DBX902 - Anyone know much about the DBX RMS module?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Curtis

Well-known member
Joined
Aug 24, 2006
Messages
305
Location
Australia
Looking at the possibility of doing a modern-day adaption of the DBX902 de-esser using the THAT components, much like Gyraf did with the G-SSL.

Diagrams for the 902 are on the DBX website. Seems the only really "sneaky" components are the DBX202 VCA module (easy enough to swap with the THAT218x plus some tweaks) and 2x DBX RMS modules.

I assume this RMS module could be subbed with the THAT2252 RMS detector with some tweaks to the support circuitry, but what I was really hoping for was if someone had any design notes or datasheets for the original DBX RMS module so I could compare some of the calculations used in the THAT2252 datasheet - things like the RMS timing cap, Ibias, symmetry trim, input resistor etc.

Already got some THAT2181's, and a few THAT2252's are on their way in the next week or so...
 
i have no info to add but i will say that i am definitely interested in something like a 902 type project. i had been looking into the 2 space, 1U dbx 900 enclosure with 2 902's but its pretty pricey for a de-esser!
 
In later 902's the RMS-converter was replaced by a simple transistor-array-diode combination on a subboard - not hard to replicate.

I think one of our 902 cards is that version - maybe I could take some pics.

Jakob E.
 
Hi Gyraf,

Thanks for the input - photo's would be a help. The diagrams from the DBX website do have a simplified schematic of the RMS module, which shows it as an array of diodes and transistors. I always figured there was a lot more going in here, but maybe not?

I'd like to make the DIY version more of an adaption of the original, so I'm not terribly fussed if I end up using the modern alternatives (THAT chips etc) and adjusting the circuit to support them, rather than build a perfect copy. I also wouldn't mind investigating the possibility of adding extra things to the circuit, such as a de-ess "listen" function so you can audition the sibilance effect in isolation of the original signal.

One other question for you specifically. I've started sketching out the PCB design and I'm leaning towards incorporating a few extra allowances for installing different components in the board - THAT1646/DRV134 vs as-per-design descrete output stage, different cap sizes, THAT2181 vs original DBX202 module etc. Would you be OK if I used your DBX202 adapter circuit on the G-SSL in my PCB?
 
Yes the same idea occurred to me now that the THAT Corporation has these useful little chips.
I was going to investigate the possibility of using the 4301 chips (combined RMS detector and VCA) as encode and decode for a BBD delay line. Same idea as Marshall Time Modulator (which uses DBX NR boards inside).
If anybody has any info on how to configure the 4301 or the 2252 to duplicate the DBX NR time constants etc. that would be very useful.
The same goes for a lot of the other stuff that uses DBX detectors and VCA's (Harmonizer, Omnipressor etc.)
Good thread !
 
'Sfunny, I was just looking into doing something similar, only with a THAT4301+2252 (or dual 4301s with the VCAs in parallel for reduced THD+N).

Random notes (which you probably figured out yourself, too):

- the '902 uses a VCA with a 50mV/dB response curve, the newer THAT VCAs are 6mV/dB.
- the time constant on the RMS-converter appears to be important. Haven't done the math or simulated that myself, yet.
- I haven't quite figured out what the purpose of JFET Q3 + entourage is.

Wouldn't a thread like this be better suited for the Drawing Board, BTW ?

JDB.
[considering integrating a 902-like de-esser with a (G)SSL-like compressor, as that would remove one VCA from the signal path. Then again, having a standalone DIY 902-workalike is pretty interesting in itself]

EDIT: nice synchro-post with barclaycon. Great minds...
 
Yep - feel free to do so - would be really nice to have a DIY-option on the 902!


Thanks Jakob :thumb: Early days yet though...


Random notes (which you probably figured out yourself, too):

- the '902 uses a VCA with a 50mV/dB response curve, the newer THAT VCAs are 6mV/dB.


Yep, and I believe Gyraf's DBX202 adapter circuit already scales the CV down to the correct mv/dB range, so no probs there.


- the time constant on the RMS-converter appears to be important. Haven't done the math or simulated that myself, yet.


