Official Forssell Limiter Build thread

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Yes, try soldering direct to the PCB,  I like to hard wire my PS wires even though it's a pain when servicing or trouble shooting.

What I experienced was with the link on it would be metering lots of compression  on one side at fire up with no audio present. Very strange?
 
Well, so far no joy on the noise frontier... I replaced R6 with a 10K Bourns conductive plastic log pot (91 series), but everytime I turn it down I get buzz building up. At full volume it's just hiss (with no input signal present), and a little bit of transformer hum.

I can't figure out what's going on, as attenuating the input should have no effect on the noise...

Maybe I'll have a chance to look at it tomorrow...

Cheers

Kris
 
I'm now starting to wonder if I may have damaged my 992 a 2604 in the servo loop...I accidentally had phantom power on when I connected it to my mixer for a listen test.

Grrr.....

Gonna try operating it without the output servo and see if it gets better....I hope so, 'cuz I'd rather that than to toss a couple 992's in the garbage!

Cheers

K
 
I'm not really getting much further on the noise front, but I'm about to try connecting it to my Alphalink A/D instead of the cheezy Behringer mixer I've been using for testing.

Turns out my op-amps are fine, but my soldering  wasn't. I had to drill the holes in the PCB a little bigger, so the 990 ones are not thru-hole anymore, which means soldering from the top side is required. That mistake resulted in my V+ connection to the output stage being 'open'. That put a pretty big DC offset on the output.

Speaking of DC offset, I get about 6mV of DC at the output of the input buffer amp. I've noticed that it's not AC coupled (appears to be a design oversight), and this appears to be resulting in some 'whoosh' sounds as I turn the makeup gain pot. Anyone else encounter this issue? Or maybe there's more problems with my soldering!

Cheers

Kris
 
Okay, I connected it to my Alphalink, powered it up via a bench supply, and voila...silence! My meters show noise down at -110 dBFS, so pretty much as good as it can get.

I can't believe I wasted 2 weeks chasing my tail on this...when it was my crappy test mixer that was the source of the problems.

I still need to try it with my own PSup, but I'm encouraged by the results thus far....

Cheers

Kris
 
Interested in this one.

Are you happy with the ones you've buildt?
What are you using it on?
Sidechain highpass-filter is usually good for a stereo-unit. How is this with the Forssell?
 
Looking at the schemo and downloading a pcb-designer,
but I really don't have the head for it. lol
 
Well, I finished my PSup...went overboard, and pre-regulated down to +/-28VDC and then into an CRC network prior to the on-card regulators. With my supply I get -109 dBFS at min gain, and -107 dBFS at max. That will work for me!

I don't think the link is really necessary. Opto isn't really what I would go for when a controlled stereo image is the desired effect.

There's no sidechain filter on the schematic, but would be pretty straightforward to add it, and IIRC there is an unused half of a 2604 available.

To understand the schematic you need to first identify the parts:
1) Sidechain
2) Input buffer
3) Output amp & servo
4) Meter
5) PSup

Start by just looking at the sidechain and come to grips with how it works. Use Fred Forssell's explanation as a guide: http://groupdiy.com/index.php?topic=16016.msg53932#msg53932

Then, draw the connections between the input buffer, audio path opto attenuator, and the output buffer. You should notice at this point that there is no makeup gain control. So you need to make a decision, do you want a fixed gain output stage and an attenuator control (log pot) feeding the output amp, or do you want to vary the gain of the output stage. I opted for the former as it keeps the bandwidth of the op-amp constant and doesn't change the authority limits of the servo. The latter will have less noise, and might be  a little easier to implement.

At this point in time you should have a good grip on the most important parts of the schematic. After that, look at the connection of the input buffer to the opto & sidecahin.

Finally, look at the meter circuit. There's lots of options there. I went for a VU because that makes me feel warm and fuzzy for an opto comp. The circuit for a VU is simpler, but hasn't been drawn out in this thread. If you go this route I can let you know what I did.

