How to fight high-gain noise?

Frequent asked about building blocks: gain stages, buffers, clipping configurations, ...
Post Reply
User avatar
The G
Grease Monkey
Information
Posts: 558
Joined: 21 Jun 2009, 11:43
Location: Bucharest, Romania
Has thanked: 408 times
Been thanked: 284 times

Post by The G »

Hi guys,

I like a lot the CMOS inverters distortions, but there is more noise at high-gain than in other circuits, at least for me.

What do you think can be done to fight the noise?

One of the more used methods seems to be limiting the bandwidth by filtering the highs pre- or/and post- clipping. My problem with this is that the lower gain sound now lacks the highs. Seems like the gain control should also control the filtering.

Another one is highs pre-boost/post-cut, either statically (like in Boss GE-7) or dynamically (I saw here one circuit using NE570/571). I'm not sure if this would work with high-gain circuits, as the boosted highs would be clipped before post-cut stage. Then again, I didn't try it yet.

Do you know more ways to fight the noise?

User avatar
jonasx26
Breadboard Brother
Information
Posts: 118
Joined: 03 Jun 2009, 17:53
Location: Sweden
Has thanked: 84 times
Been thanked: 48 times
Contact:

Post by jonasx26 »

The bandwidth and gain of the inverter varies with supply voltage.
Higher supply voltages increases the bandwidth while reducing the open loop gain.

Some of the datasheets specify inverter bandwidth vs. supply voltage when used 'properly' without feedback.
Feedback complicates things but the data still gives a rough indication.
I would suggest that you analyze these and the voltage transfer transfer characteristics graphs.

User avatar
The G
Grease Monkey
Information
Posts: 558
Joined: 21 Jun 2009, 11:43
Location: Bucharest, Romania
Has thanked: 408 times
Been thanked: 284 times

Post by The G »

OK, my research concluded that the noise in the CMOS inverter distortions is from the inverters themselves. This was a nice read.
Most of the tricks to fight noise are based on the assumption that the noise is generated outside the effect.
You cannot do much to reduce the noise, except hunting for lower noise variants - some say CD4069UBCN from Fairchild (obsolete) should be much more quiet. Maybe.
But CMOS distortions are the only effects that still have lots of noise when I shunt their input to signal reference.
So I gave up on them. It's a pity, their sound is quite nice.

User avatar
jonasx26
Breadboard Brother
Information
Posts: 118
Joined: 03 Jun 2009, 17:53
Location: Sweden
Has thanked: 84 times
Been thanked: 48 times
Contact:

Post by jonasx26 »

Seems a lot of people complain about noise issues, I've never had any problems.
I've used MC14049UBCP in most of my builds. No more noise than any opamp-clipper at stage volume.

Also experimented with different brands of 4049, 4069 and 74HCU04. There was no noticeable difference in noise..

Make sure you lower the inverters supply voltage, use proper decoupling and buffer the input and output.

EDIT: The schematic you linked looks noisy indeed. I would add small value caps across every inverter. And an additional 100µ supply filter cap.
Inverters don't like heavy loads, such as guitar cables. Output buffer would help. 47k input impedance? Input buffer would help.

User avatar
The G
Grease Monkey
Information
Posts: 558
Joined: 21 Jun 2009, 11:43
Location: Bucharest, Romania
Has thanked: 408 times
Been thanked: 284 times

Post by The G »

Seems like you're a happy guy. Maybe I should've emphasized that I'm interested in high-gain.
You can hear here a comparison I made. It's an Emma ReezaFRATzitz clone - it has an input buffer, it was powered by a battery and it has proper decoupling. All of which are supposed to exist in every pedal (at least every modern pedal) and are of no use in this particular case.
Mad Professor Stone Grey Distortion uses only two inverters, still the noise is easily audible.
Do you have any recording of your noiseless pedals? I'm very curious about them.

Also, how can be the flicker noise level be lowered by lowering the supply voltage? In my experience it's the other way around, and you said yourself in your first reply: "Higher supply voltages increases the bandwidth while reducing the open loop gain." Both effects lead to lower noise levels.

What kind of circuits did you make with 4069s? Do you care to show a schematic?

