diod compressor..will this work?..
- Thread starter Johan
- Start date
Help Support GroupDIY Audio Forum:
..ok..last try before I head off to explore that pile of PCC85/UCC85 that just landed on my doorstep..
how about this then?..the pad should be happy becouse the FET's represent a hi Z, the diodes should be happy since the pad resistors can be bigger now feeding the FET's and the dif-amp should be happy seeing lower Z at its input...and part count is still low...
...or what do you think?...
j
how about this then?..the pad should be happy becouse the FET's represent a hi Z, the diodes should be happy since the pad resistors can be bigger now feeding the FET's and the dif-amp should be happy seeing lower Z at its input...and part count is still low...
...or what do you think?...
j
This should work (although I'm not sure it would satisfy PRR) 
There are many ways to do it.
The U-pad is optimum when you need a low impedance output. The diode shunts need a high impedance source. You could U-pad and then add build-out resistors. But when you find three resistors on one node, you should wonder if you are using two stones to kill two birds, if there is a one-stone solution which gets both birds. Not that the $0.25 part-cost will kill you, just for "elegance".
This one has the transformer before the pad. That means the transfo has to handle BIG signal. If you swap them, the tranfo only has to handle little signal. That can save a lot of money/weight.
Most transfos expect a medium/low source. Therefore I would pencil the U-pad (high inpedance input, lower impedance to transfo), a small transfo, two several-K series resistors, and the diodes.
It is also possible the drive the transfo with a "high" impedance. This will reflect to the secondary. The diodes can directly shunt the secondary. An advantage is that the voltage across the winding (which causes transformer distortion) can "never" be more than dozens of mV. You could use an exceptionally cheap and small core. However since transfos are rarely used this way, you could have trouble with "minor parasitics" becoming real problems. If I was gonna build a million, I'd play with the idea. For a one-off, the thought and testing would be excessive (unless you enjoy it).
The FETs will work. But look at "3-amp instrumentation amplifier". This gives two hi-Z differential inputs in 1.5 5532s or 0.75 TL074s. There is a 2-amp variant which is even cuter, if you don't need absurd common-mode tolerance (you don't). You could also lose the coupling-caps, which may more than pay for the added silicon.
But there are many ways to do it.
The U-pad is optimum when you need a low impedance output. The diode shunts need a high impedance source. You could U-pad and then add build-out resistors. But when you find three resistors on one node, you should wonder if you are using two stones to kill two birds, if there is a one-stone solution which gets both birds. Not that the $0.25 part-cost will kill you, just for "elegance".
This one has the transformer before the pad. That means the transfo has to handle BIG signal. If you swap them, the tranfo only has to handle little signal. That can save a lot of money/weight.
Most transfos expect a medium/low source. Therefore I would pencil the U-pad (high inpedance input, lower impedance to transfo), a small transfo, two several-K series resistors, and the diodes.
It is also possible the drive the transfo with a "high" impedance. This will reflect to the secondary. The diodes can directly shunt the secondary. An advantage is that the voltage across the winding (which causes transformer distortion) can "never" be more than dozens of mV. You could use an exceptionally cheap and small core. However since transfos are rarely used this way, you could have trouble with "minor parasitics" becoming real problems. If I was gonna build a million, I'd play with the idea. For a one-off, the thought and testing would be excessive (unless you enjoy it).
The FETs will work. But look at "3-amp instrumentation amplifier". This gives two hi-Z differential inputs in 1.5 5532s or 0.75 TL074s. There is a 2-amp variant which is even cuter, if you don't need absurd common-mode tolerance (you don't). You could also lose the coupling-caps, which may more than pay for the added silicon.
But there are many ways to do it.
Tekay
Well-known member
Why not use a THAT1606 at the input and then a step up transformer after the FET's? THAT 1606 is a balanced in / balanced out driver.
Tekay, please elaborate. What advantages are you expecting from such an arrangement?
Tekay
Well-known member
As a balanced input buffer instead of an expensive transformer and to save space. And it will work balanced or unbalanced. Just a thought when I saw this thread.
so how about now then?
I could use any 1:1 600Ohm or even a 10k ground loop transformer and have two 5k series resistors to keep diodes happy..is that correct?
..then set up the amp for a combined gain of 50db ( to have gain recovery allready built into these stages) and just hang a 1k volume pot on the output...?..
j
I could use any 1:1 600Ohm or even a 10k ground loop transformer and have two 5k series resistors to keep diodes happy..is that correct?
