Author |
Message |
Uncle Krunkus
Moderator
Joined: Jul 11, 2005 Posts: 4761 Location: Sydney, Australia
Audio files: 52
G2 patch files: 1
|
Posted: Thu Jan 31, 2008 1:45 am Post subject:
Frequency multiplier idea |
|
|
I'm just wondering if someone with more theory could enlighten me about something.
If you were designing a frequency multiplier to generate 1,2,3,4 octaves up from the input signal: -
Instead of using a phase locked loop with a divider in it's feedback path, (which takes time to lock on, needs to be adjusted, etc, etc.)
Why not,
a) Full wave rectifier
b) Remove DC by decoupling with a cap
c) Low pass filter to take the edges off
d) Amplify by 2
e) return to step a)
It might make sense to put them in a different order, but you get the idea.
This would track perfectly all the time, everytime wouldn't it? You could even divide back down again for odd harmonics which tracked perfectly as well. There must be a reason this won't work. _________________ What makes a space ours, is what we put there, and what we do there. |
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Thu Jan 31, 2008 6:03 am Post subject:
|
|
|
This will work for triangle waves, but leave out the filtering. This works because the negative half of the triangle gets flipped up by the rectifier, resulting in a triangle of double the frequency and half the amplitude.
For sines you could do it using a four quadrant multiplier (or a ring modulator) to square the sine (instead of the full wave rectifier), also leave out the filtering. This works because the square of a sine is another sine lifted up to be positive only and having half the amplitude.
I use both tricks on the G2 occasionally (mostly for LFOs though).
For square waves the PLL probably is best, as any filtering used would make it depend on the actual frequency very much. _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Thu Jan 31, 2008 5:06 pm Post subject:
Re: Frequency multiplier idea |
|
|
Uncle Krunkus wrote: | If you were designing a frequency multiplier to generate 1,2,3,4 octaves up from the input signal: -
Instead of using a phase locked loop with a divider in it's feedback path, (which takes time to lock on, needs to be adjusted, etc, etc.) |
I hear this a lot - what do you mean by "takes time to lock on"? I built a PLL multiplier (http://home1.gte.net/res0658s/fatman/4046pll.html) that does 1x 1.5x 2x 3x 4x 6x and 12x. It uses a loop filter with a damping resistor so that lock time is adjustable. The damping resistor allows adjustment from so-fast-you-can't-tell to a chirp to a portamento slide. It tracks my FatMan over it's entire range with no adjustments other than the damping pot for the filter.
Besides a PLL, there's several designs on the web for at least 2x, 3x is more fun - that use triangle waves and bridge rectifiers, some that use diodes and opamps with sawtooth waves but they have their problems too. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
loss1234
Joined: Jul 24, 2007 Posts: 1536 Location: nyc
Audio files: 41
|
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Mon Apr 07, 2008 6:07 am Post subject:
|
|
|
Yes you can. The rails that supply the 4046 and associated logic set the required amplitude/level for the input signal. That is, as shown, the rails are -12v and 0v so the input signal that the PLL should track should travel from -12v to 0v. If you change the PLL supply to +15v and 0v, the input signal should travel from +15v to 0v. You can also run these logic chips from a dual supply, but unfortunately, it can't be a "standard" dual supply such as +/-12 or +/-15. It could be as high as +/- 7.5 (+/- 5v is more common). If you use +/- 5volts, then your input signal must travel between -5v and +5v.
Single ended signals like this are not commonly used in a modular, most modular audio signals are true AC and would have to be level shifted to use with this circuit if you powered it with a single ended supply. If instead you used (for example) +/-V, then you would simply need to reduce the input signal amplitude with a pot until it is the correct amplitude.
