Snow Day OD - schem, layout, demo, discussion ... and name change?

[midwayfair]

I had off work on Monday due to snow, and came up with the idea of tossing together a quick overdrive to pass the time. And because I'm not good at doing simple things like that, I instead ended up breadboarding and working on a new design all week.  

The Snow Day OD is a FET- and MOSFET-based amp simulation overdrive running on 18V that includes a switchable miniature compressor circuit and soft clipping in a "power tube" section. It goes from glassy "mostly" clean to either open or slightly compressed edge of breakup that feels very similar to one of my favorite amps, up to a medium gain compressed sound. It has lots of harmonics without sounding overly clipped, and a very wide dynamic range.

It uses some principles from several Runoff Groove circuits, Aquataur's mini compression circuit from his mods to the Umble, and Mark Hammer's Stupidly Wonderful Tone Control (which rolls off treble without changing the volume).

Although it's not directly based on any amp in particular and I wasn't going for a particular sound, I did have a couple amps in mind (both Deluxes -- a friend's 5E3 clone and a 1949 Deluxe at Invisible Sound Studios in Baltimore) while I was designing, aiming for a similar feel and gain level to them if not necessarily their tonality.

The pedal runs on 18V internally for a lot of drive but also a lot of headroom, making for plenty of volume dynamics even with the clipping.

Schematic



EDIT 3/10/14: I will add the missing Va connection to Q5 ASAP.

Demo (using my Sheraton)

Audio-only demo (using the DQ caster; no EQ, no effects, direct into the recording interface, also includes stacking with a My Little Klony and a Fatpants).

How It Works

Power section
The power section is a standard voltage doubler. The additional voltage helps coax more gain out of the FETs. FETs always clip when the input signal exceeds a certain voltage, but there is a limit to the voltage gain they can provide at any given supply voltage. They also need to be biased in a particular way to ensure the correct harmonics to emulate tube breakup. (See Runoff Groove's FETzer valve article for more information on the harmonics generated.) The higher supply voltage means that we can use 2N5457s, which have a higher input voltage threshold before clipping, while getting closer to the voltage gain of a J201 (which clips much, much sooner). This helps increase the total dynamic range of the pedal.

D2 is out polarity protection. Z1 is optional as overvoltage protection, but sometimes zeners can go bad when used with charge pumps. Otherwise the standard power filtering is there.

"Preamp"
The "preamp" section is primarily three FETzer valves in a row. Each 2N5457 is biased to 2/3 supply voltage (ended up being about 11V in mine). The bias point helps generate the correct harmonics to get the FETs sounding closer to tubes.

The gain control is right after Q1. R4 provides some minimum resistance (I didn't have room for this on my perfboard layout, but I added it to the schematic). The lowest setting will be clean even with humbuckers. Noon has a really nice edge of breakup sound -- extremely dynamic and picking attack responsive. Since the gain is simply a volume control after the first stage, rolling down your guitar's volume is pretty much identical, so the effect also cleans up nicely and smoothly.

Different FETs could be used with different results. For instance, a J201 in Q2 (with appropriate re-biasing) would be much gainier (and will also lose some clean settings). I got some good results with a 2SK170 in Q3 when using a telecaster.

Just a note: Runoff Groove's newer circuits use clipping diodes after each stage in a way that ensures smooth breakup and to avoid hard clipping the FETs, but I decided to skip that little innovation because I found they were generating clipping earlier than I liked in this particular instance.

Compression!
A miniature compressor stage, based on the one created by Aquataur for his Umble build, is created by R10-14, D1, and C6. D1 generates a negative voltage, which travels through R7 and pulls the gate of the FET negative (by about 2V at max from my measurements), dropping the input level to Q3.

The result is a bit of sag reminiscent of a tube amp at higher gain settings. The compression kicks in and becomes measurable and observable around noon on the gain dial on the "high" compression setting (at least with my strat; other guitars might trigger it earlier). There's an accompanying reduction in distortion and some added perceived sustain. The compression also swallows a bit of treble when it turns on, which creates a smoother sounding distortion sound. It also means that the effect will bright up a bit when you roll off your guitar volume, which is something I've been dying to find a way to do in overdrives that don't rely on guitar loading for a while now.

