My First Compressor/Limiter

ruffrecords · May 25, 2017

I found this great web site that has archived audio and radio magazines from decades ago. It includes Studio Sound from its inception (previously Tape Recorder and now Sound On Sound). I found the original article of the very first compressor/limiter I ever built in the July 1972 edition of Studio Sound. I actually built this and it did work. |I later incorprated it into a DJ mixer for a friend for use as a Ducker. Here is a link to a pdf I made of the article.

https://drive.google.com/open?id=0B_n67A1hN3qtdmsyR3JXcUhqRHc

Cheers

Ian

chrispbass · May 26, 2017

Any plans for a compressor project then Ian?

Best
Chris

ruffrecords · May 26, 2017

ChrisPbass said:
Any plans for a compressor project then Ian?
Best
Chris

I must admit that the motivation for finding the article was the current problems with the continued supply of Vactrols. I have pretty much abandoned the idea of a tube based optical compressor but a FET based tube compressor seems quite an enticing idea.

Cheers

Ian

chrispbass · May 26, 2017

I've only just recently learnt of the problems reg vactrols As I've been away from building gear for a few years.
Fortunately I have a small stash to finish off my projects but obviously a major pain for designers, hobbyists and the like.
How about a valve flavoured 1176 style?

ruffrecords · May 26, 2017

ChrisPbass said:
I've only just recently learnt of the problems reg vactrols As I've been away from building gear for a few years.
Fortunately I have a small stash to finish off my projects but obviously a major pain for designers, hobbyists and the like.
How about a valve flavoured 1176 style?

A possibility. The 1176 uses a FET and driving FETs with tubes is much easier than driving LEDs. The problem is the complexity of the sidechain. Easy to do when you have lots of cheap transistors to play with. With tubes you need to be much more innovative. I have studied the 1176 quite a bit. It looks like they adjust the operating point on the FET curve to change the ratio which is exactly what I would expect. However this alters the base attenuation of the FET (the attenuation when no signal is present which alters the loop gain which in turn alters the threshold. The 1176 seems to compensate for this by changing the gain in the sidechain,

The basics are not too complex. It is getting it to work with real FETs that is the hard part. And then there is the whole question of stereo linking and tracking.

Cheers

Ian

ruffrecords · May 28, 2017

You got me thinking. Here is a first pass three tube 1176.

Cheers

Ian

Michael Tibes · May 29, 2017

I find this very exciting! I just want to add that I wouldn't be too focused on using tubes in the sidechain unless there is a sonic reason for it. If it makes life easier and sounds the same? Some great sounding compressors have solid state sidechains...

Cheers,

Michael

pvision · May 29, 2017

ruffrecords said:
And then there is the whole question of stereo linking and tracking.

Just swerve that question completely by using M/S like Joe Meek compressors

Nick Froome

chrispbass · May 29, 2017

ruffrecords said:
You got me thinking. Here is a first pass three tube 1176.

Cheers

Ian

That was quick Ian

Perhaps getting a mono unit up and working would be the way to go initially as I'm sure many would not be too concerned with 'stereo tracking/matching'.
Best
Chris

ruffrecords · May 29, 2017

Stereo linking is always a problem; well the linking isn't but the tracking is. I just ordered 100 N channel FETs to do some characterisation.

Cheers

Ian

chrispbass · May 30, 2017

ruffrecords said:
Stereo linking is always a problem; well the linking isn't but the tracking is. I just ordered 100 N channel FETs to do some characterisation.

Cheers

Ian

Great stuff Ian, looking forward to the next installment

Best
Chris

PS. There are several revisions of the 1176, is one more preferable to another for 'modding'?

ruffrecords · May 31, 2017

ChrisPbass said:
Great stuff Ian, looking forward to the next installment

Best
Chris

PS. There are several revisions of the 1176, is one more preferable to another for 'modding'?

As far as I can see the revisions are mainly incremental improvements, initially to the FET drive, then the meter circuit and lastly to the input interface. The latter revisions should definitely perform better.

Cheers

Ian

Joechris · Jun 1, 2017

Could the sidechain be taken of R11? So the threshold is not affected by the output level?

Joe

ruffrecords · Jun 2, 2017

Joechris said:
Could the sidechain be taken of R11? So the threshold is not affected by the output level?

Joe

Yes. I thought about that when I drafted the schematic and I nearly changed it. The only downside is you would have to use a large value threshold pot so as not to load the FET which will indirectly affect the shape of the attenuation curve. A 1Megohm pot would do it but I am a little nervous about running high impedance signals to and from a front panel control. It would be a good idea to lay out the PCB to minimise that.

I have also looked at the FET500 version of the 1176 and I notice they have replaced the push button ratio selection with a rotary switch. I have not looked at in detail but I guess you could include a position equivalent to the all buttons pressed mode of the original 1176.

Cheers

Ian

chrispbass · Jun 2, 2017

ruffrecords said:
Yes. I thought about that when I drafted the schematic and I nearly changed it. The only downside is you would have to use a large value threshold pot so as not to load the FET which will indirectly affect the shape of the attenuation curve. A 1Megohm pot would do it but I am a little nervous about running high impedance signals to and from a front panel control. It would be a good idea to lay out the PCB to minimise that.

I have also looked at the FET500 version of the 1176 and I notice they have replaced the push button ratio selection with a rotary switch. I have not looked at in detail but I guess you could include a position equivalent to the all buttons pressed mode of the original 1176.

Cheers

Ian

Good idea Ian, was thinking about a rotary switch for the ratios too. Makes it easier to construct and saves a few pennies as well...excellent!