Here are the things I would change (or design in from the start):
1) Relay bypass. As designed right now the compressor makeup gain is active all the time, which makes A/B comparison between In and bypass harder.
2) Jumpers after the optos allowing them to be more easily calibrated (e.g. drive them with a constant tone and vary the balance pot while monitoring the resistance of each opto). Otherwise its a bit of a circular process to calibrate the meters
3) Either add a coupling cap after the input buffer, or wrap a DC servo around it (my preference, then you can get rid of all the audio path caps). I've got a wee bit of DC offset on the input amp and that results in a small 'woosh' when changing the output gain.
4) Keep the audio path short. Input buffer, audio opto, Output buffer. You should be able to fit this into about 3 inches of PCB space.
5) Consider socketing the resistor feeding the ratio pot...select it for maximum slope. Most folks just put in 100 ohms and were done with it, but at extreme settings the optos can produce negative ratios (i.e. the output actually gets quieter when the input gets louder).

Cheers

Kris
 
Thank you Kris!
I might have an offer for a stuffed working single channel.

It would be quite an undertaking for me to put up a pcb-design by myself :)
 
What is the power requirements for one board?
and
Will a smaller pwrTransformer give less noise than a bigger one?
 
The board needs 100ma at around +/-28V (feeding the regulators) for the audio circuits, and another 100ma from the +ve supply for the LED's if you are using a bar graph meter.

That's 8.4 VA for one channel. But I'd go bigger than that...probably 20 for a single channel, and 40 for two.

I used an 80VA transformer in my build, and it was overkill...

Cheers

Kris
 
DrFrankencopter said:
The board needs 100ma at around +/-28V (feeding the regulators) for the audio circuits, and another 100ma from the +ve supply for the LED's if you are using a bar graph meter.

That's 8.4 VA for one channel. But I'd go bigger than that...probably 20 for a single channel, and 40 for two.

I used an 80VA transformer in my build, and it was overkill...
Thanks a lot!
I'll need +/-15to18v for my ICs. Isn't +/-28 too much in then?
 
The boards I used have +/-24V linear regulators on them (LT1033 and LT1085), and they need at least 1.5V  more that feeding them.

What are you using as IC's? The original design was built around JFET 990's and OPA2604's for the sidechain and servos. This was to accomodate a +/-24V power supply.

If you're going with lower voltages on the rails you'll need to change the resistors in the regulator section, but you'll have a few more options for IC's.

Cheers

Kris
 
DrFrankencopter said:
The boards I used have +/-24V linear regulators on them (LT1033 and LT1085), and they need at least 1.5V  more that feeding them.

What are you using as IC's? The original design was built around JFET 990's and OPA2604's for the sidechain and servos. This was to accomodate a +/-24V power supply.

If you're going with lower voltages on the rails you'll need to change the resistors in the regulator section, but you'll have a few more options for IC's.

Cheers

Kris
Thanks!

I might be mistaken.
I have (will get) one board with OPA2604 and OPA604

They're 4,5V to 24V it seems. (my bad)

So, anything in that range, or is aiming for 24V better when following the bom?

Yes, leds.

Thinking of:
http://no.farnell.com/multicomp/mcfe015-25/transformer-15va-2-x-25v/dp/9531696
http://no.farnell.com/webapp/wcs/stores/servlet/ProductDisplay?catalogId=15001&urlLangId=47&langId=47&partNumber=1675054&storeId=10169
But, if it's better I'll go up to 30VA
Is the encapsulated one preferred?
 
Is this the right led-labels for the GR-meter?
 

Attachments

  • LM3916.png
    LM3916.png
    41.3 KB · Views: 25
And, a related question:
Will I not need a power-led then (as the gr-meter leds shows on state)?
 
G-Sun said:
Is this the right led-labels for the GR-meter?

No, that is an audio level meter.

The GR meter works backwards from 0 reduction (with no led lit) in 3dB steps of level reduction metering.  ie; 4 led's lit equals 12 dB level reduction .  My notes for this project  say 3 dB per segment using LM3916.


Labeling would look like  -30  ............. -15    -12  -9  -6  -3      ( 10 led's total)
 
hitchhiker said:
No, that is an audio level meter.

The GR meter works backwards from 0 reduction (with no led lit) in 3dB steps of level reduction metering.  ie; 4 led's lit equals 12 dB level reduction .  My notes for this project  say 3 dB per segment using LM3916.


Labeling would look like  -30  ............. -15    -12  -9  -6  -3      ( 10 led's total)
Thank you!
I'm designing a frontpanel, and would like to get this right :)
Couldn't find the gain-reduction application in the tech-paper.
So, I'll go for this labeling. Thanks!
 

Latest posts

Back
Top