An API Compressor Project

[etheory]

Hi there!

I am starting this thread to share my experience with DIY'ing a 2500/225L-inspired compressor.

I am working with a schematic I found on this board, that if you really go looking, is pretty easy to find.

I started off this crazy journey with this thread on DIY CNC'ing a 2520 I simulated in LTSpice based on information that I again picked up off this board ( and was not hard to find ): http://www.groupdiy.com/index.php?topic=46998.msg590345#msg590345
There are some basic examples of tests on that DOA at the end of that thread.

Next up, I've taken the schematic and broken it into 3 parts, 1.) The input stage, 2.) The VCA and sidechain circuitry and 3.) The output stage.

Today I verified 1.) and 3.)  Next I will work on 2.)

I wanted to share some images of the input and output stage tests.

First here is the input stage, which is a relatively simple, almost text-book, electronic de-balancing circuit based on my 2520 ( in the original it's a 2510, but I haven't built my 2510 yet ):

You can see the Xprotolab in there also assisting me in testing.  The top trace is the input, and the bottom trace is the output.  You can clearly see that this is an inverting stage with gain 0.5x when run unbalanced ( as here, since one side is grounded, when I input a proper balanced signal it'll be 1x ).

Next up I tested the output stage:

Here you can see the gorgeous Ed Anderson classicapi.com 2623-4 (http://classicapi.com/catalog/product_info.php?products_id=31) output transformer.  This will soon be replaced by a 2503 once it arrives - although throughout development I will be testing the sound of both, since I want this to be a character compressor as much as a partial 225L replica.  You can also see the rather nice Vishay output cap and WIMA bypass cap.

Here you can see that this stage is also inverting, and this time of gain 1x

Next step is to finish my Eagle board for my take on the 2510 after much tweaking on LTSpice, again entirely based on information found on this board, which I'm going to make in 2520 footprint since it allows for the most options for future expansion ( i.e. I'd love to design some other DOAs and put them in this to see how they sound too.... )

After that, which I hope to all have finished by the start of next week, is the significantly more complex VCA stage, which I'll breadboard up in stages and test also.

I think that the input, VCA and output stages will all go on one PCB just like the actual 225L, and then I can dev the sidechain and control voltages on another, which will aid both development, and simplify the interfacing at this stage.  I'm not sure if I'll combine them in the final version onto one PCB or not.  Since I'm building 4 of these in one box, that might end up the best option.

Some more images:

More to come so stay tuned!

A big thanks to Volker and Jeff Steiger, even if they don't know directly they've helped with this project, they have  8)

 

[Rob Flinn]

Nice work.  I love the mini oscilloscope did you get that from ebay & is it any good ?  I'm looking at maybe replacing my aging velleman portable scope which is a bit intermitant nowadays.

Just wondering if perhaps you could remove all the * between the letters API & 2500 etc in the thread title & the body text.  If you leave them it means the thread will be unsearchable in the future.  Sorry it's a pet peeve of mine, but let's face it you may well have found the info that started you on your quest by using the search function so ..............

 

[pucho812]

I agree the whole letter*letter* makes it hard to search and 99% of all companies do not have us in their radar or in their sights.

 

[etheory]

I love the mini oscilloscope did you get that from ebay & is it any good ?

That's an Xprotolab from gabotronics.com: http://www.gabotronics.com/development-boards/xmega-xprotolab.htm

"good" is open for interpretation ;-) I'd say it's not as good as a bench-top scope by any means - it has a TINY screen and only samples in 8-bit precision, however, it's functional, has 2 analog input channels, has a relatively high sampling rate capability, and it works well with my "tiered approach to testing" ( assume your first effort will explode, therefore connect your cheapest piece of test gear until you've verified it ).  The arbitrary waveform generator ( which I used all throughout my tests ) means that I can use it as a small, cheap, lightweight testing platform without having to run a single cable off of a breadboard to a big bulky external box.

