New GSSL Layout Proposal - Anyone Interested?

[ruckus328]

Disclaimer: that's what 2180 appnotes say - use a single 5K1 to V-. Try to ask for the "second opinion" from people who have devices configured this way or the "old way".

Also, regarding rail voltage, appnotes say that the VCAs are current-in/current-out devices and in this regard less dependant on actual rail voltage compared to voltage-out devices.

The sidechain VCA should not be as critical regarding distortion, so the "dist-trim" could be left as-is or possibly even completelly left out.

Interesting - so basically whether I feed it 12V or 15V, resistor values should remain the same.  Which causes me to start to question those values, since as you said - app notes don't say anything about them, but I would think if they were unessecarry or caused some sort of issue, somebody would have brought that up by now.  I believe Gyraf & Turbo boards both instruct to put a 3K9 resistor between pin5 and ground, not a 5K1, and Marten's page only says to change it for the audio VCA's.  Come to think of it, I seem to remember somebody actually saying to leave a 3K1 there for the sidechain VCA on one of the threads when I was building my first unit.

I do see though that Marten says the 10K between pin 3 & 5 can be removed on both the audio and sidechain VCA's (missed that).


Sidenote:

Kingston - aware if this issue - my schematic has the 100pF intact.

 

[tv]

Heres the "typical" application circuit.

If you use 5K1 @ 15V and scale down to 12V:

5K1 * 12/15 = 4,08K

3K9 is the nearest value..

 

[ruckus328]

Drrr.....Yea, gotcha.  This is one of those "no duh" moments I spoke of.  Thx.

Lukas, you really need to sig the whole "Spidey says it's going to work" deal-e-o.  That's classic.

 

[tv]

Heres a cut'n'paste info from the appnotes, regarding the dc-bias "5K1" resistor.

You asked earlier what's the "benefit" of running the sidechain VCA on +/-12V rails versus the "system" +/-15V rails. It appears to me that whatever fluctuations may occur in the MAIN 15V rails would affect the gain-cell operation in the sidechain VCA. This, in turn, would affect the whole unit, because this VCA is used as a "detector" cell (or 2 in oxford).

IOW, what gyraf did, i.e. powering the sidechain thru separate 12V rails seems like a very good decision from the "systems operating POV", ensuring stability of GR detection and CV even if the main 15V rails are sagging/dirty for whatever reason. (IOW, the main VCAs would get predictable CV no matter what is happening in the main audio path, because sidechain is isolated from it via its own rails)

Because of this, I would reccommend using 12V rails in the sidechain as well (3K9 resistor!), and paying a little extra attention to over-dimensioning the C-R-C pre-regulator filtering, so that the sidechain rails stay clean no matter what! As always, ymmv..

 

[ruckus328]

AND

I think that sums it up lol.

I see the benefits of running the sidechain at 12V.  What I'm wondering is this: do they outweigh the benefits of running the sidechain at 15V? 

From a lyout perspective - I'm in good shape either way.  If I do run it at 12V the question is whether or not to power the 12V reg's off of the 15V rail or back at the power supply off of the main rail.  If I do this, I'd have to move those 12V reg's off of the main board and put them back on the power supply (it will be tight but I can do it).  If I leave them on board and power them off of the 15V rail, I can have a plane area for 12V, tied to the main audio ground right at the 12V reg's.

I've paid alot of care to keep that 15V rail completely clean and as quite as possible, so I'm wondering if I'll have an issues with running at 15V - of course the only way to know is to power it up, and I'd hate to find out I'd made the wrong choice. 

All I know, is I have to finalize the plans on this thing, otherwise I'll never get done as it's become a massive moving target.  It's pretty much down to this issue - 12V......or.....15V.  And I really want this thing to be something that everyone will enjoy.  So far, I'm inclined to leave the SC at 12V.  I haven't really seen an actual reason to change that other than the original SSL did.

 

[tv]

Well, there's multiple scenarios (I'm not familiar with your actual layout, so this is an "academic" discourse):

The user will of course have the choice of using either 15 or 18V for the main rails - you will have no control over that, and some WILL go for the higher rails. In this case, I would (for convenience) possibly use the regs. in-series (i.e. 12V feeding off of 18V rails, decoupled via a small RC filter).

But with 15V, I would rather feed the 12V regulators directly from the graetz.. thru a hefty CRC filter. Which in you scenario means having them on the PSU board.

