THE ENGINEER'S THUMB... At last, a better compressor!

[Kevin Mitchell]

A previous question of mine has been overlooked I hope someone can shed some light.

This strip board layout has omitted the 2n9001. Would this be a problem? What purpose does this diode in it's position serve?

I would imagine the noise and overload characteristics are not 100% identical.

Thank you for elaborating.

[samhay]

Do you mean 2N7001 - an alternative to the BS170 on Merlin's schematic?

[slacker]

This strip board layout[/url] has omitted the 2n9001.

The missing diode is for power supply reverse polarity protection, providing you don't reverse the power then you don't need it and it does nothing

[Kevin Mitchell]

Do you mean 2N7001 - an alternative to the BS170 on Merlin's schematic?

I meant the 1n4001. Sorry I have way too many part numbers zipping in my head as of recently lol.

The missing diode is for power supply reverse polarity protection, providing you don't reverse the power then you don't need it and it does nothing

Thank you very much. I should have guessed.

[Kevin Mitchell]

Hey guys I'm back!

After a few bread boarding fails and many distractions I've decided to try again last night with some success.

EDIT: Fixed some connections and now compression is very noticeable. Sometimes too much compression when ratio is cranked. Only issue now is a gaining buzz when I'm not plucking any strings. I'll start from scratch again and see what happens.

I've learnt that pretty much all all controls (minus the compression (ratio) could be better set as mounted trimpots. Attack, decay and level if desired. A 1 knob compressor sounds more practical anyhow seeing how there isn't much range or use with additional 3 controls + volume.

[Kevin Mitchell]

take a feed from Pin 7 of IC2 (the opamp connecting to the PNP), hook it up to the Base of a NPN, and wire V+ through the Collector and LED from the Emitter for a simple-yet-handy compression indicator. score!

Just found this on page 8. I'll give it a shot - I hope it does what I'm looking for reliably. We'll see tonight!

[Kevin Mitchell]

I've tried the LED method I've quoted and did not have much success. It resulted in more noise and popping. I've tried different variations thinking maybe I misunderstood the connections with no luck. Had better results with B on the - instead of output but it's not stable or consistent.

Can someone please validate this picture for being a correct interpretation?

take a feed from Pin 7 of IC2 (the opamp connecting to the PNP), hook it up to the Base of a NPN, and wire V+ through the Collector and LED from the Emitter for a simple-yet-handy compression indicator. score!

Any ideas?
So the ouput of that opamp to B of a NPN. Simple enough.


Ehh I'll keep at it.

[merlinb]

I've tried the LED method I've quoted and did not have much success. Can someone please validate this picture for being a correct interpretation?

You'll need a resistor (e.g. 1k) in series with the LED. As shown it should go dimmer when there is more compression, although it sounds like you still have issues with the actual audio circuit (?)

[Kevin Mitchell]

Yeah there's something connected wrong and I can't spot it.
-The bread boarding life...

After I posted that diagram I found yours on page 9 with the 470r to vb. I've gotten results close to what you've described but a little finicky. I'll report back with a sample of the sound and the LED in action when all is well.

Thanks for the awesome compressor Merlin 

[PRR]

> the ouput of that opamp to B of a NPN. Simple enough.

That won't work. The LED is always-on and carrying infinite current. Also fouls-up the compressor.

There's a simple linear-proportional way to add an indicator. See the BC327 PNP and 1K resistor? This flows a current proportional to over-level, and closely related to amount of gain reduction. That current is weak and already committed to tickling the LM13700. But we can take the same Base signal, to a second PNP, with a smaller resistor, and get a proportionally larger current we can dump to an LED.

(This form of "current mirror" has a small 40mV mirror-error which is totally moot for a visual indicator.)



The light happens when the compressor is "doing something", which makes sense to me. If the LED is off, it is just a buffer/booster. When signal gets loud, so that the sidechain tickles the '13700 to turn-down the signal, the LED turns-on and gets brighter.