User avatar
jonasx26
Breadboard Brother
Information
Posts: 118
Joined: 03 Jun 2009, 17:53
Location: Sweden
Has thanked: 84 times
Been thanked: 48 times
Contact:

Post by jonasx26 »

I wouldn't go as far as happy :D
But my experiments with inverters went well indeed. I did focus on overdrive, though. My experience with higher gain circuits is somewhat limited.

Reducing the supply voltage reduces noise by decreasing the inverter bandwidth. For instance, with a 4069:
At 10V, open loop gain is approx. 30dB up to about 500kHz. At 3V, open loop gain is approx. 50dB up to about 100Hz.
A lower supply voltage will not remove ALL inherent noise, but limiting the bandwidth will help as high gain (or any gain at all) at 10MHz isn't necessary.

The Emma overdrive does use this trick. Do you know if the commercial units also are noisy? Seems strange. I sure wouldn't buy a noisy effect..
Try adding an additional 100nF directly across the inverter supply pins for high freq. stability.
Also, try adding a output buffer. Inverter output impedance is high due to the high feedback resistance and low open loop gain.
(to make matters worse; open loop gain also varies with load impedance and signal level (i.e. that nice compressed effect))

With many gain stages in series, noise buildup gets more difficult to avoid. And I have mostly used only two stages.


I don't have any way of recording at the moment. And the units are all sold.
But here is a video demo of a build .
Probably doesn't really say much about the noise though..

The effect in the video uses two stage inverter circuits.
Opamp follower -> 4049/4069 -> passive tone + volume -> opamp follower

Also, I used boardmounted pots. Layout is critical, especially at higher gain.

Hope you get some good results!

User avatar
The G
Grease Monkey
Information
Posts: 558
Joined: 21 Jun 2009, 11:43
Location: Bucharest, Romania
Has thanked: 408 times
Been thanked: 284 times

Post by The G »

Whoa!
Really nice pedals, dude! And beautiful case graphics. Real nice.
From your site it seems that you experimented a lot with inverters.

Still... in my view bandwidth limiting acts like a LPF. I tried filtering just to see how much filtering would it need - it nedeed so much to reach for the midrange noise that distroyed the good signal. The noise is not simply high frequency noise, so filtering is of limited use. Also lowering the voltage does limit the dynamic range and increases the noise's share in the signal. I tried it it as Reeza has a "bias" pot that does exactly that.

And about the noisy commercial pedal, buyers expect a high-gain pedal to be noisy. A metalhead's pedalboard without a noise gate is not easy to find. Actually, before I made some experiments with opamps, I thought that such a thing like a low-noise high-gain pedal doesn't exist at all.

So, to summarize my view: most of the noise is innate to CMOS electronics, the noise is across all the useful audio range (and beyond), filtering doesn't do much. I thought of using opamp gain stages in front of one or at most two CMOS inverters (that's why I was interested in Stone Grey Distortion), or (this just hit me) three stages in series, each one made of an low-noise opamp gain stage and a low-gain CMOS inverter used just for overdrive, separated by dividers.
Unfortunately (or fortunately), I started to experiment with JFETs, based on the AMT thread here on FSB, and I really like the results so far. One disadvantage of JFETs is the huge variability of its parameters and another one is that they are becoming obsolete. But I don't need more than a pedal, so no problem for me :mrgreen:.

User avatar
jonasx26
Breadboard Brother
Information
Posts: 118
Joined: 03 Jun 2009, 17:53
Location: Sweden
Has thanked: 84 times
Been thanked: 48 times
Contact:

Post by jonasx26 »

Thanks for the kind words! :D

Yes, I tried many different approaches but finally settled with the simplest one. Similar to the tube sound fuzz.

Noise gate is a solution. It doesn't even have to be very complicated. Check out the G2D Morpheus for a low parts solution.

Fets can sound good too. Each to his own.
Matching and availability doesn't really matter for oneoffs, and SMD parts aren't so scary once you get to know them..

All the best!

User avatar
juozasu
Information
Posts: 10
Joined: 03 Jul 2013, 17:33
my favorite amplifier: JCM800, Soldano SLO100
Has thanked: 1 time
Contact:

Post by juozasu »

How about using All MOSFET Op-amps like http://www.jaycar.com.au/images_uploaded/FN817.PDF; There are a lot of them now-days.

They have low enough equivalent input noise. And high gain. No need to use input/output buffers. Sound character is very similar to CMOS inverters.
I have tried several. Noise is perfect, tried at high gains about 20 000 - 30 000x (reeza gain is about 40 000x).