That's my understanding also, although THAT's design notes do have lots of examples of the 2252 and 2181 in circuits that I could perhaps borrow from. The other alternative I thought of was breadboarding an original DBX RMS module (if I can find the diagrams), comparing it against the 2252 and tweaking the 2252's response to simulate the original.


- I haven't quite figured out what the purpose of JFET Q3 + entourage is.


I think Q3 is acting as a shunt switch that turns off the CV when the de-esser's threshold is not exceeded. The gate of Q3 is connected to a comparator, which looks at the RMS of the full bandwidth signal. The comparator switches Q3 on and off if the full bandwidth signal exceeds a certain DC point, forcing the CV to 0V and giving 0dB gain reduction in the VCA. Turning off Q3 allows the sum of the RMS volatges from the highpass and full bandwidth signals to dictate the amount of gain reduction going on. I think the idea was to be able to provide the ability to de-ess signals at any level, not just some fixed threshold - the full bandwidth signal is the threshold.


Wouldn't a thread like this be better suited for the Drawing Board, BTW ?


Possibly, but being new here I will plead ignorance to not knowing any better :wink: Besides, how many "Labs" wouldn't have a "Drawing Board"? :green:
 
[quote author="Curtis"][quote author="jdbakker"]- the time constant on the RMS-converter appears to be important. Haven't done the math or simulated that myself, yet. [/quote]
That's my understanding also, although THAT's design notes do have lots of examples of the 2252 and 2181 in circuits that I could perhaps borrow from.[/quote]
As far as I can tell the internal architecture of the 2252 is significantly different from that of the RMS module in the '902; I wouldn't expect the timing capacitor charging currents to be comparable.

[quote author="Curtis"]The other alternative I thought of was breadboarding an original DBX RMS module (if I can find the diagrams), comparing it against the 2252 and tweaking the 2252's response to simulate the original.[/quote]
The diagrams are in one version of the schematics; it would appear to be a monolitic array of matched diodes/transistors (like in most log converters).

[quote author="Curtis"][quote author="jdbakker"]- I haven't quite figured out what the purpose of JFET Q3 + entourage is.[/quote]
I think Q3 is acting as a shunt switch that turns off the CV when the de-esser's threshold is not exceeded. The gate of Q3 is connected to a comparator, which looks at the RMS of the full bandwidth signal. The comparator switches Q3 on and off if the full bandwidth signal exceeds a certain DC point, forcing the CV to 0V and giving 0dB gain reduction in the VCA. Turning off Q3 allows the sum of the RMS volatges from the highpass and full bandwidth signals to dictate the amount of gain reduction going on. I think the idea was to be able to provide the ability to de-ess signals at any level, not just some fixed threshold - the full bandwidth signal is the threshold.[/quote]
I see what it does, I just don't get why the de-esser needs to be turned off if the RMS value of the full bandwidth signal is < a hardwired value. Best I can guess is that this is meant to stop the '902 from de-essing tape hiss -- engineers might get puzzled if the '902 would show full compression for 'no signal'-conditions.

[quote author="Curtis"][quote author="jdbakker"]Wouldn't a thread like this be better suited for the Drawing Board, BTW ? [/quote]
Possibly, but being new here I will plead ignorance to not knowing any better :wink: Besides, how many "Labs" wouldn't have a "Drawing Board"? :green:[/quote]
The disadvantage of having a development/analysis thread in the Lab is that it can easily get buried in the sea of 'my xxx-clone doesn't work'-type posts.

JDB.
[Oh, and welcome!]
 
As far as I can tell the internal architecture of the 2252 is significantly different from that of the RMS module in the '902; I wouldn't expect the timing capacitor charging currents to be comparable.


Nor would I, but the design notes do seem a good place to start.


The diagrams are in one version of the schematics; it would appear to be a monolitic array of matched diodes/transistors (like in most log converters).


I think I know the page you mean - I assumed that was a simplified "block diagram" of what was inside the RMS module, not showing any details on internal resistors/caps etc?

I see what it does, I just don't get why the de-esser needs to be turned off if the RMS value of the full bandwidth signal is < a hardwired value. Best I can guess is that this is meant to stop the '902 from de-essing tape hiss -- engineers might get puzzled if the '902 would show full compression for 'no signal'-conditions.