I bought the Xprotolab cause it's cheap and I won't cry if one of my creations makes it catch on fire for some reason.  My next purchase is likely to be this one: http://www.seeedstudio.com/depot/dso-quad-aluminium-alloy-black-p-1034.html?cPath=174

Which is significantly better in pretty much every way.  However I'd still kill to get a proper, accurate, large, bulky, analog oscilloscope any day.  Until my music starts paying the bills that's unlikely to happen.

I'm looking at maybe replacing my aging velleman portable scope which is a bit intermitant nowadays.

I'd probably go for the DSO Quad then.  But you'll want to check it's tolerance specifications, as it might not cut the mustard for tight tolerance work.

Just wondering if perhaps you could remove all the * between the letters API & 2500 etc in the thread title & the body text.

Done, thanks for the insight, I'm still relatively new here so I was being a bit over the top with that, understood 

 

[gemini86]

I have a 225L module sitting next to me if'n you need any info from it...

 

[hobiesound]

really cool!!!

BTW thank you for cropping/resizing those pics so i don't need to side scroll! 

greetings,

Thomas

 

[Rob Flinn]

Which is significantly better in pretty much every way.  However I'd still kill to get a proper, accurate, large, bulky, analog oscilloscope any day.  Until my music starts paying the bills that's unlikely to happen.

That's not really an issue as I have a couple of analog scopes, here.  The problem for me is when I'm working in London or abroad & I can't take a real scope with me on the train or aeroplane.    Most of the time I just need to see if a sine wave is arriving at different points in a console or other piece of gear so a cheap lightweight scope is ideal.

 

[etheory]

really cool!!!

Thanks!

BTW thank you for cropping/resizing those pics so i don't need to side scroll! 

Then get a bigger monitor  8)  It looks just fine on mine!  OK, fair enough then, I'll make them about half-size then if that helps.

 

[etheory]

The problem for me is when I'm working in London or abroad & I can't take a real scope with me on the train or aeroplane.    Most of the time I just need to see if a sine wave is arriving at different points in a console or other piece of gear so a cheap lightweight scope is ideal.

If that's the case then the DSO Quad is ideal, or, yes, you COULD use the Xprotolab, but not directly.

It needs either a mini-usb connection through which it draws power, or +5VDC and +3.3VDC rails, which you could whip up, in a box, along with it at a fraction the size of the DSO Quad if you wanted to.  Would be a really interesting project....

For checking the arrival of sine waves it's a REALLY cool tool for sure.

 

[etheory]

I have a 225L module sitting next to me if'n you need any info from it...

I think I'll take you up on that.  The schematic I have is missing a detail that's been puzzling me for a while.
As I read it, in feed-forward mode, it only accepts an external side-chain, and isn't able to use the incoming audio as a control signal.  I am assuming the real device doesn't operate this way, so it would be good to get that correction into the schematic I have in front of me.  I'll PM you.

 

[etheory]

Tentative and completely incorrect initial front panel design:

Lots of work to do....

 

[gemini86]

Just looking at the card edge, the inserts just jumper to the mult +/- fingers...

 

[Rob Flinn]

I had to open up a 2500 recently that had gone a bit wonky, left to right.  Turned out the chips needed reseating, which it seems is a known issue with these units.

What I noticed is that there is only one 2520 per channel with corresponding o/p transformer.  The gain cell is four parallel that's 2180.  Aparts from that, from memeory it just uses pretty standard DIP op amps.

 

[etheory]

What I noticed is that there is only one 2520 per channel with corresponding o/p transformer.  The gain cell is four parallel that's 2180.  Aparts from that, from memeory it just uses pretty standard DIP op amps.

Sweet, sounds like Volker's 225L schematic ( the basis for this project ) is pretty much on the money then for a 225L OR 2500 clone.  That's good to know and lines up perfectly with the info I've gathered thus far.  Thanks!

As far as I know there is a 2510 on each input, the 4 2180's as the VCA ( 225L has 2 ), and a 2520 driving a transformer on the output.  The only reason that I am testing both circuits with 2520's is that's all I have at hand at the moment.  In fact in a finished unit, I think it's better to use 2 x 2520 footprints instead of the 2510, since it's much more versatile for experimentation.

