Pellucid Compressor: an ultra low noise compressor for guitar and bass

[jonny.reckless]

Just noticed something on the schem: R16/17 aren't really 1k are they? (Consuming 4.5mA!)

Haha, yes! I run everything from a Voodoo labs pedal power 2+ so I don't typically worry about that, but they could quite happily be 10k if you were concerned about running from a PP3 battery 

[Rob Strand]

Good demo.   Despite any small points I think you achieved you goal!

1. low noise and a NE5532 biased via a 1M resistor are not entirely consistent.

It usually boils down to the size of the input cap.  Maybe increasing the 22n to 47n would improve things a small amount.

you're not using the diode bias. I know that Merlin isn't convinced these are worth the bother, but I'm not sure I entirely agree.

In practice it's hard to tell the difference.   Puzzling from a theoretical point of view but true.

Quote from: merlinb on June 19, 2019, 03:21:23 AM

    Just noticed something on the schem: R16/17 aren't really 1k are they? (Consuming 4.5mA!)


Haha, yes! I run everything from a Voodoo labs pedal power 2+ so I don't typically worry about that, but they could quite happily be 10k if you were concerned about running from a PP3 battery

That caught my eye as well. 

Perhaps some pre-emphasis/de-emphasis would help noise.

[jonny.reckless]

Perhaps some pre-emphasis/de-emphasis would help noise.

Indeed it would. I've played around with this idea: basically putting a treble booster in front of the compressor and a complementary treble cut after it. Not only does it reduce noise (assuming your treble booster is super quiet that is), but it affects the way the side-chain works so it becomes less sensitive to bass frequencies. In some ways this a better way to do compression for guitars depending on what you place after it in the signal chain. It's more subtle and natural sounding and sounds less squished even at high gain reduction, especially with humbucking pickups which are bass heavy to begin with.

In other circuits where I use an LM13700, the Reckless Vibe and the Auto de FET guitar amp, I use pre-emphasis to reduce the noise. OTAs are pretty noisy devices unless you put them in the feedback path of an opamp or use emphasis to manage hiss. It's also absolutely essential to keep the base impedance as low as possible; I find 22 ohms is the sweet spot, anything above 100 ohms and the noise floor really starts to rise especially as the bias current approaches maximum. Also, beware cheap LM13700s from China on eBay. I've been stung with some absolute crap. Noisy as hell and high distortion compared with legit TI parts In my experience the TI LM13700s are quieter than the NSC versions. Mouser sells the DIP16 TI version at $1.76 each for 10 off which is not bad. Tayda only stocks the NSC one.

I guess we should really all be using THAT corp Blackmer VCAs which are super quiet and high fidelity

[Rob Strand]

Not only does it reduce noise (assuming your treble booster is super quiet that is), but it affects the way the side-chain works so it becomes less sensitive to bass frequencies. In some ways this a better way to do compression for guitars depending on what you place after it in the signal chain. It's more subtle and natural sounding and sounds less squished even at high gain reduction, especially with humbucking pickups which are bass heavy to begin with.

It affects the side-chain if the detector is detecting to signal after the pre-emphasis and before the de-emphasis.  The feed-forward compressors don't change but the feedback compressors depend on where the circuit blocks are placed.

I find 22 ohms is the sweet spot, anything above 100 ohms and the noise floor really starts to rise especially as the bias current approaches maximum.

You did a good job hunting that down.  I've have seen that type of thing before but I don't remember going down that far.  Maybe the transistors in the LM13700's these days have lower "rbb" parameters making it worthwhile.   The NE5532 lets you drop the resistors to anything you want, so it all worked out nicely.

[bool]

Iirc TL Audio channel strips also used a ota-around-opamp compressor topology. So if you can hunt down some TLA schemos, you could also look there how they did it.

[EBK]

I guess we should really all be using THAT corp Blackmer VCAs which are super quiet and high fidelity

Once you go Blackmer, you never go back(mer).

[andy-h-h]

After reading the many very informed comments here, I'm must say that I'm a little hesitant to post this.  I built the Reckless Diode comp and liked it, so thought I might try try the Pellucid.   

So here's my vero layout - I've built it, and it works just fine.   So consider this verified.

Thanks for your design work Jonny - much appreciated.

[andy-h-h]

I think everything is OK? I've tested it on the bench with input signals from 10mVpp to 3Vpp, from 80Hz to 8kHz, with continuous tones, bursts, pink noise, and step amplitude changes, and it seems to be working acceptably. Attack time is level (and frequency) dependent but nominally measured at 3ms, release time around 200ms.

