Keeley compressor clone

[davph30]

I have increased the volume by removing the resistor across the level pot 100k pot sounds better, but am still getting quite a lot of tone sucking and am having no luck getting more treble

[Jay Doyle]

All compressors tend to suck treble. Nature of the beast unfortunately.

A couple of things you might try though: be sure that the first cap the signal sees, the cap to ground is 220 pF, if it is .0022uF like in R.G.'s schem it will suck treble. If it is already 220 pF try removing it and see if you like it that way. Otherwise you can try lowering the value of the cap right after the 3080, the one parallel to the 150k resistor.

Remove the 10k in series with the output pot for a little more output.

Is the 100k you are using log or linear taper?

[Jay Doyle]

I forgot to add, be sure to keep SOME type of cap in parallel with the 150k resistor after the 3080, it helps to stabilize the 3080.

[davph30]

Thanks Jay, I have already removed the 220pf. the The 100k is log

with thanks Dave

[davebungo]

ways to achieve less high frequency loss:

a) slow down the attack time of the compressor - transients will not be so harshly treated,
b) decrease the compression ratio,
c) increase the threshold (or gain/sensitivity as on my MXR limiter),
d) put some low pass filtering on the input to the sidechain - as a result, the sidechain will respond less to high frequency content and/or
e) put a little high frequency EQ on the output (although not too much as this may increase noise levels).

remember also that what you may be interpreting as tone sucking is probably/actually a loss of dynamics range which is the function of a compressor in the first place (so ask yourself if that is what you really want).

The best studio compressors are quite often referred to as "transparent", which is just another way of saying that they don't do an awful lot to the signal, or if they do, it is achieved in a more gradual way.

[markm]

In my experiences with compressors, if a guitar player is the type that dimes the controls on the guitar and doesn't use them for the tools that they truly are, more often then not, they will not be happy with a compressor. If on the other hand, one backs off the volume a tad and changes the signal level going into the comp, the two (guitar and comp) can work quite well together.
Usually the "tone-suck" complaints are by people who don't use the guitar's controls at all.
Beleive me, back off the volume a bit and use the two together as one instrument. Makes a difference to me anyway.
This is just my opinion however. 

[davph30]

I've tried everything I can think or and u have suggested , still no luck

i am getting a muddier sound than i expected no distortion or noise even when both controls are maxed.sustain and volume still has to be at about 3 o clock before its louder than the bypassed signal.

Im lost I have swapped all transistors and the 3080.

Any ideas?

[Jay Doyle]

Have you ever tried an actual Keeley or Ross? I ask only because you just simply may not like the sound of it. Maybe someone else can check the level at which the output exceeds the bypass but this comp isn't really meant to be a booster.

You just may not like the effect itself, unfortunately. It sounds like you have done everything possible. I don't remember the beginning of this thread. You checked to make sure the trim was set properly? You get a change in the sound when you rotate it?

[davph30]

I have checked the sound against the samples here http://www.steelbender.com/spotlightkeeleycomp.html and my pedal is much muddier against the setting used

[davph30]

Hi Jay

when i adjust the 2k trim i adjusted it to the centre it gets gated either side of this between about 11 o clock and 2 oclock

[petemoore]

  To let the volume control be set at a lower level and have the same or more highs, a small cap across it's input/wiper lets highs through...
  When the volume is decreased, resistance across the two lugs mentioned increases, this probably doesn't muddy it up much [?], but it's an opportune place to use a small cap to get a treble enhancement.
  A fixed R/C parallel affair, in the signal path just before the volume pot could be used, however this would also attenuate some signal

[markm]

The volume issue seems strange 'cause my Dynaclone and Ross comp can get quite loud on their own with no problem.
This circuit also is more single-coil friendly in my opinion as 'buckers seem a bit too dark for it sometimes.

