Bearhug Compressor - schematic, demo, layouts, & forthcoming 1776 PCB!

[midwayfair]

This is one of my design/hacks. The Bearhug is a tonally transparent, ultra-quiet, tiny, and easy-to-build compressor. It occupies some middle ground between the ultra-subtle compressors like the Flatline/Afterlife and Orange Squeezer and the super squishy compressors like the Ross.

Here is the build document from Josh at 1776 Effects (includes the etch):
https://dl.dropbox.com/u/9878279/Jon%20Patton%27s%20layouts/Bear%20Hug%20BOM.pdf

There will be a PCB run release by 1776 Effects sometime around the New Year.

Some background

Compressors are one of my favorite effects -- I've used an Orange Squeezer as an "always on" pedal in my rig for years now. I've built a lot of different compressors now, and there are a large number of good or great designs out there. I wasn't so much looking to improve on a great design as to find a little bit of unused or underused turf in the DIY community.

I had been talking with Ray Ring (Circuit Salad is his website) recently after building his compressor, which uses an opto-FET, and he mentioned the phrase "linearized FET" as one way to create compression. The use of a linearized FET frees up some interesting design space. It allows anything from a very, very tiny amount of compression to ...well, more attenuation than you'll ever need. And they're faster than an optical design -- the attack can be as little as a few milliseconds. While looking over lots and lots of different compressor schematics, I recognized the use of a linearized FET in the Rothwell Lovesqueeze. I built it and liked it, but wanted to change a few things, such as improving how it reacted to the guitar's volume and the input impedance, and getting better separation between the compression and audio levels.

Here's what I ended up with:
Josh McClarren (Gtr2/1776Effects) drew this version of the schematic.

My schematic (with notes and mods): https://dl.dropbox.com/u/9878279/Jon%20Patton%27s%20layouts/Bearhug%20compressor%20schematic.pdf

What it sounds like:

(I recommend watching in Youtube because the annotations don't show up properly in the embedded version.)

My goals with this design:

1) It had to be easy to build. So no weird resistor values or uncommon capacitors. No difficult to find or unobtainable parts. Pretty much part has a workable substitution even though they're all common. As a side benefit, it's dirt cheap. The entire circuit cost me around $7 including the 9mm pots.

2) No weird biasing or other issues that make it difficult for someone to tell if the effect is working properly. As much as I love Orange Squeezers, if the builder doesn't know what it's supposed to do or sound like, they won't bias it correctly and then it just works as a boost. I wanted to avoid this.

3) It had to be quiet. These days, honestly, if your compressor design hisses and buzzes like a Ross/Dynacomp, it's time to go back to the drawing board. I'm ridiculously happy that I succeeded with this one. This circuit is completely silent -- there's no circuit noise at all, and the high signal-to-noise ratio is identical at all compression settings. (You'll bring up the noise floor when running it as a boost, but it's not contributing any noise -- just amplifying the pickup's hum and your guitar signal with it.)

4) I wanted it small enough to fit in a 1590A. Check!

How it works

The audio path is a MOSFET booster (basically Jack Orman's with a couple adjustments) -- this gave the effect enough output to still work as a boost, but it's super quiet and has extremely high input impedance. There are a few changes to accommodate the diode voltage drop at the 9v input, and I used a negative feedback cap (C4) to cut radio frequencies rather than a cap to ground, but this is all standard stuff.

Tonally, this circuit neither cuts nor boosts any guitar frequencies. The input and output caps are sufficient for the tonal frequency of at least a 4-string bass (I don't have a 5-string to test on). C4 can be adjusted to add or cut some overall available brightness (I used 47pF in my build because, say it with me, "Jon likes his treble").

The output volume will be at unity right around noon. The highest setting will be boosted by ~10-15 dB, depending on a handful of tolerances.

There is no tone control in this design, but if you really want one, the best choice is AMZ's version 2 of Mark Hammer's Stupidly Wonderful Tone Control.

