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

[andy-h-h]

> avoids forming a voltage divider when attack is long

I tried the other version, and I was getting quite a bit of distortion on some settings.  I did use the buffer on 9 & 10, but that shouldn't matter.  Still have some distortion on some settings with this version, but it's not as bad.   

When A B testing the original vs modded, I'm preferring the original.  Might just be a case of it ain't broke, so don't fix it.   I'm sticking with the original.   

[sergiomr706]

Hallo, It s very interesting what you are posting, i was thinking about building a dual compressor, with a crossover like the one trace elliot used on their comp, and using a lows compression with an attack of about 20ms anda release of 120 and the Highs being compressed with attack of few ms, 2-5 and a 70ms release. Being fixed, could these times be implemented in that design? Thanks a lot for your efforts and research

[jonny.reckless]

Hallo, It s very interesting what you are posting, i was thinking about building a dual compressor, with a crossover like the one trace elliot used on their comp, and using a lows compression with an attack of about 20ms anda release of 120 and the Highs being compressed with attack of few ms, 2-5 and a 70ms release. Being fixed, could these times be implemented in that design? Thanks a lot for your efforts and research

You could build 2 of these and have different values of C1 to control the time constants, or try the attack / release mods as described above. Also consider putting a filter in the sidechain rather than requiring dual compressors. A treble boosted sidechain can sound great for clean guitar and bass.

You do need to be careful with the crossover design if you want the end result to sound coherent. I've dabbled with this sort of idea. Generally 4th order Linkwitz-Riley filters are considered to be the best approach since you get constant relative phase between the 2 halves.
https://en.wikipedia.org/wiki/Linkwitz%E2%80%93Riley_filter

BTW I was head of R&D at Trace Elliot for a while back in the early 90s, fond memories 

[bluebunny]

BTW I was head of R&D at Trace Elliot for a while back in the early 90s, fond memories 

So can I blame you for why my oh-so-tiny-and-cute BLX-80 weighs so freakin' much??! 

[sergiomr706]

Amazing! You were involved in the gp7sm /12 or V series? I played the  12 bands for a little time, It was not mine, but we shared rehersal place with another bassplayer who owned one, and It was such a great tone, (at least to me). Lately I started reading about the comp part and It seems that lows and Highs share a wide space btwn 250 and 900 Hz.. maybe parte of his charm? I first had thought of using a 3rd order linkwritz and the try different crossover points, 300 /400/800... So what can i day? Double thank you for your efforts and designs. This all helped shape a lot of great bass sounds!

[bool]

...
BTW I was head of R&D at Trace Elliot for a while back in the early 90s, fond memories 
...

How cool is that! TE's were my no.1 go-to amps in 90's and 00's. I still have a small "7-band eq" combo from that era somewhere in the closet (with burned mosfets and bridge I think). And everywhere I played I required a TE 400 or 500-watter (usually a combo) as my backline amp. The only other amps I tolerated were swr redheads. ("like no ampeg for me, please").

Btw, whoever did it, the mid-shape1 was a little better than mid-shape2, a sentiment that was common to other TE players around here ... but sound eng's hated it!

[Rob Strand]

BTW I was head of R&D at Trace Elliot for a while back in the early 90s, fond memories

Hey, that's pretty cool.  You must be proud of all the big names using that stuff.  The bass amps were everywhere at that time.

In the back of my mind I seem to remember the crossover on the high and low compressor bands didn't overlap symmetrically like speaker crossover.   Do you remember the reason for that?

[PRR]

> I'm preferring the original.  Might just be a case of it ain't broke, so don't fix it.

I have a feeling that, for the number of parts used, the original does about as good as can be. Tacking on "a couple more parts" asks too much of its simplicity.

A "real" compressor with full control of timing will have 2 to 12 more opamps in the sidechain. Many dozens of such things have been posted to the internet, plagiarize shamelessly.

Truly simple schemes, like the Orange Squeeze (within limits) and this Pellucid, are harder to find. It is good to have another easy-build alternative. It is hardly "a pig", but still I doubt lipstick is going to make it "prettier" than it already is.

[Ben N]

Also consider putting a filter in the sidechain rather than requiring dual compressors.

As I am given to understand, basically the approach of the emphasis control Marshall's ED-1 - a Big Muff tone control in the sidechain. Very subtle effect there.

[jonny.reckless]

In the back of my mind I seem to remember the crossover on the high and low compressor bands didn't overlap symmetrically like speaker crossover.   Do you remember the reason for that?

That was one of Clive Button's designs. He designed a lot of the early stuff, he was a one man R&D department for Trace for a long time before I joined. There was a lot of empirical tweaking involved in those products, probably it was done that way just because he liked the sound of it The graphic EQ design on the 12 band series was also "wrong" in that the filter Qs were all over the place, but it sounded good and people liked it.

I had prototyped another dual compressor with 4th order LR crossovers around the same time, I seem to recall that one went into production in 1997 just before we went bust and then got taken over by Gibson for a while, at which point I and many others were laid off.

At that time I worked with Andy Ewen (now head of R&D at Victory amps) and Paul Stevens (now senior designer at Blackstar amps) on the Triumph valve guitar amplifier designs (Bonneville, Speed Twin, Trident etc), the V type bass amps, the Tramp tube guitar hybrids (very early use of JFETs in a guitar amp circuit), Trace Acoustic PA, lots of pedals (the quad chorus was one of mine), power amps, and many others. It was a fun ride. I spent a few years as Engineering director at ARCAM (the British HiFi company) then I switched to predominantly software engineering shortly after that, which I have done ever since, but I do keep my hand in with hardware these days. I've spent the last 11 years in the San Francisco bay area.

