dbx 160

Hi folks

I'm currently working on the dbx 160 schematics for my thesis. The original 160.

I can't figure out what exactly takes place in the sidechain circuit. The RMS outputs a positive signal to a first op amp (OA3) with a "level cal" pot. Is it a "detection sensitivity control"?

Then the signal feeds a second op amp (OA 5). I have assumed that the threshold pot allows the user to set a value for a signal that will offset the RMS signal. But what happens if the RMS takes a negative value? Not sure about that

The LED circuitry is also a mystery for me.

Thanks for helping me in that quest!

Cheers

psych60s said:

But what happens if the RMS takes a negative value?

Click to expand...

The "s" in "RMS" stands for "square." Negative results are unpossible.

-a

psych60s said:

Hi folks

I'm currently working on the dbx 160 schematics for my thesis. The original 160.

Click to expand...

Thesis? what grade level or degree is this for?

I can't figure out what exactly takes place in the sidechain circuit. The RMS outputs a positive signal to a first op amp (OA3) with a "level cal" pot. Is it a "detection sensitivity control"?

Click to expand...

I could probably figure it out if I looked at it, but don't have a schematic handy.

Then the signal feeds a second op amp (OA 5). I have assumed that the threshold pot allows the user to set a value for a signal that will offset the RMS signal. But what happens if the RMS takes a negative value? Not sure about that

Click to expand...

I am not psychic (and too lazy to search out a schematic) so I can't give you specific answers, but I can answer in general.

The side-chain for a compressor needs to generate a gain control voltage proportionate to signal level and relative to a compression threshold.

DBX VCAs generally respond to a linear voltage with a logarithmic (dB) gain change, so ideally the side chain wants to convert linear signal level changes to log domain so it can directly drive the VCA.  While this detector output will generally always be a positive (or always negative) voltage that changes proportionately with signal level, by subtracting a fixed voltage from this representing the threshold, we can come up with + and - X dB gain change relative to the threshold.

RMS IMO is a bit of marketing hyperbole and I won't drag you off into the weeds with esoterica about "true RMS", so don't get hung up on that. The DBX detector performs a smoothed log conversion so that output voltage can be scaled and offset to generate +/- X dB of gain control manipulation. Generally 0V=0dB or unity gain for those generation VCAs. They may even have two gain control ports, so a negative voltage into the negative gain control port has the same effect as a positive voltage into the positive gain control port. 

You can look up the data sheet for the VCA used in that design to find out what the gain law relationship is for the voltage changes sent to the VCA (X mV/dB). Knowing that you can work backwards to figure out what the opamps and trims in the side chain are doing.

The LED circuitry is also a mystery for me.

Thanks for helping me in that quest!

Cheers

Click to expand...

The led circuitry if displaying gain reduction it will simply look at the control voltage being sent the VCA. 0Vdc = unity gain or no compression, more voltage is more dB of gain reduction. Linear voltage change at the VCA will represent a linear number of dB.

JR

Hi, and thank you for helping me !



Here's the schematic of the first dbx 160 (sorry for the size).

I haven't find data sheet for the dbx VCA 200 yet. But I have dbx 202 specs : the gain control constant is -6mV/dB. I think these VCA are quite "simples", as a positive control voltage increasing forces the VCA to reduce his gain, that's it.

JohnRoberts said:

RMS IMO is a bit of marketing hyperbole and I won't drag you off into the weeds with esoterica about "true RMS", so don't get hung up on that. The DBX detector performs a smoothed log conversion so that output voltage can be scaled and offset to generate +/- X dB of gain control manipulation. Generally 0V=0dB or unity gain for those generation VCAs.

Click to expand...

You say RMS can output a smooth log signal, so is it necessary to include a sidechain rectifier? I have to admit that I'm not an expert in Op amp-based rectifiers, but nothing looks like a rectifier to me in this circuit.

The threshold pot set a voltage substracted to the RMS signal. Higher values for the threshold pot, higher threshold offset voltage, so we need higher RMS signal to "break" the threshold signal. The result is then amplified by the following op amp and feed to the RATIO stage. I still can't figure out what the op amp before the threshold setting is for.