The compression path works like this: Your signal is tapped at the collector of Q1, where it is amplified like crazy by a high gain transistor. The comp knob sets the gain within a fairly small range (500R, with a 47R minimum). Your signal is then peak-to-peak rectified (essentially the rectifier circuit from the Lovesqueeze; apparently it's a fairly standard method), and a negative voltage swing of anywhere from about -.2v to -2v is produced at C9, with the size of the swing depending on the size of the signal after it was amplified by Q2. The negative voltage swing then drives up the resistance on the FET (Q3), which at last is connected to the emitter of Q1 through a 47uF cap. The FET's default in this setup is completely on. The resistance of a 2N5457 is only ~300Ohms, so this sets the gain of the Q1 stage a safe distance above the distortion point but still "pretty darned loud." When your input signal drives up the resistance (typically anywhere from about 100 Ohms on the lowest settings to as much as 2K with the highest settings with a really, really hard strum), that lowers the gain of the MOSFET stage, thus decreasing your guitar signal's volume and creating compression. Whew!

C9 is a hold cap, fairly small as hold caps go so that we can get subtle settings. The decay of the cap is set by a parallel resistor; here, we've used a pair in series and a High/Low switch (Sw1) that bypasses the larger resistor, giving us two decay levels. When the switch is on ("Low"), the 10K gives an extremely fast decay -- basically, it will stop compressing immediately after it stops receiving a signal. This makes it work more like peak limiting. When the switch is off ("High"), the added resistance slows down the decay, increasing the total amount of compression available and adding a "duck and swell" effect at the highest settings. 47K is the nominal setting -- the maximum "Low" setting will be about equivalent to the minimum "High" setting -- but 100K may be used for a more dramatic change between Low and High modes.

Here's my perf layout ...

PDF: https://dl.dropbox.com/u/9878279/Jon%20Patton%27s%20layouts/Bearhug%20layout.pdf

And an etch layout from Josh!

(The etchable image is in the build doc above.)

And here's my build, 1590A on perfboard:


Credit where credit is due: This circuit design uses some elements from the Lovesqueeze and Jack Orman's (AMZ) Mosfet booster. Mark Hammer and PRR helped me understand the rectifier. Joshua McClarren (Gtr2/1776 Effects) designed the etch and PCB layouts. Extra props to R O Tyree for his help in understanding compressors in general.

[Kesh]

Awesome. Been looking for a subtle, low part count, quiet comp to use as input for overdrives.

Have you tried different 2N5457s? Wondering if the manufacture variations of that part cause problems.

[WhenBoredomPeaks]

Looks cool!

How it compares to a "real" compressor/limiter? Or maybe i should ask it this way: what is the value of the fixed settings? (attack, ratio)

From your post it seems like that the attack is really fast and the release can be really fast (more limiter like) or slower. (more compressor like) I guess the pot marked as Comp is a thresh
hold pot. What about the ratio? Is it high/infinite (limiter like) or low (like compressors)?

Can i use different germanium diodes? (I have 1N34As)

The reason i am asking because i'd like to have this configured as a clean sounding limiter (fast attack, fast release, high/infinite ratio) and build an other compressor with more pots for the usual guitar stuff (like raising sustain etc.)

[~arph]

Nice!  I like your circuits.
yes compression is hard to hear on youtube, but the amount of noise in your vids is making it even harder..sounds like you're in the shower with the water running 

[midwayfair]

Looks cool!

How it compares to a "real" compressor/limiter? Or maybe i should ask it this way: what is the value of the fixed settings? (attack, ratio)

From your post it seems like that the attack is really fast and the release can be really fast (more limiter like) or slower. (more compressor like) I guess the pot marked as Comp is a thresh
hold pot. What about the ratio? Is it high/infinite (limiter like) or low (like compressors)?

Can i use different germanium diodes? (I have 1N34As)

The reason i am asking because i'd like to have this configured as a clean sounding limiter (fast attack, fast release, high/infinite ratio) and build an other compressor with more pots for the usual guitar stuff (like raising sustain etc.)

