Anyone here have a real Pearl CO-04 Compressor?

[cd]

Or successfully built a working clone?  If so, how does it sound compared to say, a Dynacomp or Ross Compressor?

[RickL]

I do. I haven't listened to either the Pearl or a Dynacomp recently but I do remember being impressed with the attack control on the Pearl. Everything from an almost clickey attack to complete mush. I also seem to remember a huge amount of apparent sustain on the Pearl without an unusual amount of noise.

I'll try to remember to dig it and my Dynacomp out in the next couple of days and do a better comparison. Remind me if I forget.

[Mark Hammer]

I forget where I cored the schem for the Pearl OC-4, but it sports a number of things worth noting:

1) A CA3080.

2) A tone control stage somewhat similar to the TS-9, but placed in front of the CA3080 (i.e., generally MORE active circuitry ahead of the gain cell in the Pearl unit, which may account for Rick's noise comments).

3) A slightly different rectifier/envelope-follower (op-amp based vs discrete transistor), but still full-wave like the Dyna, and with an attack/recovery control essentially identical to what I've been kvetching about for the past few years (varies resistance on charge-up path to 10uf cap between around 10k and 150k).  Indeed, it was from looking at the Pearl unit, the CS-2, the Nobels compressor, and I think one other, that it suddenly dawned on me that this could be easily implemented in the Ross and Dynacomp.

4) A whole lot more buffering at each end on the Pearl than on the Ross or Dynacomp, stemming from use of FET-switching.

It is entirely conceivable that the best aspects of the Pearl could be salvaged while dumping the worst aspects, if it was approached as a stompswitch-bypassed unit, rather than a FET-switched one.  You could probably lose some of the additional noise sources.

[phillip]

Here's the schematic of the Pearl CO-04 Compressor...is this the same one being discussed?

http://www.mif.pg.gda.pl/homepages/tom/files/CO-04.gif

Lots of interesting things going on in the circuit.    More "hefty" than the Ross/Dyna, but also more full-featured.  The parts count would drop quite a bit without the FET switching.

Phillip

[cd]

Philip, that's the one.  I breadboarded it the other day and it sounded terrible - no real compression effect at all.  I've put it on the backburner for the time being since I made some part substitutions, and I don't know if that makes a huge difference - a 2SK68A instead of a 2SK30A, and a 2N3906 instead of 2SA1015.  I'm thinking the 2N3906 shouldn't make a huge difference (same approximate gain) but I'm not sure what that FET is doing - it's the feedback loop of the 3080, change the gain of the 3080 and the gain of the FET changes?  If that's the case maybe that's why my subs didn't work.

I also cut out a lot of the extra crap - instead of the tone control, I just used the 10k/3k3 combination in a non-inverting stage (gain of 4) and the output driver (right after the level control) I just used the 10k/1k combination non-inverting (gain of 11).

I'll post back once I get the real deal Japanese parts and see what happens.  I wonder if there's an error in the schematic though - that 100k/470 resistor combination in the FWR looks mighty high (gain of over 200!)

[RickL]

I got a chance to play with mine tonight and I guess my memory's not what it used to be.

The attack control is fairly effective but not quite as good as I first described it. I've heard compressors with squashier attacks (the Walco Sustainer comes to mind).

The sustain is fairly good also, but again not as much as I remembered. It's also a little noisier than I remembered, especially with the tone control turned up.

The tone control works pretty well, turned all the way up this is fairly bright, but at the expense of increased noise.

And it's LOUD. You could easily overdrive the front end of an amp with this thing.

I couldn't find my Dynacomp so I couldn't make a direct comparison, but I think with the added attack control the Dynacomp would sound pretty similar to the Pearl with the tone control at about mid position.

[cd]

Argh... OK this one is going into the "frustrating junkpile" for a while.  I got all the real deal Japanese parts in, subbed them in, and no difference (see above).  I'd try a different 3080 but I don't have one, and I know the 3080 I have now is fine - I plugged it into a Dynacomp to be sure.  Sure compresses fine there.

I don't think it's my actual wiring or anything - I've built and rebuilt it on my breadboard (in different places on the board too) 5 or 6 times now.  Maybe it's time to pick apart things on the actual schematic to see what's up?  I'm not sure how the 3080 stage works, and the FWR looks funny to me:



Ignoring the FET switching for the moment, the guitar signal hits a transistor setup as an emitter follower (buffer) - nothing surprising there.  From there it goes to the TONE control, where it gets an overall gain boost of 4x, along with a high/low cut depending on where the TONE control is set... nothing surprising there.

From there it goes to that FET (source follower) which in turn sends signal to the 3080 (feedback) and the FWR.   The FWR controls the gain of the 3080.  After the LEVEL control, the signal hits an 11x gain stage, then to the output.

The 3080 is an OTA, so higher current into pin 5 = higher gain... the FWR converts the guitar signal into a DC voltage, so a higher guitar signal = more voltage, which is less current, and less gain (am I right here?)  If that's the case, shouldn't D3 and D4 be wired the other way then?

[Eb7+9]

I think the diodes are ok ... the OTA is wired to produce gain reduction with an increase in Transconductance by feeding back an inverse-phase current into the summing node at the Gate of Q2  ...

the problem could be with Q3 in the voltage-to-current converter ... if you use a device with a large-scaling current relative to original device you could have a clamped mirror diode at pin 5 of the OTA with none/any signal ... I'd measure the voltage there to see what it's doing against signal, same thing at C9 for that matter ...



you may be able to control the bias/idling current of Q3 by increasing R25 (6k8) using a trimmer ... the ATTACK path resistance set by VR1 and R24 - which dominates the discharge time of C9 assuming HfeDC of Q3 is high enough - can interefere somewhat with this adjustment scheme so it's hard to say ... probably best to find a device with similar Is specs ...