Now - what to do?


Perhaps you should wait for other's opinions, and sleep over it.

 

[ruckus328]

Sleep = Negative ;D

Well my PSU is adjustable, so yes - 18V if you want it.  For that reason, I think it is pretty obvious that the Sidechain Voltage should remain at 12V.  All of these years that the GSSL has been out - I've never heard someone say: "Man, my GSSL just sounds like garbage because the sidechain is only at 12V"... ya know?
SSC is running at 15V also, but I don't see that there could be an issue running it hotter at 18V also.

Now, if I move 12V regulation to the power supply: 

Pros:  Can run right off the Main Rail, no bogging down the 15V rail, AND - it becomes a universal GSSL Power Supply, capable of being wired into any existing GSSL already out there as well.

Cons - Couple more wires to run, crowded Power Supply, might have to add a couple more caps for some extra main board storage.

___________

Another side note:  T, I just sat down with two EE's and looked at what's going on with the PICO clean/dirty ground scheme.  None of us could figure out exactly how that's gaining you anything.  Are we missing something here?

 

[Biasrocks]

Well my PSU is adjustable, so yes - 18V if you want it.  For that reason, I think it is pretty obvious that the Sidechain Voltage should remain at 12V.  All of these years that the GSSL has been out - I've never heard someone say: "Man, my GSSL just sounds like garbage because the sidechain is only at 12V"... ya know?
SSC is running at 15V also, but I don't see that there could be an issue running it hotter at 18V also.

Use LM317/337's and a decent sized PT and be done with it. No way your design is gonna draw 1.5A's per rail. The 12volt regs in the GSSL is actually one of the weak points of the design IMO, they generate a ton of heat because they're dropping nearly 10volts and I've seen the (+) voltage rail drop below 12 volts because the reg's are taxed nearly to the breaking point. There's also poor decoupling of the 15 volt rails around the VCA's themselves. All this impacts the sound of the existing GSSL in a  negative way.

Mark

 

[ruckus328]

Bias, couldn't agree with you more on the 12V Reg's being the weekest link - in fact first thing my GSSL did when I turned it on was blow the +12 Reg.  Replaced it with an LM7812 1.5A and everything's fine.

I'm still unclear what you're suggesting here though.  I am using 3X7's for the 15V rails.

The questions is: whether or not to run the sidechain off of 12 Volts or not.  And if so should I just run the 12V regulators off of the 15V rails or off of the main +/-.  In either of these cases, I'll be using a TO-220 package, not the little 100mA TO-92's.  If you are suggesting to use 3X7's for this purpose as well - seems a little overkill to me.  Plus it will nearly double the size of the power supply, and add a bunch of extra cost and parts.  Seems much more logical to just use fixed regulators in this application.  Again though, I'm unclear if you were suggesting to use 3X7's for the 12V supply or not.

In any case, even if I do run the sidechains off of 15V, I still need 12V for the control voltage (the original SSL had this), so I'll need to put some sort of +/-12V regulation on there somewhere, whether it be on the main board or on the Power Supply.  If this is the only place I would need 12V then I see no reason not to just use the 100mA TO-92's and drive them right off of the 15V Supply.  They will only be dropping 3 volts, and sucking virtually nothing as far as current.

In any case, I really want to be done with this as well.  I'm well past the burn out point.  But I'm so close I don't want to call it quits.

 

[tv]

Another side note:  T, I just sat down with two EE's and looked at what's going on with the PICO clean/dirty ground scheme.  None of us could figure out exactly how that's gaining you anything.  Are we missing something here?

Quite possibly.

http://electronicdesign.com/article/analog-and-mixed-signal/What-s-All-This-Noise-Rejection-Stuff-Anyhow-21808.aspx

Let them dissect whats said there and apply it to the meter connection.

Hint: you have different situation than described in the article, and you have different reference point, but if they are worth their salt they will get it. With the fet opamp, and inverting connection, as in meter input you
a) escape amplifying the ground noise diff
b) don't dump hash from dirty ground into the audio ground

isn't that obvious enough?

 

[Harpo]

Bias, couldn't agree with you more on the 12V Reg's being the weekest link - in fact first thing my GSSL did when I turned it on was blow the +12 Reg.  Replaced it with an LM7812 1.5A and everything's fine.

I'm still unclear what you're suggesting here though.  I am using 3X7's for the 15V rails.