I don't have a spectroscope, so I can't give you exact numbers on something like the ratio. I can explain how it works for the most part, though.

-The ratio can be very high depending on the size of the input signal. There's no true infinite ratio design in a stompbox (nor, I think, in analog compressors, period), but a FET's maximum resistance is, what, 300M? You're need an infinitely large input signal to get there. If I had to guess based on other compressors I have lying around (lots!), I'd say that the ratio must be lower than 10:1, probably more like 5:1. Turning up the comp knob does lower the threshold, but it also in part controls the ratio: because if you turn down the gain of Q2, it won't amplify the signal as much going into the rectifier, and the decay characteristics will take care of keeping the FET's resistance from just increasing forever.

-The attack is set by the way the rectifier works: a negative wave cycle generates the negative voltage. The attack is anywhere from .5 to 1.5 cycles. That's still absurdly fast for any stompbox compressor, 10mS or less. PRR's explanation is:

"If your first transient is positive, nothing happens on that half-cycle. For a simple SIN wave, action is delayed almost 1.5 half-cycles from the start of the SIN, 1 half-cycle from the first (positive) peak."

-The decay, as noted, is set by the values of R9/R10 and C9. I believe it's ~47ms on the low decay setting and ~220ms on the high decay setting at max Comp settings.

-You can use 1N34A and it'll be okay, but that's needlessly expensive. It also might compress a little more overall, but you can just turn down the comp pot if that happens. BAT41 or 1N270 would be better if you just can't get 1N60P. However, 1N60P (Fv ~ .25) is a silicon work-alike of the 1N60, and chosen because they only cost 2c each and are available from dozens of sources: Ebay, Mouser, Taydam, etc. etc. They're the cheapest diode with Fv under .3 and they have a tight tolerance. I use them in many different circuits requiring schotkeys and germaniums.

Hopefully that's all your questions.

Awesome. Been looking for a subtle, low part count, quiet comp to use as input for overdrives.

Have you tried different 2N5457s? Wondering if the manufacture variations of that part cause problems.

I'm pretty sure it would have to be wildly out of spec, to the point of being broken, for it not to work. I test pretty much every FET I get, and I've never had a 2N5457 that was far enough off of the bell curve to be able to say what might happen. This has been built multiple times with no consideration given to the individual FET's specs and it has worked exactly the same each time. This isn't a biasing situation, which is where FETs can sometimes cause problems.

Nice!  I like your circuits.
yes compression is hard to hear on youtube, but the amount of noise in your vids is making it even harder..sounds like you're in the shower with the water running 

Sorry! It was in the middle of a hurricane! That was the only free time I really had to get it filmed.

[rutabaga bob]

NICE JOB! 

[rring]

Sounds great- solid approach and very importantly - an efficient, and clean design.

very cool

[Bill Mountain]

How well can it handle peaks?  I play a high output bass and most pedal compressors are too squishy for me.

Dirt boxes sound too compressed sometimes so I'm looking for a compressor/ limiter project.

[midwayfair]

How well can it handle peaks?  I play a high output bass and most pedal compressors are too squishy for me.

Dirt boxes sound too compressed sometimes so I'm looking for a compressor/ limiter project.

The lowest comp settings have an pretty high threshold. A strat barely triggers it except with a very hard strum. I'd say almost any setting on 10K decay would be fine for active pickups as shown, but if the minimum setting is too much, any change that lowers the gain of Q2 will also raise the threshold and reduce the total compression available. This includes simply omitting R12, which would leave open the possibility of still using it with a passive guitar or bass.

I don't have an active bass, so I can't tell you for absolutely certain -- I had to use a boost to see how it handled anything hotter than a broadcaster (which is pretty hot but not on the level of a Firebird or EMGs). It handled it well enough that I was confident that 547R was the right value for the comp pot. It only takes maybe 10 minutes to breadboard, and I'd be interesting in knowing if it CAN handle actives and still stay subtle.

I can say that it doesn't cut any bass freqs.