The attack time in mine is 10mS, and the decay is 100mS. The switch disconnects the compression in the center; one setting has a threshold-limiting resistor (220K) to ease things up on higher gain settings, and the other setting bypasses that resistor and lets through all of the signal.

It's worth noting that there's no capacitor to set the cutoff frequency of the compression, but I did consider adding it. Unfortunately there's only so much signal to go around, and in any case I like that the compression works more on chords (which have more bass and also need more clarity) than on single notes.

Tone (Treble) Control
The tone control, Mark Hammer's Stupidly Wonderful Tone Control, is after Q3. The pot is a continuous 100K series resistance between Q3 and Q4, and as the resistance increases between lugs 1 and 2 (pot turned clockwise), more treble is dumped through C8. The values were calculated to roll off treble starting at 15KHz down to 685Hz. It can get a little brighter and a little darker than the bypassed signal. It's not extreme (no "playing under water" sounds), but it's effective at controlling the harmonics going into the power tube section. R15 sets the highest cutoff frequency of the treble cut (combined with C8).

"Power Tube" -- Mosfet soft clipping stage
After the tone control comes the "power tube" section. This is a MOSFET amplifier with MOSFETs arranged as diodes in its feedback loop. The MOSFET amplifier part is a pretty standard way to hook it up without needing a Vb reference voltage. I used a MOSFET here because it didn't distort as much as a transistor, and it also sounded a little less harsh when it did overdrive.

I loved the sound of MOSFETs as clipper diodes in the Mossy Sloth, but the arrangement is a little different here. MOSFETs conduct in two directions when you connect the drain and gate together: One way is the body diode (a simple silicon diode -- and no, it has no special sound) and the other way is ... something special that soft clips over a very wide range of a couple volts before it finally hard clips. By connecting them in series at their source pins, the body diode in each one prevents the diode from conducting in reverse, and the result is a very high, soft clipping threshold (about 2V and up).

One last trick to soften the clipping was the use of a limiting resistor (R18) in series with the clipping arrangement, similar to how it's done in certain op amp based designs like the Bluesbreaker and AMZ's big muff mod with a "warp" control. Big thanks to tca on DIYSB for lots of help (and diagrams) in understanding this (plus he suggested the MOSFET as Q4!). Although I don't fully understand everything R18 is doing, but the gist of it is that once the diodes conduct, they appear in parallel with R16, and they will lower the negative feedback. R18 therefore helps set the clipping threshold and ALSO sets how "rubbery" the clipping will sound, AND just for good measure, also acts a bit like a clean blend. Values between 10K and 100K seemed to work with roughly similar overall results, but it's not quite as simple as a lower value = more distortion. I ended up not really being able to decide between 22K and 33K and went with 33K in the schematic simply because there were a bunch of them in the schematic already. So why isn't this a trimpot? I dunno, I guess it could be, but I'm not totally sure it's worth the trouble.

C9 sets the cutoff frequency of the clipping. It's bit enough to definitely pass all the frequencies on the guitar (and might be overkill, but I'm not sure how to calculate it). Smaller values (e.g. 1uF or 100nF) will let through more bass.

Output section
R20 and R21 form a voltage divider, like a volume control permanently set to half. Adjusting their ratio to each other can boost or cut the output. I found it was sufficient to get a small boost to output even at very low gain settings.

An output buffer comes after that, followed by the volume control. The output buffer could be unnecessary (after all, the MOSFET probably has lower output impedance), but I thought it sounded very slightly better with it there. I'm not sure if it added some small amount of distortion, changed the bass or treble content subtly (perhaps cutting the lowest or highest frequencies a bit extra), but I figured it was only four extra parts and I had room on the layout.

Here's my build:



And here's my perfboard layout:


If there's enough interest, I could do a small PCB run. Also, I do plan on making an etch layout, but I want to see first if I need to make a PCB layout (because I might be able to do them both in eagle instead of just making the perfboard layout into an etch image).

As always, I welcome any suggestions or questions about the design.

[Bill Mountain]

This looks fantastic.

Have you tried it with a bass at all?

[midwayfair]

This looks fantastic.

Have you tried it with a bass at all?

