Adding a wet/dry mix feature to the output of GSSL

I was wondering how difficult it might be to add a wet/dry mix feature to the output of the GSSL stereo compressor so that one could achieve parallel compression without having to deal with the hassles of a patchbay and hooking up extra cabling. In other words there would be a control that allows you to blend the dry uncompressed signal with the compressed output of the compressor. I'm thinking that you could probably make this passive since the loss from mixing two signals is not really that big a deal especially when dealing with a hybrid digital/analog setup. In my mind that's just added headroom and I don't have to worry as much about clipping the A/D converters on the way back into the computer. Having a slight loss also allows me to drive the output transformers I added to my GSSL a little harder as well.

I just found this:

http://www.groupdiy.com/index.php?topic=8360&highlight=crush++blend

That idea requires a few more parts. Is there a way to do this passively?

thanks,
Brad

[quote author="Brad McGowan"]Is there a way to do this passively?[/quote]
..if there was, I'd have saved myself some money and built one... :wink:

Keith

[quote author="Brad McGowan"]Is there a way to do this passively?

thanks,
Brad[/quote]
I guess in principle it could simply be done using transformers & some resistors. But then it won't be cheaper, which is what you were after in the first place I guess.

Regards,

Peter

[quote author="clintrubber"]I guess in principle it could simply be done using transformers & some resistors. [/quote]
It could even be done without transformers, but ANY passive approach will ruin impedance matching, any impedance-sensitive items in the chain will suffer. -Different impedance devices will work better or worse, but in ANY case, it'll be lossy and unpredictable, no matter HOW much you spend.

Basically this HAS to be active to be worth a rats ass.

Keith

[quote author="SSLtech"]Basically this HAS to be active to be worth a rats ass.

Keith[/quote]
I'm totally with you that it'd be madness to do it differently, but weren't there of those 'mixing-transformers' back then ? It's of course then sidetracking into an academic issue while the active circuit as you drew it could already have been completed in the meantime, but OK.

Regards,

Peter

Yes, but the mixing transformers had very fixed mixing combinations, and date from time when you could usually predict a known, common impedance for sources and destinations.

These days, some stuff is 600Ω, other stuff is 100kΩ input impedance. Making something work passively for every combination of load impedances is impossible, and then you still have to make it variable.

The active solution really does use a minimum number of parts, and if you can build a GSSL, then the proposed "Crush-n-Blend" is a piece of cake...

Keith

I want to play my old records, and I really don't want to spend the money on a turntable. But I already own a food processor. Can that be rigged up to do the job?
:wink:

[quote author="NewYorkDave"]I want to play my old records, and I really don't want to spend the money on a turntable. But I already own a food processor. Can that be rigged up to do the job?
:wink:[/quote]
With just a foodprocessor it will be difficult, but if you have a scanner nearby it can be done :thumb:

http://www.cs.huji.ac.il/~springer/

You can sort of do it passively. you can modify the circuit adding only passive components. If all you want to do is blend in a little dry to the compressed, this will do it. it will probably sound good on highly squashed drum tracks. you won't get a wide range of adjustment ie, mostly dry with a tiny bit of compression is impossible.

take the signal from the 5534 input buffer's output pin, cap couple thru a pot(250k?), an spst switch and a series resistor of about 20k, put that into the virtual earth point on the 5532. pin2 on the gyraf schematic.

[quote author="mikep"]You can sort of do it passively.

, put that into the virtual earth point on the 5532. pin2 on the gyraf schematic.[/quote]
That's cheating !

True of course, you're only adding passive components, but the resulting summing will still be active. It's just that you're giving the already present active stuff one more signal to process.

All fine.

I also wanted to do something like that some time ago... but it never worked properly...

if you use a passive solution.. you cannot use the make-up gain anymore, because if you turn it up you have a feed back around the vca and this affects the compression ratio...


I think it should look out for something like this...

Won't work as drawn...

The input and the output are connected on the first op-amp. To correct that part, the noninverting input needs to be lifted from ground, the iput needs to be disconnected from the inverting input and connected to the noninverting input.

Same problem with the second stage...

Lose the first stage, correct the second stage to make it noninverting, and you'll have something that passes signal. Then what you'll have is something that HALF works, because it'll never get anywhere better than a 6dB compression balance... and all it's going to do as you feed it in, is make everything LOUDER!

What you'd have to do here is look at the crush-and-blend circuit and how it's done, and incorporating THAT into it.

Essentially, just omit the unbalancing/re-balancing parts, and build the crush-n-blend. That'll give you 100% continuously variable control between uncompressed and compressed.

What I can't get over is that it was intended to be super-flexible, allowing you to use it on any compressors which you build in the future: not just the GSSL: the PRR Vari-mu, G1176, Pico Comp, Level-Lok, Forssell opto etc. all spring to mind, and that's before you even consider anything you might purchase on eBay or elsewhere... -Yet most people want to strip this ability and build a significantly less-flexible version, that can only work with the GSSL... and it's been discussed a LOT before... The fact that there was no easy way to do it was what led to the crush-n-blend being designed in the first place.

"Doomed to repeat it", I suppose...

Keith

[quote author="SSLtech"]What I can't get over is that it was intended to be super-flexible, allowing you to use it on any compressors which you build in the future:[/quote]
Can imagine your feeling. Maybe the added cabling puts people off? Or they prefer just one additional control i.s.o. having to make an additional box?
While your C&B circuit is of course easy on Veroboard, maybe a ready to go PCB would convince people ?

And they could then still de-universalizeTM it by adding it into a specific box if they really wish to do so.

regards,

Peter

I think both solutions are useful...

a external box for all the gear you don't want to modify
and an embedded solution for all the diy gear.. last one is of course better, because you can omit the balancing and debalancing stuff...

I think I'll work something out...

But there isn't a single solution for all the DIY gear... the solution for the GSSL would be different to the solution for the G1176, nothing like the PRR, which would bear no relation to the Pico Comp...

-Ya see where this is heading? -And I'm not even slightly convinced that building a stripped-down version into the GSSL even would sound better than an external box... I see no compelling reason why it should.

Is the real reason that people don't want to make another power supply and front panel? (the front panel would need to have a single potentiometer on it.. oh, and perhaps a power switch...)

Keith

SSLtech said:

But there isn't a single solution for all the DIY gear... the solution for the GSSL would be different to the solution for the G1176 [/qoute],

thank god for that principal in general, else life wouldn't be nearly as fun. but seriously, didn't the original post specifically ask about the GSSL?

Im not trying to put down your design keith, but I think this WILL sound better, as long as you don't need much flexibility.

[quote author="SSLtech"]
Is the real reason that people don't want to make another power supply and front panel?

Click to expand...

I think the original poster was being a cheapskate, which I can certainly understand.

[quote author="mikep"]...but I think this WILL sound better, as long as you don't need much flexibility.[/quote]
[quote author="SSLtech"]I see no compelling reason why it should.[/quote]

[quote author="SSLtech"]I see no compelling reason why it should.[/quote]

Its just that going thru that circuit is going to alter the normal compressed signal some small amount. I consider 2141's and 2142, and the IC opamp "significant".

and I find a hard time reasoning how a resistor + cap parallel path around the VCA is going to alter the performance or sound of the VCA one bit.

I am coming at this from a mastering perspective and I believe the sound of the unmodified SSL compressor electronics with no compression is usually too colored.