I think it is the same principle for the "above/below" LEDs. When the audio signal is higher than the threshold, the "below" LED is blocked, the "above" LED is conducting and let there be light. Same as you said, a control voltage send to the VCA allows the LED to conduct.


I'm a young french recording engineer and my thesis is about how compressors "sonic" qualities depend on electronic designs.

Quentin

psych60s said:

Hi, and thank you for helping me !


Here's the schematic of the first dbx 160 (sorry for the size).

I haven't find data sheet for the dbx VCA 200 yet. But I have dbx 202 specs : the gain control constant is -6mV/dB. I think these VCA are quite "simples", as a positive control voltage increasing forces the VCA to reduce his gain, that's it.

Click to expand...

OK.. that -6mV/dB is the control voltage that the side chain needs to send to the VCA to make it compress.

JohnRoberts said:

RMS IMO is a bit of marketing hyperbole and I won't drag you off into the weeds with esoterica about "true RMS", so don't get hung up on that. The DBX detector performs a smoothed log conversion so that output voltage can be scaled and offset to generate +/- X dB of gain control manipulation. Generally 0V=0dB or unity gain for those generation VCAs.

Click to expand...

You say RMS can output a smooth log signal, so is it necessary to include a sidechain rectifier? I have to admit that I'm not an expert in Op amp-based rectifiers, but nothing looks like a rectifier to me in this circuit.

Click to expand...

Looking at the schematic you posted they show a circuit block labelled RMS M2.  This module accepts an AC voltage feed from the input audio in parallel with the feed to the VCA. The rectifier and log conversion occurs inside this module. The smoothing is done by the capacitor connected to this module. The output of this module is a dc voltage that represents the log of the audio input level. This gets buffered by OA3. OA5 is a DC gain stage that injects a threshold voltage to determine when the the compressor starts reducing gain. This side chain output is sent to a simple divider to vary the compression ratio.

The threshold pot set a voltage substracted to the RMS signal. Higher values for the threshold pot, higher threshold offset voltage, so we need higher RMS signal to "break" the threshold signal. The result is then amplified by the following op amp and feed to the RATIO stage. I still can't figure out what the op amp before the threshold setting is for.

Click to expand...

In real world circuits there can easily be several mV of DC errors due to device matching, opamp input offsets, etc. The opamp stage incorporates a trim so they can calibrate the log voltage to accurately reflect the nominal signal level. 

I think it is the same principle for the "above/below" LEDs. When the audio signal is higher than the threshold, the "below" LED is blocked, the "above" LED is conducting and let there be light. Same as you said, a control voltage send to the VCA allows the LED to conduct.

Click to expand...

yup, the opamp treats above threshold and below threshold signal differently.  From looking at the schematic it appears they are using a VCA control port with positive control law, so negative voltage causes dB attenuation.

I'm a young french recording engineer and my thesis is about how compressors "sonic" qualities depend on electronic designs.

Quentin

Click to expand...

While I do not want to do your homework for you the sound character of compressors is generally dominated by smoothing time constants in the gain control side chain, since the input audio is multiplied by these gain changes. For this particular product the side chain character is dominated by what's inside the RMS module (and the 22uF tantalum cap attached to it.

I suspect any attempt to understand what defines that units sound character is inside that module.   

If you can't find a schematic for what's inside that module, THAT corp have published schematics for the insides of their later RMS IC sets. I could tell you but where's the fun in that?  8)

JR

JohnRoberts said:

While I do not want to do your homework for you the sound character of compressors is generally dominated by smoothing time constants in the gain control side chain, since the input audio is multiplied by these gain changes.

Click to expand...

You've helped me to undersand what happened in the sidechain, that's a lot, thank you very much for that 

I'll check for a RMS schematic.

JohnRoberts said:

yup, the opamp treats above threshold and below threshold signal differently.  From looking at the schematic it appears they are using a VCA control port with positive control law, so negative voltage causes dB attenuation.

Click to expand...

I thought the VCA should be fed with a positive control voltage to begin gain reduction. According to the dbx data sheet, gain decreases by 20 dB when 1 volt is applied to the control port.