For bench-test only, you can put your volt-meter *across* that 1K resistor. Use a 2V range (auto-ranging may be confused by dynamic guitar signals). The DC voltage is zero for no- or small-signal. When signal is excessive, voltage will increase to 1V (maybe a bit more or less), fading back to zero as the note decays.

[Kevin Mitchell]

The PNP idea worked although the LED slowly dims out at a constant rate. It does light up accordingly with the expected amount of compression. There may be an issue on the bread board still. It's obvious that as the LED dims out there's a slowly amplifying noise as there was before. Perhaps the 3080 I'm using is faulty. I'll keep at it.

[PRR]

> the LED slowly dims out at a constant rate. It does light up accordingly with the expected amount of compression.

What it probably should do *with guitar*. Guitar plucks, big start, heavy compression. Guitar fades-out, compression releases. The guitar fade-out is not "constant", depends on what strings (speaking length, gauge) and how solid the axe is; but for musical reasons the fade-out is in a narrow range.

Compare to piano, upper octave just goes "Tik", or organ which can go on forever.

So what is your present problem? (Too many pages to go back through.)

> Only issue now is a gaining buzz when I'm not plucking any strings.

The no-signal gain of this compressor (most guitar compressors) can be very high. This WILL amplify ALL the hum and buzz in the room (also hiss in the amplifier).

If you are "on breadboard", and it passes signal, and squashes, and the only problem is hiss/buzzz/hash when it goes to idle, what you need is a closed metal enclosure to keep the crap out. Time to box it up. Even if only in a bread-pan as a quick-check before you drill/polish/paint something nice. (If you have a battery power amplifier and can run the breadboard on a battery, take it out in the middle of the lawn. Not under the power-line like my lawn.)

[Kevin Mitchell]

Thanks for all the info I appreciate it.

What I mean by the LED slowly dimming is if I do a hard finger pluck and mute it quick the LED still slowly dims out even when I mute the strings so they aren't ringing out. The longer or continuous I pluck the strings or really rail on some chords it actually effects how long it takes for the LED to start to fade out again. Like it's stuck on for a long while then finally goes out. And when it does fade it amplifies any noise. If the LED is acting with the compression then something is surely out of whack if it's compressing for a way longer than than it should. I'm hoping it's not the 3080 since I was hoping to use it for the final board. I'll try the bread board again and with a switch between the 3080 and a LM13700 just to see what may happen if anything at all.

These single coils aren't really quite anyway but I do believe something isn't right here beyond interference. We'll get to the bottom of it. I'll start working on a PCB and have it pretty and boxed up come next weekend.

[PRR]

> mute it quick the LED still slowly dims out even when I mute the strings

Ah. That is also expected. A very sudden return to high gain would "breathe", pump-up the hiss too fast, as you have noted. All audio compressors have a "release" time-constant. Here it is the two 1uFd caps and the 1Meg bleed resistor, around 2 Seconds. In "normal" guitar use, the string fade-out is about this long, a good fit. If you instant-mute, you get a second or two of slowly-rising gain before the hiss comes back up. Used appropriately, the compressor brings-up the decaying tails of the notes, but if done too much the rising hiss is a problem.

Just for the experience, tack-solder a 47K across the 1Meg that comes off a 1uFd cap. That gives a good short attack. So short it may only be good for a military intercom, where we need "every syllable" at FULL loudness. Or maybe drums. 200K or so may be better for other music.

The perennial curse of compressors is that they help with old problems (overload, uneven levels) and introduce new problems (moosh, pumped hiss). It takes time to dial-in an acceptable amount of compression.

> hoping it's not the 3080

There ARE bad '3080s but they usually don't work at ALL. Some fraudster re-marks opamps or logic chips, they don't pass audio and generally show way-wrong DC readings. If it is changing gain I'd be real confident of the '3080. Solder, wiring, value, or soldering mistakes remain popular problem.

