Understanding MOSFET clipping

Frequent asked about building blocks: gain stages, buffers, clipping configurations, ...
Post Reply
User avatar
commathe
Breadboard Brother
Information
Posts: 97
Joined: 01 Jul 2013, 22:31
Has thanked: 5 times
Been thanked: 7 times

Post by commathe »

I'm having a little trouble understanding mosfet clipping. My understanding is that there are two forms: the diode-connected mosfet (tying gate and drain together) and the body diode. However, when breadboarding around I found that it worked opposite from how I expected. Examples follow.

Here are the three layouts I used. I picked a mosfet symbol that showed the body diode as on the datasheet for the 2N7000 (the n-channel enhancement mode MOSFET I used):

Image

As you can see Mosfet A uses the diode to block the body diode. This is the set-up I expected to sound the most different. I expected Mosfet B to act simply as a diode with a higher forward voltage. I used the leds as a control for comparison (the forward voltage was fairly close). My results surprised me!

First, here is a comparison of each of them used to rails after a simple Pro-co RAT clone op-amp stage:

NOTE: the board style seems to be cutting off the far side of the spectrum analysis (where the most intersting stuff is). If you want to see it just right-click then "open in new tab"
LED:
Image

Mosfet A
Image

Mosfet B
Image

As you can see, the Mosfet A (blocking the body diode) and the LED to rail are almost the same. Almost! The spectrum analysis shows a little difference, and the difference was barely audible. I expected Mosfet B to act like that though, whereas it created a much more different sound. A noticeably different sound actually.

This stands out much more when I put then in the negative-feedback loop of my op-amp:

LED (op-amp feedback)
Image

Mosfet A (op-amp feedback)
Image

Mosfet B (op-amp feedback)
Image

Here you can see again that the LED and Mosfet A are close. The difference is more pronounced this time however. Mosfet A is definitely a bit more aggressive and brighter. Not in a bad way though! Probably not enough to be actually worth it though. Mosfet B on the other hand produces something noticeably different. It's got a lot less high end, and it's generally smoother.

So, my question is: why? What is actually going on that creates this behaviour?

User avatar
bajaman
Old Solderhand
Information
Posts: 4544
Joined: 26 Jun 2007, 21:18
Location: New Brighton, Christchurch, NZ
Has thanked: 593 times
Been thanked: 2048 times

Post by bajaman »

sorry but i cannot see ANY images :!: :scratch:
be kind to all animals - especially human beings

User avatar
commathe
Breadboard Brother
Information
Posts: 97
Joined: 01 Jul 2013, 22:31
Has thanked: 5 times
Been thanked: 7 times

Post by commathe »

bajaman wrote:sorry but i cannot see ANY images :!: :scratch:
Weird! They are visible for me. Try quoting the post, the links should be visible in the text

User avatar
snofla1900
Breadboard Brother
Information
Posts: 129
Joined: 23 Sep 2012, 13:33
Has thanked: 83 times
Been thanked: 7 times

Post by snofla1900 »

sorry but i cannot see ANY images
Can't see them either !

User avatar
commathe
Breadboard Brother
Information
Posts: 97
Joined: 01 Jul 2013, 22:31
Has thanked: 5 times
Been thanked: 7 times

Post by commathe »

:cry:

I don't know what's up for you guys but I can no longer edit my post either so here are direct links:

Three clipping schematics
http://s23.postimg.org/uzwbx0xbd/Screen ... _17_16.png

Led to ground:
http://s23.postimg.org/hpd2h8w4p/Diode_to_rail.png

Mosfet A to ground:
http://s23.postimg.org/h3o3l512h/MOSFET_1_to_rail.png

Mosfet B to ground:
http://s23.postimg.org/m8pb2re09/MOSFET_2_to_rail.png

Led in op-amp loop:
http://s23.postimg.org/miwtm3smh/Diode_op_amp_loop.png

Mosfet A in op-amp loop:
http://s23.postimg.org/x70kky2ll/MOSFET ... p_loop.png

Mosfet B in op-amp loop:
http://s23.postimg.org/nveinzq21/MOSFET ... p_loop.png

User avatar
deafbutpicky
Information
Posts: 49
Joined: 03 Jun 2011, 10:07
Location: germany
Has thanked: 27 times
Been thanked: 19 times

Post by deafbutpicky »

Hi,
I'm not sure if I understand your question right. From your pics I would go for the obvious concerning the differences:

in the Mosfet B setup the clipping threshold (Vf) is lower (smoother wave shape, esp. in OPA-loop)
and more clipping induces more harmonics.

You could try some shottkys additional to the bodydiode setup to raise Vf and compare them again.

This might be interesting if you don't know already (i.e. higher C in body diodes):

http://www.muzique.com/lab/zenmos.htm
I'm using capacitors to replace my guitar effects and give more clarity to the sound.

User avatar
commathe
Breadboard Brother
Information
Posts: 97
Joined: 01 Jul 2013, 22:31
Has thanked: 5 times
Been thanked: 7 times

Post by commathe »

deafbutpicky wrote:This might be interesting if you don't know already (i.e. higher C in body diodes):

http://www.muzique.com/lab/zenmos.htm
Fantastic. I had missed this detail. This must be it! The shape of the wave is radically different from the other clipping set ups. This slower-diode/higher-capacitance must be causing that.

