the required wave form). Oh, and a square wave output will be OK btw... A simple wswitch or binary counter can just be used to choose the desired frequncy.

I'm still pondering over this. Any other ideas and ccts. and links will be very helpful!

Thanks again

 

[berkeman]

Well, you at least need frequency accuracy specs. That will determine which techniques are viable. Like, can the 17MHz square wave actually be 17.1MHz?

BTW, here's a good tutorial on frequency synthesis using direct digital techniques. Even at your level in school, it should be an interesting read:

http://www.ehb.itu.edu.tr/~eepazarc/ddstutor.html

 

[Shahil]

Thanks for the link...bit complex but informative.

Ok, let me address the accuracy spec. Basically, this project is to determine if I can build something that outputs 16 different waves. The lecturer told me to generate 1KHz, 2KHz..to 16KHz waves but if this is unworkable, any 16 waves, with a constant step, will do.

I do have two ideas...One works but is complex...the other, well I aint sure of it yet.

1. Use a 555 timer to generate a wave that is an LCM of 1 to 16. To get a desired freq. this is fed into a counter that acts as a divider and outputs a wave of a desired freq. ie. If you require a 7KHz wave, the wave is fed into a binary counter that divides the generated wave and outputs a 7KHz wave. A switch is used to choose the desired freq.

he problem is that a very high freq. wave is required as the LCM of 1 to 16 is a big number (I haven't worked it out tho...sorry!) and 16 counters will be required. It will work, but it is inefficient.

2. Generate 1KHz, 2KHz, 4KHz and 8KHz waves. To get a desired wave, all you do is choose the desired freq. via a switch and the desired wave will be generated as a combination of the generated waves. ie. If a 7KHz wave is required, the switch combines the 1KHz, 2KHz and 4KHz waves and outputs a 7KHz wave.

I don't know how exactly to implement it or if the waves can be generated this way.

Any ideas?

 

[Averagesupernova]

You still haven't addressed the accuracy spec. You've told us several times the basic functionality of it, but not accuracy. It has to be 16 Khz +/- how much? Does the thing have to be accurate to within a tenth of a hertz of can it be off as much as 100 hertz? Give us a part per million spec. Concerning your counters, why do you need separate counters? I don't know what your expertise is in some of these areas, but have you considered programmable counters?

 

[Shahil]

I don't think I made it clear but the accuracy of the step size is really unimportant. It is a projects and the important thing is to have, basically, 16 frequency steps.

Also, is accuracy the real important issue here? If I do use the first method, surely the counters will divide the wave, that you have made pretty accurately with the 555, into whatever the required freq. is?

I'll re-iterate though...this is not a freq. generator that will be used to control some physical system...It is a class project that requires an understanding of the concepts behind it rather than building something that has a 1% accuracy...I hope that clears things up.

And thanks fr the idea about the programmable counters...I discussed the idea this morning with a friend...I haven't used them before but I'll learn them:)

So...basically, the cct. would consist of a 555timer generating a waveform...this way will be i/p into a programmable counter and the o/p of the counter will give out the desired waveform...seems rather simple. Just hope it works!

 

[Averagesupernova]

I don't think I made it clear but the accuracy of the step size is really unimportant.

That is not in question. If the accuracy of each individual frequency is right on, then obviously the step size will be too.

Also, is accuracy the real important issue here?

You tell me. I'm the one asking. Haven't gotten an answer yet.

I'll re-iterate though...this is not a freq. generator that will be used to control some physical system...It is a class project that requires an understanding of the concepts behind it rather than building something that has a 1% accuracy...I hope that clears things up.

I'm still not clear on things. You want a frequency synthesizer. The definition of that is somewhat broad. Have you been given any requirements of what you NEED to use? Are you required to use frequency dividers? Can't you just build an oscillator and switch in different resistors or capacitors to bump the frequency? If the accuracy of the thing is not real critical then this is what I would do.

So...basically, the cct. would consist of a 555timer generating a waveform...this way will be i/p into a programmable counter and the o/p of the counter will give out the desired waveform...seems rather simple. Just hope it works!

A bit more complex than necessary as you will find later on. If counters/dividers are not necessary, go with my previous suggestion. I'd appreciate a list of requirements from your instructor. I could be a lot more help than I have so far if I knew this.

 

[Shahil]

I'm still not clear on things. You want a frequency synthesizer. The definition of that is somewhat broad. Have you been given any requirements of what you NEED to use? Are you required to use frequency dividers? Can't you just build an oscillator and switch in different resistors or capacitors to bump the frequency? If the accuracy of the thing is not real critical then this is what I would do.
...
A bit more complex than necessary as you will find later on. If counters/dividers are not necessary, go with my previous suggestion. I'd appreciate a list of requirements from your instructor. I could be a lot more help than I have so far if I knew this.