Seems logical. I assume that there must be some absolute minimum threshold that must be overcome in order for de-essing to commence - hence the comparator/Q3 arrangement maybe? - otherwise, as you say, with no input signal at all you may show full GR, or even have the HPF and FBW signals "chattering" causing excess noise in the CV.


The disadvantage of having a development/analysis thread in the Lab is that it can easily get buried in the sea of 'my xxx-clone doesn't work'-type posts.


Point taken :wink:

Errmm...Don't suppose a moderator could move this thread into the Drawing Board Forum for me please?
 
[quote author="barclaycon"]Yes the same idea occurred to me now that the THAT Corporation has these useful little chips.
I was going to investigate the possibility of using the 4301 chips (combined RMS detector and VCA) as encode and decode for a BBD delay line. Same idea as Marshall Time Modulator (which uses DBX NR boards inside).
If anybody has any info on how to configure the 4301 or the 2252 to duplicate the DBX NR time constants etc. that would be very useful.
The same goes for a lot of the other stuff that uses DBX detectors and VCA's (Harmonizer, Omnipressor etc.)
Good thread ![/quote]

I have a lot of experience with companding NR and BBD delays (back in '70s). I found NE572 compandor chips not really a limiting factor for noise floor or distortion with BBDs. I don't know that you want to mimic all of the DBX response shaping which are designed mainly for tape.

The fixed HF pre-de emphasis is useful, but you need to carefully integrate it all into anti-alias and anti-image filters. You want to grab compressor rectifier input just before the BBD and expander rectifier input just after, but be careful to get all BBD clock residuals out of that expander control loop.
-------

I believe DBX published an AES paper on de-esser design back in 80's (?). I don't know how much benefit RMS detection really offers, if any in a de-esser, but it was part of DBX's secret sauce so they used it everywhere. The modern chip sets are cheap enough to play with and see.

Have fun..

JR
 
[quote author="JohnRoberts"]I believe DBX published an AES paper on de-esser design back in 80's (?). I don't know how much benefit RMS detection really offers, if any in a de-esser, but it was part of DBX's secret sauce so they used it everywhere. The modern chip sets are cheap enough to play with and see. [/quote]

Makes the math easy on a feedforward, though. I've thought about combining a de-esser and an RMS compressor in one box, you still need the same stuff as in the 902, plus one or two op-amps for the compressor function in the CV loop.

I was wondering about that JFET but these reasons make a lot of sense.

-Dale
 
Indeed the de-esser is a wideband gain manipulation, only frequency shaping is in it's side chain, so one VCA path could de-ess, compress, limit, expand/gate, whatever...

Using log/dB math, gain controls are pretty straightforward to combine and manipulate.

JR
 
I have a lot of experience with companding NR and BBD delays (back in '70s). I found NE572 compandor chips not really a limiting factor for noise floor or distortion with BBDs. I don't know that you want to mimic all of the DBX response shaping which are designed mainly for tape.
There is a great deal of difference in how delay lines like the Marshall and say, the BEL sound. I am convinced that a lot of this is due to the NR schemes employed. The Time Modulator and effects like the Harmonizer sound big and fat and, if you like, 'tape-like'. That's a very desirable attribute! It's especially true when you use feedback (spin) with these delay lines. The signal doesn't go thin and sharp like it does on NE570 type schemes. If the criteria was simply noise floor and distortion around a straight un-modulated, non fed-back delay then I guess a NE572 would be fine.
I'm basing my findings on gear that I've used which sounds great.
A BBD companded with 2 x THAT chips in a DBX type configuration would, I think, be a fantastic project.
If memory serves me correctly I think Mr. St Croix marketed such a device and called it a Tape-Delay Simulator.
 
I believe you're making an inappropriate logical leap to assume the NR was the magic in Steve's delay line. There are lots of not so subtle interactions especially when you involve regenerative feedback.

For now I will respectfully agree to disagree.

Good luck, and I hope you find the mojo you are searching for.

I have no desire to revisit BBDs while they were very good to me back in the '70s and 'early '80s. Then is then and now is now...

