| Electronics > Projects, Designs, and Technical Stuff |
| Generating High Frequency Triangular Wave. Which way is the best? |
| (1/1) |
| syntax333:
Hi, I am currently working on a project which I need to generate high frequency triangular wave to modulate CVBS signal (PWM). CVBS signal has max. frequency of 3.57MHz thus I need to pick triangular wave frequency of 3.57 * 5 (Higher than 5th harmonic) ≈ 18MHz. I am planning to compare CVBS signal with triangular wave to get PWM signal at output. However, I don't know which way of creating triangular wave would be appropriate. I listed the ways of creating triangular wave which I found online below. I have no experiance with the listed way. So can you give me your opinions about it. Also if you know a better way I would be really appreciate it if you share with me. * Using AD9102 DDS Waveform Generator chip. (180MHz, has cut-off approximately at 90MHz, 90MHz/5 = 18MHz for triangular waveform) * Using Crystal or oscillator to generate sinusoidal with high frequency then using comparator to get square wave then converting it to triangular wave with op-amp integrator. |
| DaJMasta:
Is this something on the bench or for inside a product? On a bench, an arbitrary waveform gen is generally going to have the bandwidth you're looking at and should be a fine choice. Integrated into something, the integrator method would be a good choice, but some sort of DDS or digital generator may be a better choice for stability/consistency over many units. |
| duak:
What amplitude stability do you need? If the frequency doesn't vary significantly, then using an integrator driven from a voltage reference through analog switches will have good stability. The amplifude of the waveform is inversely related to the frequency and will decrease as the frequency increases. You'll have to find an op-amp for the integrator that has the bandwidth and slew rate and is unity gain stable. There are tri-wave generators that use switched current sources to charge and discharge a timing cap. This article shows the general idea: https://www.planetanalog.com/function-generator-circuit-concepts-part-2-current-source-function-generators-fgs/ It is self oscillating so if you have a square wave frequency reference, you will have to break the loop and drive the current switches from the frequency reference. |
| T3sl4co1l:
Modulating, to what end? I've never heard of composite being PWM modulated before, and I can't think of any obvious applications... If you're recovering baseband after filtering anyway, it will surely be easier to simply amplify it, or use a more traditional RF modulation if transmission over any distance is required? Also, NTSC is closer to 6MHz BW, 3.58 is the colorburst frequency which basically separates luma (below) from chroma (above). And after a DSB mixing process (like PWM), you need to consider twice the bandwidth (since the LSB extends down from the PWM carrier by the same BW), which means for a 6MHz BW you need a carrier over 12MHz, and that's with a "brick wall" filter. (Probably, 18 would be tolerable; a filter cutting sharply between 6MHz and 18-6 = 12MHz, with low group delay, will require a high order type though.) Tim |
| David Hess:
The second example Duak linked is the highest performance analog method and can achieve bandwidths of 100s of MHz relatively easy. The integrator method might be possible now with fast operational amplifiers but the requirements for unity gain stability and high speed conflict. I know a truly marvelous way to make an integrator using a current feedback operational amplifier which this post is too small to contain. |
| Navigation |
| Message Index |