With a modular, it is possible that the input voltage could be quite a bit higher amplitude than these CMOS ICs want. It is important to know that there are diodes built into the CMOS ICs (for static electricity protection) that will turn on if the input signal goes above the Vdd supply or goes below the Vss supply. Turning these diodes on will at least clip the input signal if not cause other problems. This will never happen in a FatMan because the VCO is powered using the same rails as the PLL. In a modular, you can't guarantee that so you have to design the circuit so that the input will never be above Vdd or below Vss. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
loss1234
Joined: Jul 24, 2007 Posts: 1536 Location: nyc
Audio files: 41
|
Posted: Mon Apr 07, 2008 7:18 am Post subject:
|
|
|
let me ask you something...i have been building some of the "lunetta" circuits lately, basically cmos chips wired straight to jacks. now in the process i have been connecting both inputs and outputs from 4040's, 4020's and 4070's powered off of +15 and 0 grnd in combination with my modulars ins and outs which are off of +/-15volts...now i dont think ive had any clipping problems (that i know of-but i am quite new to this)
so would the 4046 behave in a similar manner (internally) to say, a 4040? if so, i dont think i am too worried about clipping. If i fry a 4046, i suppose id need to look more into level shifting and or using a +/- 5volt supply.
however, level shifiting in general is at this point a mystery to me (besides just using a voltage divider pot) so id like to avoid it if i can.
could i maybe just use a 7905/7805 setup hung off of my +/- 15v supply to feed them? or would that add too much noise to my modular setup?
anyway, great info and circuit. i plan on adding it to the "BUILD SOON" Heap. i notice your site doesnt seem to have any recent schematics. why not?
thanks a lot!! _________________ -------------------------------------------- check out various dan music at: http://www.myspace.com/lossnyc
http://www.myspace.com/snazelle
http://www.soundclick.com/lossnyc.htm http://www.indie911.com/dan-snazelle |
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Mon Apr 07, 2008 8:11 am Post subject:
|
|
|
These diodes are like all diodes, they have a current limit before they will smoke. If the current is controlled, by resistors for example, then they can conduct without damage. However, even if they are not damaged, they will still clip the signal which you may not actually see since the clipped signal will be seen only internal to the IC and the output will be turned into a rectangular logic level signal anyway.
But remember that the 4046 is a bit different than a 4040 or most of the 4xxx family in that most of the ICs are purely digital and don't expect analog inputs, least of all ones that cause these diodes to conduct. The 4046 has a VCO that while made up of CMOS transistors internally, is really an analog thing living in an otherwise digital IC. The input to the 4046 can certainly be rectangular, but doesn't have to be. The output of the 4046 will always be rectangular.
Bottom line is that if the current caused by turning these diodes on is excessive, the IC will smoke. If it's not quite high enough to smoke it, it can still heat it up quite a bit which may adversely effect it's operation.
What I would do is to protect the diodes by including a large series resistance between the driving signal and the 4046 input. A CMOS transistor is voltage activated, not current, so this large resistor shouldn't cause problems by itself. For example, if 100K is used and the signal can be as high as 15 volts, then the maximum current by Ohm's law is 0.15 mA (or 150 uA). I just looked at the datasheet, they don't say how much current those diodes can take - principally because an engineer would not normally want these diodes to EVER conduct given the device's intended application. So I can't say if a 100K will protect you or not.
I would do this experiment: Set up a PLL with whatever supply you want to use. Make sure that it works correctly (tracks the input signal) with an input that does NOT cause these diodes to conduct. If it works, you know you have the IC set up correctly.
Now you can mess with more real signals that you would apply. Add a 10 Meg resistor in series with the 4046 input pin and apply your more real amplitude signal through it (use a signal that you know will be higher than Vdd or lower than Vss and is in the normal modular range). If the IC works and does not heat up, you're fine. If the PLL doesn't track, then it's possible you need a smaller resistor. I'd take it down by half to 5 megs and see, then 2 megs, etc. until you find the largest resistor that allows it to work without heating up the IC.