JR
 
I'm not sure that document exists although there is a similar one for the VCAs.


Yeah, I getting that impression :? I suppose I shouldn't be surprised really, it was a manufacturer's specific part.

I have found some more detailed diagrams for the internals of the 208, but there appear to be a few differences between what they show and what the 902 schematics show, so I'm not entirely convinced I can trust them.


Taking a quick look at the 902 schematic it appears that rms detectors are log ratio-ing the filtered and unfiltered input. I think - and this is just after a quick look - that if you built it using 2252s with the datasheet recommended input resistors, Ibias, symmetry etc. you'd be close. Most of the minor "log 0" level differences would wash out in the log ratio and by adjusting the input R and possibly the threshold of OA6 could be made close.


After rummaging through the 2252 datasheet and some of THAT's other design notes I'm coming to a similar conclusion. It would seem that the most critical parts to get right would be the series Rin resistor and the timing cap Ct. More likely than not if I just went with the datasheet example values I'd end up with something that works. Whether it'd replace the original RMS module like-for-like might be unlikely, but also might not be too important anyway.

Looking at the 902 diagrams, my thinking is that I could probably delete R60, R64-73, C19-23, OA4 and QA2 and just drop in the 2252 typical circuit in its place and have a working replacement for the HPF RMS section, no? Same deal with the FBW RMS section too.


That leaves the possible VCA scale factor difference of 6.1 mV/dB (2180) 6.6 mV/dB (4301 VCA) vs 50 mV/db (202) which would require scaling of R52, R49, R55 etc. depending on the VCAs used.


That was the reason I was looking at using Jakob's DBX202 adapter circuit - it will scale the CV down from 50mV/dB to 6mV/dB. So provided I leave the existing CV at 50mV/dB the end user has the option of installing a DBX202 or a modified THAT2181.

Of course, providing such an option for the DBX202 may be a bit of overkill in this application. Perhaps I should really just be designing this for direct interface with the 2181, in which case I can leave out the CV scaling circuitry of the adpater and scale those resistors you mention?


Here's the discrete version of the rms detector I cloned circa 1982:


Ooooo, interesting. I'll have to have a closer look at that - thanks!

One other thing - how important do you think the ability to trim the symmetry of the RMS detectors is? Would I be better off configuring the 2252 to "self-trim" as in figure 9 in the datasheet (page 6), and save a bit more room on the PCB?
 
I believe you're making an inappropriate logical leap to assume the NR was the magic in Steve's delay line. There are lots of not so subtle interactions especially when you involve regenerative feedback.

Yeah, well I guess my 31 years of making records don't really mean that much.
I own a Marshall Time Modulator, a 949 Harmonizer and more BBD - based effects than you can shake a stick at.
I go for the great sound first and then try and find out what's involved.
Generally the compander chip types come second as far as I'm concerned
(including some of the MicMix ones I'm afraid).
I spoke to the great man several times before he died in 2006 and definitely got the impression that the NR scheme he used was a VERY important part of the Time Modulator sound.
Thanks for your opinion John, inappropriate logical leaps notwithstanding, I respectfully disagree.
 
Maybe we should be looking at the same drawing. I get the feeling the designators on the 902 drawing I have are different.


OK, well the drawings I'm working from are here:

http://www.dbxpro.com/Download/discontinued.htm

Under the dropdown list "Schematics" there is a zip file with the diagrams. There's a better resolution version under the "Service Manuals" section too.


A 202 would be overkill and a 2180 would out-perform it anyway.


Yeah, fair enough. Makes the PCB design a little easier too. I'm trying to squeeze it into a single-sided Eurocard board (160mmx100mm) and finding the going a little tough at the moment, so any space I can save will be good.


I would include it (doesn't use a lot of room) but not having one might not hurt too bad particularly in the HF detector as the ripple from asymmetry manifests itself at low frequencies with the timing cap values you're using. In a compressor it minimizes LF third harmonic. This is due to the detector ripple at 2F multiplying the input to produce 3F at the VCA output.


I'll be able to find out more when I receive the 2252's I've got on order - nothing beats real-life experimentation.

Thanks for your help thus far :guinness:
 

Latest posts

Back
Top