1. Dear Guest, if you haven't already... enter to WIN Samplitude Pro X4!
    Dismiss Notice

Building an opto compressor from a kit

Discussion in 'Compressors / Limiters (analog)' started by DonnyThompson, Oct 16, 2017.

  1. DonnyThompson

    DonnyThompson Distinguished Member

    Joined:
    Nov 25, 2012
    Location:
    Akron/Cleveland, OH
    Home Page:
    Yeah, cuz I'm definitely the guy you want soldering while hanging upside down. ;)
     
  2. DonnyThompson

    DonnyThompson Distinguished Member

    Joined:
    Nov 25, 2012
    Location:
    Akron/Cleveland, OH
    Home Page:
    Well, the only advantage I could think of at that price point difference is that you might be able to choose transformers other than Cinemag, which seems to appear in all the Warm Audio gear... I'm assuming that if you got a kit from Hairball they might offer optional transformer choices. ??
    I'm not "knocking" Cinemag XFO's... my only experience with them was that it was an optional XFO choice I had with the ADK pop and swap mic pre I reviewed here a couple years ago. During that shootout, the Cinemag was one of the XFOs I used, and it came in next to last on the list of preferences you guys mentioned. But, that's just one XFO model. I'm willing to listen to what they are now using in their moderately priced line of compressors. And maybe, even if it's the same tranny, it's possible it sounds better when used in GR scenarios.
    :)
     
  3. Boswell

    Boswell Moderator Distinguished Member

    Joined:
    Apr 19, 2006
    Location:
    UK
    Home Page:
    My memory of hearing one is that it's as least as good as many of the other FET compressor clones. That doesn't necessarily make it as good as the original 1176. Reviews have judged that the low-mid range is a little muddier in the Hairball, but that can suit some sources.

    However, at this level, it's my opinion that the objective is not to compare how exactly it replicates the performance of the original piece but whether it does just as good a job in its own way.

    I note that Hairball makes a big point that the transformers they use are modelled after the ones in the 1176 rather than going for more modern transformer designs, on the grounds that the transformers were a major contributer to the sound of the original equipment. Without carrying out a lot more test and analysis, it's hard to argue with that.
     
    kmetal and pcrecord like this.
  4. pcrecord

    pcrecord Quality recording seeker ! Distinguished Member

    Joined:
    Feb 21, 2013
    Location:
    Quebec, Canada
    Home Page:
    Wow this is good info here.. I know the wa76 uses cinemag transformers which shouldn't be that bad.. Always hard to say how things gonna sound before trying.

    The funny thing is, those classic gear that we crave for their sound today were almost all build to have the most transparent signal path..
    It's mostly the aging of the components which put them out of specs that made some become legends.
    I'm sure some clones could sound even better than some beat up originals and vice versa. ;)

    The dream would be to enter a liquidation center and try 100 old units and buy a few that sounds just like we want.
    Guess what ? No doubt, I would pick different ones than other RO members...
    It makes me think that to a certain degree we need to have a bit of faith and just buy the gear and decide on what it's gonna shine once we hear it instead of chasing a sound that's almost impossible to describe and reproduce.
    I had no idea how the T47 mic would sound. After many tests I now know it's my go to for guitar cabs and some vocals.
    If someone said ' hey they are good on cabs and vocal ' it wouldn't give me the proof or confidence to buy it before I tried it in my studio. . .
    We all have our own taste, experience and ear training !
     
  5. pcrecord

    pcrecord Quality recording seeker ! Distinguished Member

    Joined:
    Feb 21, 2013
    Location:
    Quebec, Canada
    Home Page:
    Back to OPTO, sorry I stepped on your thread with the 1176 Donny...
    The La2a DIY seems to be more difficult to find. Maybe it's because tubes works on high voltage and it's not at a level of amateurs.
    But again the Warm audio version is only 899 USD !!
     
