I'm one half of 'Expat Audio', and for several years now I've been looking into the problems with the majority of GSSL's (SSL clone buss limiters).
One problem which most builders and users simply didn't understand is that the GSSL does NOT behave like a genuine SSL buss limiter. about 8 years ago I made a video trying to show the differences when you pan a mono signal around.
Since people already noticed that "it compresses the kick and bass to hard" -and with those usually being high-amplitude, low-frequency-heavy instruments, the "band-aid" solution was to add a high-pass filter in the sidechain.
This was entirely the wrong way to fix the problem. Now the unit STILL doesn't compress centrally-panned instruments correctly, and in addition, it stops working with the correct sensitivity at low frequencies... Two wrongs don't make a right.
The CORRECT solution is to add a SECOND sidechain detector, which corrects the stereo panning inaccuracies of the GSSL design.
We built it, we offered it up to the DIY community, people bought it, built it and installed it into their existing builds, and the result was that every single one of them LOVED the result. http://expataudio.myshopify.com/products/gssl-turbo-pcb
That corrects the compression misbehavior... then people started to ask us about correcting the other issues which they were having, in particular the high noise floor. (Usually 120Hz hum in the US, and 100Hz hum in the rest of the world.) We came up with the 'CRC' board, which just solves the problem absolutely.
next came a metering board, so that people had the option of using a true VU meter instead of a milliamp meter, for those that wanted to view the gain reduction in the same manner that you would see GR displayed on an LA-2a, or an 1176 etc. -That of course doesn't actually affect the SOUND of the unit, but DOES significantly affect used perception... an interesting option, but not strictly 'sonic' on nature.
Now you have to understand that the signal path on an SSL limiter is very simple. The signal comes in, is buffered by an op-amp chip, then pases through a VCA, then goes through one final op-amp chip and hits the output. That's it. -If you want to affect "how the unit sounds", as opposed to 'how it modulates gain', then you have to take this into consideration.
Personally, I don't go for simple IC-swapping "magic", and despite the fact that people do things like strapping a transformer across the output and declaring it sonically "transformed" (get it?), this is not an approach I recommend. Not at all. You need to make some changes in order for transformers to work happily. Most of the sonic differences a user may notice probably come from the fact that a GSSL -as originally built- was never meant to feed a transformer.
If you want transformers in the signal path, you really do have to change the circuit in order to DRIVE the transformer correctly... And for that reason, we came up with another add-on board called "the Cavendish".
The Cavendish is a board which REPLACES the input and output op-amps with circuitry that essentially mimics an API signal path. The input (a differential receiver) stays largely in place, but the chip is fed through a discrete op-amp (2520 or your choice of compatible alternatives) instead. The output circuitry is basically completely bypassed. the signal is 'grabbed' coming out of the VCA, where it then goes to an API output driver circuit, with a 2520 (or compatible alternative) which then drives a 25030 output transformer, which then goes to the output XLR's.
So, basically we've looked at all the various comments which people have had about the GSSL, and made "bolt-on" boards to address its issues. We've never offered anything for sale which is "voodoo" or offers "imaginary improvement". All of our modifications are test-repeatable and significant in terms of their difference, with the possible exception of the Cavendish, which is more of a "preference" thing.
Our other offerings like the DIY mic preamp stuff have been more of a way to offer other things on the webstore, but -speaking personally here- my primary fascination was always looking for ways to improve what was initially a rather poorly-executed copy of an excellent original design (the G384 SSL limiter was actually vastly better than the 'GSSL' clone... for example, simply add a turbo board and a CRC, and marvel at the improvement in noise floor and stereo limiting detection!)
I posted a couple of videos on YouTube -in my horribly mid-Atlantic accent- (search "expat audio turbo") which show things like the metering board and the turbo board. I'm happy to answer any other questions.
Thought I'd take a moment to describe some of the efforts that are taking place to get get a small production flow set up here at expat audio (namely, my garage... but I have eyes on more if you guys start buying!)
Since the dawn of the industrial revolution, engineers, scientists and anyone self employed have made efforts to streamline repetitive tasks associated with their product. By streamlining, we create greater efficiency and lower the time and effort required to create a product.
The next few pictures are credited to user "bluesky636" on the Fender Guitar Amplifiers forum.
Electronic circuits were originally assembled using a point to point topology. Components were carefully selected, and connected to each other carefully, and hand soldered together. This can still be seen in some very high-end boutique guitar amplifiers. Obviously, this is painstaking work, and in today's economy only reserved for the most expensive of products. It's also limited to larger discrete products. Fine pitch IC's need not apply.