But it should be very good supply bypassing and star grounding.

Also could be used low input impedanse - less hum from pickup's is additional benefit. For example ROG "Mockman V2.0".

User avatar
The G
Grease Monkey
Information
Posts: 558
Joined: 21 Jun 2009, 11:43
Location: Bucharest, Romania
Has thanked: 408 times
Been thanked: 284 times

Post by The G »

Whoa, that's an old thread!

You may have something here and I'll try it. I'll make some corrections, though:
- CA3130 is not an all-CMOS op amp. You have a schematic on page 4 and it is also written right at the beginning of the datasheet:
"CA3130A and CA3130 are op amps that combine the advantage of both CMOS and bipolar transistors."
- even if it were, it is an op amp. I have yet to find an op amp that overdrives in an nice way to my ear. For example, I'm a big fan of TLC2262/2272, which are closer to an all-CMOS op amp. They are MUCH quieter than CMOS inverters, but the sound you get by overdriving them is significantly different. And they have the stuttering decay problem of most of the op amps.

- the problem that originated this thread was the CMOS specific hiss, not hum or ground loops. I still believe - and aparently I'm not the only one - that you can have a extreme gain CMOS distortion without hum or buzz, but not without hiss.

User avatar
jonasx26
Breadboard Brother
Information
Posts: 118
Joined: 03 Jun 2009, 17:53
Location: Sweden
Has thanked: 84 times
Been thanked: 48 times
Contact:

Post by jonasx26 »

ggedamed wrote:I still believe - and aparently I'm not the only one - that you can have a extreme gain CMOS distortion without hum or buzz, but not without hiss.
I'd second this. It'd be very difficult to get around the 'built-in' inherent noise in the inverters. Won't ever ever come near any op amps in terms of noise. But they sure do clip A LOT nicer.

User avatar
juozasu
Information
Posts: 10
Joined: 03 Jul 2013, 17:33
my favorite amplifier: JCM800, Soldano SLO100
Has thanked: 1 time
Contact:

Post by juozasu »

ggedamed wrote: - even if it were, it is an op amp. I have yet to find an op amp that overdrives in an nice way to my ear. For example, I'm a big fan of TLC2262/2272, which are closer to an all-CMOS op amp. They are MUCH quieter than CMOS inverters, but the sound you get by overdriving them is significantly different. And they have the stuttering decay problem of most of the op amps.
CA3130 was only an example. There are a lot's of "True MOSFET" opamp's.

Have You tried TS912? I did not notice any of decay problems. Clipping is very similar to CMOS inverters.
ggedamed wrote: - the problem that originated this thread was the CMOS specific hiss, not hum or ground loops. I still believe - and aparently I'm not the only one - that you can have a extreme gain CMOS distortion without hum or buzz, but not without hiss.
Of course, thats why I am suggesting to try a good alternative. Because:
ggedamed wrote:I started to experiment with JFETs
J-FETs clipping characteristics are very different from CMOS. But CMOS opamp's clipping (TS912 "CMOS DUAL OPERATIONAL AMPLIFIER") is very similar to CMOS inverters.

Also, have You tried MC14069UB (in more than 2 gain stages), it should be a little bit less noise than ordinary 4069. I didn't.

User avatar
The G
Grease Monkey
Information
Posts: 558
Joined: 21 Jun 2009, 11:43
Location: Bucharest, Romania
Has thanked: 408 times
Been thanked: 284 times

Post by The G »

Hmm, TS912 look really interesting - latch-up imunity, very good output load handling, low supply current, ESD protection and so on. I definitely need to try it.

I only tried HCF4069UBE, HEF4069UBP, MMC 4096E (an old Romanian CD4069 copy) and I'm pretty sure I tried a CD4069UBE. I have yet to try CD4007. I gave up when I decided that the noise is coming from the CMOS structure itself.
I still want to try a circuit where the CMOS inverter is fed with an already amplified signal, but my to-do list is way ahead of me.

BTW, I wasn't looking specifically for CMOS inverter clipping, but more for a nice-to-my-ear clipping. JFETs are quite fine and at this moment my favorite distortion is a JFET based one. There is always a better one, though :mrgreen:.

