New GSSL Layout Proposal - Anyone Interested?

[tv]

I know what you have been thinking! BUT consider that if you connected the "dirty" ground to the "main" ground "at" or before the first reservoir cap's ground point, they wouldn't contaminate the clean audio ground at all (all LED return paths would terminate before the actual regulated rails/ground contact points).
If you used a fet-input opamp to "ground-sense" the CV for meters, the clean ground stays clean. IOW, this is both a matter of layout AND circuit design. At least to me, this seems very clear..

 

[Harpo]

Thats the usual approach as you describe it. Drawback is the current demand from a single most often positive rail by these relais and optical gimmics, sometimes drawing more than the rest of the circuit, and returning all current to 0V/audio gnd. A floating supply for the non-audio stuff might have the additional advantage of a more equal loading of the mains transformer and bridge rectifier, but obviously coming with its own drawbacks.

 

[tv]

It's easier to drive relais from floated supply than LED meters who need to be driven by gnd-referenced signals. Which in essence leads to a conclusion to use a mixed approach ..

 

[ruckus328]

It's easier to drive relais from floated supply than LED meters who need to be driven by gnd-referenced signals. Which in essence leads to a conclusion to use a mixed approach ..

Yes, I've come to the conclusion that mixed approach is what needs to be done as well.  My original plan isn't going to work.  Just when I think I'm done lol.  Just to be clear, there's 2 things going on here:

1) Power LED's, Pushbutton LED's, relays:

These are all running off of a dedicated LM7812 with a dedicated supply and return path, terminating all the way back at the power supply at the cap bank, resistor between the input and ground terminal to the +/- rails off of the bridge.  Simple, effective, life is good

2) Here's the PITA: LED Level Meter - not so simple (LM3916 driving it).  This is where I'm weighing the options of how to do this.  It's not for GR - it's input/output only, so it's pulling directly off of the audio signals, rectified through a TL074.  It's going to potentially pull about 100mA of current, so obviously best bet is to keep this out of the audio return path.  I'm just wondering what the simplest solution is for this, and if going down the "ground sensing" route is overkill and uneccessary.

This is pretty much the last piece of the puzzle, and I just wanted to say thanks for the input and suggestions so far, it is very much appreciated and won't be forgotten.

 

[Ptownkid]

Why not just use an off board adjustable multi rail psu...

 

[tommypiper]

It's easier to drive relais from floated supply than LED meters who need to be driven by gnd-referenced signals. Which in essence leads to a conclusion to use a mixed approach ..

Yes, I've come to the conclusion that mixed approach is what needs to be done as well.  My original plan isn't going to work.  Just when I think I'm done lol.  Just to be clear, there's 2 things going on here:

1) Power LED's, Pushbutton LED's, relays:

These are all running off of a dedicated LM7812 with a dedicated supply and return path, terminating all the way back at the power supply at the cap bank, resistor between the input and ground terminal to the +/- rails off of the bridge.  Simple, effective, life is good

2) Here's the PITA: LED Level Meter - not so simple (LM3916 driving it).  This is where I'm weighing the options of how to do this.  It's not for GR - it's input/output only, so it's pulling directly off of the audio signals, rectified through a TL074.  It's going to potentially pull about 100mA of current, so obviously best bet is to keep this out of the audio return path.  I'm just wondering what the simplest solution is for this, and if going down the "ground sensing" route is overkill and uneccessary.

This is pretty much the last piece of the puzzle, and I just wanted to say thanks for the input and suggestions so far, it is very much appreciated and won't be forgotten.

IMO, forget the LED Level Meter.  It's extra frill we don't need.  We've never neeeded or wanted it on the GSSL before.  It's not on the original SSL.  It's potential problems and noise for the audio path.  It's adding unnecessary wiring and possible noise and complication to the audio outputs.  Please, just say no, stop trying to put everything including the kitchen sink into this...  

This is an analog box.... remember.  I prefer the old skool look of an analog meter.  I hate LED meters on anything besides computer interfaces.

I would also forget relays, but I'm sure I'll hear push-back on that one...    Relays offer no audio advantage.  Research, well respected audio pros and the solid (not C-47 type  ;D) audiophile community all forgo relays in favor of quality toggle switches.  Respected engineers have told me relays often add noise.  Toggle switches are passive, can be very high quality, do not add noise or complication, and they were used all over vintage gear!  Even with crappy contacts in old vintage gear they usually work great.  A modern toggle with gold contacts solves almost all relay applications with elegance and simplicity.  I simply do not understand the obsession with relays.  There is always a simpler solution.  And relays complicate everything, especially the power supply.  

Look at all your posts on this...  You guys are making things waaaay too complicated.  This is an analog device, guys.  Come on... Analog.  Hello?  No digital.  No relays and LED meter crap, paaalease.  Let those who want to do that stuff DIY...

