GSSL add-on help thread

[Potato Cakes]

Ok so l got to the bottom of my problem, and i had bugger all to do with the operation of the Turbo PCB....It was me...or to be more specific "The stupid balanced signal I have been feeding my GSSL from my mobile phone all this time" FACE PALM!!

Since my soldering/projects desk is some distance from my computer...and Cubase, I was playing audio from my mobile via a small Stereo jack going to two large jacks...just for convenience. so I decide to check my meter calibration whilst I was fault finding the turbo issue, so I hooked everything to Cubase, suddenly my feeble meter readings became hefty meter readings! another FACE PALM! hooked up the Turbo again and the needle is wiggling away nicely.

Now just to button up this sexy beast and put it to work.

I was going to remove this rather long winded post but it does end with a rather poignant "What not to do" so I will leave it for other GSSL builders.

And thanks again to Harpo for his help.

You had a question on GSSL page about the LCR compression characteristics of the the Turbo board, so not to pollute the stock build only thread with more mod questions, I'm replying here. If you viewed the video on Expat's website regarding this circuit, you get a full explanation of what it does and why. In the original SSL buss compressor circuit, the panning of the material source in the stereo field does not affect gain reduction. If I recall your post on the other thread, you mentioned you saw someone had posted a video of the turbo and the gain reduction changed with panning the material, and if that is the case this is not correct. If your GSSL with the turbo mod has the same gain reduction when applying tone and panning it between the left and right inputs, then you have a working mod. Anymore I when I build anymore of these compressors I permanently wire the Turbo circuit in as this gives the GSSL the compression characteristic why I would use an original buss compressor in the first place.

Thanks!

Paul

 

[aaronmcphee]

Hi Paul.

Yes that made more sense to me as the turbo was designed to give an equal side compression as to match the centre, it was a YouTuber known as "really nice audio" perhaps go take a look? My query related more to the fact that with the external sidechain engaged I got a imbalance when panning, I have since decided not to worry about it as the difference is small...it's just annoying.

Thanks for replying.

 

[Potato Cakes]

Historically, doing Turbo and SSC boards (which requires two of them as you know) has been a little tricky to get a perfect balance. If there is an unbalance with just the external in on the SSCs, then either the signal getting to both boards is not the same or the signal getting to the ICs on each board has some sort of variation if indeed there is balanced GR on all of other settings.

On a side note, I tend to agree with the Expat guys that an HPF for the sidechain in the GSSL is not necessary with the Turbo function. A parallel mix function gets much more use (at least for myself), which seems most any DAW incorporates in their external insert functions. Possibly I could see using an SSC for the Thrust characteristics, but I have a MixBuzz that I use for FOH mixing and I have yet to want to use that function, so I'm not sure if I'll ever actually want to have it for any future GSSL builds. An external input for the sidechain could be very helpful in the studio, but I would want to explore a simpler method to do this rather than the extra work (and expense) of wiring up two SSC boards.

Glad you've sorted everything out and enjoy your now working GSSL.

Thanks!

Paul

 

[aaronmcphee]

Historically, doing Turbo and SSC boards (which requires two of them as you know) has been a little tricky to get a perfect balance. If there is an unbalance with just the external in on the SSCs, then either the signal getting to both boards is not the same or the signal getting to the ICs on each board has some sort of variation if indeed there is balanced GR on all of other settings.

On a side note, I tend to agree with the Expat guys that an HPF for the sidechain in the GSSL is not necessary with the Turbo function. A parallel mix function gets much more use (at least for myself), which seems most any DAW incorporates in their external insert functions. Possibly I could see using an SSC for the Thrust characteristics, but I have a MixBuzz that I use for FOH mixing and I have yet to want to use that function, so I'm not sure if I'll ever actually want to have it for any future GSSL builds. An external input for the sidechain could be very helpful in the studio, but I would want to explore a simpler method to do this rather than the extra work (and expense) of wiring up two SSC boards.

Glad you've sorted everything out and enjoy your now working GSSL.

Thanks!

Paul

Yeah I did actually build the two SSC's in readiness but have decided its probably not worth the hassle, or potentially giving myself more problems to sort out.