Okay...lets clear this up first. Accuracy is not important. The basis of this design project is to show that we can generate a 16 different waves that can be chosen by using some type of switching device. I don't think it'll matter if I have a device which outputs 16 waves in 2Khz steps...At completion, the cct constructed would generated the 16 different wave forms in equal steps. To "correct" this, as such, would just require some tweaking of component values to drop the steps from 2KHz to 1KHz...that can be done quite simply. The challenge is to get a cct that can actually generate these waves.

Ok, unfortunately, my communication with my lecturer lasted around 1 minute. He told me that for my practical next week, I must design and build a frequency synthesizer that will generate 16 different waves from 1KHz to 16KHz in 1Khz steps. That's basically it. I was given no other specs. Whatever I now do is up to me. Yes, it is very vague, but that's what I got.

The counter idea was one idea I came up with. I do realize it might be complex but I've worked with counters before...thats why I asked you guys for some help on any simpler designs! Any more help with the oscillator idea would be rather helpful. The course is based around the workings of simple components (resistors, caps, transistors and diodes) so I think the lecturer would prefer these to be used.

Oh, and we do have like a day to build it...It's an 8hr practical so ideally it should be done in a day. Our other pracs for this subject have been known to run for 2weeks so ja...

 

[Averagesupernova]

Ok. That's what I'm looking for. Thanks. Do you have any experience with op-amps? It is not too difficult to set up a triangle wave generator with an op-amp. Switch in different resistor values for different frequencies. A quad package op-amp would work very nicely. If you have some experience with ramp generators and such I'll help you out. But if you don't have any op-amp experience we'll leave that option out. If not, the 555 isn't a bad idea if you just switch in different resistors and capacitors. I mention the 555 simply because you already have mentioned it.

 

[Averagesupernova]

I really hope this thread doesn't turn into another one of these: https://www.physicsforums.com/showthread.php?t=77130

 

[Shahil]

Sorry for taking a while to reply...I've been busy with my MAIN design project that's due in two days time (RFID access control) .. that's kinda the reason why I posted and asked for help with this cct...I'm so cut for time right now.

Anyway, ja...i think the op-amp idea will work. I've googled and found some information on sine wave generation...Not really sure how they work really (shoulda paid attention when we covered this in 2 years ago!)

http://ourworld.cs.com/gknott5413/elect496.htm
http://uhaweb.hartford.edu/jmhill/ee360/lab/opamp/opamp1.htm

here's the links...um...im guessing the basic idea using different resistor and cap combinations wired up to the op-amp and providing feedback to generate different waves...it might end up a bit messy if i do wire up 16 different combination to a single op-amp though! The second one looks usuable although I'm a bit unsure about which resistors and cap values to modify. (Worse, I know at this stage in my degree, I should know)

 

[Averagesupernova]

I had something more in mind that would generate triangle waves. Set up an op-amp to generate a ramp signal. A current source through a capacitor will get a linear ramping voltage across the capacitor. Knowing this, you should be able incorporate a capacitor in the feedback path of an op-amp and get a desired output. Also, set up a feedback path with a comparator and some hysteresis and a switch to be able to select how much current is going into your capacitor. This switching will determine how fast the voltage ramps up and down. Do you see where this is going? Basically I've explained the whole thing to you but left out details because after all, this IS homework right? You can do it with one capacitor, 20 resistors, 3 op-amps, and of course a 16 position switch. So a quad op-amp package would be fine. Of course it is set up to run on a split power supply and I didn't include power supply bypass capacitors in my parts count.

Edit: Some of the resistors, oh, say 15 or 16 of them, will be custom. Either use pots or combine several resistors up to get the right value.

 

[Shahil]

Hmmm...this is going to be fun...

Thanks for the help...I'll discuss this with my partner now...See what he has to say.

I get where you going with it though...thanks again...I will post soon. I promise

Time to go dig in my old electronic notes for the exact info.

 

[Shahil]

Something like this?

http://images.cjb.net/12818.jpg

 

[Averagesupernova]

You're close. You don't really need a resistor in parallel with the capacitor in the left op-amp. Pay attention to the polarity of the feedback. You need to do something different on the second op-amp. A resistor hanging on the input of the op-amp with nothing else attached doesn't really do anything. Make one stage ONLY a schmitt trigger. It's only job is to square up the signal with hysteresis. The upper and lower trip points determine at what voltage the ramp switches direction. Then feed this square wave output into a stage that you can use your switch to vary the current into the ramp generator. I have mine drawn out but I won't show it to you... yet. Give me something more, maybe you'll come up with something different or even simpler. You're doing great.

 

[Shahil]

The cct i gave u was just something brief that i quickly drew up in Visio so to just give and get an idea of how to make the cct...

I will take your advice and give you the cct i am going to use soon.

Kinda got to get my mjor project done today...no time or anything else, not even sleep:yuck:

 

[Averagesupernova]

Sounds great. I'll be waiting.

 

[Shahil]

Sounds great. I'll be waiting.

Sorry for taking so long...Been busy with some other work and ended up building an instrumentation amp instead for the prac

I am interseted in the design of this synthesizer though...I'll give it a shot when i have some spare time

 

[Averagesupernova]

I should have known.