..then set up the amp for a combined gain of 50db ( to have gain recovery allready built into these stages) and just hang a 1k volume pot on the output...?..
j
juniorhifikit
Well-known member
PRR said:
Could you elaborate on this a please? I see for example the SSL channel line input 82E01 is a one opamp differential input, while the 82E149 is a two opamp differential input. What's the difference? Why is the 1 opamp design a cheat? Not questioning your advice at all, just trying to learn the difference.
The single opamp diff amp has terribly different input impedance on the + and - inputs. With R resistors all around, the + in impedance is 2R, the - in is 1R, but the differential input is not the sum, it is also 2R; and the common mode impedance is 1R, which may be considered too low.
The two-opamp diffamp is much more predictable. In that case, the + in Z is 1R, identical to the - in, and the differential input Z is exactly the sum, 2R. Common mode Z is lower though, at R/2. In this particular application, the effect of the diode's dynamic resistance variation will have different action on the + and - side of the diff input. This dynamic balance variation may create problems, in particular regarding the Control Voltage rejection, which is a major problem in a diode VCA.
There's a possibility to tweak the one-opamp diff amp by making the resistors in the - branch (2G+1) greater than the ones in the + branch, but it is not really applicable to the current situation, because the gain reduction causes the loop gain to vary.
The two-opamp diffamp is much more predictable. In that case, the + in Z is 1R, identical to the - in, and the differential input Z is exactly the sum, 2R. Common mode Z is lower though, at R/2. In this particular application, the effect of the diode's dynamic resistance variation will have different action on the + and - side of the diff input. This dynamic balance variation may create problems, in particular regarding the Control Voltage rejection, which is a major problem in a diode VCA.
There's a possibility to tweak the one-opamp diff amp by making the resistors in the - branch (2G+1) greater than the ones in the + branch, but it is not really applicable to the current situation, because the gain reduction causes the loop gain to vary.
> The single opamp diff amp has terribly different input impedance on the + and - inputs.
AND different impedance for Differential and Common-mode signals.
It not possible to have "equal impedance" for all conditions.
It does work quite fine for de-referencing nearby sources, such as inside a large studio.
The two plans you show both have fairly low input impedance, around 30K. This may be OK. In fact it is "good" when sources may become disconnected: the channel tends to go quiet, not amplify room-buzz.
They add add significant noise-resistance. Not an issue at +4dBm, but at the few mV of the diodes, enough to dominate idle hiss.
The 1-amp and 2-amp plans you show are set up for unity gain or some loss. You have to do that when taking +18dBm signals from studio sources. If you can accept a scheme which has significant gain (you need gain), there is another plan using just two opamps which has "infinite" input impedance. . And of course the cannonical 3-amp plan.
AND different impedance for Differential and Common-mode signals.
It not possible to have "equal impedance" for all conditions.
It does work quite fine for de-referencing nearby sources, such as inside a large studio.
The two plans you show both have fairly low input impedance, around 30K. This may be OK. In fact it is "good" when sources may become disconnected: the channel tends to go quiet, not amplify room-buzz.
They add add significant noise-resistance. Not an issue at +4dBm, but at the few mV of the diodes, enough to dominate idle hiss.
The 1-amp and 2-amp plans you show are set up for unity gain or some loss. You have to do that when taking +18dBm signals from studio sources. If you can accept a scheme which has significant gain (you need gain), there is another plan using just two opamps which has "infinite" input impedance. . And of course the cannonical 3-amp plan.
juniorhifikit
Well-known member
Thanks gentlemen! Sorry to take this off-track. I was interested for my own situation, which IS line level and has nothing to do with the diode bridge compressor. Now back to your regularly scheduled program...
FWIW, I'm still very interested in the differences, sonically, between those two SSL line input circuits, as I don't have enough back-to-back experience on the two, and may not have even been aware at the time which I was mixing through.
FWIW, I'm still very interested in the differences, sonically, between those two SSL line input circuits, as I don't have enough back-to-back experience on the two, and may not have even been aware at the time which I was mixing through.
..it works..it works...IT WORKS....
I buillt one chanel on a stripboard today, and with some minor changes to the sidechain bias and meter circuit, it really works...the tracking of the meter still does fall off when adjusting ratio and some tweaking will be necessary before I'm happy...but it works..
and thank's to PRR and Abbey, I did learn something too ( really the whole point of this exercise..)..I hope others did too..