Once developers realized the error of their ways, and looked for more automation and efficiency in building equipment, along came "turret board". Again, a similar technology to point to point, Turett board was designed for large discrete components (resistors, capacitors, large transistors etc) but rather than needing to know exactly what connects to what, a simpler layout is created, that can documented for re-use by trained manufacturing staff.
Again, this can be found in some boutique guitar amps again.
In the 1950's following world war two, Printed Circuit Boards started making their way in to modern consumer goods. (Thanks wikipedia!). Printed Circuit boards originally still required human beings to stuff all the parts, as almost everything require pushing the leads of components through specifically placed holes on the circuit board. It's this technology that many of the Pro Audio clones we make at groupdiy were originally assembled in. Assembly Operators (people who stuff the circuits) were now able to have minimum training to populate components into circuit boards.
In fact manual assembly is still quite popular for circuits that must use through hole (devices will leads) components. Most large connectors (such as XLR's or 1/4" jacks) are hand soldered to circuit boards, even if everythign else is machine assembled!
An honourable mention at this point needs to be made for a technique known as wire-wrapping. This is very similar to point-to-point, but small wire is used to make the connection between one device and another. Devices are normally held in place in a fine 0.1" array of holes. Wire wrapping actually created a very strong bond between the wire and the device being held. It is a manual process, but seeing as the devices are held in a specific place, documentation for assembly operators is possible. In fact, the Apollo Guidance Computer, used to take man to the moon, was entirely assembled using wire wrap technology.
Anyway, back to consumer goods and PCB's!
There are some really clever ways that factories trained their operators to know where to put the right components. In my travels in the semiconductor industry, I visited a factory where components were kept in automated component bins, that would pop out automatically, as the product was assembled. Then above the workstation, a specific light would shine on the PCB to show the operator where to position the component that was in the tray. (really wish I'd taken a picture now!). In short, only one component is available to the operator at any one time, and pre-programmed slideshow would shine down on the board to tell the operator where to place that part.
Through hole components on a printed circuit boards can either have their leads soldered manually by operators, or can be run through a solder wave.As the name hints, this is literlly a wave of molten solder that the circuit boards is run over, which solders the legs of through hole components to the circuit board. This process is still used regularly today.
By the 70's and 80's smart machines had been trained to pick up and place the thru hole components, but they still weren't perfect. In a quest for continued efficiency, much of the electronic design community moved to surface mount components or SMD. This more or less brings us up to date.
Today, surface mount devices compromise the lions share of consumer electronic products on the market. Without SMD components, small portable electronic products really wouldn't be possible, as the assembly machines can be much more accurate than a human being trying to assemble at speed. (yes, prototyping with SMD is possible, and manufacturing in small volume is possible - but VERY tedious!).
Surface mount boards typically go through 3 processes -
Solder paste (rather than solder wire!) is applied to the pcb's through a stencil.
The PCB is then put into a "Pick and Place" Machine -- a machine will all the components held in reels or tubes, that has an vacuum nozzle on an XY axis. The vacuum nozzle picks up the part from a reel, then moves to a preprogrammed location on the pcb, and drops the part (on to the solder paste)
PCB's with solder paste and components in place now go through a process calls "reflow soldering", where the entire board is heated to the point where the solder paste liquifies and the tiny balls of solder in the paste turn into a liquid of their own. This solder connects the legs of devices to the pads on the PCB.
The next blog entry is going to talk about how we (at Expat Audio) will set up our own SMT production flow, to create a family of modules that can be used in your DIY systems!
Expat Audio has existed for the last few years, simply shipping PCB's that folks would buy and install parts themselves. Moving forward, that will change somewhat.
This week saw the first INA163 Mic Pre boards come back from the fab, get thrown on our pick and place machine and assembled here in Dallas Texas, then tested and pushed to their limits in Orlando, Florida.
(In the picture; 1x Uber Power Supply, 4x INA163 boards)
These boards are expected to be available in the next few weeks - PCB's and Solder Paste Stencils have been ordered.
In short, prototype boards are getting 0.0002% THD up to almost 70dB of gain (wooohooo!). Combined with the Uber Power Supply, Mic Pre FE and 3 button mic controller boards, audio DIY'ers can really get ultra high performance, for little money.
Expat Audio will continue selling the traditional GSSL mod's moving forward, but will look to integrate more and more module based products (assembled AND tested!) into the product line... helping audio DIY'ers get to a final product faster and easier. (and focus on the rest of the signal chain!)