  6. Davedog

    Davedog Distinguished Member

    Joined:
    Dec 10, 2001
    Location:
    Pacific NW
    I know someone who has the Blue Stripe 1176 kit. It sounds FANTASTIC. I know someone else who built a stereo LA2A from an early Drip kit. I own an Opto 6 Drip. I know someone who has a REDD47 kit built and OMG. I know someone who built two U87 kits.
    The point is, these things can built if you have some good skills with a soldering iron and a REALLY GOOD soldering iron station. This is NOT something you want to try with a simple soldering iron/wood burner kit. You need all the proper stuff.....bench....light....orbital vise...constant and accurate temperature controlled iron with several tip sizes.......solder sponge....desoldering braid....ECT ETC....and, GREAT solder....
     
    audiokid and kmetal like this.
  7. Kurt Foster

    Kurt Foster Distinguished Member

    Joined:
    Jul 2, 2002
    Location:
    77 Sunset Lane.
    the solution is easy. just buy everything you can. especially if you work itb. i'm serious.

    from what i've heard, an LA2a does an LA2a "thing" but i don't think any two LA2's sound exactly alike, even when they are new. you can get ones that are close but as they age, components will drift in spec one way while in anther LA2a, different components will age/ drift / change in performance. it's the same with all gear.

    if a Warm LA2a or 1176 does the "thing" associated with the legacy piece, then they are doing all that can be expected. all recordists have their preferences when it comes to certain pieces of equipment. this is why guys like Chris Lord -Alge, have comps and eq's in their racks dedicated and set for particular tasks and they never even change the settings. it's like this LA2 is better on VX than the others and i have permanently dedicated it for that task, and these two 1176's ar the best sounding ones i have to strap across the 2 mix, kind of thing.
     
    kmetal and DonnyThompson like this.
  8. DonnyThompson

    DonnyThompson Distinguished Member

    Joined:
    Nov 25, 2012
    Location:
    Akron/Cleveland, OH
    Home Page:
    And I think that comes simply from experience with working with what you have. As you've said, you could put two LA2's side by side, buss the exact signal through both that had the exact settings, and one could sound different from the other. Not necessarily that one would sound "better" than the other, just different. Which makes sense in your reference to CLA ... having certain same models dedicated to certain tasks. For whatever reason,maybe one sounds a little better for guitars than it does vocals. Lol. I'd love to be in the position to be able to test that theory myself - "hmmm, which of my four LA2's will sound best on a B3?" ... ;)
     
    kmetal and Kurt Foster like this.
  9. audiokid

    audiokid Chris Staff

    Joined:
    Mar 20, 2000
    Location:
    BC, Canada
    Home Page:
    The last two LA2A I owned sounded so close, I couldn’t tell them apart. Used them in a variety of 2 bus tasks.
    I’m on my 4th UA LA2A now and the new UA LA2A are super smooth.

    I have swapped out the tubes with special ones and although those do sound different, I could live with the stock ones with no problem.
     
    kmetal and pcrecord like this.
  10. John S Dyson

    John S Dyson Active Member

    Joined:
    Feb 1, 2017
    Location:
    Fishers, Indiana (near Indy)
    I have a suggestion (if it can fit the application) -- the best flexibility (assuming a good audio interface) is to do the compressor in software on a convienient OS (e.g. Linux). I used to do HW compressors in the past, and found it so much easier to match multiple channels in software :). Of course, the VCA chips (like the DBX heritage THATCORP stuff) have the best quality -- but it isn't trivial to get fantastic/ideal behavior. I guess an OPTO has lots of advantages in itself (assuming a HW design.) The problem with compression/expansion/etc, is that unless the device has a limited purpose, an ideal unit can be somewhat complex and tricky. (Attack/decay time constants, doing multiband to get one set of advantages, but getting alot of disadvantages along with it, avoiding intermod -- which also occurs in HW, etc.) I think that one big advantage of OPTOS is that intermod is/can be more easily controlled, even though they are a little slower to control.

    I really learned my intermod lesson one time when doing a multi-band design, yet muting all but the LF channel on the output. So, without post-filtering,you can actually hear the results of the control signal mixing with the LF audio. There are interesting solutions to that problem, and an interesting analog HALF solution (not quite a full solution) using I/Q techniques, but the only 90% solution that I know of is in SW. (The SW solution can be made to be asymptotically better and better, but once the bands are brought down to an octave in size, and they are post-filtered, most of the intermod advantage is attained.