The questions is: whether or not to run the sidechain off of 12 Volts or not.  And if so should I just run the 12V regulators off of the 15V rails or off of the main +/-.  In either of these cases, I'll be using a TO-220 package, not the little 100mA TO-92's.  If you are suggesting to use 3X7's for this purpose as well - seems a little overkill to me.  Plus it will nearly double the size of the power supply, and add a bunch of extra cost and parts.  Seems much more logical to just use fixed regulators in this application.  Again though, I'm unclear if you were suggesting to use 3X7's for the 12V supply or not.

In any case, even if I do run the sidechains off of 15V, I still need 12V for the control voltage (the original SSL had this), so I'll need to put some sort of +/-12V regulation on there somewhere, whether it be on the main board or on the Power Supply.  If this is the only place I would need 12V then I see no reason not to just use the 100mA TO-92's and drive them right off of the 15V Supply.  They will only be dropping 3 volts, and sucking virtually nothing as far as current.

In any case, I really want to be done with this as well.  I'm well past the burn out point.  But I'm so close I don't want to call it quits.

Only my humble opinion, but I would only supply the sidechain VCA, ratio network, threshold and makeup control from +/-12V by these 7xL12, fed from the 3x7 regulators main supply. These come with lower dropout voltage than 3x7 regulators, take less boardspace, are cheap and easily availiable. From datasheet they need 330nF caps close to input and at least 100nF on output. Placing some anti-latching diodes might be useful as well.
The sidechain VCAs abs.max.current is 10mA, the ratio network drawing less than 1mA, so in dual sidechain/turbo config drawing 22mA. Add 0.72mA for threshold and makeup and add 2x6.5mA regulators quiescent current giving a worst case total of 36mA for both regulators. Maybe less than a 3x7 regulators config would require to meet min.current load.
Using 7x12 regulators, coming with higher quiescent current and dropout voltage than the 7xL12 parts might give poorer results in this config with +/-15V preregulators, so the TV suggested feed from a separate CRC might fit better in this case.
The sidechain opamp stages could better be supplied from the higher voltage rails.
As always YMMV.

Heres the "typical" application circuit.

If you use 5K1 @ 15V and scale down to 12V:

5K1 * 12/15 = 4,08K

3K9 is the nearest value..

From THAT2180 datasheet I'd calculate it a little different for 2.4mA bias current.
(-2.85V - -15V)/2.4mA=5K0625 or (-2.85V - -12V)/2.4mA=3K8125. These -2.85V are the '4 diode drops below 0V' or 'voltage at V- with no signal'.
For these -15V or -12V rail voltages the nearest R values will be the same as your 5K1 or 3K9 though.

 

[ruckus328]

Harpo, thx for your input - this is just the kind of stuff I'm looking to hear, and your raise some good points.

It know it seems at this point like it's becomming overly complicated, but I want to make sure what I implement is something we can all be happy with - I mean, I could make a decision on this in 5 seconds flat and call it a day, but it doesn't neccessarily mean what I settle on will be the most logical solution, which is why I am all for hearing some opinions on this.

Since control voltages/Sidechain VCA is pulling so little, I can't justify using anything other than the L12's and supplying them off the +/-15V rails through a C/R/C if I go with this scenereo.  Like you said, simple cheap, readily available, and little board space.  Voltage drop is exactly what you would want it to be (3V).  And I can keep them on the main board.  We won't have any of the issues we have with the current GSSL boards as far as them being stressed (by your calculations worse case they are at 30% of there capability) - Nothing, and I repeat nothing other than what you said will be running off of the 12V rails.  All LED's/relays/etc are on their own rail with plenty of juice to supply whatever they need.  If anyone wants to add additional Lights,relays, whatever, they can tie into this dedicated rail.  There's even some wiring Pads for them to do so.  Seems to me, it makes really no difference whether the Sidechain Opamps are ran off of 12V or 15V, other than running them off of 15V will take the load off of the 12V rails.  Seems logical.

Actually, with the exception of the Opamps running off of 15V, this scenereo is what I currently have implemented in copper atm.  And as luck may have it, changing the opamps to 15V will take all of about 2 minutes. They're placement is such that I won't have any issues adjusting the plan accordingly. 

I'm gonna sleep on this, but seems like this is the most logical way to go, unless anyone has some concrete reasons why it wouldn't be.

Ohh and I do have protection diodes across the L12's already.

Thx again everyone.

