Post Go back to editing

AD725 oscillator options


I want to use AD725 and made circuit to automatically detect NTSC and PAL, so that after that it will feed correct crystal oscillator to AD725.

I have 2 options:

1- get 2 oscillators, feed them through SPDT analog switch and make the circuit activate either one automatically.

2- as seen in this image. I got LTC1799 and made 2 resistor lines according to its datasheet to have 14.318180MHz or 17.734475MHz (please check if my calculations are correct). Here this part will deliver the correct frequency instead of having 2 oscillators now I got only one. Still going to use the analog switch as seen.

My issue is that option 2 is expensive and I would like to have option 1 or anything cheaper than option 2. Most oscillators here are 3.3v and most of them are out of stock.

I selected these 2 oscillators:


are they good? they are 3.3v and do not go to 5v, are they good enough with ad725 or the 2nd option is better?


  • FormerMember
    0 FormerMember
on Nov 22, 2021 12:23 PM

The LTC1799 is a good part however it is not designed to generate color burst frequencies.  Color burst frequencies must be stable and accurate, else you might get color artifacts in the final image.

  • so what do you suggest? LTC1799 was my best attempt to avoid crystal oscillators. Kindly specify which crystal oscillator I should use so I can find them on digikey. I posted 2 oscillators one for ntsc and another for pal, kindly check if they are good or not.

    also, did you see my analog switch? will it be ok with passing the clock signal from oscillators? no other way to do the switching.

  • FormerMember
    0 FormerMember on Nov 23, 2021 9:36 PM in reply to VEGETA

    The AD725 needs a TTL input crystal source of 4 times the color burst frequency of the output format.  As shown in the data sheet this would be 14.318180MHz or 17.734475MHz.  

    Normally I would use crystal oscillators.  The AD725 is designed for +5V rails.  I can't find 5V oscillators for these frequencies.  However there are plenty of 3.3V oscillators available and they'll drive the AD725 just fine, just need a 3.3V rail besides the 5V rail.  

    Search for oscillator with Function=enable/disable, voltage supply=3.3V, Part Status=In stock, Frequency=14.31818MHz and base resonator=crystal

    Do the same except for 17.734475MHz

    I'm seeing stock problems on the PAL oscillator.  You can use the enable/disable pins on the oscillator to switch between the 2 frequencies.

    In the data sheet there's a low cost crystal oscillator circuit, Figure 20 which use readily available crystals from digikey.  Just might take a bit of cap tuning to get them to oscillate at first.  You can change U1B to a nand gate and use it as an enable/disable switch.  Any of the Epson FA-238 or any equivalent series will work

    Watch availability, it seems to come and go on most any of these parts.  check availability over manufacturer.

    Do not use MEMS oscillators, they have poor jitter and temp tolerance.  Always go with a crystal or crystal based oscillator.  Do not use FPGA oscillators even if you could get the right frequency.  They have similar problems that MEMs have, especially jitter.

    No analog switch needed.  You do need stable 4FSC so this leads to a crystal somewhere

  • Hello,

    I have the following:

    NTSC: < which is the previous item. has enable feature (called tri-state) and also 3.3v crystal. this is digikey marketplace, i don't prefer this. < from QTM, same with enable feature as well. till now, this is the to be used one.

    PAL: << MEMS, could not find crystal one.

    here is a search:

    it shows 17.73447 and 17.734475 and 17.734476 which I think all can do it right? still, even with being crystal, they are all 0 quantity. the only ones available are MEMS type.

    can you tell me what is wrong with LTC1799 exactly? I posted my circuit with resistors which could deliver the needed frequencies. please take a look at it

    I see no suitable PAL oscillator till now and I won't go to the route of figure 20 since it needs manual adjustments which is a no-go for me unfortunately.


  • I have used these parts:

    ASE-14.31818MHZ-LC-T for NTSC -> 3.3v crystal-based.

    SIT8208AI-82-33E-17.734475 for PAL -> 3.3v, mems based. there is no crystal based one. this one says it is suitable for video in its datasheet.

    please check the circuit for enabling them here:

    circuit operation: both crystal oscillators EN\OE\ST are tied to ground via a pull-down resistor of 10k to make sure both are off. then analog switch chooses which one to activate based on "STND" signal. if STND is ON, then NTSC is the way to go and vice versa. Here analog switch won't affect the circuit since crystal outputs are directly tied to AD725 pin... the 2 crystal oscillators outputs are tied together but I don't think this will cause anything since one of them will on and the other is hi-z

    this is the best I could do besides LTC1799 which you didn't recommend.

    what do you think?

    notice that I can't do any manual adjustments or so on.

  • FormerMember
    0 FormerMember on Nov 29, 2021 10:51 AM in reply to VEGETA

    The color burst frequency must be very stable and accurate to work correctly.  MEMS and adjustable devices like the LTC1799 or FPGA oscillators will have problems with this.  This is why I recommended crystal based sources to drive the clock inputs.  Please look again at the simple oscillator circuit in the data sheet.  IT is the lowest cost version with available parts in stock.  

    Tying two oscillator outputs together will work fine as long as one of the outputs is in Hi-Z mode.

    The LTC1799 is an excellent part to replace oscillators that drive uP systems where absolute accuracy and stability is not as important as needed for the color burst frequency.  The LTC1799 uses resistors to set the frequency and delta frequency at best is 1.5%.  This 1.5% plus resistor tolerance of 1% do not meet the accuracy needed for color burst frequency.

    The SiT2008 might work well, I've never used MEMs devices however the DS does show the accuracy needed.  Interesting they don't list the 14.38Mhz one for NTSC, it is a programmable part.

  • ok so I found this crystal:

    and this inverter IC:

    inverter IC doesn't have enable pin so only option to turn it off is to switch off its power pin itself.

    are they good?

    I still think the mems one i posted is good.

  • FormerMember
    0 FormerMember on Nov 29, 2021 3:02 PM in reply to VEGETA

    Crystal is good

    HC04 is good

    Another way to do this is create 2 crystal oscillators as shown in Figure 20 from the data sheet using crystals and instead of in U1B gate use this circuit.

    This implements a digital switch between 14 and 17 MHz based on whether SW is high or low.  Probably the cheapest path overall.

    Or implement the oscillators directly from the data sheet and use the analog switch from before.  

    MEMs is good.  I'd like to see the PAL color matching when you get it working.  Again I've never used a MEMs device for color burst.

  • hello,

    your circuit seems simple and very cheap.. this way I can use 2 crystals with one inverter chip since the one I posted has 6 inverters in it, will be using 4 to deliver the 2 oscillator outputs. then one of these NAND gates:

    to activate the desired input.

    kindly check the newer circuit which has a physical SP3T switch to have these options: auto, ntsc, pal. auto is activated by "STND" signal as seen which comes from ntsc\pal auto detect signal, while the other 2 switch options are for forcing ntsc or pal in case auto circuit didn't work. ntsc being 3.3v and pal being 0v.

    newer circuit:

    so here I can remove the analog switch, then take sp3t output signal and deliver it to the NAND input as seen in your circuit.

  • FormerMember
    0 FormerMember on Nov 29, 2021 3:40 PM in reply to VEGETA

    The SP3T looks OK.  The link still shows the analog switch.  You can remove analog switch and replace with the nand circuit.  Run through the logic of the nand circuit is understand how it works.  

  • hello,

    well, i decided to stick with that mems for pal and make a simple prototype and see what happens. i will inform you with results and if we need to proceed to do the other suggested circuit.

  • Reply Children
    No Data