Quentin

psych60s said:

JohnRoberts said:

While I do not want to do your homework for you the sound character of compressors is generally dominated by smoothing time constants in the gain control side chain, since the input audio is multiplied by these gain changes.

Click to expand...

You've helped me to undersand what happened in the sidechain, that's a lot, thank you very much for that 

I'll check for a RMS schematic.

JohnRoberts said:

yup, the opamp treats above threshold and below threshold signal differently.  From looking at the schematic it appears they are using a VCA control port with positive control law, so negative voltage causes dB attenuation.

Click to expand...

I thought the VCA should be fed with a positive control voltage to begin gain reduction. According to the dbx data sheet, gain decreases by 20 dB when 1 volt is applied to the control port.


Quentin

Click to expand...

I've already mentioned this but the VCAs generally have two control ports, one responds to positive voltages with gain, one with attenuation.

There is plenty of evidence in that schematic to support my observation.

The gain pot is labelled as +20dB at  +15V end,  (-) 20dB at -15V end.

The VCA has a NI pin (negative control) shorted to ground (0v), and a PI pin that gets control voltage.

Finally looking at the threshold opamp and above/below threshold LEDs, the above threshol LED lights when opamp output swings negative.

JR

PS for TMI in premium designs you can drive both VCA control port with symmetrical opposite polarity control voltages. This is useful when commanding large amounts of attenuation (like for fader automation in consoles), so VCA audio path doesn't become imbalanced by too much DC on just one control port, For compressors only commanding a couple ten dB not an issue. 

If I understand correctly what you've said :

The RMS detector provides a negative, log-converted signal (I can't find the schematic but I think there is an inverter somewhere). I may be wrong but there is a diode (CR19) connected to ground I think the positive signal would be routed to ground and negative signal to the op amp anyway.

But the VCA should return to a fixed gain in dB once the signal is below the threshold. Because when the signal is below the compressor has a linear amplification.
The control signal should be blocked or something to avoid the signal usually amplified by the threshold op amp to be fed to the VCA. Do you know how it is performed ?


Quentin

psych60s said:

If I understand correctly what you've said :

The RMS detector provides a negative, log-converted signal (I can't find the schematic but I think there is an inverter somewhere). I may be wrong but there is a diode (CR19) connected to ground I think the positive signal would be routed to ground and negative signal to the op amp anyway.

Click to expand...

Yes looks like negative log voltage.

Log conversion is generally performed by rectifying the audio, then converting that level information to a linear current. Finally sending that linear full wave rectified current through a diode junction (or transistor base-emitter junction), which converts that linear current to a log voltage. 

But the VCA should return to a fixed gain in dB once the signal is below the threshold. Because when the signal is below the compressor has a linear amplification.
The control signal should be blocked or something to avoid the signal usually amplified by the threshold op amp to be fed to the VCA. Do you know how it is performed ?

Click to expand...

A steering diode in series with output of opamp oa5, CR13 blocks positive control voltage, and only allows negative control voltage past that point.

Quentin

Click to expand...

I wouldn't get too lost in the weeds of the control voltage path. In the context of your original question or investigation about what affects the "sonic" quality of this compressor it is mostly inside that RMS module. Perhaps 10% VCA distortion, 90% side chain dynamic response, maybe 15/85 for early dirty VCAs.

Perhaps to understand what this compressor is doing, you need to understand what it isn't doing. Many compressors perform separate attack and release time constants on the rectified side chain. DBX IIRC places a single smoothing capacitor directly across the current driven diode junction performing the log conversion giving a unique smoothing characteristic that is not a simple single time constant but varies with level as the junction impedance varies (faster for louder signals, slower for softer signals).

A rigorous inspection of this would be interesting, perhaps you can look at how it responds to different level tone bursts, compared to other compressor side chain approaches.

JR

PS: I feel like I should get a grade for this.

The 200 VCA is in the tech docs section and Themionic posted this link to the schematics in an earlier discussion on the 208 module in the 160

thermionic said:

Matthias just reminded me that I posted the schematics at another group here: http://gearslutz.com/board/showthread.php?t=38317

I didn't try the schematic as I needed to sell the compressor ASAP, so I fitted a factory unit (it was a fairly valuable comp, so I wanted it to look original).

