PLL question, will a 4046 clocked at 512Hz be able to reliably generate 131,072Hz ?

[blueroomelectronics]

I've not needed to use PLLs in the past but I need to multiply a 512Hz TTL clock (from a PIC I/O pin) to get 131,072kHx (256x) will a 4046 or 74HCT4046 clocked at 512Hz be able to reliably generate 131,072kHz ?
Are there simpler methods? Frequency doublers?

 

[ronsimpson]

512 x 256 = 131072hz? 131072khz?
A 4046 of any type can work at 131072hz. You will need a divide by 256. (8-bit divider)
The time constants in the feedback need to be slow.
Should work fine.

 

[blueroomelectronics]

Thanks Ron, got a comma where a decimal should have been. Once a PLL has lock does it stay locked if the input frequency is constant? I may have made an error in my calculations and may have to use 256Hz x 512PLL to get the 131,072Hz I require. What happens to PLLs at higher multiplication factors?

 

[Roff]

You need to use phase detector II, which guarantees phase AND frequency lock. For lowest jitter, use a high-quality (low leakage, low dielectric absorption) capacitor such as polypropylene or polystyrene in your loop filter. Keep in mind that your VCO will be running basically open loop between positive transitions of your input signal (512Hz). RC oscillators inherently have more phase jitter than LC (better) or crystal (best) oscillators. If you can live with a little jitter, you should be OK.

 

[blueroomelectronics]

Are there better PLLs than the old 74HC4046? And since it's a power of two are their any frequency doubler circuits that work well? How does one toss a crystal on the 4046? I have a very accurate 20MHz OCXO but I need an equally accurate 131,072Hz. I can divide the 20MHz 78125 times to get the 256Hz but can't think of another way to get the desired 131,072Hz clock.

Hmm jitter, I want my cake and eat it too

 

[ronsimpson]

Are you just trying to make 1331khz or must it be 'locked' to 512?
Look at the family MC14515x and it's many other brothers.

 

[blueroomelectronics]

I'll check the MC14515 out. It just has to be 131... KHz duty cycle not important but accuracy is. It will become the master sampling clock so jitter might be a problem.

Is it the MC145106 PLL?

 

[Roff]

I'll check the MC14515 out. It just has to be 131... KHz duty cycle not important but accuracy is. It will become the master sampling clock so jitter might be a problem.

Is it the MC145106 PLL?

Ron's question is important. Does your 131,072Hz clock need to be phaselocked to your PIC clock, or can it stand alone?

 

[blueroomelectronics]

Sorry about that. It can standalone, it will become the master sampling clock.

 

[audioguru]

I tried a 74HC4046 in my sine-wave generator. It sync'd the variable output frequency to the clock for switched-capacitor lowpass filter ICs which is 128 times higher. I used phase-detector-II for its wide frequency range.

The output sounded and looked fine but the jitter caused distortion to be about 0.5% instead of the 0.005% without the PLL.

The phase-detector-II suddenly detects the frequency too high and forces it lower, then detects it too low and forces it higher. The jitter is less when the loop filter takes all day to change frequencies.

 

[Roff]

Sorry about that. It can standalone, it will become the master sampling clock.

But it's not sampling something locked to the PIC clock? Were you just going to use the PIC as a frequency divider for a stable source?
Can you tell us what you are sampling, without having to become a serial killer?

 

[ronsimpson]

Just for fun I pulled up Excel and Digikey (citizen oscillators) and did some math.

Oscillator divide by
19.6608mhz 150
16.384 125
13.5 103
11.2896 86
You could run the CPU at 19.6608 not 20mhz and add one IC to divide by 150.

 

[Roff]

The 4046 type II phase detector has a small dead zone, which causes phase "hunting" (jitter). The HC/HCT4046 is better in this respect than the CD4046, because of the faster internal logic circuitry.

 

[blueroomelectronics]

It's not top secret, it's a timebase for sampling four 24 bit A/D converters. OCXOs are expensive but their accuracy meets the spec required in the design. I could use 19.6608 and have the oscillators cut (this is something that does appeal to me)
I do need these three frequencies
43690.6666 (32768*4/3)
40000
32768

Any one crystal timebase that's not over 30MHz come to mind? 20MHz was chosen by me because it also is one of the few frequencies that works with USB but I have no problem using a separate crystal for the 18F4550.
The Delta Sigma A/D (an ADS1210) likes 10MHz (20/2) and supports a synchronous start clock ie the three frequencies shown above.
As for the three frequencies it's designed to work with existing gear so I'm stuck with them.

 

[ronsimpson]

I looked at digikey for 'off the shielf' frequencies. That is where 19.6608 came from. They clame there are 25 different kinds of 19.6608 osc. in stock today and 77 different kinds of crystals in stock.
19.6608/150=131,072
19.6608/600=32768
19.6608/450=43690.76

Will USB work at 2% off in friequencies. I think it's OK.

 

[blueroomelectronics]

Actually the USB part is very fussy (it will PLL to 96MHz inside the PIC) and the datasheet shows only about 10 possible crystal speeds. The 40,000Hz is for the sampling not the USB and of course it won't divide evenly from 19.6608MHz to 40kHz that I can figure. I do LIKE 19.6608MHz crystals, and anything else that divides into 9600. I use them in the WASP kit that's coming out soon.

 

[blueroomelectronics]

Well it turns out to be a smaller PLL multiplier with a 9.8304MHz OCXO. I can divide down to 1.6kHz and PLL it back 25x to 40,000Hz I'm assuming a smaller PLL multiplier will be less prone to jitter?

 

[Diver300]

Have you looked at using a DDS IC straight from the 20 MHz clock?

 

[blueroomelectronics]

I'm not familar with DDS I'll google it. Any particular IC?

 

[Diver300]

I've used the DDS ICs from Analog devices (Analog Devices, Inc.: Converters Amplifiers Processors MEMS A/D Converters Analog to Digital Video Converter Temperature Sensors Analog Device RF Amplifiers Differential Amplifiers Digital Signal Processing Thermal Management D to A Converters Microc)

You program them serially, and the output is a current source, and needs some output filtering.