Need oscillator circuit

[yllawwally]

What would be a good way to get an oscillator at 42.954540MHz?

[Chris Smith]

Crystal

[russlk]

CMOS logic makes a simple oscillator. Check out the 74HC4060 or any of the CMOS frequency dividers. The data sheet should show how to hook up the crystal, then you can use any inverting logic gate to implement the oscillator if you don't need the divider.

[yllawwally]

Thanks looks like the 74HC4060 will be perfect, with a 3.579545MHz crystal.

[Chris Foley]

(NOTE: TYPO ALERT -- I called out A CD74ACT161 as a CD74HC161 below. Typo corrected -- sorry. HCMOS chips don't go that fast -- counter chips, either)<p>I'm not sure you're heading in the right direction here. The 74HC4060 has a couple of pins that can be used with either an RC or a crystal as an oscillator, but even the HC version isn't good for 42 MHz for the oscillator or divider input. Also, if you use a 3.579545 MHz crystal as the input to the divider, the possible output frequencies will be lower instead of higher frequency.<p>'HC4060 Data Sheet.pdf<p>If you can live with 40 MHz, TTL-compatible oscillators are widely available which are a one part solution. Just apply 5V/3.3V/whatever, and voilla! you gots your frequency. Rock solid stable, usually with controlled rise and fall times for reasonable loads and little or no ringing.<p>You might want to try holding on to that TV crystal, though, if you need that exact frequency. Fortunately, 42.954540MHz is exactly 12 times the frequency of the TV crystal. So, you might want to set up your crystal with an inverter, a resistor and a couple of small caps as a 3.579545 MHz oscillator, and feed that into a PLL paired with a counter IC set as a divide-by-12 counter. Both of these concepts are covered in detail in Don Lancaster's CMOS Cookbook, but with the 4046 PLL. It's available in libraries, and also at his website. You need to own this book. It's the perfect introduction to digital logic.<p>Don Lancaster's Web Page<p>The only problem is, the standard 4000-series CMOS covered in the Cookbook won't get you up to 42.954540MHz. In order to go there, you might want to use the TI CD74ACT279 digital PLL IC, along with a CD74<act>161 presettable counter chip. They will be fast enough at 5V to handle the frequency of interest.<p>Texas Instruments CD74ACT297 Data Sheet.pdf<p>If you're doing this as a hobbyist-type setup, whatever ICs you use, you might want to try the "dead bug" method of construction. Glue the chips upside-down on a small piece of copper-clad FR-4, and use the copper as a ground plane. You can tack-solder multi-layer ceramic .01uF SMT caps directly between the Vcc pins and the ground plane for bypass on each IC. Use 30AWG solid for point-to-point wiring, and keep the high frequency output from the PLL to the counter as short as possible. Use an inverter/buffer to pick off the high frequency output, to isolate the PLL from the capacitance of your signal wire. Get a _good_ 10uF SMT tantalum cap for bypassing where the power is applied to your board. If you use an electrolytic, you're doomed. If you do it this way, you have a good chance of getting it right the first time. Don't try to debug this setup without having a good scope and a X10 probe matched to the scope, both rated for at least 100 MHz. Oh, yes -- use a ground strap at all times. If you nuke it, you'll know it right away, but trying to figure out why it's kind of wrong is something you wouldn't wish on anybody.<p>Happy hunting. Feel free to email me if you have any questions.<p>[ August 12, 2003: Message edited by: Chris Foley ]<p>[ August 17, 2003: Message edited by: Chris Foley ]</p>

[yllawwally]

You're right that chip will only divide. I just wasn't thinking. The project hooks up to a tv, so I think it needs to be pretty accurate. I couldn't find a crystal an a multiple of the frequency I need, other than lower frequencies.

[rshayes]

Since you want a frequency 12 times the color subcarrier frequency, the 3.579545 MHz crystal is probably the easiest and cheapest to get by far. Multiplying this by twelve can be done with a phase locked loop. This is a little complicated, but the divide by 12 needed can probably be absorbed into your other logic. The 45 Mhz oscillator can be an LC oscillator with a varicap to allow frequency control. One or two transistors can do this. A phase detector is needed, this can probably be part of a 74HC4046. The classic phase detector was the MC4044, but this is a TTL device, and may no longer be available. The phase detector operates at 3.58 MHz so speed should not be a problem. A basic ap note on phase locked loops is Motorola AN-535 (www.motorola.com). Another basic ap note was published by Signetics, which was bought by Phillips. I don't know if this one is still available. The MECL Data Book (www.onsemi.com) may still have some phase locked loop information.

[Chris Foley]

Maximum VCO frequency for the 'HC4046 is 12 MHz at TTL voltage levels. Plain HCMOS just doesn't go there. If anyone else has a better PLL chip that'll do 42MHz, I'd like to know.<p>[ August 14, 2003: Message edited by: Chris Foley ]</p>

[rshayes]

The phase detector only has to work at 3.58 MHz. The oscillator is at 45 MHz and does represent a problem. Voltage controlled oscillators can be built using either a multivibrator circuit or an LC circuit with a varicap.<p>The classic multivibrator part was the MC4024, which may be obselete. I think Texas Instruments may also make a couple of multivibrator type parts. Its hard to get much over about 20 MHz out of these parts, because the capacitor starts to get smaller than the stray board capacitance.<p>The classic LC part was the MC1648. This was an emitter coupled logic part. This meant that the output was not at either TTL or CMOS levels and some kind of level shifting was necessary. The tuned circuit was also on an internally set bias voltage, which had to be allowed for as it represented a fixed bias on the varicap. I believe that the MCF1648 is also obselete, but there are other similar parts in production.<p>This may be heresy these days, but I have had fairly good luck with a hartley oscillator using a single transistor. The coil can be a few turns on a powdered iron or ferrite toroid. A second transistor can be used to buffer the output and convert the signal to logic levels.

[Timothy Rasch]

Hi ,I am an electronic technician ,I see in Digikeys catalog in the cmos/ttl oscillator [crystal osc.section] there is a 40 mhz part# [se1907-nd];42mhz part# [se1908-nd]available as cosmos as a 4 lead dip package [sg-51ph]$3.00 each.[complete osc.&crystal] You can search their recent catalog at www.digi-key.com for more information. You can also write me at [email protected] for more ideas.If you can find Digi-keys programable crystal osc. section you might find another alternative to your problem.

[yllawwally]

It looks like SG-8002DC-STM-ND, from digi-key will be a suitable device. They will program it before they ship it out. So maybe I finally found a good solution.<p>http://rocky.digikey.com/scripts/Produc ... 8002DC-PWT