Best
Chris

PS, any thoughts regarding the iron at this stage? I recall an oep being used in one version or is it best to stay close to the original? I'm not sure how much of the character of the 1176 is influenced by the tx, although obviously the circuit is going a bit 'off piste' now anyway

ruffrecords · Jun 2, 2017

I have been looking more at the ohmic region of a FET. Vishay has a useful paper on the subject. They show the attached graph of normalised rDS vs VGS:

The normal bias point is at VGS = 1 when rDS is a maximum. Reducing VGS reduces the resistance and hence the gain (when used in a pot divider).

The ratio you get depends on the slope of this curve. As VGS varies from 1.0 to 0.8, rDS varies over two decades but as VGS varies from 0.8 to 0.0, rDS varies by a single decade.

The actual attenuation you get modifies this even further. I need to get my head around exactly how attenuation varies with VGS

Cheers

Ian

Michael Tibes · Jun 2, 2017

ruffrecords said:
...I guess you could include a position equivalent to the all buttons pressed mode of the original 1176.

There is a detailed description for that on mnats 1176 page. I vaguely remember that 4:1 and 20:1 need to be connected in order to achieve the 'all button in - effect'...

Michael

ruffrecords · Jun 2, 2017

ChrisPbass said:
PS, any thoughts regarding the iron at this stage? I recall an oep being used in one version or is it best to stay close to the original? I'm not sure how much of the character of the 1176 is influenced by the tx, although obviously the circuit is going a bit 'off piste' now anyway

The main audio path at present is based on one half of my twin line amp design so for output iron you can use any 600:600 transformer provided the load is 10K bridging. For a true 600 ohm load drive it needs a 2:1 transformer for which I normally use a Carnhill VTB2291 or an Edcor XSM2.4K/600. For the inout, any good quality 10K:10K transformer will do.

Cheers

Ian

ruffrecords · Jun 4, 2017

I have been looking at loads of schematics of FET compressors from stomp boxes to gain brains. As a result I have made a few tweaks to the original schematic and filled in a few component values where I am reasonably confident of the.

I have moved the side chain take off point to just after the FET so the output stage is now an independent gain make up amplifier. I have looked at lots of schemes for setting ratio and biasing the FET and combined a couple in away that I hope is fairly straightforward to understand and operate. For biasing I have a simple dc voltage i series with the source of the FET but decoupled at ac so it does not affect the gain..As a rule, FeETs are biased so they are only just starting to reduce gain with no signal in the side chain - in the 1176 the bias is set for a 1dB drop in gain.

Ratio generally appears to be altered by moving the bias so the FET works on a different part of its curve so that its dB per volt changes. The problem with this method is it relies on knowing the FET characteristics which means selecting FETs. I have seen another method may be less dependent on FET characteristics. It works by altering the total gain reduction the FET can make. In the schematic attached this is achieved by inserting a 1K variable resistor in the source of the FET and using a 10K series resistor in the top of the voltage divider. With the resistor set to 1K, the total gain reduction from off bias to fully on bias is 20dB. This represents the lowest ratio. If the bias ranges from -4V to 0V this represents 5dB/volt but the point is the FET works over its entire range to produce 20dB of gain reduction. If the resistor is changed to 100 ohms, the FET can provide up to 40dB of attenuation in the same voltage range so it produces 10dB/volt gain reduction. You might think this leads only to a doubling in the ratio but it also means the FET only works in the first 20dB of its range for the same 20dB gain reduction. In this region the slop is much steeper to the actual ration will be a lot higher. Right now I do ot know exactly what the relationship is between this resistor value and the actual ratio.

I do not plan to incorporate the complex ratio and threshold switching in the 1176 as I have no intention of making a tube clone of it.

Cheers

Ian

ruffrecords · Jun 8, 2017

I have been running some sims of the side chain and FET circuits from which some of the 1176 design decisions begin to make sense. First, they recommend you start with the input control at the half way point. As this is a log control it typically gives you 20dB attenuation at the mid point. This means that a 0dBu signal is presented to the FET as about 80mV and even a +4dBu input is only 123mV. This is important because the FET will distort badly above these levels. In the 1176, this FET output is followed by an amp with about 26dB gain which feeds the side chain. This means a 35mV signal from the FET is enough to start the diodes conducting and produce gain reduction. If the biggest signal the FET is allowed to see is 100mV then this will be 2V at the output of the 26dB amp and the side chain will be producing well over a volt of gain reduction control voltage. With nearly all FETs biased close to pinch off, a one volt change in bias will produce as much as 40dB gain reduction. Bottom line is there is no way this compressor will let the FET signal get beyond 100mV (probably a lot less).This is probably also the reason there is no separate threshold control in the side chain of the 1176; instead the input drive control is the threshold. The threshold is changed with the ratio because with a higher ratio you get more gain reduction so you can allow the threshold to be higher; but for the lowest ratio the threshold is little more than 35mV across the FET.

So, I have abetter understanding of what is going on. It is clear that all the tweaking is going to happen in the FET biasing and side chain gain areas so for now I plan to put these on a separate front panel mounted PCB along with their associated controls. With this in mind, I have modified a TwinLineAmp PCB layout to incorporate the sidechain amplifier and full wave rectifier which seems to simulate OK. I have ordered some prototype PCBs.

Cheers

ian

My First Compressor/Limiter

Help Support GroupDIY Audio Forum:

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Attachments

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Attachments

Well-known member

Well-known member

Well-known member

Attachments

Well-known member

Similar threads

Join the conversation!

Join today and get all the highlights of this community direct to your inbox. It's FREE!