New Effect: Microtubes 2k (Cmos Design)

[zencafe]

Hello everyone!

I want to introduce you my Microtubes 2K, which is a 4049UBE based distortion that I recently designed.
You'll see many similarities with many others 4049 projects (such as the Double D, or Hot Harmonics) but honestly. I didn't based on any of them. I haven't even built them (Someday I will though ).
The of building and designing it was manly done by ear. I deffined some ranges of gain I wanted to pull out of the circuit, and tried to stick to it.
The Mesa amps) first Jfet stage is a booster, similar as the ones found in the earlier circuits I mentioned. It has way more filtering though, this in order to cut as much bass as possible (like Engl and , to get a  more tight and not farty distortion.
After the jfet, there's this 1M resistor, which limit the signal going into the cmos, this will produce a huge decrease of gain, and therefore amount of distortion. If you bypass it (SW1)  the gain increases. This allows you to use this unit as a quite clean Booster, increasing versatility.
After the cmos gain stages there are two lowpass filters: The reason I added this was that in my opinion, the 4049UBE generates too many high harmonics when pushed to high gain settings, this is the reason I think it sounds so tubey, cuz of the rich harmonic content it has. There are 2 ways of controlling it:

1- Increasing the value of the feedback caps: This is quite handy and simple solution, but I had problems by tweaking the values too much, when the cap was big enough to bring non-frizzy distortion (1.2nF) it attenuated the "good" harmonics, so the sound gets a bit dark and "wrapped".
2- Cutting the very high harmonic content with filters: This was my option, instead of limiting the generation of harmonics, I thought it was better to let them generate all the rich harmonic content, and then cut the ones I didn't want. Thats why I added two lpf at the end of the circuit, of one's cap is switcheable, so that you stay just with one filter, this would increase the high harmonics content, therefore I called it Bright switch.



So, here it is!

[bool]

I wonder why nobody seem to use a mosfet boost as a front-end in these 4049 designs? Wouldn't that be more logical than using a jfet? I mean "bof" ...

Another thing that caught my eye: how come nobody (afaik) used a g-d "diode" connected mosfet (2n7000-ish) as a "voltage-sag" element for the 4049? Maybe in-line with the ubiquitous "220R" resistor ... or connected to a pot (s-g-d) to use it as a "variac" control ... i.e. connected similar to the quiescent current biasing BJTs in poweramps (pot lugs to s and d, wiper to gate). A mosfet should easily drop the required 2-2,5V, just perfect for the sweetspot.

[zencafe]

Very interesting points!
could you explain better the sag emulator thing? (do you have any links to read about it) About the variac control I've tried it, and I got awful oscillations with the voltages between VDD and 1/2VDD... At 9 and 4,5 it worked as expected, but in 7V for example it oscillated... weird. What I've done with other "designs" is to use a toggle to switch the VDD of the 4049 between 9 and 4,5V.
About he mosfet, yes, I've thought about it a lot! But haven't had enough time to research and experiment.
It actually would be better to use a Mosfets for other reason: Since you can get more gain from a Mosfet than a jfet, you could basically have the same amount of gain, without increasing the feedback resistors too much... this would mean a lower output impedance.
I should do the experiments soon, I'll let you know .

D.

[dschwartz]

I wonder why nobody seem to use a mosfet boost as a front-end in these 4049 designs? Wouldn't that be more logical than using a jfet? I mean "bof" ...

Another thing that caught my eye: how come nobody (afaik) used a g-d "diode" connected mosfet (2n7000-ish) as a "voltage-sag" element for the 4049? Maybe in-line with the ubiquitous "220R" resistor ... or connected to a pot (s-g-d) to use it as a "variac" control ... i.e. connected similar to the quiescent current biasing BJTs in poweramps (pot lugs to s and d, wiper to gate). A mosfet should easily drop the required 2-2,5V, just perfect for the sweetspot.

Cause fets sound better than mosfets, ..they clip nicer.. If gain is the issue here, then go with opamps or darlingtons instead..
the sag thing look like a nice idea..

[zencafe]

I wonder why nobody seem to use a mosfet boost as a front-end in these 4049 designs? Wouldn't that be more logical than using a jfet? I mean "bof" ...

Another thing that caught my eye: how come nobody (afaik) used a g-d "diode" connected mosfet (2n7000-ish) as a "voltage-sag" element for the 4049? Maybe in-line with the ubiquitous "220R" resistor ... or connected to a pot (s-g-d) to use it as a "variac" control ... i.e. connected similar to the quiescent current biasing BJTs in poweramps (pot lugs to s and d, wiper to gate). A mosfet should easily drop the required 2-2,5V, just perfect for the sweetspot.

Cause fets sound better than mosfets, ..they clip nicer.. If gain is the issue here, then go with opamps or darlingtons instead..
the sag thing look like a nice idea..

Yeah, I've tried with OpAmps as well, didn't like so much, but I should try with better boosters before say anything seriously, BJT are next also.
Is true that jfets clip nicer, but that is not an issue here, since the first stage is used as a booster in order to crank up the CMOS harder, however, the signal is mean to remain clean at that point.
Dunno why, but I have the feeling that mosfets are the perfect match for 4049.

