Moon LNDer overdrive distortion - depletion mode MOSFET pedal design

[jonny.reckless]

In response to the Little Jim MOSFET distortion circuit https://www.diystompboxes.com/smfforum/index.php?topic=125174.msg1190435#msg1190435, I was asked if I could maybe use the LND150 depletion mode MOSFET in a distortion pedal design, as it might be less hissy. The Moon LNDer is what I managed to come up with. It comprises 3 common source gain stages in series, with a James style EQ at the end of the chain. It has less gain than the Little Jim, and is more from clean to mid gain than heavy distortion, but it responds nicely to playing dynamics. The two band James EQ is more flexible than a simple tone control and lets you match it to the input of your amp more easily. There's noticeably less hiss than when using BS170 MOSFETs which is nice. The LND150 has much lower input capacitance than the BS170 so I was able to get away without a unity gain source follower on the input, instead just connecting to the first gain stage directly, without too much tone suck.






Here is a video demo of the Moon LNDer in action:

The signal chain is guitar -> moon LNDer -> Two notes Torpedo CAB M (loaded with Celestion G12M impulse response) -> Strymon reverb -> PC.

[iainpunk]

SOUNDS GREAT!!!
wouldn't you place the bass control before the distortion/clipping instead of after? i always wondered why no one seems to do that, especially with lower gain overdrive circuits, to give it more versatility in the character of the drive.

cheers, Iain

edit: what range would i need to expect the source voltages to be? 0,8-ish volt?
i wonder what diode bias would sound like on a MOSFET circuit, i really like the sound on JFET circuits, a certain growl that resistive bias doesn't do.

[Fancy Lime]

SOUNDS GREAT!!!
wouldn't you place the bass control before the distortion/clipping instead of after? i always wondered why no one seems to do that, especially with lower gain overdrive circuits, to give it more versatility in the character of the drive.

cheers, Iain

edit: what range would i need to expect the source voltages to be? 0,8-ish volt?
i wonder what diode bias would sound like on a MOSFET circuit, i really like the sound on JFET circuits, a certain growl that resistive bias doesn't do.

I would not say no one does that thing with bass control before clipping. I always do that, if I have a bass control at all.

Care to share a design with diode source bias?

@Johnny
Great design! I've been meaning to track down some useful DMOS devices for use as variable resistors. Seems like these may be useable for my purpose, although I whish the on-resistance was a bit lower but I can probably design around that. Maybe.

Andy

[iainpunk]

this is a SRPP gain stage which is really cool, it has high gain and soft clipping, great as a starting point for most of my designs. the biggest downside is the relatively high quensient current and the need for somewhat matched-ish jfets. if you put 2 after each other, you get a nice gainey distortion. i have only tried it with BF245's since i have a bunch of those (near 50) and no other Jfet's, but i believe that if they are somewhat close in Beta, they should work.


cheers, Iain

[rankot]

In response to the Little Jim MOSFET distortion circuit https://www.diystompboxes.com/smfforum/index.php?topic=125174.msg1190435#msg1190435, I was asked if I could maybe use the LND150 depletion mode MOSFET in a distortion pedal design, as it might be less hissy. The Moon LNDer is what I managed to come up with. It comprises 3 common source gain stages in series, with a James style EQ at the end of the chain. It has less gain than the Little Jim, and is more from clean to mid gain than heavy distortion, but it responds nicely to playing dynamics. The two band James EQ is more flexible than a simple tone control and lets you match it to the input of your amp more easily. There's noticeably less hiss than when using BS170 MOSFETs which is nice. The LND150 has much lower input capacitance than the BS170 so I was able to get away without a unity gain source follower on the input, instead just connecting to the first gain stage directly, without too much tone suck.

I'm amazed how quickly you managed to make a prototype of this! 

[jonny.reckless]

I bought a bag of 50 LND150 from mouser. They were very closely matched, all within 50mV of each other for VGSoff. Much closer matching than you normally get with JFETs. In this circuit I'm running around -0.8V on the gate for a current of approx 200uA. This seemed to be the sweet spot for these devices in terms of getting a nice looking waveform with lots of second harmonic. The lack of gate current means you don't need to worry about blocking distortion.

[niektb]

I noticed the audio is a bit out of sync with the video, no big deal for a pedal demo of course, but a small tip is to clap at the beginning of the video and use that distinctive point to sync the A/V (and then remove the beginning of the video, to just before the actual video starts) 

[rankot]

Nice demo!

Have you tried to run it at 18V, for example?

[Ben N]

Sounds great! Plenty of gain to my ear, and really nice as an MIAB. Time to order some of those LNDs.

[jonny.reckless]

Nice demo!
Have you tried to run it at 18V, for example?

I did briefly, it didn't sound as good as with 9V, I presume I would need to readjust the bias points for 18V operation.

[Radical CJ]

SOUNDS GREAT!!!

