convert sine wave to other wave forms

It's usually easier to start with the triangle and go from there. A PLL can fix up the frequency.

Assuming you have a clean sine wave, you can easily generate the square wave with a "suitable" comparator. You can get the triangle with an integrator and amplitude leveling loop(s).

Be aware that you need 3X the output bandwidth to get ANY difference at all. 5X is better. And that's without significant phase shift so the harmonics sum properly. I expect you're doing all this with chips and wire bonds. Calculate how many nanohenries of series inductance you can have in your integrating cap and how much C you can have in your current switch.

I've assumed that you really want triangles and squares at 1GHz. If you don't care that they all look like sinewaves, your problem is much simpler. mike

--
Return address is VALID but some sites block emails
with links.  Delete this sig when replying.
.
Wanted, PCMCIA SCSI Card for HP m820 CDRW.
FS 500MHz Tek DSOscilloscope TDS540 Make Offer
Wanted, 12.1" LCD for Gateway Solo 5300. Samsung LT121SU-121
Bunch of stuff For Sale and Wanted at the link below.
MAKE THE OBVIOUS CHANGES TO THE LINK
http://www.geocities.com/SiliconValley/Monitor/4710/

Do you mean that your DDS generates sine-weighted digital data, without a DAC? If so, what's the data clock rate... 2 GHz?

What DDS is it? What are the logic levels?

Why do you want sine/square/sawtooth/triangle waveforms at 1 GHz?

John

You do realize, don't you, that you will have to have at least 3 to 5 of the first odd harmonics for reasonable fidelity (low THD)? This translates to 7-11 GHz for a 1 GHz fundamental. And this holds regardless of how you choose to create the non-sinusoidal waveform. Your easiest solution is probably to wave shape starting with a triangle waveform. "The Art of Electronics" has a circuit for doing this using a diode shaping network. The disadvantage of this circuit is that it works properly with only 1 signal amplitude.

How many bits of resolution and linearity do you want on those other waveforms? Will 16 do? And will it be sufficient to have amplitude limited to 100V into 50 ohms?

Thanks for the replies. I am absolute a beginner in circuits. I am trying to build a frequency generator since it is cheaper than buying one. This freq. generator will be used to generage RF freq. in the

930Mhz bandwidth. Here is the info for the analog.ocm AN9858 DDS chip I plan to use.

Can this DDS chip produce sine waves at upto 1Ghz? Can someone point me to the right direction?

Thanks,

george snipped-for-privacy@yahoo.com wrote:

able

of

generate

Hi, George,

This chip includes a dac, but it can only synthesize frequencies up to Fclk/2 in theory, somewhat less in practice. Figure maybe 400 MHz sinewave with a good output filter.

Frankly, designing a good 1 GHz sinewave oscillator is not for beginners, sort of like tackling Everest alone on your first climb. You might consider some used gear on ebay or something. Non-sine waves at this speed are *seriously* difficult.

John

The Analog AD9858 DDS chip has a max of 2,000MSPS speed when using a divide by 2 thing. So, does this mean that the maximum sine wave frequency is 2,000M / 2 = 1GHZ Sine Wave?

I am looking for a DDS that can output upto a 1GHZ Sine Wave. Does anyone know of a chip that can do this?

Thanks,

John Lark>

buying

chip

point

to

waves

[top-post re-arranged]

No. The input clock can be up to 2 GHz, but when it is, it is mandatory to use the divide by 2 feature, which slows the internal clock back down to 1 GHz. I have used that chip, but I am not re-reading the datasheet right now. So if you think I am wrong, please quote me chapter and verse from the datasheet.

And the 9858 is the fastest DDS that I know of. I don't think it is much good beyond 400 MHz output.

You should explore the possibility of using frequency multipliers and/or mixers to attain your goals. For example, if you clock the DDS at 1 GHz, you can mix the 1 GHz with the DDS output to obtain either 1 GHz - Fdds, or 1 GHz + Fdds, depending on how you filter the output. Fdds is the DDS output frequency.

The DDS chip actually has a mixer built into it, but I think it is set up for down-conversion only. So you probably want to find a mixer elsewhere. Realistically, if you want to operate somewhere near 930 MHz, you should probably operate the DDS at, say, 700 MHZ, then output 230 MHz, and upconvert by mixing the 700 MHz with the DDS output to obtain 930 MHz and 470 MHz. A high-pass filter should get rid of the 470 MHz. Note, however, that I am not really an RF guy, so...

Also, please don't top-post. It forces others to read the thread out-of-order.

--Mac

Read the data sheet, for Pete's sake.

John

'I am an absolute beginner...'. There is a reason a 1 GHz generator is expensive. I suspect you will find that even at minimum wage rates, this is going to eat some serious time for a beginner (lots less for a Larkin but still significant). A good lesson I learned a long time ago is-- Don't re-invent the wheel unless you can REALLY improve it. Great buys sometimes on eBay.

GG