..if there is any interest, I could draw up the schematics with the values I used on one sheet...but lets wait and see...
thank's for all your patiens and help, Gentlemen..hat's off..
j
I buillt one chanel on a stripboard today, and with some minor changes to the sidechain bias and meter circuit, it really works...the tracking of the meter still does fall off when adjusting ratio and some tweaking will be necessary before I'm happy...but it works..
and thank's to PRR and Abbey, I did learn something too ( really the whole point of this exercise..)..I hope others did too..
..if there is any interest, I could draw up the schematics with the values I used on one sheet...but lets wait and see...
thank's for all your patiens and help, Gentlemen..hat's off..
j
just to wrap this one up..
..here is an accoustic guitar recorded through a sm57, laying on the table in front of me, straight into my laptop..first 20 sec without compressor, following 20 sec the same track throught this comp with rather heavy compression. there is some ugly background noise on both samples...could be the tv or the room or perhaps some compander thingy in laptop...I dont know, but it's on both samples, so it's not the diode comp..quick and dirty..take it for what it is...
http://www.aronnelson.com/DIYFiles/up/compressortest.mp3
j
..here is an accoustic guitar recorded through a sm57, laying on the table in front of me, straight into my laptop..first 20 sec without compressor, following 20 sec the same track throught this comp with rather heavy compression. there is some ugly background noise on both samples...could be the tv or the room or perhaps some compander thingy in laptop...I dont know, but it's on both samples, so it's not the diode comp..quick and dirty..take it for what it is...
http://www.aronnelson.com/DIYFiles/up/compressortest.mp3
j
skipwave
Well-known member
Johan said:
Please do! I can't get enough dynamics fx options, and this one looks very intriguing.
kooma
Well-known member
think this could sound wicked on squashing drums
ok..here it is..
and here as a link to a bigger pic..
http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=40958&g2_serialNumber=1
the meter worked better this way. it should be a uA DC meter and the trimer across it sets the sensitivity...set this way, it works more like a GSSL meter than the usual GR meter( it goes up with GR, not down..)...it was easier to get this to work than the other way...and it also helps when seting the bias..just lift the trimmer, set ratio pot(50k) at 0 Ohm, adjust bias until the meter just moves..no more..all diodes are 1N4148 and the sidechain opamp could be any opamp that doesnt oscillate at gain=1..an 4558 is just fine here...
the transformer is working at -30db, so ANY 1:1 transformer will work..I used a "tip of my thumb" sized 600Ohm.. feed it + - 15v..or as much as your opamps can take..
have fun...
and thank's to PRR and Abbey for helping me with this one..
j
and here as a link to a bigger pic..
http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=40958&g2_serialNumber=1
the meter worked better this way. it should be a uA DC meter and the trimer across it sets the sensitivity...set this way, it works more like a GSSL meter than the usual GR meter( it goes up with GR, not down..)...it was easier to get this to work than the other way...and it also helps when seting the bias..just lift the trimmer, set ratio pot(50k) at 0 Ohm, adjust bias until the meter just moves..no more..all diodes are 1N4148 and the sidechain opamp could be any opamp that doesnt oscillate at gain=1..an 4558 is just fine here...
the transformer is working at -30db, so ANY 1:1 transformer will work..I used a "tip of my thumb" sized 600Ohm.. feed it + - 15v..or as much as your opamps can take..
have fun...
and thank's to PRR and Abbey for helping me with this one..
j
skipwave
Well-known member
that is mighty strange seeing the meter in series with the ratio pot. Why not have the ratio pot before the buffer Emitter Follower and have a separate buffer for the meter?
Hmmm... Maybe I just answered my own question, that's considerably more parts.
Thanks for sharing. This is one I definitely want to tinker with.
Hmmm... Maybe I just answered my own question, that's considerably more parts.
Thanks for sharing. This is one I definitely want to tinker with.
audiovisceral
Well-known member
- Joined
- Mar 25, 2008
- Messages
- 150
I think it sounds great. I've always loved the sound of a diode bridge compressor, and this looks like a really nice, simple one.
What controls does it have? Could you post back if you get the metering or other quirks ironed out? Also, how's the hiss/noise level?
This could be a very cool little project.
What controls does it have? Could you post back if you get the metering or other quirks ironed out? Also, how's the hiss/noise level?
This could be a very cool little project.
