Input clipping on Orange Squeezer question-Build/Mod report

Started by vortex, February 20, 2005, 04:28:00 PM

Previous topic - Next topic

vortex

Man, i am jazzed with the OS comp! I love the sweet and lively nature of the compression and the tonal color enhancements.

I am using the 5457 tranys and a 4558 op (prefered over the tlo72). 10K bias trimmer.

Additionally, in search of adjusting and slowing the attack characterisics I  substituted a 10K pot for R11 and a 500K pot for R12. I am still experimenting with these, but have found that R11 has an effect on the compression amount and R12 effects the attack and release speed. The 500K may be overkill!

I also added a switch for C7 that selects either 4.7uF of 100uF. The 100uF cap does a fantastic job of slowing the attack and release. IMHO this is a worthwhile mod! I love compression with a slow attack!

Which brings me to my question: On the strongest attacks from my guitars I get clipping, even in the slow attack setting. If I turn down the guitar volume some, it still is there. I have tweaked the bias on the OS and have checked to see if it is a guitar pickup height issue. I am doing this at unity gain so I don't think it is an amp thing either.
So, I am thinking it is related to the 4558 being driven a hair too hard. How do i adjust the gain or input to the 4558? Or is there something else i should be looking at?

Thanks in advance!

MartyMart

Try reducing the stock 220k to a 200k or even 180k, I went with 200k and dont have any clipping problems with a strat  :D
Bias pot setting is crucial also, the "sweet spot" sounds great.
cool mods by the way !!

Marty 8)
"Success is the ability to go from one failure to another with no loss of enthusiasm"
My Website www.martinlister.com

vortex

Thanks for the info Marty, I appreciate your quick response!

I subbed a 250K pot for R9 (220K) and ended up liking 180K in that position. That did help with the clipping.

I also noticed that the 500K pot that I subbed in for R12 would induce a bit of farting out when the low frequencies were pushed. (Only at the very end of it's travel :0-15K range). That was part of the problem. I guess I could add a resistor to curb that, but the Squeezer sounds really rich in the high end up there!

I have to say that I think that this is a fascinating circuit just begging to be tweaked. Dan Armstrong built a winner even without external knobs and switches.

I am not experiencing much in the way of noise but I am going to build another with metal film resistors just for fun.

MartyMart

I dont know if you saw the post, but I built mine in an "original" orange squeezer box, so it had to be "small" with bias/vol as trim pots
My only other "sub" apart from the 200k was a 330k rather than 390k from the power rail, as thats all I had !
Doesn't seem to have made any difference !
To have unity gain when engaged, my "vol" trimmer is set to about one third of its travel.
I could have a wicked boost otherwise  :twisted:

Cheers,
Marty.
"Success is the ability to go from one failure to another with no loss of enthusiasm"
My Website www.martinlister.com

inverseroom

I built the "Build Your Own Clone" kit version of this circuit, and I really love it.  That one has a gain knob and a trimpot for, I believe, threshhold.  An unsubtle but really, really useful sound!

moosapotamus

Here's a (untested) scheme with a bunch of other OS mods that I collected from a thread where one of the topics was how to optimize the OS for bass. This should still work great (or, maybe a littler better) for guitar, too.

Orange Squeezer +++

Rob Strand contributed the full-wave rectifier mod with the intent of improving tracking low frequencies. I still have yet to try any of the mods out, myself, but I'd be interested to know if anyone else does.
... just another one of those things hovering near the top of my to-do list.  :wink:

~ Charlie
moosapotamus.net
"I tend to like anything that I think sounds good."

Mark Hammer

Note that the input to the op-amp is biased asymmetrically.  That is, the two resistors straddling V+ and ground at the input to the op-amp do not divide (or at least in principal, do not appear to divide) the supply voltage in half.  This means that the signal has less headroom in one direction than in the other (the biasing provides less headroom for the positive-going portion of the signal, which is, incidentally, the portion of the signal which will pass through the diode).  I'm not sure what the rationale was for this originally - perhaps having something to do with the envelope follower and anticipating signal loss through the diode - but that may be a source of the distortion you hear.  Initially I thought it was envelope ripple, but you noted that the same distortion is present independent of attack/decay settings, suggesting its origins are elsewhere.  I'm wondering if more symmetrical biasing might not solve your "problem".  Certainly, if there is risk of clipping from a hefty signal, then envelope signal loss through the diode would not be a concern and - if my theory is correct - the precompensation not needed.  Finally, as you yourself have considered, dropping the gain down a bit might be useful.

As a side note, and just to spark some thinking on an otherwise dreary Monday (bummer about Hunter S. Thompson), cast your collective mind back to the "What" compressor.  The designer of that unit noted that what made the unit feel so transparent was the double envelope follower that utilized two different time constants (one fast, one slow) to anticipate how the ear reacts to transients.

The Orange Squeezer uses a FET as the ground leg of a voltage divider comprised of the FET and the 82k input resistor.  Where many compressors alter the gain of the circuit, the OS maintains a fixed gain, but variably attenuates the signal level fed to the gain stage.  There is, in principle, nothing that says a second cascaded attenuator can't be added, whether before or after the gain stage, governed by a second envelope follower with different time constants, to achieve what the "What" achieves.  

In the OS++ that Charlie links to, the signal from the gain stage is sent off in two directions, with one of the taps being inverted and the envelope combined with the first, to form what is essentially a full-wave-rectified envelope signal (preferred because of less envelope ripple).  In this case, the two envelope signals merge and share the same time constants.

But suppose the 2nd path is NOT inverted, and uses different time constants, AND is mixed with the first at the gate of the FET?  Perhaps that might yield the sort of complex envelope function that occurs in thw "What".

Just thinking out loud.  Jump in any time.