~jc

[Eb7+9]

... as an alternative, try using a variable bias source for the FWR and VR1 - a 10k trimmer with a 10uF cap to ground on the wiper should work ... this way you can adjust the idling bias of Q3 and not screw up the dynamics of the averaging filter ...

[cd]

JC, a lot what you wrote went way over my head, but anyway  I tried a 25k trimmer instead of R25 (the 6.8k going to the base of the transistor), made no difference.  Then I tested the voltage at C9 (the 10u tantalum cap).  With no signal, and with the 25k trimmer at max (25k), I get 5.5V.  Bash a chord, and the voltage drops to 3.0V, which then slow rises as the chord dies out until it gets back to 5.5V.  So I presume the FWR is working right, and the cap is charging as expected, so it must be something with the voltage to current converter, right?  I tested the current going into pin 5, if I bash a chord it doesn't change, it's a steady 18uA.

[Eb7+9]

Sounds like the FWR is doing it's thing on the cap ... and Q3 is not happy ...

I ran a DC Transfer sim with AIM-spice using 2SA1015 and 2n3906 models for Q3, 1n4148 model for the OTA diode and an ideal voltage source replacing the cap ... the current output  profile was identical for both Q3 models ... at 5.5v the sim shows a current of around 750uA (with VR4=0) ...

The sim'd circuit puts out a linear current from ground up to a voltage of around 8.2volts and the transfer slope is set by a combination of R25 and the sum of VR4 and R26 - you can't make R25 too small in the circuit since it acts to buffer the diode of Q3 against the low drive of C9 ...

It would appear the circuit uses the BE junction of Q3 as a diode and relies on HFE specs to output a corresponding current at the collector (the 2SA1015 data mentions a linear HFE against current) ... similarly the collector load also affects the transfer of current - so VR4 and R26 together also set the voltage-to-current transfer rate as well ...

Code Select Expand


Q2SA1015 test

vin 22 0 5.5
vcc 99 0 dc 9

Q1 1 2 99 Q2SA1015

Rb 2 22 6.8k
*Rb 2 22 1k
*Rb 2 22 68k

Rchoke 1 11 22k
*Rchoke 1 11 222k

Dota 11 0 D1N4148

.model Q2SA1015 PNP(Is=295.1E-18 Xti=3 Eg=1.11 Vaf=100 Bf=110 Ne=1.5 Ise=0
+               Ikf=0 Xtb=1.5 Br=10.45 Nc=2 Isc=0 Ikr=0 Rc=15 Cjc=66.2p
+               Mjc=1.054 Vjc=.75 Fc=.5 Cje=5p Mje=.3333 Vje=.75 Tr=10n
+               Tf=1.661n Itf=0 Vtf=0 Xtf=0)

.MODEL q2n3906 pnp
+IS=7.75521e-12 BF=194.093 NF=1.35509 VAF=156.436
+IKF=0.0660057 ISE=1.88546e-12 NE=1.81673 BR=3.41317
+NR=1.5 VAR=5.86061 IKR=1.70599 ISC=7.64281e-10
+NC=1.92376 RB=6.48961 IRB=0.1 RBM=0.1
+RE=0.0001 RC=2.45044 XTB=0.1 XTI=1
+EG=1.05 CJE=6.11928e-12 VJE=0.4 MJE=0.248812
+TF=5.21843e-10 XTF=0.932702 VTF=9.1046 ITF=0.0106472
+CJC=6.85007e-12 VJC=0.4 MJC=0.279018 XCJC=0.9
+FC=0.478887 CJS=0 VJS=0.75 MJS=0.5
+TR=4.30605e-07 PTF=0 KF=0 AF=1

.MODEL D1N4148 D(IS=0.1P RS=16 CJO=2P TT=12N BV=100 IBV=3.867E-10)

All I can say at this point is to make sure your small-signal PNP device  is connected properly and try several devices - I'd keep trying along with the 25k trimmer for R25 but don't go too low ... and from the wide linearity range it would appear you don't need to resort to bias control ...

~jc

[cd]

Thanks for keeping on chiming in, JC  OK unfortunately I only have 5 2SA1015s, I didn't get that many since I thought it would work the first time.  With all 5, no dice.  Of course I quadruple checked all the wiring and everything, by eye AND with a multimeter just to be sure my breadboard isn't wonky.  THEN I tried some other PNP transistors: 2N3906s, 2SB476s, 2N4126s, 2SC495s, 2N5087s, even a couple of OC44s and AC128s - no dice.  I kept checking the current going into the 3080 and it was always a steady 14-18uA, never changed more than 1uA with signal level.

On a whim, I replaced the PNP with a 2N5210 (NPN) and wired it like in a Dynacomp, with the emitter driving pin 5 of the 3080, the collector to 9V, and base to R25 and C9.  NOW I got a small current change on pin 5 of the 3080 - from 2.7uA to 8uA.  Not a lot but I should hear some change, right?  WRONG!!  Still no difference (same as all the PNP transistors above).

OK so I decide to check all my components, maybe I fried my 3080 (I only have one to play around with).  So I used all the same parts (where possible) and re-wired everything to a Dynacomp.  Same pots, same wire, same 3080, same transistors, etc.  WORKS PERFECTLY!   Argh.  Needless to say I'm putting this one to bed for a long, long time.  Maybe I'll follow my usual advice and just buy the real thing

[Eb7+9]

... the only other thing I can think of that wouldn't make Q3 sink the right amount of current is if the OTA rails aren't connected between +9/gnd but rather +9/bias ...

C'est la vie !!