MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

heptaphasis - 3HP 7-phase LFO
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author heptaphasis - 3HP 7-phase LFO
charonme



http://charon.sk/heptaphasis/
https://github.com/charonme/heptaphasis

based on the Yusynth Quadrature LFO


eagle schematic
eagle PCB
charonme
PCBs ordered at pcbway! hyper


oberkorn
thank you. will get a couple of these next time I order PCBs
charonme
Chugging Beers
but wait till mine arrive so that I can test whether it even works smile
cackland
Nice work. I've recently been looking at circuits that phase shift an incoming LFO, rather than generate.

Was thinking of using 2164 vca, but considering the LM13700.
joem
charonme wrote:
Chugging Beers
but wait till mine arrive so that I can test whether it even works smile


Do let us know! I'm intrigued!
cackland
Any update on this project? Working / improvements / redesigns...
charonme
still waiting for the PCBs from china, I will update as soon as they arrive
charonme
tonight I'm soldering

cackland
Nice. Hopefully no oversights and everything goes to plan smile
charonme
well the THT schottkys barely fit in their holes, the THT solder pads seem too small to comfortably solder the parts and the SMD solder pads are larger than eagle led me to believe, so they cover parts of the silkscreen and some letters are not that easy to read

also this looks dangerously close together:
guest
the silkscreen seems consistently off. perhaps an error with the manufacturer? you should have a look at your gerbers in another program to see what they look like. you can upload them to oshpark and they have a free view that lets you look at them before purchase.
charonme
well... it doesn't work, the oscillations die off when lowering the frequency with the pot. using polystyrene THT caps instead of the 470nF SMD X7R helped only a little
cackland
At what frequency do the oscillations stop?

Wonder if the yusynth version had the same problem at the same frequency.
ersatzplanet
No room for the 8th phase?
BugBrand
charonme wrote:
well... it doesn't work, the oscillations die off when lowering the frequency with the pot. using polystyrene THT caps instead of the 470nF SMD X7R helped only a little


It is an unfortunate thing with such quadratures - I've encountered such frustrations and believe that in part it is down to how clean the board is (flux could lead to stray leakage). I know that when I was prototyping I had to do a lot of cleaning - so next tests I'm going to try with water-soluble flux (easier to remove than regular 'no clean').

I'd be interested to hear comments on other possible issues...


PS - I haven't seen uses like the bits around TR8 & TR9 - are these meant for starting up - pushing things high so oscillations start?
guest
are T8,9 backwards in your schematic? shouldnt it be the collectors that go to the summing junctions?
frijitz
My switchable 6/8 phase LFO is here:
http://www.electro-music.com/forum/topic-19841.html
and also at my site.

Runs over a range of 0.02 Hz to beyond 3 kHz. It’s configured as a limit-cycle oscillator (damping on each stage) rather than as an oscillating filter.

Ian
nigel
ersatzplanet wrote:
No room for the 8th phase?

Asked and answered in the original thread. There's no room for another socket on the front panel.
charonme
guest wrote:
are T8,9 backwards in your schematic? shouldnt it be the collectors that go to the summing junctions?
I don't know much about transistors, so maybe, but it works well on the breadboard and in falstad simulation like this.

nigel wrote:
ersatzplanet wrote:
No room for the 8th phase?

Asked and answered in the original thread. There's no room for another socket on the front panel.
maybe in the next revision with a 4hp panel and two columns of output jacks. Or a jumper on the pcb could turn the frequency CV input jack into the 8th phase output

BugBrand wrote:
I haven't seen uses like the bits around TR8 & TR9 - are these meant for starting up - pushing things high so oscillations start?
yes, the capacitor-transistor design comes from guest. The circuit around T8 introduces a rough nanosecond disturbance to which the filter responds with starting the oscillations quickly and the circuit around T9 increases the frequency to max for a couple of seconds so that the oscillations stabilize themselves quickly