I am still confused as to why I am not seeing much in the way of the gate-tied-to-drain effect though. In theory, it shouldn't behave like a normal diode at all as it will "slowly" switch itself on until the gate is eating up all the current... yet it is acting almost identically to my led clippers :? I'm wondering if putting a small resistance between the gate and drain will have an effect or not (slowing down the turn on of the gate). Definitely need to go back to the breadboard and do some more tests.

I guess my real issue is: Why is the the Mosfet A set-up behaving nearly identically to the LEDs? This set up *should* be the most different, and yet it is not.

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 »

Maybe LEDs are behaving differently than the standard diode clippers. Try the usual silicon diodes so you have the whole spectrum.

User avatar
commathe
Breadboard Brother
Information
Posts: 97
Joined: 01 Jul 2013, 22:31
Has thanked: 5 times
Been thanked: 7 times

Post by commathe »

ggedamed wrote:Maybe LEDs are behaving differently than the standard diode clippers. Try the usual silicon diodes so you have the whole spectrum.
Even if this is the case, the Mosfets are still not behaving like they should. In theory they will switch on slowly, which should give a very rounded and soft clipped waveform. Any circuit simulation I run of mosfets does this, I'm confused as to why my real life ones are not.

I guess I should add that pretty much every commercial pedal I see that uses mosfets is also using the higher capacitance body diode (OCD, Fulldrive, Zendrive etc.). Even the AMZ Shakka Bradda does. Although I read about the theoretical soft clipping mosfets could provide when you block the body diode and use the gate tied to drain configuration (a number of R.G. posts on here about it), I see no pedals using it. So something is up there too. :?

User avatar
deafbutpicky
Information
Posts: 49
Joined: 03 Jun 2011, 10:07
Location: germany
Has thanked: 27 times
Been thanked: 19 times

Post by deafbutpicky »

The problem with the Mosfet A type clippers is their high Vf (1.2V-2n7000 - 2.5V-BS170).
They need a lot of gain and headroom to be used as heavy clippers.
I'm using capacitors to replace my guitar effects and give more clarity to the sound.

User avatar
commathe
Breadboard Brother
Information
Posts: 97
Joined: 01 Jul 2013, 22:31
Has thanked: 5 times
Been thanked: 7 times

Post by commathe »

Thanks. I didn't consider this. I wasn't using the maximum possible gain in that set-up so it definitely warrants a going back to try it as such. I also found that the diode loop use of Mosfet A seems to behaving like it should. I might try putting some resistors at the source or something similar and see what effect this has too.

Lots of fun experimentation to do!

User avatar
Intripped
Cap Cooler
Information
Posts: 670
Joined: 25 Aug 2009, 01:03
Location: Italy
Has thanked: 985 times
Been thanked: 305 times

Post by Intripped »

hey! thanks for this, very interesting.
have you done some further experiments?
what is the best sounding configuration on your opinion?

here's a video from proguitarshop where Andy makes a demo of the new highly modded YATS from Earth Quacker Devices:

from 3:44 you can listen to different clipping options, mosfets included.
(there's a nice shot on the clipping section on the PCB too (...what do you guys think of D8? what is it for?))

@commathe: can you confirm these sonic differences, expecially between LEDs and mosfets? (what's the corresponding configuration?)

thanks!

User avatar
microbailey
Breadboard Brother
Information
Posts: 99
Joined: 05 Dec 2017, 23:19
Location: United Kingdom
Has thanked: 21 times
Been thanked: 55 times

Post by microbailey »

I've been looking at the Hermida Zendrive traced schematics which are posted in various places and have a circuit analysis question about the clipping arrangement.
First. my assumption, looking at how the circuit is connected around the op-amp is that the MOSFETs are diode-connected (Gate to Drain) so giving square-law I-V diode behaviour above their Vth. That's a neat twist on the the usual exponential diode I-V and I guess gives this pedal its overdrive sound.

The question is about the diodes in series with the MOSFETs - can anyone explain to me why there are two diodes in one direction and one diode in the other?
Is it deliberate to make the clipping assymetrical? If so, does that affect the odd/even harmonics produced?

I like the sound of this pedal and looking at the schematic I like the neat way the designer used the 2 caps (feedback and voice) to roll-off the op-amp gain at the low and high ends.
(I''m mainly a digital designer and so I'm easily impressed by analog "tricks" :applause: )
"There's something about a Gucci loafer kicking on a fuzz pedal" Alex Turner, Arctic Monkeys

User avatar
ppluis0
Diode Debunker
Information
Posts: 915
Joined: 14 Jul 2010, 18:33
Location: Argentina
Has thanked: 127 times
Been thanked: 299 times

Post by ppluis0 »

microbailey wrote:The question is about the diodes in series with the MOSFETs - can anyone explain to me why there are two diodes in one direction and one diode in the other?
Is it deliberate to make the clipping assymetrical? If so, does that affect the odd/even harmonics produced?
Exactly, that is the reason... 8)

User avatar
microbailey
Breadboard Brother
Information
Posts: 99
Joined: 05 Dec 2017, 23:19
Location: United Kingdom
Has thanked: 21 times
Been thanked: 55 times

Post by microbailey »

ppluis0 wrote:
microbailey wrote:The question is about the diodes in series with the MOSFETs - can anyone explain to me why there are two diodes in one direction and one diode in the other?
Is it deliberate to make the clipping assymetrical? If so, does that affect the odd/even harmonics produced?
Exactly, that is the reason... 8)
Great. Thanks ppluis0. I think I'm getting the hang of this :D
"There's something about a Gucci loafer kicking on a fuzz pedal" Alex Turner, Arctic Monkeys

Post Reply