Sorry, Bill, the best I can do for a bass is an octave pedal after the guitar. The bass player in my band has a perpetual loan of my bass (for all intents and purposes, I've given it to him).

It has a cut at 72Hz cut "sort of" at C5 and a 67Hz cut at C2. Those are really the only places with meaningful bass loss; every other cut is below 20Hz (I think). So I think you'd have to double both of those caps.

In any case, I imagine it would be quite a bit dirtier with bass, so I'm not sure what other changes would have to be made. Probably raising R19 at the very , perhaps even using a FET with higher Idss in Q3 for less dirt in the preamp.

I'm probably not the best guy to ask about modding some effects for bass, since I'm quite bad at playing that instrument.

[samhay]

Looks interesting Jon.

The MOSFET arangement looks pretty cool. What's the thought process behind tapping the drain in the middle of the two 10k resistors?

I haven't seen that sag/compression circuit before - is it useful enough to warrant its own effect, e.g. a Fetzer stage with a sag control?

In your schematic, it looks like you are missing a connection to V+ from the collector of Q5.

[midwayfair]

Looks interesting Jon.

The MOSFET arangement looks pretty cool. What's the thought process behind tapping the drain in the middle of the two 10k resistors?

The pedal was INCREDIBLY loud without that. It's just the basic arrangement in the Fuzz Face, Rangemaster, Fuzzstainer, etc. It seems to be nicer to the volume knob behavior and treble content than a resistor in series with the volume pot.

I've also read that doing a divider like that results in a small drop in midrange. I couldn't find enough reliable information about it, though, so I didn't want to put my foot in my mouth in the "How It Works" section. It does sound a little different when I change the ratio; it gets fatter as you tilt it toward the transistor. So there must be something going on there.

I haven't seen that sag/compression circuit before - is it useful enough to warrant its own effect, e.g. a Fetzer stage with a sag control?

Unfortunately, not really, unless you're okay with adding another active device. You really need a lot of drive before it kicks in, enough to overdrive a typical FET. So I would say either at least two stages (it COULD go after Q2 with a bypass cap on either or both FETs) or just add a transistor to drive the envelope.

Oh, the other thing is, you need a small gate resistor to let through the small amount of negative voltage, so if it were right at the input, you'd need an input cap AND it would heavily load the guitar pickups. No idea how it would act. That has you up to at least two active devices regardless, with enough drive that it's probably not totally clean anyway. And increasing the decay will do funny things to the tone ... There are too many good simple compressors for me to really think this one's worth porting to its own effect!

In your schematic, it looks like you are missing a connection to V+ from the collector of Q5.

Yup, I'll fix it ASAP! Thanks for looking out.

[Kipper4]

Another great project. Thanks for the backlog Jon.
It must have a fair output to direct record to the interface. A nice feature and a great help to those recording in apartments and small studios that want a tube sound without the decibel count from a real amp. Nice job.
Lots of tone variation too. The compressor is a nice touch too.
I really like the zen drive with the mosfets in the feed back loop. I understand it's a different set up seeing as the zen uses op amps. But I like the touch responsiveness of this and your set project seems to do similar.
Great work man.
Rich
Also it reminded me of some of the tones in Snowy Whites Bird of Paradise.
I can't post you tube links from this machine else I would. Sorry.

[sajy_ho]

Great work man and nice sound too.
I had a question; can I bias all JFETs using fetzer?

[midwayfair]

I had a question; can I bias all JFETs using fetzer?

The FETzer valve setup will bias any FET, yes. You can't necessarily just use any FET with the bias trimpot and source resistor values shown, though. The Runoff Groove article has more information than I can possibly regurgitate here: runoffgroove.com/fetzervalve.html

[Bill Mountain]

Is Q5 missing a V+ connection?

[midwayfair]

Is Q5 missing a V+ connection?

I really will fix that, I promise. 

EDIT 3/10/14: I will add the missing Va connection to Q5 ASAP.

In your schematic, it looks like you are missing a connection to V+ from the collector of Q5.

[Jopn]

Hey Jon,

I'm playing with a guitarist who's currently recording his clean electric straight into his digital recorder.  It sounds godawful so I'm putting together a preamp pedal for him to run through first.  Thoughts on using this, possibly with a condor sim after it?