Please don't yell back at me, just my opinion... trying to help steer the boat back to reality.

JMHO....    

 

[ruckus328]

IMO, forget the LED Level Meter.  It's extra frill we don't need.  We've never neeeded or wanted it on the GSSL before.  It's not on the original SSL.  It's potential problems and noise for the audio path.  It's adding unnecessary wiring and possible noise and complication to the audio outputs.  Please, just say no, stop trying to put everything including the kitchen sink into this...  

Honestly (and nobody's yelling at you, we all have the right to our opinion), I think you're in the minority here.  The inability to know what my input/output levels or if I'm peaking to me is one of the biggest things to address on this unit.  And alot of things (like hpf, mono sc, etc.) are not on the original SSL.

This is an analog box.... remember.  I prefer the old skool look of an analog meter.  I hate LED meters on anything besides computer interfaces.

As do I and most others - which is why it's still there  You'll see.

I would also forget relays, but I'm sure I'll hear push-back on that one...    Relays offer no audio advantage.  Research, well respected audio pros and the solid (not C-47 type  ;D) audiophile community all forgo relays in favor of quality toggle switches.  Respected engineers have told me relays often add noise.  Toggle switches are passive, can be very high quality, do not add noise or complication, and they were used all over vintage gear!  Even with crappy contacts in old vintage gear they usually work great.  A modern toggle with gold contacts solves almost all relay applications with elegance and simplicity.  I simply do not understand the obsession with relays.  There is always a simpler solution.  And relays complicate everything, especially the power supply.  

On the contrary, it litterally would not have been possible to do everything I've done without the relays, and it would have resulted in about 4x's the wiring.  Think about it - If I want to switch 4 signals simultaneously (like in the case of a balanced audio bypass circuit), that means 4 poles and 3 wires per pole (so 12 wires), plus now the audio is being brought off board (risk of noise).  Or, you can do it with relays right on the board right at the audio input/output, and all you need is 2 wires from a pushbutton switch to turn the relays on.  Much simpler.

Like I said, nobody's yelling, just trying to point out the flip side of the coin.

 

[ruckus328]

Why not just use an off board adjustable multi rail psu...

I am, that's not the issue.  The issue is keeping the LED Meter driver (LM3916) isloated from audio ground and on its own return path.  Problem is - the signal that's feeding it is referenced audio ground so you can't just isolate the ground connection on the LM3916.

As I see it - there's 2 options here:

1) Let the return path of the 3916/Meter LED's dump onto audio ground.  Keep in mind this thing now has a ground plane and proper Power Supply.  Will it cause noise?  I dunno.  Possibly.  Only way is to test it.

2) Insert some sort of "Ground Sensing" so I can keep the 3916/LED Meter isolated.  Problem here - I am waaaaaaaay out of board space.  Like really really out of space.  When you guys see it you'll know what I mean.  It's about as effecient as it can be, so with the features I've incorporated, it's as small as possible.  Only way to go any smaller would be to start smashing parts like the Input/Output area of the existing board (something I made sure to avoid).

 

[Biasrocks]

The GSSL sidechain VCA used to be implemented as DBX2150. This Chip has a recomended supply voltage of +/-12V or 24V across IC and absolute max.30V across IC.

I beg to differ.

The datasheet for the 2150 series says it's capable of +/-18V (36V), the 2180 series is capable of +/-20volts (40V). So were not even close to maximum on these guys @ 30V.


http://www.wiseguysynth.com/larry/schematics/dbx_vca/2150data.pdf

http://thatcorp.com/datashts/THAT_2180-Series_Datasheet.pdf

The original SSL used +/-15 volts on the DBX202C VCA in the main audio path and side chain which makes total sense since you want the sidechain to react similarly to the main audio VCA.  The +/-12 volts was used for control voltage purposes and the audio opamps ran +/-18volts.

It would be great to get back to the original design as much as possible.

I have the original SSL schematic if you don't have a copy.

Mark

 

[Ptownkid]

regarding having in/out and gr metering, the easiest thing to do is use the expat pcb and an analog meter...but that's just my opinion.

 

[Harpo]

The GSSL sidechain VCA used to be implemented as DBX2150. This Chip has a recomended supply voltage of +/-12V or 24V across IC and absolute max.30V across IC.

I beg to differ.

The datasheet for the 2150 series says it's capable of +/-18V (36V), the 2180 series is capable of +/-20volts (40V). So were not even close to maximum on these guys @ 30V.


http://www.wiseguysynth.com/larry/schematics/dbx_vca/2150data.pdf

http://thatcorp.com/datashts/THAT_2180-Series_Datasheet.pdf

The original SSL used +/-15 volts on the DBX202C VCA in the main audio and side chain paths.  The +/-12 volts was used for control voltage purposes and the audio opamps ran +/-18volts.

