Skip to main content

I am going to undertake an experiment, and I need the help of knowledgeable people. In return, I will share my results freely in another thread for all to use. So if you contribute, you stand to gain something. Perhaps a perfect recreation of the API525, perhaps an imperfect one, but most assuredly, a new idea on how to achieve the sounds of your dreams.

This is a big promise to make to experienced engineers. But I will try. I will do my best. And I will keep my ego in check.

The only thing I don't want is discouragement. Please don't drop by to say it's a waste of time or impossible. This is an experiment; as such, it may fail. We already know that.

Here's what I want to do: Use free VST plugins, along with routing/busing/sends-returns,EQ, etc., to mimic the legendary API-525 compressor. In order to do this, I need some key pieces of information:

1. Refer to this thread for general info:

2. What is the actual attack time?

3. What is the crossover point between lo-freq compression and mid-upper freq compression?

4. What does a "feedback" circuit (in this compressor) actually do?

5. Does the de-ess function operate by simply inverting the hi-freq portion 180 out of phase and sending some back in?

I propose that some sort of parallel compression is taking place to process the lows separate from the hi-freq. That is where I will start. Problem is, I have no API525 to compare to, so your ears and experience are a must. I'll post sound clips and we'll see how it goes.


Topic Tags


anonymous Sat, 03/02/2013 - 08:40


I snagged this from API's website:

Originally released in the 70's, the API 525 is a "feedback" type compressor. An input (threshold) control and output (make-up gain) control are provided, along with an additional "Ceiling" fine-tune function which can increase gain reduction (lower threshold) while simultaneously raising output level to match, so the user can alter dynamics "on the fly" without level changes.

A compression/limit switch sets ratio at either 2:1 or 20:1.

Attack time is as fast as 15u/s.

Four auto release modes are offered via two switches, 0.1s, 0.5s, 1.5s, & 2.0s.

A De-Ess function inserts an inverse vocal energy curve filter in the detector for effective sibilance/pop reduction. The 525 design has been taken from the original blue prints and spec control drawings from the API archives.

Release times vary with frequency, with high frequency / full bandwidth content released faster than lower frequencies for natural envelope tracking. Attack time is fixed, chosen to catch the fastest peaks, without "pumping". Equally useful as a tracking, mixdown or program compressor/limiter, the 525 utilizes "vintage" dynamics control and an easy to use multi-function control set. The 525 has a hard-wire bypass switch, a balanced input, and a pinout that is the same as all the other 500 series modules. Two or more 525s can be linked for multi-channel compression.

source: [[url=http://[/URL]="http://www.apiaudio…"]API 525 Discrete Compressor, Re-Issue[/]="http://www.apiaudio…"]API 525 Discrete Compressor, Re-Issue[/]

Paul999 Sat, 03/02/2013 - 08:44

That is really tough! The 525 like all API toys have a sound. It is reacts a little like tape where the harder you hit it it compresses a little. The highs are really cool sounding. They get this slightly rolled off but airy sound. This is a really bad description.LOL

It is not earth shattering contrary to all the hype but it is worth it. The waves API collection is good and useful but did not get the API sound.

mberry593 Sat, 03/02/2013 - 09:31

I had an API console back in the 1970s with two 525s on it. The biggest problems with them were the combination of the small knobs and small meters made them very difficult to use. I think that I still have a copy of the schematic---I'll look for it later today.
Please understand that these are not parallel compressors. I don't want to be discouraging. If you want to try to replicate a 525, that's a good project & I am happy to help. If you want to experiment with parallel compression (I have), that's also a good project & I do have some experience that may be helpful. But they are separate projects.

JohnTodd Sat, 03/02/2013 - 10:43

I see that they are not parallel compressors - but to mimic the different response to lo vs upper frequencies I propose to use parallel to process the two areas differently. Really, the 2 compressors are set the same except for the release time on the lows.

Im experimenting in session right now. More l8r.

JohnTodd Sat, 03/02/2013 - 11:01

First round

OK, I've done the first round.
That link should get you into the Dropbox folder where the files are.