I had even moded a new front plate to accommodate the SSC frequency rotary but that rotary is now going to become a glorified Turbo on/off switch....yes I know why not have the turbo fixed? well ive got to fill the hole with something

Oh on the subject of external sidechain,  can you tell me how to wire a ts jack to this filter?, wouldn't mind having the option to feed a signal into it.

Cheers.

 

[Script]

Turbo board problem:
different control voltages between L and R sidechains

Removed all bells and whistles from the circuit for now. Thus, a basic and 'reasonably' working gssl plus Turbo board flywired at 47K SC IN junction for now.
- Only modification still in place right now is the '27K unity gain' trimmers feeding the audio path VCAs instead of the 15K (as specified the rev #7 board by pcbgrinder).
- As well as a '100K resistor disconnect switch' (labelled 'Feedforward' on the front panel) before pin 2 of the 1/2 TL072. This resistor is in place for all measurements.
- And two hardbypass relays (powered from 15V rail) wired so that the unit always compresses but with the INs switched directly to OUT in default/OFF.

SC VCA (2181B) test:
3K9 resistors installed to set the SC VCA voltage (as the SC 2181Bs are both powered from 12V). Pin 4 disconnected, 10K between pins 3 and 5 disconnected.


RATIO MEASUREMENT
(across R6 [1meg] on Turbo board and equivalent on main board). Switching ratio on front panel yields this:
      main / turbo
1:10  582 / 618
1:4  270 / 296
1:2  191 / 155
RATIO switch connected correctly but ratios are off.

SC MEASUREMENTS (all DC in mV)
SC 2181B (steady mono IN signal on L & R)
(gssl board / turbo board)
pin 3: -9.8 /-9.8                  (CV to SC connection)   
pin 5: -2.9V / -2.9V  Ec-  (both 3K9s definitely installed, voltage almost ideal as per datasheet at -2.85)
pin 6: -2.7 / -2.7      Gnd
pin 7: 12.10V / 12.10V
pin 8: -2.5 / -3.3      OUT
Looking good. Although OUTs are a bit off.

TL074 (steady mono IN signal on L & R)
gssl board / turbo board
pin 1: -267.1 /  14.3 !!!
pin14: -256.1 / +159  !!! @ 2:1
Now that is seriously off.

Am I totally off in suspecting the rectifier diodes around pins 1 and 2 of the TL074?

 

[Script]

I put in new diodes around the TL074, but it didn't fix it. I have checked all solder joints  and connections , the placement and orientation of components, as well as for broken traces. No problems there as far as I can see.

I did troubleshooting using Keith's run-through
https://groupdiy.com/index.php?topic=28113.msg339348#msg339348

With no input signal, AC at pins 1 and 7 are fine.
However, DC looks suspicious.
Pin 1 DC -99mV /  -139mV

And at pin 1 with input signal it all goes bonkers (threshold CCW ~ CW) )
DC -243mV ~ -347mV / +2mV ~ +61mV

 

[Script]

Turbo board problem found and fixed.

Suspicion that the Turbo board had probably gotten fried once (all on crimp connectors), so I exchanged all diodes and the electrolytic on the T board, as well as the TL074 just in case. Still CV at pin 14 was going all bonkers.

The T board is an exact copy of the main board 'dummy' VCA & buffer & rectifier & ratio resistor network. So I connected all cabling and then measured and compared resistors in place. (--This is not the way to do it usually, cos of component interaction, but here it works cos the sidechain circuitries are exactly mirrored -- so whatever the metre shows as resistance, it should read the same on both boards.--) Turned out resistor readings around the rectifier didn't match.

So I traced the boards and behold:  a tiny solder joint on the main board, which had botched up the main rectifier, causing severe CV offset btw main and T, but not to a degree that it was noticeable on the front panel controls and stereo audio (as I have no experience with SSLs). Basically, the unit had compressed with 'reasonable' CV coming only from the main board sidechain (effectively reacting to Audio IN from one channel only) -- and even that CV was wrong.

Anyway, Turbo mode is now working

 

[gyraf]

Aah, well spotted

This a good example of troubleshooting irl. - hard work, no magic needed

- and thanks for reporting back..!

/Jakob E.