[RobertKeeley]

Hi Jay

when i adjust the 2k trim i adjusted it to the centre it gets gated either side of this between about 11 o clock and 2 oclock

This MAY be a clue as to what is wrong.  I've found in MANY (all even?!?!) cases that you can hardly notice a difference to the sound even when adjusting that trim pot quite a bit from one extreme to the other...sure at the very ends, there will be a problem... but in general I think I've noticed much more play "allowed" before 11 oclock and 2 oclock.  So if it only works at the noon position but no where else, this is an indication that something may be wrong.  I've seen it before a bunch as we've built them...but I honestly can't remember what now...except that it was a part (resistor or cap) that is wrong.  For example a 15K where a 150K should be, etc, etc... IF you built the board, check that of course again.
Funny that you mention changing the transistors and chip... I have to beat my techs up on that subject... YOU DON'T ALWAYS START BY CHANGING THE THING WITH THE MOST LEGS ON IT!!!! 
LOL
I undertand, trust me, that is how I started many, many a broken device!
Robert

ps, I'll check a comp tomorrow and see what the play is in the 2k trimmer.  I bet it is pretty large if memory serves.
ps again...yes, the MPSA18 were chosen because of high gain.     I still think I could do better on the 'ultimate' transistor selection, just haven't gotten back to that idea yet!

rk

[R.G.]

The 2K trimmer is substantially useless. It can be effectively replaced by two 1K resistors.

The 2K trimmer was put in there to balance the two inputs to the 3080. This was back when all semiconductors, the 3080 included, were worse than they are today. It's only purpose is to keep the gain signal on pin 5 from feeding through to the output by nulling it out in the first diffamp pair inside the 3080. Modern 3080's don't usually need it.

Having a trimmer there for an uninformed DIY effects maker is an unendurable temptation to tweak it and then to worry when it's right. Set it to the middle and ignore it, or replace it by two 1K's - or tweak it infinitely until it's somehow more perfect to your ears.

The 3080 has a flat response well up into the MHz region. Treble loss is not coming from the 3080. The treble is deliberately rolled off at a couple of places in the circuit. Go there.

One of the worst problems with the Dynacomp and the Ross copy of it is that the same emitter follower which runs the gain feedback loop also runs the audio output. This was done to keep the thing simple, but it represents a conflict of interest for that transistor. It can't really do a good job of both at the same time - it's a compromise. The higher the current gain of that transistor, the better it does, which is why the 5088/5089/A18 series with their very high current gains produce better results. They isolate the high impedance 3080 output from the losses on the envelope detection side better. But using a separate audio output and envelope driver would work better yet.

In the end, it's not the semiconductors or the passives that are the problem, it's the design compromises that limit you.

[puretube]

been biting my tongue on that one...  - tnx for explaining that, R.G. 

[vanhansen]

Great thread here.  I've been wanting to build a compressor for a while, or purchase one, and just couldn't see myself doing so because every compressor I ever tried out I couldn't make work for me.  That changed last year at the Arlington Guitar Show.  I tried out Keeley's Compressor and was floored.  It's the only compressor that I actually liked out of all the ones I've tried.  I'd love to snag one this year.  Who knows, maybe there will be some show specials.    And Robert is a really nice guy.  He took time out of the craziness at the booth (and it was crazy) to say hi and talk for a little bit.  I'll be visiting the booth this year if he's there.

Right now, the idea of building one is a little intimidating.  My DIY hobby has been on the backburner so I'm still working with overdrives and distortions.    Hopefully I can dedicate more time to it soon but in the meantime, it's changing up what I have done and buying what I'd like to have.

[davph30]

Good news!!!
I've found the problem, the 100k pot was faulty if i pressed on the connection with a screw driver the sound brightened up and gave me much more play on the 2k pot, can now adjustit between 9 and 3 o'clock without the sound gating.

[RobertKeeley]

We stopped using the 2k trimmer a long time ago.  There was no need for it.  Look at our pedals that were produced late 2005 I think.
The gain of the transistors solved many problems, so that was an easy transistion.  I still think I might be able to find a better selection of transistors.  I think working on the input buffer and other parts of that stage would be a productive thing.
Yes, there are many things you can do to that circuit to make it better... the first thing is anything better than the 3080.
Thanks
Robert

[Dirk_Hendrik]

Modern 3080's don't usually need it.

Hmm RG, You might want to reconsider this:
http://www.till.com/blog/archives/2005/06/last_of_the_ota.html

[R.G.]

I'm aware of that.

In this case "modern" means "the ones still not used in something and available for builds".

The Dynacomp was designed when the 3080 was still new and when the fabs could just barely build them. There is a continuous tweaking process at every fab to make them run better. The last-produced 3080's - and every chip, for that matter - are or can be better, faster, lower noise, higher performance etc.

I have real world experience with my statement. The trimmer is unnecessary for the Dynacomp circuit.

Don's comments about making a discrete OTA are well founded. I posted about this some ways back. You can make up an OK-ish OTA if you can just get monolithic dual transistors. You need a well matched pair for the input diffamp, and then modestly matched ones for the current mirrors.