+1

[jonny.reckless]

By increasing R7 to 100k and R6 to 47k, I was able to get a bit higher gain out of the circuit without having to add an extra gain stage. Here's a quick demo, recorded with my phone camera and mic, into a Blackstar HT20 set to clean.

[willienillie]

By increasing R7 to 100k and R6 to 47k, I was able to get a bit higher gain out of the circuit without having to add an extra gain stage.

Did you have to rebias?  If not, double-pole switch, Rock/Metal.

[jonny.reckless]

By increasing R7 to 100k and R6 to 47k, I was able to get a bit higher gain out of the circuit without having to add an extra gain stage.

Did you have to rebias?  If not, double-pole switch, Rock/Metal.

Just those two resistor changes, so it could be done with a DPDT switch. It gives about another 6dB or so of gain; just enough to take it into classic lead territory.

[PRR]

By increasing R7 to 100k and R6 to 47k, I was able to get a bit higher gain out of the circuit without having to add an extra gain stage..... another 6dB or so of gain....

In JFETs, voltage gain goes about as inverse of square-root of current. For similar voltage bias, R7 from 10k to 100k implies 1/10th the current and 3.16 times the gain, 10dB. R8 will shave that some, and this is not a JFET, but yeah "6dB or so" is expectable. And a clever place to find free gain.

The "drawback" is less extended treble. But the LND150 has 12AX7-like capacitances (it's not one of those monster MOSFETs), and 100k will support well beyond the guitar band.

[jonny.reckless]

In JFETs, voltage gain goes about as inverse of square-root of current. For similar voltage bias, R7 from 10k to 100k implies 1/10th the current and 3.16 times the gain, 10dB. R8 will shave that some, and this is not a JFET, but yeah "6dB or so" is expectable. And a clever place to find free gain.
The "drawback" is less extended treble. But the LND150 has 12AX7-like capacitances (it's not one of those monster MOSFETs), and 100k will support well beyond the guitar band.

Thanks. I didn't really notice any more treble roll off, the output stage rolls off pretty early anyway to tame the fizz. I actually like the LND150 - it's kind of JFET like in its behavior and tone, low capacitance (at least compared with the BS170) so you can couple the guitar straight into a common source gain stage and not get killed by Miller capacitance. Also you don't have to worry about blocking distortion or weird bias shifts due to the gate current as you would in a JFET or BJT design. And from my limited experience, they do seem pretty quiet too - certainly much less hissy than the BS170 / 2N7000, and comparable to J113s which have been my go to FET for guitar preamp circuits for about 25 years. I might start developing a few preamp ideas around the LND150

[PRR]

Readily sold as thru-hole for a half-buck. Worth putting on the idiots palette. My idiot is old so needs syntax tweaks.

Code Select Expand

*LND150 MODEL, from MicroChip, edited for Cadence/ORcad Pspice 9.1, (C)1999
*
.MODEL LND150   NMOS (
LEVEL=3
RS=150.00       
NSUB=5.0E13
DELTA=0.1     
KAPPA=1.O     
TPG=1           
CGDO=2.1716E-12
RD=40.0       
VTO=-2.0     
VMAX=1.0E8       
ETA=0.1
NFS=6.6E10   
TOX=1.0E-7     
LD=1.698E-9     
UO=862.425
XJ=6.4666E-7 
THETA=1.0E-5   
CGSO=5.09E-10   
L=10.0E-6
W=600E-6)
* .ENDS
*


There IS a well-known MOS model quirk right at VTO: a kink in what should be a monotonic curve. But it is at 1 micro-Amp so should be moot for audio. (The model takes VTO too literally while the crystal is more gradual.)

[PRR]

...I didn't really notice any more treble roll off...

You wouldn't. Because you follow Leo's feetsteps.

High-Mu triode, plate resistor 470k gives AM-radio treble (we don't like static). 270k gives a little less gain and more treble, Often seen on Gibson. After some fluttering, Leo Fender tended to use the 100k plate resistors. Taking 220k and 100pFd load this runs about -3dB @ 40KC (kHz), or 1dB at 20kHz, for one stage. "Too good" except a g-amp is multiple stages. Apparently Leo or his cohort could hear the difference 100k or something higher. Probably by a rising response not rising through cascaded -1dB@5KC stages.

[11-90-an]

Was googling about replacing tubes with bjts (you know, I try to make everything not BJT into BJT... due to availability ) and found this comment by merlin...

https://www.diystompboxes.com/smfforum/index.php?topic=108690.20
Check out reply 28...

Sorry not as related to topic, but wanted to share...

[rankot]

I know this is out of topic, but I can't find the answer myself, and I need your help - I plan to use LND150 as a source follower in my build of Alembic F2b, so I can have lower output impedance. Will it be OK to put it there, no hi freq loss? I tried to simulate this, and it seems to be fine.