Look at the jpgs to see the settings.

The "dry" file is the reference for all. I used Addictive Drums and made a rudimentary beat. I then tweaked AD to be as dry and lifeless as possible, ie, no compression in AD, not even Room mics. Just individual EQs left at the default setting for each channel.

Then I set up the compressors using a 2:1 ratio and then an 8:1, which is teh highest this plugin will go.

Each compressor track has an EQ on it for HP or LP. On the screengrabs you can see which is which.

Compare the processed tracks to the dry track. I did nothing else to the tracks, didn't even normalize them.


PS. It'll take a few minutes for them to all upload into Dropbox.

mberry593 Sat, 03/02/2013 - 13:21

I can't find my schematic but that's ok as I found one on the internet. Here is a link.

Note that the gain control is fet Q1 (just under A1). "feedback" in this use just means that the output drives the gain control. Almost all dynamics devices work that way including LA-3, 1176, BA-6, Audimax, etc. The only thing that I can think of that was not strictly feedback was the old Volumax that sampled in the middle.

The DS is just about exactly what you suggested. The switch inserts r47, C11, R45, & C10 into A3's inverting (180 degree) input.

You can see all of the release time determination being done in the lower right of the drawing. I won't try to figure this all out---that would make my hair hurt!

API states that the attack time was 15 us. I really doubt that. I believe it was much slower....something like 200.

Now, on to parallel compression.....

I have fooled around with this a little. First, you need to get something like the Waves In Phase to see what is happening.

[=""]InPhase - Phase Alignment Plugin | Waves[/]="…"]InPhase - Phase Alignment Plugin | Waves[/]

You are about to experience what really frustrates loudspeaker crossover designers. In order to eliminate a messy disturbance at the crossover, you often will end up with the high and low bands out of phase with each other. Yes, you can fix this with the all-pass filters but it isn't easy. Be prepared to spend a great deal of time adjusting them. I started by sending noise into the equalizers and looking at the combination on a spectrum analyzer. (NOTE: You MUST have a spectrum analyzer to do this but if you don't have one, don't worry, Blue Cat has an excellent one that you can download free! [[url=http://="http://www.bluecata…"]Blue Cat's FreqAnalyst - Real Time Spectrum Analysis Plug-in (VST, Audio Unit, RTAS, AAX, DX) (Freeware)[/]="http://www.bluecata…"]Blue Cat's FreqAnalyst - Real Time Spectrum Analysis Plug-in (VST, Audio Unit, RTAS, AAX, DX) (Freeware)[/] Yes, it will do VST.)

If you really don't want to delve into all of this geek stuff, you need linear phase equalizers.....I don't know Cubase...maybe what you already are using is linear phase. [[url=http://[/URL]="http://theproaudiof…"]Linear Phase vs. Minimum Phase EQ[/]="http://theproaudiof…"]Linear Phase vs. Minimum Phase EQ[/]

OK, I'm trying hard not to be discouraging here because I like what you are doing. Please understand that IMO, you aren't just trying to emulate a 525C here, you are on a path to make something better! It is going to be really tough to model those program-dependent release times. Although the parallel approach isn't really a pure emulation approach, I can't think of any other way to do this. So, I think you are on the right track. Just be prepared for some work!

Good luck!

audiokid Sat, 03/02/2013 - 17:08

This aught to be interesting. I'll start off saying, listen to the music on the radio. That aught to be the first clue.
And people summing OTB on budget summing systems in pro tools doing the round trip are about the worst it gets. Problem people have is, they don't know its happening because it is one bit at a time. Mixerman calls it the curse. I like that term. It means a lot of things to me.

I have know idea how the numbers prove or disprove all scientifically but I hear how "in phase" my analog rig is compared to the majority of mixes I hear. I think we are all getting accustomed to that phasy sound.

mberry593 Sat, 03/02/2013 - 20:14

JohnTodd, post: 401612 wrote: How do things get out of phase ITB? ----------
Or am I not understanding the phasing issue? When speaking of phasing, are we talking about the waveform crests and troughs not lining up properly? are understanding & it is about the crests and troughs not lining up.