Here is my latest addition to the project, my take on a 2510 op amp ( OK OK so if you read my other thread then you would know that mitred tracks are nothing but BS as far as electrons go, but come on, they look SOOO nice ;-), so I'm going to leave them that way ), also note the rounded corners for the milling bit, a suggestion of a friend of mine, which I can't wait to see on a board in real-life:

And here is my refined new 2nd edition 2520:

 

[etheory]

The 2503's JUST arrived for the output stage 8)  Thanks Ed Anderson and Jeff Steiger, they are BEAUTIFUL looking heavy little buggers.

Unfortunately, however, I'm going to have to park this thread for a month.
I've got to travel interstate for a freelance gig, so I can't continue until then :'(

 

[etheory]

OK, so the expected freelance gig didn't happen, but something else has come up in it's place.  So I have a couple more days to go nuts on this project, before I start working on another film project.

So firstly, here is a temporary PSU board I just designed and will print tomorrow to replace the mess currently on the breadboard:

And I'm currently designing the main board which has the 2510 input, 2520 output, the VCA's and the basic processing circuitry for the control voltages.  This will "run off" to the breadboard so I can tweak and test the side-chain circuitry.

Back soon!

 

[Harpo]

So firstly, here is a temporary PSU board I just designed and will print tomorrow to replace the mess currently on the breadboard:

http://www.evolutionarytheory.com/wp-content/uploads/2012/01/etheory_AC_To_Dual_DC_PSU_v01_01.jpg

A bridge rectifier might be a nice to have on board.
Instead of D1 and D2, a bypass cap closer to regulators inputs might be of more use.
Routing input of LM337 from behind could get you to a single layer board or spare a wire link.
Linking the trimmers wiper with its CW pin instead of its CCW pin seems more intuitive for increasing output voltage when turning clockwise.
Just my 2ct and YMMV.

 

[etheory]

A bridge rectifier might be a nice to have on board.
Instead of D1 and D2, a bypass cap closer to regulators inputs might be of more use.
Routing input of LM337 from behind could get you to a single layer board or spare a wire link.
Linking the trimmers wiper with its CW pin instead of its CCW pin seems more intuitive for increasing output voltage when turning clockwise.
Just my 2ct and YMMV.

All good points!  Although this is a somewhat "special" board.  D1 and D2 are there as a half-wave rectifier.  For my specific purposes a bridge rectifier won't work.  The basic premise is that you feed AC in ( not centre-tapped from a CT transformer, just from an earth wall-wart that I have, single ended ) and get DC +-V out.  It's purely for testing stuff on a breadboard ( which I still have a lot to do of for this project ), so it's by no means a "final" PSU, but more a very specific one I wanna use for testing.  One I am not afraid to have blow up a couple of times during tests ;-)

The routing for the 337 is an approach I tried, but then had a bit of letterheads with a mate of mine who I always use for mutual checking of work, who insisted that breaking the "earth" ( it's not really earth anyway ;-) ) loop was potentially bad.  However I'll likely ignore him since his previous example of this being an issue was in a multiple MHz digital device, not a low frequency analog PSU like this.  So yeah, it's still an option for removing the link.

The trimmer idea is perfect, and I'll change that now.  TBHO I always intended on testing that facet before committing to a CNC of the board, so thanks for the reminder, and for the time to give it a look over!

 

[etheory]

Hmmm, I just checked the trimmers I have here, and the way it's shown on the previous schematic does give clockwise increase in voltage.  I think my trimmers are just different to the Eagle symbol I am using.

 

[Harpo]

D1 and D2 are there as a half-wave rectifier.

Maybe missing something, but as drawn they only short out the transformer every 2nd. half cycle. Dunno in what sequence parts will blow, probably diode 1st., followed by transformer when diodes fail short, or the reversed biased caps when diodes fail open.