I'll be debugging it this weekend
frijitz
charonme wrote:
BugBrand wrote:
I haven't seen uses like the bits around TR8 & TR9 - are these meant for starting up - pushing things high so oscillations start?
yes, the capacitor-transistor design comes from guest. The circuit around T8 introduces a rough nanosecond disturbance to which the filter responds with starting the oscillations quickly and the circuit around T9 increases the frequency to max for a couple of seconds so that the oscillations stabilize themselves quickly

In the work I have done on this system, if it is running at a high frequency and you quickly turn the frequency down, then there is no settling time required. The signals are always at the correct relative phase and just start running slower. So it seems to me all you need for startup is a voltage pulse at the CV input. Sure, drive it with a transistor switch if you want, but I say try just a series RC from V+ to the base of the input transistor.

Ian
charonme
frijitz wrote:
In the work I have done on this system, if it is running at a high frequency and you quickly turn the frequency down, then there is no settling time required. The signals are always at the correct relative phase and just start running slower. So it seems to me all you need for startup is a voltage pulse at the CV input. Sure, drive it with a transistor switch if you want, but I say try just a series RC from V+ to the base of the input transistor.


if my LFO had a high enough max frequency then yes, this would work, but my LFO's top frequency is just <20Hz, so without the pulse through T8 it would look like this:
So it appears that the frequency rise alone is insufficient to reach the stable amplitude (set to +-5v)
guest
on your previous design you had diodes in series with the transistors. maybe that would help? perhaps try removing R38 and seeing if the low frequency performance improves.
frijitz
charonme wrote:
frijitz wrote:
In the work I have done on this system, if it is running at a high frequency and you quickly turn the frequency down, then there is no settling time required. The signals are always at the correct relative phase and just start running slower. So it seems to me all you need for startup is a voltage pulse at the CV input. Sure, drive it with a transistor switch if you want, but I say try just a series RC from V+ to the base of the input transistor.

if my LFO had a high enough max frequency then yes, this would work, but my LFO's top frequency is just <20Hz,

Ah, I see.

But I have a couple of comments/questions:

Why are you using such a huge value for the integrating caps? Yves uses 10nF and I usually use 1nF. (And where do you even get a 470nF polystyrene part?)

If you use a smaller value then you can still bring the operating freq range down <20Hz by adjusting R51. (Note that Yves has a trimmer for that.) Then you would be able to start up with a pulse at the input to push the system into proper quadrature.

This design has pretty low distortion. I just measured it again and found the 3d harmonic down by ~40db and nothing else significant. If you can tolerate a bit more distortion, then you can lower R27 to provide a larger positive feedback.

Ian
charonme
I have 20Hz as the top end of my frequency range. I want it to go down to ~20 minutes

wait why did I say polystyrene :( Sorry for the mistake, I have these polyester caps

I need very low distortion because the 45° (+n*180°) is created by adding 0° and 90°, this wouldn't work as well if the added shapes were not exact sines

btw what is the function of R27 (or that whole negative feedback loop)?
charonme
I found out what the problem is: I tested T8 on a breadboard with 2N3906 and it worked perfectly, but for the PCB I used BC857. I soldered some legs to a BC857 and tried it on the breadboard instead of 2N3906 and now it behaves the same (broken) way as the PCB.

On mouser I noticed there are versions of the 2N3906 with different values of VCEO, VCBO, VEBO, Collector-Emitter Saturation Voltage and Maximum DC Collector Current. I have no idea what these are and which version I have

But when I add the original 1n4148 diode in series with the 10Ohm R38 it seems to be working fine. I originally had that diode there because it somehow seemed intuitively logical to me, but it was working (with 2n3906) even wihout it, so I removed it from the design d'oh!

