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

sinusoidal wavefolder
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page Previous  1, 2, 3, 4, 5  Next [all]
Author sinusoidal wavefolder
megaohm
Really cool circuit you have!
Love how it sounds.

Went searching for the paper and came across this which may relate and be interesting (page 12)?

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=4&cad=rj a&uact=8&ved=0ahUKEwivkrXppbHSAhUL34MKHWlVD44QFggoMAM&url=http%3A%2F%2 Fdocs-europe.electrocomponents.com%2Fwebdocs%2F0025%2F0900766b80025499 .pdf&usg=AFQjCNEQr_rCCzPF0RdVd6xQ-_o7fgIUIw&sig2=7y6xVUx_ndGz2QYIDbL3E g

Of course that chip is extinct!
Dead Banana
guest
yes, that was the chip they made from the original. it would be cool if it still existed.
guest
Isaiah wrote:
Also, what's the ideal input voltage range?


i havent yet set all of that, but the main circuit core runs at ~1mV/degree if i recall correctly. the input and output need to be scaled.
wsy
guest wrote:
i put together that barrie gilbert sine-shaper circuit i had discovered while researching for the OTA sine wave shaper paper:
.


EXCELLENT!!!

I hope you realize that this is making me rethink continuing work on the FMO, don't you?

- Bill
nd595
guest wrote:

this then brings up the question: what would you prefer? a THMT PCB with transistor matching required, or an SMT PCB that is already populated?

i think im going to do a kit for this one, as it seems pretty useful. the circuit can be used to: 1. do low distortion sinewave shaping, 2. act as a quadruple wavefolder, 3. do PM or pseudo FM (there would need to be some leakage on the integrator to avoid slight offset voltages from saturating the modulator).


I would rather prefer SMT PCB, populated or not - can do all the soldering with a great pleasure! And tinkering with the thing - what can be better?
guest
wsy wrote:

I hope you realize that this is making me rethink continuing work on the FMO, don't you?


i sure hope not! as much as i love analog, digital is the way to go for FM.
wsy
guest wrote:
wsy wrote:

I hope you realize that this is making me rethink continuing work on the FMO, don't you?


i sure hope not! as much as i love analog, digital is the way to go for FM.


But there's flaws in digital FM / PM.... some people complain about the "low fi" or th "sawtooth". The FM Ogre samples it's
inputs at something well over 40 KHz, so the phase angle in TZPM has a "jump" that often. The jump is minimized by the software (there's
a predictor/corrector) but it's still nonzero.

You'd say "but it's up at 40 KHz or more" and you're right - but if you have some high frequencies already in the signal, you can get
aliasing frequencies that come on down into the audible range. Very minor, mind you, and audible only at very high phase angle mod
rates, but nonzero. Whether or not it's less audible than the distortion of this circuit is arguable.

- Bill
guest
i put some better opamps in (TL072) and its much, much cleaner. the distortion is down to -70dB on the harmonics on the middle range, and the clicks are inaudible for the vast majority of the range. at really high modulation you can hear them a bit in the background. at really low frequencies its a slight click, and at really high freuqencies its a bit of a hiss. i found some fun operating points while mucking around. overdriving it sounds really good with a high frequency saw wave. it can also be used to produce pure overtones at 2x, 3x, and 4x the fundamental. im tempted to put another transistor pair in there so the 5th can be achieved.

i think the remaining inconsistencies are from transitions between matched pairs, which is a bit hard to avoid in a discrete like this. im considering seeing what a pair of LM3046 can do for me. the matching is probably worse on that, though.
batchas
This is the kind of thread I like very much here on Muffwiggler applause

I like the videos of the tests too, like "test 1" for instance thumbs up
guest
ok, i did some tests with amplitude modulation using the current control port on the bottom. as it turns out, the emitter voltages dont go below 0V, so a ground referenced exponential converter can be used to create the current. the modulation bleedthrough isnt that great. i got it down to -60dB, but it was quite finicky, and -50dB will probably be more likely in practice. with a bunch more trimming it might get down lower, but i dont think thats worth it, considering how easy it would be to add a 2164 at the end. but, for envelopes and such, it seems to work pretty well. with it full off its -90dB or better for carrier supression. at really high (distorting) input ranges, the carrier gets forced through at full volume, but the final design will have clamping diodes to prevent this (the transistors shouldnt be driven like that anyways).
Clive
Well done, guest applause

The first video example shows the transfer function, which looks very similar to the Serge middle wavefolder. Can I expect that to do PM too?
gbiz
Fantastic work guest applause

edit: SMD please
xahdrez
Wow, love how this sounds. This is fun!

