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.
etheory said:
This is what most plug-in developers do, and it really annoys me.
It's simply not enough.

To get the proper operation, you need to model the rectifier, since that is a part of the threshold (since you lose one diode drop at that point), along with the proper release and attack circuits, then the LED for the opto needs to be properly modelled, since it's the second part of the threshold (another diode drop) and is highly non-linear (it follows a Lambert W curve). Then you need to model the optical element properly (which I've done by measuring the real thing) to get the right gain response.

I still don't see it. Sure, the diode drop establishes the threshold for compression, I totally see that, but does it need a spice model or do you just set a thresh value of 0.7V and get on with it. As for attack and release, they are simple RC circuits, and can be modelled with 1st order low pass transfer functions.

The Vactrol is nonlinear, and is deserving of a more elaborate treatment, but does it need a physical model? What is wrong with simply using the mfg's curves? How well does your model match the mfg data, or your own data?

I'm down with the idea of modelling the functional bits of the circuit, but I don't think SPICE level detail models are necessary for a sidechain circuit. I'd save the detailed modelling for the audio path....

But hey, like you say YMMV....

Cheers

Kris
 
You make some good points DrFrankencopter.

I'm not using a full spice model (well, mathematically it's identical, but practically, much simpler to evaluate), since obviously, for real-time, you need as few operations as possible. What I AM doing, however, is using the same Modified Nodal Analysis that SPICE uses, to get my basic equations.

For instance, to model the two-diode bipolar rectifier used in this side-chain, you need to model two diodes charging a capacitor with a drain resistor for release. This can be modelled in simple ways, or using Newton-Raphson iteration (which is what I am doing) to get a result that matches the SPICE simulator (and my measurements of the real thing) to a precision I doubt I could get better than in software. When you do this you see that there are a bunch of subtle effects that a simpler modelling approach doesn't display. And to me, that's important. It might not be to other people, but really this is, to some degree, an experiment in "is it really worth it?" and "can we really do better". To which, so far, I think the answer is yes.

As for the RC circuits for release and attack, for the attack, yes, you can simply use a trapezoidal-integrated low pass filter model (standard stuff), but the release is the combination of a low pass filter with the rectifier feeding it, which has a far more complex response. In some ways, you could say it's a voltage controlled low pass filter, since the higher the input voltage, the lower the diode resistance. But in order to solve it as the real circuit does, it's a balancing equation (current in = current out), so you need to either iterate, or you can directly solve it using the Lambert W function (which I also do sometimes).

The Opto model I'm using I'm still working on. So far I've modelled the LED directly  by measurement, but the Photoresistor is something I'm still working on. I have a static model, but I am additionally working on a dynamic model (to include the attack and release properties of the optical cell itself).

I'll post some audio once I'm done and you can all decide for yourself if I've been wasting my time ;-) The fact is, I've learned LOADS about circuit analysis by doing this, and it's going to massively help me with my analog designs in the future.

cheers for the comments.
 
Cool...anytime you dig in and try to understand something fully you're bound to learn tons. I applaud your efforts, and perhaps I can learn from your experience.

I'd be interested in hearing how you characterize some of these subtle effects...for example the Diode resistance effect on the release cap. If the diode is forward biased, then current is flowing into, and charging the release cap,  in this case does this effect even matter? Sure, it will affect the charge level slightly...perhaps this is what you're referring to. What is the magnitude of this effect?

Cheers

Kris
 
You guys kill me,  this compressor is not grabby or difficult to use, it dials in easily.  Maybe too much modelling etc is grabbing your review ?  The only drag about this design is that the stereo link has found no success as far as I know.  Maybe build it with the real parts and then post opinions?
 
I didn't model the ratio control (the 5K pot before the opto LED), so it's basically just a very very high ratio (just that 500R resistor there by itself), which explains the insane jumpiness....
Once I put that in and can dial it back, I imagine it won't be so twitchy once it gets past threshold. Hopefully....

But yes, I'll definitely be putting this together for real for reference. I have all the parts for it. Can't wait too!
 
etheory said:
I didn't model the ratio control (the 5K pot before the opto LED), so it's basically just a very very high ratio (just that 500R resistor there by itself), which explains the insane jumpiness....

When I tested the relationship between resistance and ratio using a Rmin of 100 ohms I found that at certain low resistances the compression ratio is beyond infinity to 1 (i.e. as the signal gets louder the output gets quieter).

Stereo linking this comp would be a pain, as the Vactrol's are so variable...unless you selected them for similar characteristics.

Cheers

K
 
Does anybody have a copy of Fred's original VU meter circuit? I've got my front panel all set up for VU metering, but it seems all the schematics I have are LEDs. I suppose I could figure it out, but am hoping someone can save some effort for me.

Cheers

Kris
 
Okay, so it's been 10 years since I started this project, but it's finally alive!

My build:
  • Forssell DLR-1 on the input
  • Forssell JFET 992 on the output
  • VU Meters
  • 2U front panel
  • 12 position rotary switch for makeup gain control
  • Impedance balanced output (50 ohm resistor connecting pin 3 to gnd
  • 6 ratios selectable from a rotary switch (2:1, 4:1, 6:1, 8:1, 10:1)

I haven't completed the meter section yet, as I just wanted to get it passing audio and compressing...finally have success at that. A few screw ups on the way, and not having a complete (or correct) schematic made it a bit of a challenge, but I'm well on my way.