[samhay]

Thanks for your thoughts Jon.
Sounds like I need to play with the MOSFET drain tap approach, but will add the sag control to the long list.

[midwayfair]

I'm playing with a guitarist who's currently recording his clean electric straight into his digital recorder.  It sounds godawful so I'm putting together a preamp pedal for him to run through first.  Thoughts on using this, possibly with a condor sim after it?

Well, check the audio only recording for the straight-to-console sound to make sure you like the sound of this particular pedal. I would still want to EQ the sound for direct recording, so he's going to need a decent DAW to go with it -- and most decent DAWs these days have at least competent amp simulation. I really like Logic's amps for clean sounds, but they do get kind of gross on the dirty sounds. Basically, this pedal is like everything up to the speaker hook up, and as we all know there's a lot more going on after that point!

FWIW, I tried this into the tweed, vox, and blackface sims on my computer and it sounded much more realistic than the pedal by itself, but the distortion from the pedal was better than anything I was getting out of the sims.

I can also HIGHLY recommend the Azabache and Britannia for what you're talking about -- they are both more specifically made to sound like particular amps and they both sound amazing. And they both already have some cabinet simulation.

My goal with this project was less to emulate the sound of any particular amp than to add the feel and breakup of an amp to an existing platform. Even using a solid state amp is not the same as direct.

I can't answer the question about how it will pair with the Condor specifically, as I haven't built that project, but I'm guessing it will be a big help and a good alternative to a plugin amp sim -- the EQ response of the Condor is very, very similar to the EQ shape on many of the amp plugins.

Lastly, I would venture a guess that even just a buffer will be a big help with your friend -- chances are the input impedance of the recorder is too low to handle a guitar. Keep that in mind for your preamp.

[Jopn]

Thanks a tonne Jon!  I had Britannia half built for this purpose when I saw Snow Day pop up.  I'll definitely be building both, but I'll probably just continue on with Britannia for this current (and urgently needed!) build.

Cheers!

John

[pinkjimiphoton]

[bluebunny]

Another great project. Thanks for the backlog Jon.

Ha ha!   

Code Select Expand

todo_list = todo_list + 1

[Ben N]

Very nicely done, Jon, as always, and the out-of-the-box thinking is most welcome. But, honestly, I'll never get to this if someone doesn't do a PCB. 

[midwayfair]

I just wanted to give people a heads-up that there will be a PCB for this. I won't be the one making the layout. That's the most I can say about it for now, though; I don't want to spill the beans about who is going to be working on it. But I will say that as usual, it will be someone in the community with more talent than me.

[sajy_ho]

Hey Jon, hope you're OK
I'm trying to breadboard your circuit, but as I couldn't find any 2n5457 around, I'm using J201s for all the fetzer stages; my problem is the high gain. It sounds very brittle and I want a bluesy type OD like yours.
So what do you think? Can I use 9V voltage presented in the circuit for fetzers to lower the gain and run the rest of it on 18V, wouldn't it interact with the saturation switch? or just use a voltage dividder to do that?

[midwayfair]

Hey Jon, hope you're OK
I'm trying to breadboard your circuit, but as I couldn't find any 2n5457 around, I'm using J201s for all the fetzer stages; my problem is the high gain. It sounds very brittle and I want a bluesy type OD like yours.
So what do you think? Can I use 9V voltage presented in the circuit for fetzers to lower the gain and run the rest of it on 18V, wouldn't it interact with the saturation switch? or just use a voltage dividder to do that?

Do you have any other FETs around? You could mix and match a couple lower gain FETs with a single J201 in Q2 and get something really close. J201s more than twice as much gain. If you're going to use those, I'd say run it on 9v, and increase the 33K between each stage a little (I'm not sure how much, but try 47K). You might lose a bit of treble, but the gain will be similar.

The big thing that will be affected is that you're going to be VASTLY reducing the dynamic range. You're going from an average of -1.58V Vp (the voltage at which it clips) to only 0.83V.

The compression shouldn't be affected. You just want to get down to approximately the same gains.

EDIT: I'm not sure what'll happen with the MOSFET stage. I finalized that after moving to 18V.