If I get some time, I might try it, failing that, anyone here want a try?


Justin

Click to expand...

Thanks Justin!

tc

JohnRoberts said:

A steering diode in series with output of opamp oa5, CR13 blocks positive control voltage, and only allows negative control voltage past that point.

Click to expand...

It rang a bell just after I posted the message! 

JohnRoberts said:

PS: I feel like I should get a grade for this.

Click to expand...

I think too, thanks for your great help.

Thank you for the link topcat! Many subjects on the 208 on Gearslutz, I hadn't seen that one!

Here's the schematic for the RMS Detector / Log converter used in the 160.

In complement to what JR wrote earlier, it is interesting to note that in the Blackmer* rms detector, log conversion happens before rectification (one can argue that it happens simultaneously, though).
Before Blackmer, RMS converters did rectify before log conversion. The time-constant circuit used a single RC, which resulted in equal attack and release times. Since compressors need fast attack and long release, an additional circuit had to be grafted. As JR said, putting the capacitor after a diode results in a short attack time and a long release time. Note that the attack time is dependant on the dynamic resistance of the diode. Since this resistance decreases when the voltage across it increases, the larger the variation in input signal, the quicker the response. As a result, the attack time only doubles for each 20dB increase. With a linear rectifier, it would increase 10x for each 20dB. The important consequence is that the compression effect happens very fast when needed, and slow when the level doesn't change much, which is good for reducing distortion.

* I know that officially, only Blackmer VCA's are patented; David Blackmer did not submit his rms detector for patent, apparently because he did not understand well enough how it worked to put it in writing.

BTW, most of the "diodes" in the Blackmer rms detector are actually transistors with base and collector connected. That allowed taking advantage of existing multiple transistors in monolithic form, which resulted in much better matching.

Nine years late Abbey But good info to add to the thread!

Yes, I didn't notice. I have been asked by a member to comment on this thread.

As far as I know (and maybe I'm wrong) there are two versions of the RMS converter, 207 and 208 (9) and often under the pictures for 207 one can find the wrong explanation that it is 208. 207 block unlike 208 uses discrete transistors instead of diodes, provided that the critical transistors are thermally coupled with a ceramic cooler.

moamps said:

As far as I know (and maybe I'm wrong) there are two versions of the RMS converter, 207 and 208 (9) and often under the pictures for 207 one can find the wrong explanation that it is 208. 207 block unlike 208 uses discrete transistors instead of diodes, provided that the critical transistors are thermally coupled with a ceramic cooler.

Click to expand...

Generally discrete transistors heatsinked together for thermal tracking was for making VCAs. Modern (That corp) VCA ICs exhibit superior thermal tracking (and everything else). Ironic perhaps that VCAs get so much better just before becoming almost obsolete.

JR

Actually, transistor or diode matching in the rms detector is not as important as in the VCA. The original 2252 had a symmetry adjustment, that the new chips (4301, 4305...) don't have. The sonic imprint of a badly adjusted detector is a very secondary effect.
It may matter on a very asymmetric waveform, but I've never met such a case.
It may affect reproductibility/threshold stability, though, which may have been an issue whern used in a broadcast final limiter.

abbey road d enfer said:

Actually, transistor or diode matching in the rms detector is not as important as in the VCA. The original 2252 had a symmetry adjustment, that the new chips (4301, 4305...) don't have. The sonic imprint of a badly adjusted detector is a very secondary effect.
It may matter on a very asymmetric waveform, but I've never met such a case.
It may affect reproductibility/threshold stability, though, which may have been an issue whern used in a broadcast final limiter.

Click to expand...

Back in the day I made my log conversion using integrated circuit transistor arrays (perhaps obsolete now). The squared calculation in RMS is peformed by 2x log....

JR

Hi folks... I m new in this field... I move faders and knobs...
But in my poor experience the way the 207 rms detector tracks material is unique...
At first use.. I loved dbx 165a...
After a lot of findings what I understand is that the heart of this device is just dbx 207 and the way it detects.... You can play with timing controls but whit a detector like that life is easy
In electronic terms i really don t know what s happening but you real feel it is a different and unique one to have in your rack
I think I will try to build one
Best
Richi