Help me MOD - D&R FET STEREO limiter

[altfidelity]

Hey everyone,

I'm looking to tweak the compression ratio in the D&R FET limiter and could use some guidance. Would really appreciate any help offered. If I'm getting the circuit right, would adding an extra dual-pot next to the R60 and R30 resistor to ground do the trick?

Also, I've been curious - is it possible that this sidechain is in a feed-forward configuration? Normally, FET circuits tend to be feed-back right?

Thanks in advance for any insights or advice you can share!

 

[gyraf]

In general, it's easiest to think of ratio as after-rectifier or DC-control-voltage gain. Changing the R27/R57 (10K, setting DC amplifier gain) is probably a good bet. Perhaps try a 2x100K Lin pot here?

Yeah, fet-comps are usually FB, but I don't see any problem in changing to FF (other than for calculating exact ratios, which we don't really care about, do we?) - just take your sidechain after C1/C15 - or add another pot in parallel with the input level pot and name this one "Threshold".

I think a 10K log pot would make more sense as input potentiometer(s) - the 100K seems to be begging for unneeded noise

/Jakob E.

 

[altfidelity]

Changing the R27/R57 (10K, setting DC amplifier gain) is probably a good bet. Perhaps try a 2x100K Lin pot here?

Great, thanks Jakob!

10K log pot would make more sense as input potentiometer

Noted.


I will report back as soon as I start tinkering with it.

 

[tommia]

R18/r38 change to a switch with lower values (lowest 330 ohm) . You will get 6db/oct hp filtering on side chain

 

[altfidelity]

You will get 6db/oct hp filtering on side chain

SC hp, that's also a good one.
Just curious, how did you come up with limit 330ohm value?

 

[gyraf]

Lowering R18/R38 from 10K to something lower, to form a HPF with the existing C10/C14 (2u2) electrolytics might be not a optimal, as this would also load down the sidechain input impedance significantly: As-is, in fb mode, this part is driven from the audio amplifier 5532, which will drive down to 1K or so but not much less than that in a good way.

Resistance Frequency Capacitance Calculator gives you, 2u2 and 1K, some 72Hz hpf - which may or may not be high enough for your purpose

..worse, when this circuit set to FF mode, these sidechain inputs would significantly load down the input of the circuit, requiring a lot of juice from the preceding stage

It would probably better to keep the 10K for R18/R38 to avoid this relatively heavy loading, and then switch C10/C14 value for sidechain HPF functionality: A 220nF would give you about 72Hz, a 100nF would be a quite usable 160Hz (re-using above calculator). A three-pos toggle with center-off should be able to handle these three options..

/Jakob E.

 

[gyraf]

..recalculating: the input impedance of sidechain is actually 10K(Rin)//10K(to inv.input)//20K (to inv. input 2), total of 4K Ohms

Now 470nF gives you 85Hz, 220nF gives you 180Hz hpf

/Jakob E.

 

[altfidelity]

Changing the R27/R57 (10K, setting DC amplifier gain) is probably a good bet. Perhaps try a 2x100K Lin pot here?

I'm finally picking up 100k dual gang pots in the afternoon (delivery service is sooo slow around Xmas).

2 semi-dumb questions:

- is the correlation between resistance values R27/R57 and effective ratio linear? i.e. halving/doubling resistance is halving/doubling ratio?
- should I put same resistors after the pots? is it OK to have there no resistance at all, when the pots will be turned all the way to min?

 

[Hubbub]

Just a few observations. Remember, you often get what you pay for, so ignore me if it suits you.

If you are looking to change the compression ratio the goal is to change the sidechain gain between the audio path and the gain reduction FETs. More gain = more compression, less gain = softer compression. If you said your goal was one or the other I missed it, sorry. But if you goal is both more and less, I have a suggestion.

I see only one spot where more and less can be done somewhat cleanly by replacing a fixed resistor with a pot that goes from more resistance (higher sidechain gain / compression) to less resistance (lower sidechain gain / compression). The spot in question is R11(R43). Even that is a bit iffy as any change to R11(R43) should be balanced with a matching increase in R61(R67) to keep the zero signal level of A4(A6) at the 2.2V point. Any DC shift here effects a change in the point of compression start. Just modifying R11(R43) might provide an acceptable change in compression start point with compression ratio, you might want to give it a try. Maybe use a 68K resistor in series with your 100K pot for both more and less compression. I'd also consider increasing R61(R67) by a fixed value of 390K or 430K so at max compression the output of A4(A6) rests at 2.2V with no audio in.

Changing R27(R57) will also affect sidechain gain, but it also shifts the DC bias on the JFETS unless R29(R59) also tracks the change. Tweaking only R27(R57) changes the onset of compression as well as the compression ratio, the results may not be good since this mucks directly with JFET bias. Here I would try maybe a 3.3K resistor in series with a 10K pot for both more and less compression.

Additionally yes, you can replace R18(R38) with a pot, but it only gives less compression, not more. If that's the goal it can work. But the pot wiper would see a 6.66K Ohm load, so using a 100K pot may prove unsatisfactory with a less than optimal working range on the control. 50K or 25K might work better. And all the previous comments about the HPF corner changing with pot position apply, so frequency sensitivity interaction happens for this approach.