In almost all cases when we adjust sound with an equalizer, it not only changes the frequency response but also the phase response. This is true for ITB digital, hardware analog, and even natural acoustic filters like listening through a cardboard tube. In most cases, this is not a problem at fact it is a very natural effect that the ear expects when the frequency response is changed.

It is possible to build devices called all-pass filters to compensate for this. It is usually considered to be economically unreasonable to do this for analog hardware although it has been done....I believe Weiss & Cedar make them. It is very doable for digital ITB. But, it is rarely done because there isn't much call for it. It is usually perfectly acceptable to have phase shift when you isn't a problem.

It is, however a big problem when you try to do parallel compression. The idea here is to take apart the spectrum and later put it back together without altering it in any way except for the compression. You are not really trying to equalize. Forget the compressors for a moment. When you took a high-pass and a low-pass and then summed them back together again, you could jockey things around with noise on the spectrum analyzer until you got the frequency response to be flat, but you would find when listening to the result that it sounded different. That is because the two bands that you created were offset in phase. You can correct for this. The Waves plug-in that I linked isn't just a measurement device. It also includes some all-pass filters that you can insert to correct just the kind of thing that you are doing. But, rather than working to correct it, why not just avoid the entire issue from the start by using linear phase equalizers to do the splitting. Not only will you save yourself a great amount of work doing phase correction but you will also find that the frequency response summing is significantly easier to do.

I work with RTAS & AAX plug-ins so I don't know what is available with VST, but a quick google for linear phase equalizers should give you some ideas.

Again, good luck! I wish you well with your project. Even if you don't achieve an acceptable 525 emulation, you may well end up with an excellent parallel compressor which is well-worth doing.

mberry593 Sat, 03/02/2013 - 20:25


At the risk of boring you to death, this is the heart of why we need to compensate op-amps for stability. All op-amps have an eventual high frequency rolloff. This is accompanied by an inevitable phase shift. If the phase rotates around enough to make negative feedback turn positive before the gain crosses unity, the circuit will become unstable. That's why you see those 22 pf capacitors in the feedback loops of 5534s. The idea is to get the high freq gain down below unity before negative feedback becomes positive due to the phase shift. You might think, why stop at 22, why not just limit it to the audio band. The reason is that the op-amp has to quickly charge & discharge the compensation capacitor and if you go too far, slew rate will suffer.


RemyRAD Sun, 03/03/2013 - 05:05

I might point out that certain plug-ins, for whatever reasons, sometimes invert phase. The BBE, Sonic Maximizer plug-in is a good example of a phase inverting processor. Not sure what the thought process was behind that? So when I use that plug-in, I frequently then have to invert phase, as well. Otherwise, I started to get some pretty peculiar phase cancellation when that track is combined with all of the other tracks from a live event recording. So I really have to be careful about the use of some of these plug-ins that happen to invert phase. And you might be experiencing something similar? So when ya try to do some kind of spectral dynamic processing, everything is going to go screwy. It might force you to process the other channels with the actual processing bypassed? Or ya invert that processed track to play well, with the others. So you're not doing anything wrong necessarily it's the software doing it. I just keep trying to figure out the reasons why they're doing it?

You just can't tell with some of this software?
Mx. Remy Ann David

mberry593 Sun, 03/03/2013 - 10:07

Now that you have the parallel thing conquered, it is time to take on the most difficult part. I spent awhile googling around and looking for compressors that allow you to play with not just the release time but also the shape of the release time. I was overwhelmed by the sheer number of devices available but very few allowed for any customization of the control. I did find one.


In addition to a release time control, this has a 'RMS length' control. That might be the kind of thing that is called for here. You can play with this as it does do VST but I can't as it won't do RTAS or AAX.

This really needs even more research. I sort of remember seeing something else that was adjustable, but I can't remember. I'll keep looking.

.............anyone else know of anything with tweakble release beyond just time?


User login