[Bill Mountain]

I've got everything except the 1k pot and the zener.  I'll have to use generic Ge's and a 2N5458 but that shouldn't be a problem.  Any benefits to discrete over IC's like the Rothwell.

[midwayfair]

Josh at 1776 Effects got the PCBs in today!

Here's the link if anyone is interested and doesn't etch or build on perfboard:
http://1776effects.com/store/

The build document on his site also contains the etch image (same as the document in my top post).

I've got everything except the 1k pot and the zener.  I'll have to use generic Ge's and a 2N5458 but that shouldn't be a problem.

Hi, Bill, sorry I never saw your questions. The 5458 and any generic Ge will both work, though the 1N60P was chosen because it's ridiculously cheap. The 5457 seemed a little more responsive (I tried both).

Any benefits to discrete over IC's like the Rothwell.

Mostly it's a flavor thing. I don't really claim that this is better than something like the Rothwell, just that it's another way of doing something similar. I think it has its good points, however.

The BS170 has a little bit of its own compression. I really liked the super high input impedance. It meant fewer capacitors/filtering and fewer parts in general. The MOSFET audio stage has a little less noise than the op amp in the Rothwell. Although I didn't really run into distortion with either one, theoretically, the Rothwell could distort with a boosted input signal at max compression on the front end of the note (the gain pot in the Rothwell is limited to 5K because any higher and it distorts). Not that distortion is desirable in any compressor, but at least if someone does manage to overdrive the audio path in a MOSFET (I couldn't with this ... 300Ohms on the source is a lot), the MOSFET's distortion sounds nicer. Finally, it's a little more modular, since there are multiple substitutions that work for most of the parts, and transistor stages always seem to be a little more tweakable than op amp stages. With an op amp design, you can't do something weird like ... stick a germanium transistor in the audio stage and turn it into an overdrive with sag ... (... this just occurred to me and now I need to go try it)

I know some people prefer working with op amps; I prefer to work with transistors.

Obviously one huge disadvantage is that the parts are either obsolete or getting there. I don't think there's a SMD version of the 5457, right?

[bean]

Ordered! I'm looking forward to this

[midwayfair]

Ordered! I'm looking forward to this

[Keppy]

Cool! One question, though: C9 looks backwards to me, and the notes in one of your links refer to it as "reversed." What's with that?

[maartendh]

Negative voltage from the signal (filtered out by the 1n60p rectifierdiodes) controls the fet; so with respect to ground the polarity of this cap is correct.

Nice design Jon! Will try it soon.
Maarten

[Keppy]

Ah, I see now. I was neglecting C7 & C8, the DC blocking caps. Thank you.

[midwayfair]

Hey, if anyone's interested, 1776 Effects current has the PCBs for this design discounted to $6. Just in case anyone was on the fence about building it.

[bluebunny]

I posted this in the Pictures thread a while back:

Today's offerings:
Jon's Bearhug compressor, on my Vero layout.

Then Jon said this:

Aww, that's an adorable graphic. Thanks for building it. Hope you like it!

And I replied:

It is indeed very cute.  It was a pleasure building it - thanks for sharing with us, Jon - but it doesn't quite work yet!     It's a great little distortion pedal, but not so good as a compressor just yet.  Most likely self-inflicted...    Need to do a little debugging...

And Jon added:

Start a debug thread and I'll help you out.

And I said:

Thanks Jon - will do.  Just want to go over my homebrew Vero layout first.  I'm hoping it's something obvious and stooopid.  Otherwise, a debug thread is on its way!

Edit: found a bug in my Vero layout - missing track cut!      Will fix when I get a chance and report back...

Well, long story short - the missing track cut was indeed the culprit - all fixed now!  This is a great little compressor.  And cute.     Thanks again to Jon for sharing with us.

[midwayfair]

Glad you got yours working!

[bluebunny]

Thanks Jon.  Yeah, I'm pleased too.  I like your design a lot and the results are great.  Very quiet and gets the job done nicely.  I'll be trying it out on my bass next.