DS1077

You might as well ask "how do I build an airplane?" or "how do I become a golf pro?"

John

Sorry for such a vague question I was rather annoyed as I had just written quite a long question and lost it when my laptop crashed. I was wondering, if I managed to produce a frequency of say, 100MHz using the pic and the frequency synthesizor. How do I modulate the 100MHz signal with an audio signal...

Thanks for any help

Richard

As far as I can see, Phil Allison is correct. You can't modulate the signal.

I always thought it would be interesting to use an Analog Devices DDS to generate an FM signal without any mixers. You would still need a low-pass filter, since the output is a DAC and since you would want to keep second and third harmonics out of the signal. But if you use one of the faster DDS's, the harmonics might be very low, and the reconstruction filter wouldn't need to be high order. Note that I very much doubt that this is the best way to do this. I just think it would be an interesting way to do it.

--Mac

OK, so I don't think using a DDS is feasable from the information you gave me. Phil, what do you mean look elsewhere, where else should I look. How could I do this more simply

Thanks

Richard

Ok so what if I use the max2606. It seems to be a pretty good VCO and I just need a PLL to change frequency.. Any example circuits or good chips to use for a pll.

Thanks

OTH, AM modulation of a DDS is practical to consider, since many of them have a current-controlled output-level programming pin.

Somebody has to stick up for AM radio! :>)

I think Phil meant, not a DDS by itself. But a DDS is one way (not necessarily the best) to create the FM-station frequency of your choice. Thinking out loud... To make an FM modulated transmitter, you could add a PLL oscillator to track the DDS, with a fairly low 20Hz PLL loop bandwidth, etc., and directly ac-couple modulate the PLL's VCO input for the FM signal. An immediate issue that comes to mind is the VCO voltage-frequency transfer function. An RF VCO's gain changes with frequency, an aspect that's normally not very important, but this would make it difficult to reliably set the FM-modulation index.

If you aim is to produce a simple, broadcast band FM transmitter with programmable frequency etc - look elsewhere.

........ Phil

What are you trying to do? I've never designed an FM transmitter, but it seems like the basics are a VCO plus some kind of RF source, plus some mixers and filters.

I seem to remember that frequency modulation schemes (for commercial US FM radio) start by modulating a 10 MHz carrier and up-convert from there.

You would probably have to up-convert it in more than one step to keep the filters manageable.

IMO, my DDS idea is attractive for someone who doesn't already know how to design a frequency modulator, and isn't trying to make a commercial product.

The way it would work is that you start by figuring out the voltage to frequency rule you would like to use (e.g., perhaps a +1V input pushes the frequency up by 50 kHz) then figure out what this means arithmetically for the frequency tuning word (FTW) of the DDS, and you are most of the way there. You just sample your analog input at some reasonable rate (40 kHz?), then for each sample, you calculate a new FTW and load it into the DDS.

If you are only going to use an 8-bit ADC, then a lookup table would be an attractive way to handle the input voltage-to-FTW conversion. For music broadcasting, you would have to go to more than 8-bits, and calculating the tuning word would probably be the way to go.

I wonder if Analog Devices would pay me to design one of these things using all (or mostly) Analog Devices products and publish the design on the web?

Seems like you could connect an ADC and DDS to a DSP plus maybe a little glue logic, and you would have a working system.

;-)

regards, Mac

Winfield Hill skrev:

Analog Devices has an app-note on making a stereo fm transmitter with a dsp and a DDS. The DSP takes audio samples and calculates frequency words for the DDS. The input frequency for the DDS and the frequencies programmed such the the first alias falls in the FM band i.e. 125MHz DDS programmed for 35MHz, first alias at 90MHz

-Lasse

You're right. I am on a local AM radio station fairly often to discuss Veteran's needs, both on the local and national level. AM is more suited to local news and talk than FM, and this has always been the case. It just seems odd to not be the broadcast engineer, after so many years. :(

So, you want to modulate a 10 Mhz signal +/- 75 KHz and upconvert it, then try to remove the unwanted image 20 MHz away? Good luck.

The two common methods are a phase modulated crystal oscillator followed by a chain of frequency doublers, This is usually a *12 or *24 multiplication to get the needed modulation index and reduce distortion.

The second method is a PLL that has the VCO tuning voltage modulated by the audio to produce FM.

One final point is that you need the proper pre emphasis network for the audio. What you use depends on the system used in your country.

Why not do a search for fm broadcast transmitters and see how its done? There are a lot of good schematics and other information online, if you'll just look for it.

.......... Phil

I read in sci.electronics.design that Michael A. Terrell wrote (in ) about 'DS1077', on Sun, 25 Sep 2005:

With modern filter design methods, it's not so dreadfully difficult. But far from trivial.

So very 20th century. (;-) Marconi FMQ system.

That's the modern way.

Compared with getting the FM right, that's trivial. You don't even need it if you accept a usable rather than a very good signal-to-noise ratio. It's only a treble boost circuit, +3 dB at 3183 Hz or 2122 Hz. Even if you pick the wrong one, it will just sound a bit peculiar.

......... Phil

Well, I would up-convert it in steps. Like I said, I've never designed an FM transmitter.

How is channel selection handled? I mean, they don't use a different crystal for each station on the dial, I am sure.

Right, and stereo encoding.

The more I read this thread, the more I think the DDS + DSP or uController is a good idea. Channel selection would be done by just choosing the correct base FTW, and the modulation would be done by adding or subtracting a delta (which would not depend on channel) to the base.

--Mac

True, but there are still a lot of the old transmitters in use that use this method.

Its rather ironic that most applications for PLLs require extra care to remove FM and phase noise by putting traps and filters in the tuning voltage line.

The brute-force DDS approach. SFAIK all of Analog's 400MHz DDS chips, which can create the FM frequencies directly, are programmed by means of a serial digital-data input. For example, the AD9953 has a 32-bit FTW, or frequency tuning word. The FTW must be proceeded by an 8-bit instruction address, which means 40 clocks are required per sample. To achieve a say 48kHz sample rate, the serial-control bus clock must be faster than 1.92MHz. This is below the chip's 25MHz control-bus spec, so a 48kHz brute-force FM-modulator approach should work, if the controlling uP with its A-D or data source can keep up with the task.