Still a few issues to contend with though:
  • Audible buzz appears to be coming from the LEDs that I have powered off the rectifier (pre regulators)...buzz level drops significantly when I disconnect the meter LEDS. Not sure what's up here.
  • The makeup gain switch causes a massive transient during switching...I'm not sure if this is caused in part by the servo catching up with changes in DC offset, but I might just change it out for a 100k pot and be done with it
  • Bypass leaves the makeup gain active. Maybe not an issue, but not my preference. The bypass also seems a little slow...maybe it takes a while for the release cap to bleed off

Cheers for now...I'll keep this thread posted as I work out the issues.

Kris
 

Attachments

  • Comp1.jpg
    Comp1.jpg
    47.2 KB · Views: 101
Ok, some progress on the LED issue....


I was powering the LEDs from +35V (post rectifier, pre-regulator), and returning to ground. But it seems that the 40 mA of current was upsetting the audio ground. So, to get around this I have powered the LEDs across the +/-35V rails (i.e. returning to the negative supply). This has made a significant improvement.

Cheers

Kris

(edited Dec 31 corrected pre-rectifier to pre-regulator)
 
Hello gents!
I have a pair of Rev 1.b and i am looking for a BOM of it. I have searched around and found that all of the links are dead.  :-[
Does anyone have any info on these old boards?

Thanks in advance!
 
tflanagan2678 said:
Hello gents!
I have a pair of Rev 1.b and i am looking for a BOM of it. I have searched around and found that all of the links are dead.  :-[
Does anyone have any info on these old boards?

Thanks in advance!

I'll take a look at work next week to see if I saved the BOM somewhere...I may have printed it out and put it in my Audio binder there. No promises though!

Cheers

Kris
 
Could you say what changes were made between the V1 and V1.b boards ?
  There was one error with the V1 board that I remember,  IIRC it was the attack pot being backwards marked on  the pcb.  Could have been another control pot though? ( long time now!)

If that's the only difference then  the BOM will be unchanged.
 
I'm pretty sure I was in on the original board order from TommyTones, and my boards say rev 1.b

I'm pretty sure that the ratio pot is backwards if you use PCB mount pots. Also, there is one cap that is missing from the BOM...if you go back through my posts on this thread I point it out somewhere.

I got the meters working yesterday! I took +24V feeding the wiper of a 500k trim pot that was connected across the meter. This was the zero adjust. On one end of the 500k pot connect to the meter LDR. On the other end of the LDR disconnect it from the +24V and tie it to GND. On the other end of the 500k pot connect a 20k resistor to GND. The 20k resistor value needs to be tuned a bit to set the meter sensitivity (tracking), but it seems to work pretty nicely.

I used the header for the LED's as a way to get my connections for makeup gain, and meters off the board.

The attached photo details it...

Cheers

Kris

 

Attachments

  • OptoClose.JPG
    OptoClose.JPG
    140.8 KB · Views: 57
Well, my initial listening impressions are that this is a smooth sounding compressor. I can hear a little bit of high end softening when gain reduction is taking place, and the attack is definitely on the slow side. It's not what you want for transient control, but I think it will do very well at overall levelling. I seem to prefer the 4:1 and 6:1 ratios rather than the higher slopes, but have only really subjected it to pre-mixed/mastered material being sourced from my phone.

I still have to balance the otto's, select the meter sensitivity resistors, and figure out the stereo linking (super low priority, as I doubt I'll ever use it in a stereo application). But it's getting close to being done!

Cheers

Kris
 

Attachments

  • OptoFront.JPG
    OptoFront.JPG
    92.3 KB · Views: 53
Nice build Kris. I'd say your initial assessment  is spot on. Two of my builds use the same DOA's you  have in yours. It'll be interesting to see how you find the stereo linking.  I got some strange results at turn on if the link was left on. 
Cheers, Lance
 
I looked through my files and couldn't find a BOM for the opto comp. Its probably the least documented project I've done...

But I did find a bunch of stuff I wrote about the SSL comp. I think I'll re-post it, and maybe even bread board some of the circuits. I think there were some good ideas in there.

Cheers

Kris
 
I got the link wired up, and strange behaviour at start up indeed. I only tried a couple times, but I managed to get both GR meters indicating 10dB of reduction with no signal present. Interesting....

I managed to get the meters calibrated today, but managed to mess up my power supply connectors in the process. I'm getting 120Hz buzz coming through the molex connector I used to power up to  post rectifier. I think I'm going to abandon the molex in favour of either a soldered connection, or a terminal strip. It seems that messing around with the connector can reduce the noise level, so I'm pretty sure those connections aren't as low impedance as I'd like.

I replaced the 100k stepped makeup gain switches with 10k pots, and the output is nice and smooth now, but I notice that the noise is higher when the pot is attenuating. Not sure what's up...but I think psup and grounding may be playing into it.

Cheers

Kris
 
Back
Top