[electricco]

I built it and... WoW ! 
Super compressor, the best stomp comp I ever used so far. Really good job Merlin !!! 

I built the 5 knobs one and in some settings I can recreate either CS3 or dynacomp sound as well 

the pro is the versatility of this stomp, but at the same time in some settings there is no musical application range...
I think 510Kresistence + 500KB Pot will be better for release POT ( in some setting 1M will make sort of volume swell)

I think also Ratio+Threshold should be limited as well.. but I do not know well the right mod for these (any suggestion is welcome!)

What I feel the need for is a tone Control (the bright switch works but not enough), using it for the arpeggio in 1° position on Guitar bass chords there is too much loss of high.
Can the bright switch turned into a tone control in style of TS9 or similar ? (changing the resistor with a pot)
Or can be used the other OP of LM13700 for tone purpose (if so, how...  ) ?

Thanks again for sharing!!
Looking forward for any suggestion.

[patricks]

Hi all,
I ordered the parts for the ET a while ago and I'm looking forward to building it, but now I've seen the GTi version! Is there a vero layout for the new version floating around anywhere?
Cheers
Pat

[merlinb]

Can the bright switch turned into a tone control in style of TS9 or similar ? (changing the resistor with a pot)

You could replace the 10k resistor (the one in series with the 4.7n tone cap) with a pot wired as a variable resistor. 47k or 100k would probably do. Maybe with a 1k to 5k resistor in series with it, to limit the range a bit -you'll have to experiment.

[patricks]

Hi again,
A simple question to begin with - to add a threshold control to the GTi, would I just replace R10 with a 1M log pot?

Now the other questions, first about the LED bar graph display - is it possible to quantify the amount of compression at each step? I like to put numbers on things
The TI datasheet has a table on page 4 showing threshold voltages and signal strength, but I'm not sure how to translate it to "dB reduction" for a compression meter. The LM3915 might be easier to implement, since it's 3dB per step. The 3915 is easier for me to get a hold of, too, which is a bonus.

Speaking of putting numbers on things, is there an easy way to work out some values for the attack, release, and ratio at various stages of pot rotation? It wouldn't be too tricky to do in a DAW after I've built it, but if there's a way to work it out beforehand that'd be great.

Cheers
Pat

[merlinb]

Hi again,
A simple question to begin with - to add a threshold control to the GTi, would I just replace R10 with a 1M log pot?

Which schem version are you referring to? I recommend a linear pot, rather than log.

Now the other questions, first about the LED bar graph display - is it possible to quantify the amount of compression at each step?

I don't think so, because the numbers will be different depending on the ratio setting. It's so long since I did this that I can't even find the GTI schematic! But I think the bar graph gives only a relative indication between no compression and max compression, whatever max happens to be for that ratio setting.

Speaking of putting numbers on things, is there an easy way to work out some values for the attack, release, and ratio at various stages of pot rotation? It wouldn't be too tricky to do in a DAW after I've built it, but if there's a way to work it out beforehand that'd be great.

Hmm, I doubt you can do it in advance. The attack/release times are somewhat non-linear (incalculable) owing to the type of precision rectifier. I think you'll have to measure it the old fashioned way!

[patricks]

Thanks for the feedback, Merlin

The schematic is in the pdf posted on page 8 of the discussion:

Would u be so kind to share the 2 layer 5 knobs PCB layout here ? Thank you in advance. 

Here you go, a single-sided version:
https://dl.dropboxusercontent.com/u/57831278/115x60mm.bmp
https://dl.dropboxusercontent.com/u/57831278/TheEngineersThumbGTI.pdf

If the bar graph is just a relative indication, I'll probably go with the LM3915. That'd produce a display that's less detailed when the signal's being compressed the most, but should have ever so slightly better resolution when the signal's just starting to become compressed. All academic, really, though.

Good to know about the attack/release times, too. It'll be easy to do some measurements after it's built. I'll probably just get distracted and start playing, though
Cheers
Pat