I'd definitely be interested in a PCB & Panel - I'd probably prefer not to have to match transistors, but it wouldn't stop me - either version would be appealing to me! Great work!
guest
Clive wrote:
The first video example shows the transfer function, which looks very similar to the Serge middle wavefolder. Can I expect that to do PM too?


yes, you can do PM with a wavefolder and a (optional) sine shaper afterwards. there was some discussion of this in the "analog phase modulator" thread that is currently active.
wsy
xahdrez wrote:
Wow, love how this sounds. This is fun!

I'd definitely be interested in a PCB & Panel - I'd probably prefer not to have to match transistors, but it wouldn't stop me - either version would be appealing to me! Great work!


There's about 20 different quad matched NPN arrays 'active in stock' at DigiKey:

https://www.digikey.com/products/en/discrete-semiconductor-products/tr ansistors-bipolar-bjt-arrays/277?k=transistors&k=&pkeyword=transistors &pv74=85&FV=ffe00115%2C1f140000&mnonly=0&newproducts=0&ColumnSort=0&pa ge=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=25

so it wouldn't be such a stretch to use these as the matched pairs (remembering to use them in the right order!)

- Bill
jhulk
yep you can i used that 300 series for the core chip design


so its easily done with quads
guest
ive been thinking about the matched quads. a lot of the 4-transistor dies are not matched, and have low beta transistors (3904, 2222). the main contenders for matching are THAT300 (6$), MAT04 (10$), and LM3046 (0.75$). the LM3046 isnt actually a matched quad, though, but might be as good as a bunch of matched pairs. there still ends up being at least one transition with the LM3046, and 2 with the quads, as the core takes 10 transistors. if i used quads, id bump it up to 12. by contrast, the PMP4201 cost 0.20$ and have better transistors and better matching than the LM3046. they are pretty close the the THAT300 in performance, but the MAT14 is amazing, and way better than them both.
jhulk
i chose that because they do npn pnp and dual npn/pnp quads

and are matched most of the new quad arrays are not matched but i have tested the 3904/6 varieties and they are close matched on the die a bit like using the same transistors next to each other on a tape which seem to be evenly matched as its done by robots now rather than a human tolerances are a lot higher.

i had one transistor left over in the quad npn all the others are dual npn/pnp matched pairs

on the later design which i also have a chip it uses even more as it has a phase control and for +voltage and -voltage and has a current biasing circuit for this that attaches to this basic core by the emitters connected transistor.

and attaches to the pnp current mirrors either end .

so i split it into 2 chips.

i like the mat chips do they do equivalent types like the that series.

i also like the bcm847/57 chips and the ls318/358 pairs.

the other reason i like using the that series and ls is because micross uk has the bare dies they have the octal matched transistors from that and lots of dies so that an asic can be produced so its easier to do the schematic capture using those devices as they have the dies so easy to add to a sub straight and inter wire weld on the prototyping machine which is a wonder to watch. on a magnified screen.
J3RK
I second the Linear Systems LS318 and LS358! They're excellent, and half the price of SSM22xx and a quarter the price of the MAT series. Nicely specced too. The LS318 and LS358 are the log-conformance parts. (great for exponential converters and such) They also have ultramatched pairs that aren't specced for this if I remember correctly. You can order them directly from Linear Systems, through Trendsetter electronics, and I think there are a few places in Europe that distribute them as well.
jhulk
micross uk has them

and why i like using them as im currently doing some chip designs that they are prototyping for me.
guest
quick question:

should the VCAs for this be linear or exponential? im thinking exponential so it more closely matches the FM algorithm, but linear would be easier, and would have a more clearly defined "off" state.
jhulk
exponential

the phase can be linear
Isaiah
guest
Which one will feel most natural when you manually open up the audio VCA?
To clarify, when I say "natural", I mean to avoid the situation you often find if a linear pot is used as an audio attenuator (where most of the pot's effective range is only across a small part of the pot's rotation, not spread out across the pot's whole rotation.)

Would it be difficult to implement both linear and exponential and enable the user to select either with a jumper, or header and shorting pin on the PCB?

Is it especially necessary that the (audio) VCA can close completely?
I'm not for or against it, just wondered. Obviously the PM VCA should be able to fully close.
guest
so, it depends a lot upon what people want to drive it with. linear envelope generators would work better with an exponential VCA, and vice versa. the attack portion is the part that sounds different between the two.

im leaning towards a dual scheme, like you suggest. basically a linear VCA driven with a current source that can be either linear or exponential, depending upon whether the signal is applied to the base or the collector of the reference transistor. it will require thermistors for exponential mode, which isnt too bad, except that panasonic is end of lifing the ERAV series thermistors that are cheap.
Isaiah
guest
Any news on this?
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page Previous  1, 2, 3, 4, 5  Next [all]
Page 2 of 5
Powered by phpBB © phpBB Group