[awitee]

is there a layout or project pdf available for this? i just might replace my hot tubes clone with this one, hopefully it wouldnt be as farty/gated/fuzzy

[bool]

maybe something like this would work as a front-end? add your own protection gate-source zener; caps may need tweaking to work in your circuit - and you can insert some 10K resistor between the pot lug and out (100nF cap) so the volume won't go totally down.

[gigimarga]

hello,

i've built tonight the microtubes 2k (thx zencafe) and it sounds very well.

the only BIG problem i had was when i bypass R5(1M) using SW1 i got a very strong oscillation (but it can be controlled by the pots of the guitar  ).

a little big problem is that SW3 (i think is misnamed...logically it name could be SW2) have a too subtle effect for my taste, but it's working good.

thx zencafe again!

[spud]

Hi - anyway someone could post some clips?  I'd like to hear it. 

G - How does it compare to the other 4049 circuits sound wise? 

spud

[gigimarga]

i will try to record some soundclips tomorrow, but the quality will be poor because i must to record them straight through my cheap soundcard.

it sounds very well, maybe the best...please give me sometime to test it more

[mdaudet]

Zencafe's Microtubes is a very very great sounding distortion. It respects the dynamics and tone of your guitar without adding any more but great tubish gain. It cleans with the volume knob in a fantastic way. With a strat it sounds like classic rock, heavy blues, fusion spot. With humbuckers or High output coils, Microtubes will take you to the metal area without any problem. We tested along with Zencafe and Dschwartz at my home and the overall opinion was this: FANTASTIC

Is the first time I hear a 4049 based pedal without the fuzzy end.

Using a mosfet in the booster part is a great idea, maybe a BS170 will sound better than a 2N7000. Since the booster would no be the clipping part of the circuit, is good to try different transistors, like a Darlington MPSA14, 2N5088, etc.

Congrats my friend for your nice circuit.

see ya soon.

Matt

[awitee]

wow! cant wait for the layout

[dschwartz]

you should be proud zencafe..i think you have a winner here.. the next BSIAB.
I have to add to mdaudet´s comment that the touch sensitivity was awesome..like it helped you to play better, this is a really nice distortion for shredders! 

hmm..you have to hurry up with the layout though..we´re getting impatient!!

[gez]

Another thing that caught my eye: how come nobody (afaik) used a g-d "diode" connected mosfet (2n7000-ish) as a "voltage-sag" element for the 4049? Maybe in-line with the ubiquitous "220R" resistor ... or connected to a pot (s-g-d) to use it as a "variac" control ... i.e. connected similar to the quiescent current biasing BJTs in poweramps (pot lugs to s and d, wiper to gate). A mosfet should easily drop the required 2-2,5V, just perfect for the sweetspot.

I've dropped supply voltage to the Vdd pin using a 3V3 zener and Si diode.  Only works well if a few CMOS stages are blazing, though, as you need a certain amount of current through a zener to do what it says on the tin.

I've also used a MOSFET follower (source connects to Vdd pin, gate to a trimpot across the rails) to dial in that sweet spot.  These days, I just use a 5V regulator.

[bool]

I did see 6-ish volt supplies in commercial cmos "abuses" - didn't fender use a 4007 in some of their ss-amps? Doesn't a _resistor-connected_ cmos chip actually provoke a "voltage sagging" when driven hard (in a sense that voltage drop is varying somewhat proportionally to how much the inverters "sweat")? Therefore I imagine a single small-signal mosfet could be a "less-hard" provider of a voltage, compared to a regulated stabilized supply. More "organic" if I may say so.

Otoh, Gez, you are the author of a 4007 "nutcruncher", right? (I was always wondering if it would be possible and __useful__ to connect the transistors in a 4007 as a sort-of mos-face, therefore I asked you about the fuzz in the other thread).

What strikes me in all these cmos-designs is the following:
- nobody seems to include a self-biasing mosfet front-end. This could be a "harder" driver than fet, and signal feeding the next cmos stage could be more "pre-filtered" without an opamp (not to mention being more "in spirit of the design")
- there are __always__ a couple of invertors being wasted - why on earth nobody designed a simple speaker simulator (being on the same power line it could sag with the rest of the circuit ... ). Imo ROG team provided some building blocks already, additionally the "linkwitz transform" circuit (built around an inverting opamp) could be a nice starter to convert into a one-inverter speakersim. IMO a 50Hz hipass, 90Hz peak, 400Hz notch, 2,8kHz peak and 3,5kHz lowpass is needed. YMMV.

see here:
- it's sortof a double  "bridged-t"
- in fact, a hipass @ the input is needed so there would be a "bump" @ 90~ish Hz (a c-rtoground-c "t" before the input ) . A notch @ 400Hz should be calculated and the "t" in NFB loop should get rid of the "R2" and provide for the 2'8kHz~ish peak and the lowpass functionality.

NOTE: imagine there's a nice sweaty cmos invertor instead of the opamp (!!!)

http://sound.westhost.com/p71-f2.gif

some formulas from Mr.L. himself:

http://www.linkwitzlab.com/images/graphics/shlv-lp2.gif

a "doctored" idea:



The main problem (to me) seems that there isn't a linear-analog model of 4049/4069UB suitable for ltspice etc. available?

