4046 Phase Lock Loop dilemma

h-tone · Joined: Mar 23, 2006 Posts: 8 Location: Minneapolis

Hi. I'm new here. Thought I'd introduce myself by unburdening my horrible problem onto people who know more than me about these things Very Happy

I'm trying to make a sort of sampling oscillator based on the 4046 that can tune into an input note. I also want the option of the oscillator locking in on the input note and droning it forever, even if the input changes. Inspired by the Boss DF-2 Distortion Feedbacker.

Problem is the lock indicator (made of 2 NOR gates, resistor, diode, capacitor attached to pins 1 & 2 of 4046) that hits a latch ( 2 NOR gates) that opens an analog switch at pin 13 that freezes the oscillator.
Too much false triggering-- it locks onto notes well outside the input note. I don't understand the lock indicator circuit (the only one I could find in books & internet) enough to clean it up or design a different one. Anyone???

I suspect the secret may lie in the DF-2 schematic, which I cant understand barely at all.

Cheers.

blue hell · Posted: Thu Mar 23, 2006 6:02 pm Post subject:

h-tone.

Could it be that your problem is with low signal levels ?

I can imagine it to be necessary to have the lock detect to be valid only above a certain minimal input signal amplitude. But what do I know ... it's more than 20 years ago I did stuff like this Smile

v-un-v · Posted: Fri Mar 24, 2006 2:03 am Post subject:

Jan's right- but expect no miracles either. The 4046 was never ever intended to be a guitar tracking synthesiser. Also you need to try and isolate a just one guitar string as much as possible- otherwise you will get mayhem and mayhem only!!

hope that helps Smile

oh yes...like Jan mentioned.... welcome

to electro-music.com Very Happy

bugbrand · Posted: Fri Mar 24, 2006 4:18 am Post subject:

Ideas for audio tracking::::

Yeah, like Blue Hell said, getting the signal level right is probably important.. You may be best to square up the input signal - turn it into a square wave 'cos thats what digi 4000 chips want right?? ((maybe I don't know exactly what the 4046 wants..)) -- use a comparator for that -- and, once you've got the audio to sqaure then you can use standard clock dividers to give suboctaves. RIpping audio!

Yeah, treating the audio -- for these sort of applications you need to get the fundamental frequency of any note - so that means single notes and try to cut all the high freqs from them.

Another thing I wanna look at is the Analog Devices AD650 Freq to Voltage and Voltage to Freq converter (..you can get free samples from www.analog.com) -- maybe this would be a useful thing to figure a circuit for - I've got a couple of samples, just need to find the time (..recurring theme - time time time its never enough!!!)

h-tone · Joined: Mar 23, 2006 Posts: 8 Location: Minneapolis

Thanks for the replies...
I've got the input signal pretty well squared out, using Craig Anderton's Superfuzz design that has served me well for other octave divider (analog signal meets digital IC) stuff I've built-- that with a tie-up resistor before the final capacitor makes it pretty much go rail to rail. Tracking really isn't a problem now. Problem is in the lock detection circuit-- it's really just meant to light up an LED for visual confirmation that the circuit is in lock, so a couple of
false sputters aren't a big deal. But in my use as the switch for a sort of sample and hold it won't work (but I can do it by ear and detach it at the right time to sample the correct note).
Even if I can't figure this feature out it will still be pretty cool without it I think--
my idea is to have at least 3 oscillators that are modular so they can track an incoming note and output that note or an interval above it, or have the oscillators track each other back to back, or any combination thereof, with the tracking rate able to be set so you can hear it sliding up and down searching for the note or make it almost instantaneous. It is still a little dirty with a guitar in, but with keyboards in it works pretty good.

blue hell · Posted: Fri Mar 24, 2006 1:54 pm Post subject:

Ok .. then the next thing I'd look into would be a spike filter on the lock output, starting with a simple RC filter and then see what the nature of the failing is.

I'm not an expert in this but an inherently stable lock detector seems difficult

As far as I understand it looks at the phase difference between incoming and locked signal, so evey a-periodic edge in the input will give a lock-off spike (which then will immediately be followed by a lock-on probably), and you'll have a phase glitch in the locked signal then resulting from a voltage jump for the vco control signal.

What could also help is to filter the output voltage (but probably not the VCO control voltage, as having large time constants in the feedback loop will make the output jumpy).