Additionally you can make multiple changes starting with R18(R38) becoming a pot. Use that for the variability, then boost the max compression with R11(R43) being raised but bias corrected via R61(R67) being raised by the same percentage. Lots of options here.

Interesting project, best of luck with it!

 

[richiyobs]

Hi folks
What about adding a balanced input with variable gain?
So if more compression Is needed you can drive the jfet... Also this way the jfet should see the same impedance and not a variable one
I got some simulation and noticed the release Is not varying so much... It s a long constant so maybe It s ok not seeing variations...
At test point at thr amp i got 2.4v with 2meg at v- .... At ratio amp i m using 3.6meg and here CV out Is from -2 up to 0.4
As Hubbub said altering thr or ratio amps values the bias has to be adjusted cause of strage behavior and strange wave out....
Setting the test points in the right manner the wave out Is well ducked and the First transient very round smoothed respect to other jfet compressor where the initial transient Is usually left passing and very High in level... This Is what i ve seen with 1khz @ 4dBu

Interesting design for me

Best

 

[RuudNL]

Funny, this is a very old design by D&R, made in The Netherlands. (The country where I live.)
Even in The Netherlands, D&R is often referred to as 'Dutch Rubbish'...

 

[altfidelity]

Funny, this is a very old design by D&R, made in The Netherlands. (The country where I live.)
Even in The Netherlands, D&R is often referred to as 'Dutch Rubbish'...

Its NOT rubbish ... but its not masterpiece either.
I've bought compressors triple the price, that have been more disappointing. For what I've paid, it's a steal.

It's dead silent, it doesn't losses the top or low end, and is fast and effective at being an limiter.
The knee on a limiter is RAZOR sharp, it goes from zero reduction to absolute limiting in a moment.
And it doesn't "pump" audibly like some vca limiters, it cuts(saturates?) those transients right off in a pleasant grainy/tape-ish way.

There is not a lot of tone/fatness/sweetening going on though, I give you that.

I will post some pictures & measurements in the evening, I haven't got as much time as I hoped for this weekend.

On the mods side of things, I will replaced pots, and adding power switch, but mostly I wonder how would it sound as a compressor with lower ratios. Also maybe adding (GSSL) style gain reduction ammeter, would that work?

There is also something slightly wrong on one channel, the ratio goes negative ratio for a couple of moments. (hand-drawn sketch below)

What about adding a balanced input with variable gain?

The pot labelled "threshold" actually increases output input already.

but it only gives less compression, not more. If that's the goal it can work.

Yup.

I got some simulation and noticed the release Is not varying so much

It is, the difference is audible, the release can actually be quite long.

 

[richiyobs]

attached simulation i was running....
psu is dual 16 cause i d like to build it in 500 format one day....
@altfidelity could you please upload any sound samples, i m just very curious about its performance

also attached an old article, 1947, about peak limiting amplifier
you can see a similar behavior and well explained pictures

thanks for your attention

best

 

[altfidelity]

would you please upload any sound samples, i m just very curious about its performance

Second half is wet, fastest attack, fastest release. Both section are normalised to absolute peak value.

 

[richiyobs]

Second half is wet, fastest attack, fastest release. Both section are normalised to absolute peak value.

i get error code....it can t be listened
believe the mp3 has to be zipped to be downloaded

 

[altfidelity]

Sound examples here groupdiy - Google Drive
Second half is wet, both section are normalised to absolute peak value. File "fast" is a bit over-compressed, file "medium" is a bit more moderately compressed. If you want more examples, send me "dry" files.


Ratio & knee measurement, some envelope shape measurements, three different attack times (slow, med, fast) and release shapes.

 

[richiyobs]

Cool ... Fast Attack the First transient not overshoot in the positive part

Thanks for sharing sounds and shapes

Best

 

[altfidelity]

Yeah, but you can see that one channel is slightly faulty (possibly just mismatched) see attached picture. Purple is left, blue is right channel.
Could your sim predict values for R27/R57 , for lower ratio (2/4/8/16/limiter(original))?

i d like to build it in 500 format one day....

It would be perfect fit, very little parts and even in original state is probably physically small enough to fit in 500 chassis.
Its half rack width, but the pcb is even smaller.

 

[richiyobs]

The channel with more ratio could be fixed adjusting the CV trim at ratio amp...
In the schemo there s a note to mach channels

 

[richiyobs]

Yeah, but you can see that one channel is slightly faulty (possibly just mismatched) see attached picture. Purple is left, blue is right channel.
Could your sim predict values for R27/R57 , for lower ratio (2/4/8/16/limiter(original))?

It would be perfect fit, very little parts and even in original state is probably physically small enough to fit in 500 chassis.
Its half rack width, but the pcb is even smaller.

Yes i altered R27 up to 100k... And ratio changes.... But Is not 1:1 a bit of tension Is there and so signal affected

I m drawing a version with balanced input with gain ... And some mods to Attack and release net
A different behavior ... Now release Is faster and seems produce beefy tails
Also added autorelease...
I got inspired from a vca compressor for this approach