 

[Script]

Spent some more time on the unit on the bench.

-- Variable Attack/Release fixed --

Unit has variable 'Attack' & 'Release 1' & 'Release 2'. However, the VCA crapped out at fastest release settings on both releases.  Apart from there being some confusion about the  placement of the 91K and 750K and their respective caps in the 'Auto' release setting, something else was wrong. So:

(a) Here's a link to a schematic to the time constants of an SSL, hosted by Gyraf.
http://gyraf.dk/gy_pd/ssl/ssl_82e27.gif
(b) Here's a discussion by PRR on what caps make sense with which resistor.
https://groupdiy.com/index.php?topic=63477.msg804502#msg804502
(c) And here's a link to a post about 'variable timing" with a link there to a sketch by forum member Harpo.
https://groupdiy.com/index.php?topic=47.msg522397#msg522397
http://www.hausverwaltung-heger.de/al_leck_trick/VariTiming.gif

Turns out that the 820R for minimum attack (pot fully CCW) was installed, but the 91K resistor for minimum release (fully CCW) had been forgotten. The 1M 'REL 1' pot was installed in parallel to the 0.47uf; but erraneously instead of the 91K.

-->----0.47uf------ --------------- ------6.8uf-- ------------->GND
|                                                                      |
-->--[[[[91K]]]]---1M pot (REL1)-------1M pot (REL 2)->GND

Soldered a 91K on the wire from the PCB to the IN pin of the 1M pot.  Without it, the timing network was effectively shorted to GND when REL 1 and REL 2 were both fully CCW.

-------------------
The two release pots take up space on the (still provisional) faceplate, but are quite nifty in use. They allow to easily dial in the classic 'Auto' setting -- i.e., when REL 1 is fully CCW (at 91K, not 100K, but close) and REL 2 is at around 3 o'clock (at 750K). Plus they give a plethora of other double-release-time settings  inbetween and beyond. Though not fully sure they all make sense

 

[Script]

SSC (Filter) board (fixed).

Unit has Turbo and two SSC filter boards with TL and TM (boards by barclycon and/or PCBgrinder).

Problem: TL was 'loud' -- not as in perceived loudnes but in sillily dB loud, as compared to the other filter settings, including OFF. While TM was OK or maybe too quiet.

Check:
The SSC boards replace the 47Ks on the main board, so on the SSC boards the two 20Ks at the input had been correctly solderjumped in parallel to achieve 10K for SSC unity gain. L into one SSC board, R into the other SSC board.

Used SSC board calibration procedure as described by forum member Harpo. 1kH signal (about 1V) into the unit and then measure for unity gain at the SSC board (before 10K in and before 47K out) through all filter positions.

- Turns out the trimmers for the  TL were only off. Adjusted the trimmers to achieve unity gain (same voltage) from SSC In to Out. All good, easy fix.

- The TM trimmers, however, had been replaced by what turned out to be 12K resistors. But the signal was to quiet. So on one channel I put a 10K trimmer in and a 6k8 (both as written on the board) and set for unity gain from IN to OUT. All good.

- Same procedure for the other channel. I took out the 12K and behold : under the ugly solder blobs, the solder pads had come off. Apparently the SSC board doesn't like resoldering. Not sure it was me who wrecked the pads or they had already been off. Anyway, I had to wirejump connections using the legs of the 6k8 resistor. Checked all traces for connectivity. All good. Fixed.

Hooked the unit back up and had a very quick play only. All filter settings work apparently, but I'm not sure I fully understand the hype around the TM and TL settings   Will frequency sweep the filters again, maybe there's still something going on. Also, Rightmark Audio Analyzer had told me that there was a lot of noise in the TM and TL with a headroom of only 32dB !?? That's normal?

 

[Script]

Not (really) an add-on.

Had isolated the main board (gssl rev 7) from the metal standoffs. Had no plastic standoffs at hand, so I widened the holes on the board with a drill and used shrinktube for isolation, before putting the nuts back on.  Board sucessfully isolated from case.

Therefore, one of the 0V connections at the audio INs and OUTs on the board goes to the starground screw next to the main power in. The other 0V is clipped (not connected to anything). In/Out wires are twisted.