[Rob Strand]

That was one of Clive Button's designs. He designed a lot of the early stuff, he was a one man R&D department for Trace for a long time before I joined. There was a lot of empirical tweaking involved in those products, probably it was done that way just because he liked the sound of it The graphic EQ design on the 12 band series was also "wrong" in that the filter Qs were all over the place, but it sounded good and people liked it.

I guess that's the difference between "studio perfect" designs and purpose-built instrument preamps/amps.  I've seen the funky Q thing on a few products.

I had prototyped another dual compressor with 4th order LR crossovers around the same time, I seem to recall that one went into production in 1997 just before we went bust and then got taken over by Gibson for a while, at which point I and many others were laid off.

I don't think I've seen that one.  There was one design, probably around that era, which used only a first-order crossover.  I vaguely remember the design was feed-forward instead of feedback like the older designs.

before we went bust and then got taken over by Gibson for a while, at which point I and many others were laid off.

I'm not sure what it is with the music instrument/amp business but there's a pattern like that in many companies. Ernie Ball seems to have a done a good job with Music-Man.

At that time I worked with Andy Ewen (now head of R&D at Victory amps) and Paul Stevens (now senior designer at Blackstar amps) on the Triumph valve guitar amplifier designs (Bonneville, Speed Twin, Trident etc), the V type bass amps, the Tramp tube guitar hybrids (very early use of JFETs in a guitar amp circuit), Trace Acoustic PA, lots of pedals (the quad chorus was one of mine), power amps, and many others. It was a fun ride. I spent a few years as Engineering director at ARCAM (the British HiFi company) then I switched to predominantly software engineering shortly after that, which I have done ever since, but I do keep my hand in with hardware these days. I've spent the last 11 years in the San Francisco bay area.

I see analog electronics as a dying art.   Software engineering is pretty stable, although there's quite a gap between web-design and real-time processing.

Cool post, interesting stuff.

[Rob Strand]

I had prototyped another dual compressor with 4th order LR crossovers around the same time, I seem to recall that one went into production in 1997

I had a poke around in my schematics and I can't find that one.  My collection is by no means complete though.

[jonny.reckless]

I don't have a copy of any of them any more so I can't help there, sorry.

[Rob Strand]

I don't have a copy of any of them any more so I can't help there, sorry.

No worries.  I was interested to see the evolution.

I don't actually keep any circuits from any place I've worked at.  It's just a policy I've adopted.

[GreedyFly]

Does it sound good with humbuckers?

[Fancy Lime]

Hi there,

I tried to analyze the current noise on this thing because my experience with very similar input stages (though non-compressing) is also that they are very quiet compared to, say, a TL072. But with the relatively high loop impedance and the 10k series resistor on the input, I would seem to think that there is room for improvement, no? Unfortunately, I am not an EE and was unable to dig up a useful explanation of a rigorous noise analysis. All I did find, and I found a lot, exhausts itself in oversimplified guesstimations and narrow examples, which doesn't really help me understand how to do the analysis properly. Would someone be so kind to point me to a good source? Especially for how to treat a guitar pickup in a current noise calculation and why the input cap seems to make such a big noise difference.

Thanks,
Andy

[Fancy Lime]

Anyone?

[PRR]

> exhausts itself in oversimplified guesstimations and narrow examples

OK, I'm sure you know what you are asking, but I don't know the question or even the references.

> very similar input stages

Similar to which one? Jonny's first post? Or some later proposal?

The first post in this thread IS a BJT input, in a hiss-critical application (idle compressors hiss), and Jonny is no fool in these matters. He just puts a 1Meg on a '5532, and lives with the DC offset. He can't have been panicked about current hiss in guitar context because he has that 10k in front. Also 10k in the NFB loop.

[Fancy Lime]

Hi Paul,

I was referring to the first schematic. By similar I mean NE 5532 as a non inverting first boost stage with a 1M bias resistor and a few kiloohms of parallel resistance in the NFB loop.

Thing is, Jonny certainly does know what he's doing and when he claims that this thing is not noisy I'll take his word for it. But then there is this 10k input resistor and almost 10k parallel resistance in the NFB loop at the highest setting and these should contribute rather substantial noise in an opamp stage with substantial input bias current and current noise, like the 5532. The Johnson noise of the input resistor is amplified by the gain of the stage, which is 40db when maxed.

My question is: does any of this matter or is it all being swamped by the inherent noise of the pickups? To evaluate this, I would need to know how to calculate all the noise contributions. There are formulas for that and they are easy enough to find. However, all of the ones I found only consider resistance. But I know from personal experimentation that in a stage like this, the input cap makes a very clearly audible difference. Anything below 220n before a 1M bias resistor adds audible noise. How does Jonny get away with 22n here? Why does the cap even matter for the noise? What about the inductance of the pickups, does that interact with the input current of the opamp and if so how?

It's not a simple question with a simple answer. Hence my asking for a source to teach me the full rigorous treatment of these matters instead of the simplified versions you find in many app notes and manufacturer reports.

Thanks,
Andy

[PRR]

> Anything below 220n before a 1M bias resistor adds audible noise.

I've seen subsonic rumble from small input caps, but you seem to be describing something else. I don't know.

For most purposes we can get away with calling a guitar "50k"; the volume pot dominates all else except full-up.

That plus 10k in input and 10k in NFB makes 70k, which is 1.4dB higher than just 50k.

OSI for '5532 is like 4k (but quite broad) so 70k is well off optimum. It's too late here to figure what that comes out to.

TL072 has "no" hiss current but is way off from ideal hiss voltage. There are better choices today. Including, for many cases, the old fashioned '5532, if only because prime parts won't have excess hiss.