I replaced the 10Ohm R38 on the PCB with a LL4148 and it works now, except at very low frequencies it starts noticeably interfering with the oscillation and introduces an offset. I'm considering completely removing this part of the circuit and lengthen the increased startup frequency by using a higher value cap for C24
guest
yeah, you were getting reverse leakage through the transistor (Vbe breakdown), and the diode acted as a backup to this.
charonme

guest
looks good!
cackland
Very nice smile

Couldn't see the LL4148 on the pcb...

A few questions:

- so you removed R38, R39, C21 and T8 and it still works? Wasn't this sub circuit of components meant to help with the initial voltage start up?

When you started it up in your video, why did it act the way it did if the circuit was removed?

Also, why would polyster through hole caps work over smd mlcc caps you replaced?
charonme
the 4148 was a temporary fix of the subcircuit around T8 which I finally removed because it still affected the oscillations, caused an offset etc. The T9 subcircuit that causes a temporary rise of the frequency at the beginning is still there, I just prolonged the period during which the freq is maxed out so that there is enough time for the amplitude to stabilize. However it still isn't ideal because if you turn it off and back on again the 10uF cap is still charged and the freq.rise period is too short for it to work, so you need to manually increase the frequency with the pot. In rev2 I might add a jfet that would discharge the cap when power is removed

cackland wrote:
why would polyster through hole caps work over smd mlcc caps you replaced?


turns out mlccs have some technological imperfections that make them unsuitable for this kind of application (for "timing caps"). I did often see in various DIY build documents and BOMs a note like "timing cap - use a good one", but I didn't know which are the "good ones" nor that there is such a huge difference in practice. Perhaps there even is a thread about this somewhere here or on electro-music

some info might be here:
https://www.muffwiggler.com/forum/viewtopic.php?t=82443
https://www.muffwiggler.com/forum/viewtopic.php?t=142469
http://electro-music.com/forum/topic-41821.html
I'd be interested in any other good info on this
charonme
frijitz
charonme wrote:
the 4148 was a temporary fix of the subcircuit around T8 which I finally removed because it still affected the oscillations, caused an offset etc.

Yeah, I never understood that one. Seems to me it forces the system into a non-quadrature state. What would be good about that?
Quote:
turns out mlccs have some technological imperfections that make them unsuitable for this kind of application (for "timing caps"). I did often see in various DIY build documents and BOMs a note like "timing cap - use a good one", but I didn't know which are the "good ones" nor that there is such a huge difference in practice. Perhaps there even is a thread about this somewhere here or on electro-music

It's pretty standard to use silver micas for timing caps. What "technological imperfections" are you refering to?

Ian
charonme
frijitz wrote:
it forces the system into a non-quadrature state
it just introduced a nanosecond disturbance that was supposed to instantly make the circuit to oscillate at a higher amplitude like this:
it worked well at some specific conditions, but created other problems because the transistor leaked even after it was shut off.


frijitz wrote:
What "technological imperfections" are you refering to?

nothing specific, I don't know anything about caps as was clearly demonstrated when I assumed the SMD caps I bought would work smile I just referred to the fact that it didn't work there properly and not because it would be a faulty product
charonme
frijitz wrote:
This design has pretty low distortion. I just measured it again

how did you measure it and what circuit exactly did you measure? I'd like to measure the distortion of the one I have here, but the frequencies are well below what my scope's spectrum analyzer could display nor do I have a dc coupled computer interface
frijitz
charonme wrote:
frijitz wrote:
This design has pretty low distortion. I just measured it again

how did you measure it and what circuit exactly did you measure? I'd like to measure the distortion of the one I have here, but the frequencies are well below what my scope's spectrum analyzer could display nor do I have a dc coupled computer interface

It was the original design with manual control — the one Yves modified. I recorded it at a couple hundred Hz and analyzed it with software.
EATyourGUITAR
I think it would be way more punk if you named it "hepatitis" and put it on a rusty iron panel SlayerBadger!
charonme
lol yes, but it would be much harder to google it wink
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 2
Powered by phpBB © phpBB Group