multi-phase oscillator

[puretube]

...people are even still patenting tube circuits nowadays;
...be assured I`ve been researching for a long time;
...don`t want no digital spikes anywhere, that might cause clicking/zippering;
...don`t wanna need filtering;
...want it utmost simple and superbly featuresque!
(haven`t found the latter anywhere).

[puretube]

http://focus.ti.com/lit/an/sloa060/sloa060.pdf

[puretube]

one of the most beautiful schemos ever:
http://v3.espacenet.com/origdoc?DB=EPODOC&IDX=US2017708&F=0&QPN=US2017708#


another nice one:

http://v3.espacenet.com/origdoc?DB=EPODOC&IDX=US2791694&F=0&QPN=US2791694

[zachary vex]

they both open as complete jibberish for me!  8^(

[puretube]

http://www.pat2pdf.org/patents/pat2017708.pdf

http://www.pat2pdf.org/patents/pat2791694.pdf

better?

[gez]

The other answer is - get a PIC. A PIC can output you multi-phase LFOs directly in PWM format, as well as reading the speed pot.

Apologies for resurrecting this old thread, but analogue techniques for doing what I want to do are getting a tad complex. 

Trying to get my head round the above.  Are you saying that 4 outputs of the PIC are used, each one giving a PWM output?  If so, does this work along the lines of a Class-D amp, i.e. a simple LC filter at the output blocks the carrier frequency and the average DC is developed across the cap??

[gez]

Anyone want to take a stab at it?

[The Tone God]

Trying to get my head round the above.  Are you saying that 4 outputs of the PIC are used, each one giving a PWM output?  If so, does this work along the lines of a Class-D amp, i.e. a simple LC filter at the output blocks the carrier frequency and the average DC is developed across the cap??

You can get uCs with four PWM outputs. Each PWM output is, in general, controlled by a timer. All it would take is the proper manipulation of the timer(s) so each output would represent a phase. Fairly simple to do.

Andrew

[no one ever]

ah hem. NOW is this what we think it is? (is this thread ringing any bells)

[Sir H C]

To get 90 degree phase shifts from an oscillator, you can either do a phase shift oscillator with 4 sections but this often has multiple places where the gain can cause oscillation, or you can look at the fun quadriture oscillators as are used for most modern mixers.  You just need an all pass network that can create the 90 degree phase shift over the needed band, and then you can use the signal and the 90 degree signal and with inversions get all the signals you need.

This is how we do it for RF to get image reject mixers, so I think it can work for audio frequencies.

[TELEFUNKON]

IIRC, gez posted a 4phase triangle oscillator himself, a while ago?

[gez]

You can get uCs with four PWM outputs. Each PWM output is, in general, controlled by a timer. All it would take is the proper manipulation of the timer(s) so each output would represent a phase. Fairly simple to do.

Thanks for the reply Andrew.  I'm still finding my way around PICs so it'll take a while until I'm up to scratch! 

How do you convert the PWM outputs to something (resembling?) analogue?  Is this like the class D approach as I outlined in my earlier post?  Thanks.

[gez]

IIRC, gez posted a 4phase triangle oscillator himself, a while ago?

Indeed I did Mr Funk-on, but I need a rectified sine in quadrature.  Number of ways of doing this in analogue but the parts count (and with some approaches the IC count) is a tad high for my liking. 

I've already found one approach that uses weighted resistors at the output of a PIC to get a stepped waveform - this has a lower parts count than doing the stepped thing with CMOS chips - but I'd prefer something a little smoother.

Just did a spell check and apparently I should be called fez...

[TELEFUNKON]

gez: did you get the PM?
the forum crapped out again right after writing...

[Nasse]

http://ww1.microchip.com/downloads/en/AppNotes/00538c.pdf


Found this when goooogling around, dunno if this kind of principle (around 4015 cmos chip) could be used for lfo but cant see why not
(tried to find some shift register/resitor matrix like circuits but did not find any better, i have some circuits but could not find those just now)
http://home.att.net/~theremin1/Circuit_Library/wheel.htm

I wish luck for your projects

[gez]

gez: did you get the PM?
the forum crapped out again right after writing...

Yup, thanks.  I've replied. 

[gez]

http://ww1.microchip.com/downloads/en/AppNotes/00538c.pdf

Ah, thank you Nasse!  It obviously is along similar lines to the Class-D approach then. 

Question.  With these PICs that have PWM output.  Do they PWM a waveform presented to the input (via built in A-to-D converter) or does one program the chip to produce the desired waveform, which is then outputted in PWM format?

[The Tone God]

I've already found one approach that uses weighted resistors at the output of a PIC to get a stepped waveform - this has a lower parts count than doing the stepped thing with CMOS chips - but I'd prefer something a little smoother.

That would be a varition of DSS (Direct Signal Synthisis). Interesting stuff but if you are doing it with discreets then it eats up alot of outputs.

How do you convert the PWM outputs to something (resembling?) analogue?

Usually the common way is to stick on a simple RC low pass network. Depending on the application sometimes you can even get away without needing the filter. The filter should be tunned for the frequency of the PWM output.

Question.  With these PICs that have PWM output.  Do they PWM a waveform presented to the input (via built in A-to-D converter) or does one program the chip to produce the desired waveform, which is then outputted in PWM format?

The PWM is just a way of generating a variable output voltage. It is done in the software/hardware with no relation to an input unless you program it so. So it closer to the later.

Andrew

[gez]

Thanks for your help Andrew, much appreciated.  Any chance of giving me the code of one of these PICs that can do the PWM thing, so I can look at the datasheet and see what's in store? [/gulp] 

[gez]

Incedentally, the PIC stepped approach is outlined here (about half way down the page):

http://www.ke4nyv.com/picprojects.htm