As configured, each OTA requires approximately 400uA into its bias pin to hard limit the amplitude at maximum compression with a maximum signal level of 3Vpp. This causes the collectors of TR1 and TR2 to rise to about 5V at maximum gain reduction, due to the 10k current sharing resistors, well short of saturation. Release time is set by C1, with 47uF giving a release time of around 200ms measured on the bench. I quite like fast attack and release for guitar compression, it makes it feel more 'alive' when you are playing, especially with single coils. Most dynacomp style compressors have way too slow release for my taste. You might want to increase C1 to 100uF for a bass compressor though.

Hi Jonny,

How might one go about adjusting the attack, if this is at all possible?   I put a 100uf cap on a switch with the 47uf for the release rate, I wouldn't mind to have the option to play with the attack a little.  If not, I might just add a blend pot.

Thanks for your work, it's a very nice little compressor (along with your diode comp).

Andy

[jonny.reckless]

How might one go about adjusting the attack, if this is at all possible?   I put a 100uf cap on a switch with the 47uf for the release rate, I wouldn't mind to have the option to play with the attack a little.  If not, I might just add a blend pot.
Thanks for your work, it's a very nice little compressor (along with your diode comp).

Thanks, I am glad you like it.
I don't think you could easily adjust the attack using this two transistor peak detecting sidechain. Maybe increase the 10k resistors (R2, R4) feeding the transistor bases? Or add a resistor (say 1k) between the collectors of TR1 and TR2 and the 47uF release cap C1. It would have some effect, but also probably make the knee softer.
Jonny

[PRR]

> adjust the attack



??

[jonny.reckless]

Yes, like that  although 10k might be a bit too long? I guess try it and see how it sounds. You might get overcompression with slow attack in a feedback limiter like this.

[andy-h-h]

> adjust the attack



??

I'll try it on the weekend.  I like the idea of having the option to soften the knee too, so if that's an unintended result of this, I'll be happy with that.

Thanks both. 

Still think I might try a blend pot as well, just to see how it sounds.  I have a feeling that I'm going to play with things, and then just decide that it works just fine as is....   

[PRR]

> 10k might be a bit too long?

I would counter that 47uFd (against 100k release) is "long".

However a 10k LIN will easily set to 1k, 50ms, which leaves only a small blip of un-limited signal. A 10k AUD will trim to a few hundred ohms for few-mS attack. Further timing tweaks should be obvious.

[jonny.reckless]

The 100k is just there to pull the collectors down the last volt or so to ground when there is no signal. The release time is defined by the 47uF cap and the 2 x 10k resistors into the bias pins, which look like 2 diodes to ground. Theoretically 235ms but measured to be less than this, somewhere around 150ms on my unit by eye on the scope, which is about where I like release time for guitar compression. It's a subjective thing but most commercial compressors have too slow release for my taste. The guitar feels more alive and responsive to me when you play with a shorter release, especially with a strat.

[andy-h-h]

The release time is defined by the 47uF cap and the 2 x 10k resistors into the bias pins, which look like 2 diodes to ground.

So could one replace the two 10Ks with a dual gang pot and have more adjustment on the release?  I also play bass, which is why I'm interested in both attack and release settings. 

I'm enjoying this and your diode comp.   

[rankot]

Thanks Jonny, this is quite interesting! Will try to build one once I finish current projects (15+)

[PRR]

> release time is defined by the 47uF cap and the 2 x 10k resistors

Yes; I was blind.

> replace the two 10Ks with a dual gang pot

No. If you turn pots to zero the OTA burns up. You could add series resistance to both. But this is coming down to not having much spare gain in the sidechain. Dinking with resistors is likely to just make the action softer without so much control of timing.

As a general thing, to change *both* attack and release the same way, change the 47uFd.

[jonny.reckless]

If you really want to be able to control both release and attack, you probably want to implement Merlin's suggestion of using one of the spare buffers in the LM13700 to buffer the control voltage from R9 and R10, then implement Paul's suggestion of a 10k pot in series with the collectors of TR1 and TR2. When you have done this, at this point, R3 defines the release time (in conjunction with C1) and the 10k series pot defines the attack time. I would keep R9 and R10 both as 10k.

This will definitely change the character of the compressor quite a lot, and will probably sound a bit weird with slow attack times and short release due to the sidechain architecture and values involved, but I'd be interested to see how (if) it works out for you.

Jonny

[Rob Strand]

Jonny
[pic]

In that type of configation, sometimes moving the release cap before the attack is better since it avoids forming a voltage divider when attack is long (large Rattack) and release short (small Rrelease).

[PRR]

> avoids forming a voltage divider when attack is long