 

[tv]

From THAT2180 datasheet I'd calculate it a little different for 2.4mA bias current.
(-2.85V - -15V)/2.4mA=5K0625 or (-2.85V - -12V)/2.4mA=3K8125. These -2.85V are the '4 diode drops below 0V' or 'voltage at V- with no signal'.
For these -15V or -12V rail voltages the nearest R values will be the same as your 5K1 or 3K9 though.

Yes! but, for brewity and transparency... (I used my patented toilet-paper math approach..)

For the inquiring minds: BOTH Harpos "precise" and my "toilet-paper" method would lead one to use 6K2 resistor in case of using 18V rails.

So there 8)

 

[Harpo]

Yes! but, for brewity and transparency... (I used my patented toilet-paper math approach..)

For the inquiring minds: BOTH Harpos "precise" and my "toilet-paper" method would lead one to use 6K2 resistor in case of using 18V rails.

So there 8)

LOL. patent#123456789 on 'Improving ones multitasking abilities by ballpark figuring resistor values and the academic use of toilet-paper'.
Differences for same current will increase for lower supply voltages, thinking +/-9V or split 9V for guitar pedals. (your patented method 3K06 or 1K53 vs mine 2K56 or 687R). At some point this might get your method back to its original use  ;D.

 

[tv]

Of course, but you can't get pass recognizing that it works surprisingly well AND fast at the "sweet" voltages used most commonly in audio, +/-12 - 18V.

Who's talking guitar pedals, eh?
;D

 

[ruckus328]

Curiousity question - are you guys all using film's for your PS decoupling (aka bypass) caps?

 

[tv]

I used film, multilayer cer. and tantalum (only in 9 or 12-volt guitar pres).

Perhaps I liked the multilayer cer. x7r's best - but if you look at diyaudio.com, theres a lot of use of film caps in PSU and rail decoupling..

 

[ruckus328]

Yea, me too as far as the ceramics.  Left enough room to use film's if you like though.  It starts to get complicated since there's .1uF's in the audio chain for this, cine films would be desirable there.  For that reason people may just wish to use Film's across the board for .1uF's, just to simplify things.  As you guys like to put it though ymmv.

Back to this thing - I was all done with my changes when paranoia set in again regarding those L12 regulators.  I just can't seem to get the bad taste of those things out of my mouth.  And I think alot of people frown upon them as well with all the issue they've caused on the existing design.  Even though I've addressed the issue of LED's and what not bogging them down, they still make me nervous.  Since thier reliablility is debatable, and since whether powering them off of the 15V rail might cause some additional stress on the 15V Regulators (even though probably not), I've decided to go the safe route here.  This is the LAST TIME I will be changing it lol.  The kicker is I spent HOURS trying to cram those +/-12V Regulator circuits on the main board.  All wasted time.  Ohh well.

In any case - my final version:

All regulation is now off board, driven directly off of the Bridge.  +/-12V rails will be powered through a CRC off of 1A TO-220 LM7812/7912's.  That means the 12V Regulators will be at about 10% load worse case.  Plenty of overhead.  Overkill really, but safe & reliable = yes.  Dedicated 12V rail for LED's & relays also powered off of a 7812 through a CRC.
So, no more blowing regulators, no more worrying about stressing any of them, period.  The PS Boards will be part of the kit for the new design, however I will likely order additional Power Supplies Boards in case anyone wishes to use them with existing GSSL's (now that they have both +/- 12V & 15V, they will be backwards compatable)

Sidechain Voltage will remain intact at 12V.  Maybe down the road I will experiment with upped voltages here, but if I roll the dice on this one - this thing may never get done.  12V is proven.  And sidechain will behave the same if you up the Voltages on the Audio Path, which now that the main 15V supply is adjustable - you are free to do.

 

[Ptownkid]

Are you dead set on the 78/79's....317 and 337's are much cheaper are they not?

 

[ruckus328]

Are you dead set on the 78/79's....317 and 337's are much cheaper are they not?

Nah, 3X7's are more expensive, take up alot more real estate because of added components needed to make em work, and for this application there might not even be enough current draw to even turn them on.  This is only for the 12V for the sidechain & control voltages.

For the audio rails, I am using 317/337's though.

SO:

+15V - LM317
-15V - LM337
+12V - LM7812
-12V - LM7912
ISO +12V (For LED's, Relays, etc) - LM7812


PS: You got mail