Unit is wired for balanced audio INs and OUTs, meaning there's a 6dB loss in volume when using unbalanced cables.

------------
Unit has the Expat CRCRCR board installed. Sits between toroidal and main gssl board power in. No changes in the PSU, meaning that there are two AC recifiers now. Guess I could get rid of the gssl onboard one, but low priority.

Can't find documentation/installation guide for CRCRCR board online. Is keeping the second AC rectifier detrimental?

 

[Potato Cakes]

Can't find documentation/installation guide for CRCRCR board online. Is keeping the second AC rectifier detrimental?

Supposedly it is not, but on all of ones I've build with the CRC I use jumpers in the spot for the rectifier on the main board. You just have to make sure the polarity of the DC to the main board matches the markings where the old rectifier is silkscreened.

Thanks!

Paul

 

[Script]

SSC EXT IN (barclycon board) didn't work. Fix almost trivial.

Unit uses jack INs to feed external 'Key' signal (e.g. a kick drum) into the sidechain(s).

Jack tip goes to 'RET' / 'EXT IN'. While jack ground needs to be 'referenced' to 0V. The used  jack plugs are full plastic, so no connection btw. jack ground and case. All it needed was running a cable from jack plug ground to case starground.

Skipped the SSC 'Send', cos it either requires the use of four jack plugs (2xSend & 2xReturn) or wiring the plug for Y-cables when using only two jack plugs. -- Or as an overkill project for a rainy day: drill two more holes for two extra 'Send' jacks or make the two installed balanced jacks switchable on the back faceplate  :

Instead, might use the 'Send' on the SSC PCB to later add more filters internally to the installed 2-deck 12-position switch.

 

[Script]

And some more time with the unit.

Hooking up 2 x SSC boards (Baclaycon/PCBgrinder) and accounting for difference of 6dB in SC level between modes

Installed was a 2D2P toggle switch with one deck to sum L&R after the filter boards and the other deck to activate an indicator LED.  Problem:  difference in SC signal level, which makes comparing the two modes impossible, as it requires adjusting the THR knob. The following fix is not the 'ultimate' way of doing it, but it worked for me without redoing too much.

Oxford mode clearly needs two SC filter boards to filter L & R independently. But I found at least three, if not four, different ways on this forum of how to hook up GSSL mode with the two SCF boards, including contradicting numbers of switch decks needed (from 1 deck to 3 decks).
(a) Simply summing L & R before the SC filter board, effectively using only one filter board.
(b) Summing L & R before the SC filter boards and using both boards
(c) Summing L & R after the filter boards
(d) Igor's mixbuzz 500 method of summing L & R after the filter boards via an additional buffer and some gain staging -- however, I can't claim I have fully understood this from description only. Maybe someone can help out here...

Anyway, for now, I have kept the post-SCF board summing link, cos I don't want to rip out the existing wires, don't want to resolder the SCF boards again to insert different resistor values and, above all, I don't want to redo all the crimp connectors at the filter board inputs/outputs. Also, I want to keep the OXF mode indicator LED.

So for GSSL mode, I ran two wires per board (soldered on the board underside to the 20K input resistors) leading to two switch decks to parallel the two installed 20K input resistors on each of the Barclaycon SC filter boards. This yields attenuated inputs through 2x20K in parallel = 10K per board for Oxford mode, and attenuated inputs through 1 x 20K = 20K for GSSL mode per board.

Effectively, the L & R audio signals are now filtered independently also in Aarhus mode before summing , but filtered at lower levels than in Oxford mode, in order to account for the dB buildup when mono'ing the audio signals post-SCF.

In my case, including the LED, making it all switchable required a 3D2P anyway. The way I did it now requires a 4D2P and just some extra wire.
SCF1 output is hardwired to the main board.
Switch:
- deck 1: switches SCF2 output EITHER to turbo board input OR main board input (there it links with SCF1 output for mono summing)
- deck 2: parallels 2 x 20K resistors to 10K on SCF1 board for OXF mode
- deck 3: parallels 2 x 20K resistors to 10K on SCF2 board  for OXF mode
- deck 4: activates the OXF mode indicator LED powered from the 15V rail.