User avatar
jonasx26
Breadboard Brother
Information
Posts: 118
Joined: 03 Jun 2009, 17:53
Location: Sweden
Has thanked: 84 times
Been thanked: 48 times
Contact:

Post by jonasx26 »

Still .. It's apples and oranges. Two very different devices operating in very different ways.
See the internal schematics. But hey, if it sounds sweet it sounds sweet.
Attachments
Internal schematics
Internal schematics

User avatar
juozasu
Information
Posts: 10
Joined: 03 Jul 2013, 17:33
my favorite amplifier: JCM800, Soldano SLO100
Has thanked: 1 time
Contact:

Post by juozasu »

ggedamed wrote:BTW, I wasn't looking specifically for CMOS inverter clipping, but more for a nice-to-my-ear clipping. JFETs are quite fine and at this moment my favorite distortion is a JFET based one. There is always a better one, though :mrgreen:.
Several years ago I was trying various clipping stages (Mostly high gain): Diodes, jFET, Tube's (GROOVE tubes trio preamp, JCM800 preamp and so on), I found that there is nothing better than i.e. modified BSIAB 2 clipping ("Minibooster" stages). ()
But when I heard reezafratzitz, now I can say that there is nothing better :). Now I experimented with TS912, and found very similar clipping sound characteristics as 4069, but without extreme hiss noise.

User avatar
The G
Grease Monkey
Information
Posts: 558
Joined: 21 Jun 2009, 11:43
Location: Bucharest, Romania
Has thanked: 408 times
Been thanked: 284 times

Post by The G »

jonasx26 wrote:Still .. It's apples and oranges. Two very different devices operating in very different ways.
See the internal schematics. But hey, if it sounds sweet it sounds sweet.
Hi, Jonas! I know that TS912 is way different from a CMOS inverter, I just said it's an interesting part. I'm not searching a certain sound, so I try everything it appear interesting. And if it doesn't have the usual op amp decay problem, all the better.
juozasu wrote:[...]
Several years ago I was trying various clipping stages (Mostly high gain): Diodes, jFET, Tube's (GROOVE tubes trio preamp, JCM800 preamp and so on), I found that there is nothing better than i.e. modified BSIAB 2 clipping ("Minibooster" stages). ()
But when I heard reezafratzitz, now I can say that there is nothing better :). Now I experimented with TS912, and found very similar clipping sound characteristics as 4069, but without extreme hiss noise.
OK, here's mine :lol::

User avatar
jonasx26
Breadboard Brother
Information
Posts: 118
Joined: 03 Jun 2009, 17:53
Location: Sweden
Has thanked: 84 times
Been thanked: 48 times
Contact:

Post by jonasx26 »

ggedamed wrote: Hi, Jonas! I know that TS912 is way different from a CMOS inverter, I just said it's an interesting part. I'm not searching a certain sound, so I try everything it appear interesting. And if it doesn't have the usual op amp decay problem, all the better.
Hi :D
Yeah I know you know. But the characteristics that make the CMOS inverters clip way sweet is inherent to CMOS inverters alone.
A CMOS op amp behaves nothing like an inverter when clipping. Sure both can sound good, but as far as audio amplification is concerned, all they have in common is "MOS" in their name.
I'm sure you're already fully aware of this. Thing is, I assumed the thread was about inverters and confusion followed..
I'd suggest Jfets for 'similar' clipping and lower noise. Carry on gentlemen.

User avatar
juozasu
Information
Posts: 10
Joined: 03 Jul 2013, 17:33
my favorite amplifier: JCM800, Soldano SLO100
Has thanked: 1 time
Contact:

Post by juozasu »

ggedamed wrote:OK, here's mine :lol::
Good sounding overdrive/distortion! But I still hear specific J-Fets clipping sound not similar to MOS.
Do You use "mini-booster's" type gain stages?

Video is very interesting and informative.

User avatar
juozasu
Information
Posts: 10
Joined: 03 Jul 2013, 17:33
my favorite amplifier: JCM800, Soldano SLO100
Has thanked: 1 time
Contact:

Post by juozasu »

I'd also recommend not to give up up with CMOS, and try the simplest noise gate to use with high gain setting.

download/file.php?id=21750&mode=view
Noise gate is "formed" by D5, D6;

Effectiveness of this simple gate:

From 7:10

Post Reply