It would be great to get back to the original design as much as possible.

I have the original SSL schematic if you don't have a copy.

Mark

So whoever is right, this DBX2150 app note claims abs.max. 30V.
IIRC the SSL comp used a DBX2151 VCA in the sidechain, similar to DBX2150A or DBX2155, not DBX2150, that might explain this difference. Probably splitting hairs, as you most probably will use a THAT218x instead.

 

[livingnote]

Aha - yeah. How to do the thing with the grounds between the meter in point to still keep them somewhat isolated,
here we go. Who dumps what in what, right? Same deal with el beasto here, it's so funny how you keep running
into the same stuff. Anybody?

 

[ruckus328]

regarding having in/out and gr metering, the easiest thing to do is use the expat pcb and an analog meter...but that's just my opinion.

Yes, but you can't compress and monitor levels at the same time.......

 

[ruckus328]

Aha - yeah. How to do the thing with the grounds between the meter in point to still keep them somewhat isolated,
here we go. Who dumps what in what, right? Same deal with el beasto here, it's so funny how you keep running
into the same stuff. Anybody?

Haha.  Yup.  So are you just dumping your meter onto audio ground in your current setup?

 

[warpie]

regarding having in/out and gr metering, the easiest thing to do is use the expat pcb and an analog meter...but that's just my opinion.

Yes, but you can't compress and monitor levels at the same time.......

I suppose you can use a meter for GR and a second meter, together with the expat pcb, for out metering  ???

 

[livingnote]

Well the way it is actually not - I'm dumping the meter into LED ground and the signal gets to see the meter's
input opamp through a resistor, but one thing I don't know is what the approximate error voltage will be between
the two grounds where it counts.

Then again brutalizing your sig metering into 10 LEDs doesn't sound like it's gonna even really matter, eh?

Ah yeah - expat rocks, but the two of us painted ourselves into a corner here

 

[ruckus328]

the two of us painted ourselves into a corner here

Understatement of the day.  LOL  For me anyways.

 

[Biasrocks]

So whoever is right, this DBX2150 app note claims abs.max. 30V.
IIRC the SSL comp used a DBX2151 VCA in the sidechain, similar to DBX2150A or DBX2155, not DBX2150, that might explain this difference. Probably splitting hairs, as you most probably will use a THAT218x instead.

Believe it or not the SSL used a 202C in the main path and in the sidechain. I haven't actually verified this against an actual unit, but that's what the schematic calls for.

Your app note for the 2150 says that higher voltages can be accommodated easily.

"Since the maximum voltage which may be applied across the IC is 30V, supply voltages of up to ±24V can be accommodated by appropriate scaling of R5, R8, and the R6/R7 voltage divider."

Apples and oranges I suppose because as you say not many if anyone will be using a 2150 vca.

I don't see any reason to potentially cripple the sidechain detector with +/-12 volts vs the proper +/-15v as called for in the SSL schematic.

Mark

 

[ruckus328]

The GSSL sidechain VCA used to be implemented as DBX2150. This Chip has a recomended supply voltage of +/-12V or 24V across IC and absolute max.30V across IC.

I beg to differ.

The datasheet for the 2150 series says it's capable of +/-18V (36V), the 2180 series is capable of +/-20volts (40V). So were not even close to maximum on these guys @ 30V.


http://www.wiseguysynth.com/larry/schematics/dbx_vca/2150data.pdf

http://thatcorp.com/datashts/THAT_2180-Series_Datasheet.pdf

The original SSL used +/-15 volts on the DBX202C VCA in the main audio path and side chain which makes total sense since you want the sidechain to react similarly to the main audio VCA.  The +/-12 volts was used for control voltage purposes and the audio opamps ran +/-18volts.

It would be great to get back to the original design as much as possible.

I have the original SSL schematic if you don't have a copy.

Mark

Changing the sidechain voltage to +/-15V wouldn't really be an issue from a layout perspective, I'm not sure what other adjustments would need to be made then though if I do, since yes, it will be THAT218X in the sidechain position - I don't know why someone would want to use a 2150/51 there these days anyways.

 

[tv]

Aha - yeah. How to do the thing with the grounds between the meter in point to still keep them somewhat isolated,
here we go. Who dumps what in what, right? Same deal with el beasto here, it's so funny how you keep running
into the same stuff. Anybody?

Jesus Lovin' ChristTM!

RED goes to clean,
GREEN goes to dirty ground

(imo the simplest way to "ground-sense" the signal referenced to the clean/audio ground)

Courtesy of me and pico-meter.