Oxford and Aarhus mode are now comparable when switched without having to back off on the THR in Aarhus mode. Thresholds are virtually equal, albeit the amount of compression in dB can differ with Mode depending on audio material.

---------------
Just tried highly 'phasey' bass audio material, as mentioned by Igor, and wow. In practice, Oxford is for surgical stereo material applications, while Aarhus mode is now useful for that extra oumph -- especially in combination with the filters it's almost a compressor and stereo enhancer. No wonder, cos the GSSL in Aarhus mode can't compress what it can't 'see', namely stereo or 'phasey' stuff. A really big kudos to the original GSSL idea !

 

[Script]

1.5:1 Ratio modification added

Info plus calibration procedure and schematic is here:
https://groupdiy.com/index.php?topic=49854.msg630595#msg630595

It requires only four additional resistors and a diode (per channel), plus three connections soldered directly to the underside of the Turbo SC board. I placed the resistors and diode on vero board attached to the Turbo board next to the ratio resistor network (see picture). For the main channel they are placed close to the left side of the RAT/ATT/REL/GAIN subboard.

The difference between 2:1 and 1.5:1 is not that huge after all. Still, in use, personally, I found that the 2:1 setting now hardly sees much use any more.

 

[Script]

ATT REL switch for Soft and Tight

Stole this idea from Igor's Mixbuzz. It swaps the order of the Attack and Release. As is, on the GSSL, the order is ATT followed by REL (soft). The unit I work on has variable ATT and REL, so there are many cables running from the subboard. But there is a jumper on the GSSLrev7 subboard ((the jumper that often gets forgotten in new builds) that connects the release network post-attack (see picture). It should be easy to just sub the jumper with two wires connecting to a 1D2P toggle switch and run a third wire to connect the release network pre-attack.

Changing the order of ATT and REL has dramatic effect, turning this compressor into a completely different unit (with 'subjectively' different THR behaviour, so not directly comparable). Therefore, I am not sure yet whether this is a keeper. For now the toggle is only dangling on wires.

 

[lars on]

Hi every GSSL addict !

I have this comp with one SSC working very well.

I'm trying to wire a Turbo switchable mod with 2 SSC boards (fighting with my brain...)

So I installed a turbo board (from expat with 2180 VCA) and two SSC boards (from PCB Grinder) following the schematics from @flipside on post #15 of this thread (https://groupdiy.com/index.php?topic=35450.msg503923#msg503923) but nothing happened .

There is no difference between Turbo and Oxford.

My voltages are good.
My second SSC board is fine and calibrated.
My IC's are good too.

Maybe I have to clarify somthing with 47K or 20K resistors.

Any little clue is very welcome.


Some test later... and simplifyng the thing :
I tryed with one SSC board and Turbo.
I plugged it like holger advise (what I understood) like the pics attached or like this from Reply#13 of stereokillah https://groupdiy.com/index.php?topic=36643.msg450463#msg450463

Few minutes later :
I removed the TL074 from mainboard and no distortion anymore.
But I don't think he is damaged because in classic gssl config (without any add-on) everything is fine...

 

[lars on]

Just to clarify something.

When I send a 0dbu 1Khz in the unit, I set the threshold to see 10 db of reduction (when I'm on 10:1 ratio position) and when I change ratio the reduction remain the same on the meter and on output ?

Is it normal ?

I'm having headhache 

 

[Potato Cakes]

Just to clarify something.

When I send a 0dbu 1Khz in the unit, I set the threshold to see 10 db of reduction (when I'm on 10:1 ratio position) and when I change ratio the reduction remain the same on the meter and on output ?

Is it normal ?

I'm having headhache 

The threshold is not a fixed point in relation to the ratios like you are probably used to seeing on the graphic representation of a plugin. When you lower ratio, the threshold point shifts down with it, so it is possible that reduction from a 1kHz test tone stays the same based on which add on boards you are using. Run some actual program that is drum heavy through it and then cycle through the ratios. That will be a much better gauge on how the compressor is functioning.

Thanks!

Paul

 

[Script]

The two SSC filter boards (pcbgrinder) fully replace the two 47K's on the main board.

Your ratio switch wirings are good?