ANYBODY?

[zencafe]

Thanks a lot for your replies!

And thanks also to gigamarga for building it.
Yeah, SW3 is missnamed, I recycled another schem that had 3 switches, one worked with to BMP tone control in order to have a scooped, or flat mids response...
I didn't like the way that tone control sounded with this circuit so I removed it... I'm currently working on a version with a 3 band active control. So that you could model the distortion in a More accurate and versatile way.

gigamarga, how did it sound with your guitar? and what guitar (pickups) do you use?
I'm very interested on how it sounds with different guitars, I've so far tried it with a Fender Strat, a Ibanez with DiMarzio and a Les Paul, all of them, sounded very very very different through the Microtubes...

Thanks again for all the support to mdaudet a dschwartz (gracias amigos!!) and thank you all for the advises!
Gez: I'm really considering about adding a LM75L05... I think it would be very useful to be able to change the voltage feeding the 4049.
I hope gigamarga uploads some clips, I'm very curious!!!

Thanks again to you all

Best Regards from Chile

Doug.

[zencafe]

By the way, If you think the effect of the ultrabirght switch (Sw3) is too subtle... You could use a DPDP toggle switch and cut but caps instead of just one, the you'd get much more high harmonics, or better yet, use a 3 posicion DPDT to choose between both, one or non lpf, this would give you a lot more options and versatility, specially at lower gain settings were you may miss a richer higher harmonic content.

I'll post later a schem with how I think you could implement this idea.

D.

[dschwartz]

maybe the frecuency response of the amplifier gigimarga was using was below the filter cut range, so switching the cap woulkd do little effect since that frequencies are already cut.

[zencafe]

maybe the frequency response of the amplifier gigimarga was using was below the filter cut range, so switching the cap would do little effect since that frequencies are already cut.

Yeah, Daniel that's a possibility!

About the oscillation: Did you try changing the IC? When first breadboarded the prototype I got awful oscillations as well, after checking all the components, realised it was the IC, even though it sounded as expected, it oscillated... try with a new 4049.

D.

[zencafe]

I did see 6-ish volt supplies in commercial cmos "abuses" - didn't fender use a 4007 in some of their ss-amps? Doesn't a _resistor-connected_ cmos chip actually provoke a "voltage sagging" when driven hard (in a sense that voltage drop is varying somewhat proportionally to how much the inverters "sweat")? Therefore I imagine a single small-signal mosfet could be a "less-hard" provider of a voltage, compared to a regulated stabilized supply. More "organic" if I may say so.

Otoh, Gez, you are the author of a 4007 "nutcruncher", right? (I was always wondering if it would be possible and __useful__ to connect the transistors in a 4007 as a sort-of mos-face, therefore I asked you about the fuzz in the other thread).

What strikes me in all these cmos-designs is the following:
- nobody seems to include a self-biasing mosfet front-end. This could be a "harder" driver than fet, and signal feeding the next cmos stage could be more "pre-filtered" without an opamp (not to mention being more "in spirit of the design")
- there are __always__ a couple of invertors being wasted - why on earth nobody designed a simple speaker simulator (being on the same power line it could sag with the rest of the circuit ... ). Imo ROG team provided some building blocks already, additionally the "linkwitz transform" circuit (built around an inverting opamp) could be a nice starter to convert into a one-inverter speakersim. IMO a 50Hz hipass, 90Hz peak, 400Hz notch, 2,8kHz peak and 3,5kHz lowpass is needed. YMMV.

see here:
- it's sortof a double  "bridged-t"
- in fact, a hipass @ the input is needed so there would be a "bump" @ 90~ish Hz (a c-rtoground-c "t" before the input ) . A notch @ 400Hz should be calculated and the "t" in NFB loop should get rid of the "R2" and provide for the 2'8kHz~ish peak and the lowpass functionality.

NOTE: imagine there's a nice sweaty cmos invertor instead of the opamp (!!!)

http://sound.westhost.com/p71-f2.gif

some formulas from Mr.L. himself:

http://www.linkwitzlab.com/images/graphics/shlv-lp2.gif

a "doctored" idea:



The main problem (to me) seems that there isn't a linear-analog model of 4049/4069UB suitable for ltspice etc. available?

ANYBODY?

Great info bool!

By the way, in the filter you posted you can indeed swap the OpAmo for a Cmos invertor, but there's the issue of the very high noise present on this type of curcuits, dunno why, but when using more than 3 stages (not necesarily gain stages) the noise is reeaaaally nasty, I tried using one of the invertors as a buffer after the gain stages and I got awful noise and oscillations, it was too much. I think the only way of using the 6 of them in a same circuit would be what ROG team did with their Double D by making a 2 channel stompbox... You could also try connecting them in parallel in order to increase the current output, this may be useful to use the unit with a small speaker... just guessing: 2 stages (+ an input booster) to get a clean or crunch "preamp" stage, and then you could use the 4 remaining invertors as a small power (or should I say Micropower  ?).