----------
As for the supply, you can use 7805/7905 like that. But you will still need the resistor method I described if the signals will be more than +/- 5. Using a split supply for the 4046 will eliminate problems with level shifting. As far as noise goes, it may cause noise, I don't know. I think it depends on the supply, filtering and how much current is currently being used from it. The same +/-5 volt regulator could alternatively be powered with a wallwart and dual halfwave rectifier - cheesey, yes, but it should work.
----------
The FatMan site hasn't been updated because my design focus has shifted to FPGA synthesizers. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
loss1234
Joined: Jul 24, 2007 Posts: 1536 Location: nyc
Audio files: 41
|
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Mon Apr 07, 2008 10:38 am Post subject:
|
|
|
ScottG wrote: | I just looked at the datasheet, they don't say how much current those diodes can take - principally because an engineer would not normally want these diodes to EVER conduct given the device's intended application. So I can't say if a 100K will protect you or not. |
I looked it up for a 4001 the other day ... 15 uA _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Mon Apr 07, 2008 11:02 am Post subject:
|
|
|
Blue Hell wrote: | I looked it up for a 4001 the other day ... 15 uA |
that's good to know. If his 4046 supply is +/-5 volts and if signals can be +/- 15 volts, it has to deal with 10v above or below. At 15uA, that's 666,666 ohms. Or one could say to be safe, 1 Meg should provide burnout protection. The input impedance of these ICs is much greater than 1 Meg, so I think that should work. I'd bet that 10 Megs works. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Mon Apr 07, 2008 11:47 am Post subject:
|
|
|
Yes that would work, although maybe not when the PCB is a bit dirty
And when the protection diodes are gone already it will not work, as the diodes tend to fry into a state of low resistance.
External Schottky diodes is the way to go IMHO _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Mon Apr 07, 2008 12:32 pm Post subject:
|
|
|
Blue Hell wrote: | External Schottky diodes is the way to go IMHO |
A clamp? Yes, that would be the best approach, essentially bypassing the IC diodes with faster external ones. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
Peake
Joined: Jun 29, 2007 Posts: 1113 Location: Loss Angeles
Audio files: 3
|
Posted: Tue Apr 08, 2008 5:28 pm Post subject:
|
|
|
Ian Fritz also has a multiplier using an LED driver. These are very interesting ideas, especially for clocks. I'm looking at the Flame MIDI divider with real-time selection of beat division/multiplication, and know that one could create a DIY box that is similar. |
|
Back to top
|
|
|
loss1234
Joined: Jul 24, 2007 Posts: 1536 Location: nyc
Audio files: 41
|
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Wed Apr 09, 2008 5:55 am Post subject:
|
|
|
I'll mention this in case it matters - some frequency multiplier schemes produce a rectangular output that is not 50% duty cycle. One of the reasons I built my PLL circuit is that I wanted 50% duty cycle output. All 7 outputs are 50% duty cycle true square waves. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
Peake
Joined: Jun 29, 2007 Posts: 1113 Location: Loss Angeles
Audio files: 3
|
Posted: Wed Apr 09, 2008 9:18 am Post subject:
|
|
|
loss1234 wrote: | which fritz project is it?? |
http://electro-music.com/forum/topic-19084.html
I believe that The Bridechamber are going to sell that and other Fritz PCBs.
I haven't done the math to see if the ratios produced are what are expected in standard rhythmic division/multiplication, ex: 8th triplets, etc. |
|
Back to top
|
|
|
loss1234
Joined: Jul 24, 2007 Posts: 1536 Location: nyc
Audio files: 41
|
|
Back to top
|
|
|
Peake
Joined: Jun 29, 2007 Posts: 1113 Location: Loss Angeles
Audio files: 3
|
Posted: Wed Apr 09, 2008 8:02 pm Post subject:
|
|
|
I've truly only taken the most cursory of glances at it so I should shut up and let the thread go back on topic... |
|
Back to top
|
|
|
|