    John Dyson
     
  11. pcrecord

    pcrecord Quality recording seeker ! Distinguished Member

    Joined:
    Feb 21, 2013
    Location:
    Quebec, Canada
    Home Page:
    Most of us use software compressors when mixing and most of the time it's a better way then doing roundtrip setups with external hardwares. Specially if our converters aren't of pristine quality.
    Where hardware gear gets handy is on the way in when recording. Some analog gear are unique sounding and those are worth using even if software alternative exist.
    When recording vocals, I can't imagine not using a channel strip including a compressor.
    I don't get vocalist that knows how to work a mic often. Most of my customers are amateurs who can ruin a track with out of control dynamics.
    Of course If they had the budget I could chart the song and ride the gain manually, that would be the best way to record. But most of them are unpredictable so that compressor is my life saver.

    About opto cell compressor type, I'm not sure I follow you. There isn't much to control but the gain and reduction.
    The good thing about them is that they are very forgiving if they suddenly get hit hard compared to other compressors who will introduce more artifacts.
     
  12. John S Dyson

    John S Dyson Active Member

    Joined:
    Feb 1, 2017
    Location:
    Fishers, Indiana (near Indy)
    Here is what I mean about 'intermod' when talking about 'OPTO's. Anytime one does gain control, it is almost like 'modulating' an AM carrier. In this case, the audio is the 'carrier', and the gain control is the 'modulation'. So, by changing the gain in a way that the frequency spectrum of the gain control comes in the range of the audio frequencies, then there will be additional sidebands (distortion) added to the audio. One fortunate thing is that much of the time, these 'sidebands' reside close to the various frequency components of the audio signal, so are not very audible. However, when the spectrum of the gain control signal along with the audio mix together (intermodulate) into the audio band in a way that the signal is isolated from the rest of the normal audio spectrum -- then that intermod can become more audible.

    * My statement about 'gain control in the range of audio frequencies' is correct, but incomplete -- it is possible for the mixing products to appear into the audio band without the gain control spectrum being exactly in the audio spectrum -- but the actual residence of the spectrum in the audio band does make it worse.
    * Also -- even if the circuitry is perfectly balanced in a traditional sense, there will still be intermod products due the very nature of multiplying the audio signal by a gain control signal.
    * There are ways to maintain a different kind of balance which can partially cancel some of the mixing products, but the mathematics and technique are beyond what I can describe in a few paragraphs.​

    In the case of the OPTO, the gain control is naturally slowed down significantly -- so the intermod products at higher frequencies are naturally reduced. I suspect that there are higher order behaviors WRT optos that further attenuate the intermod products, thereby further reducing the undesired intermod products.

    There are ways to mitigate a lot of the intermod products ( a perfect example of terrible intermod is the beginning of ABBA's SuperTrouper on many versions of the commercial releases.) They seem to be using limiters/compressors which allow the voices to mix together, thereby producing MUCH lower frequency products (lower than the middle/high voice frequencies) and become audible as an ugly mess. On my distribution site, I have an example of fully processed SuperTrouper without audible intermod -- because there are some reasonable (and not so reasonable) SW and HW solutions.

    My distribution/example site: https://spaces.hightail.com/space/bOPBXTkeeT

    The easiest mitigation of the worst intermod (but not a 100% solution) is to split the audio into multiple bands, apply the agc gain control, filter each channel and then recombine (all must be in time/delay sychronization), and much of the 'far away' intermod will be removed. There are other solutions like I/Q technqiues which can remove part of the intermod (the most audible part), yet leaves some of the less audible intermod. RDolby, Burwen or someone like that came up with that cool idea (easily practical and cheap in SW nowadays.)

    Here is my distribution site, and listen to the 'processed' versions of the music... You'll notice no obvious intermod (uses several techniques to rid the signal of the nasty mess.)

    If I was doing a compressor/limiter in HW nowadays, I'd either use some 'THATCORP trinkets" or perhaps an OPTO. I'd tend to stay away from FETS due to the possibie need to tweak to mitigate some effects of component mismatches (which causes more difficulty in handling multiple channels.) Using normal Bipolar transistor or diode methods are also out of the question nowadays (in a way, the THATCORP/DBX approach is a 'bipolar' approach, but has done lots of stuff to linearize the signal, and provides logarithmic control.) * If you look at the design of a DolbyA unit, they have lots of component selection needed to make the units to internally and externally match -- far beyond what we might want to do today. In those units, there are LOTS AND LOTS of 'tweaks' considering that it is basically comprised of 4 channels of approx 7 transistor (two gain control JFETS) each. R Dolby was a very bright man -- he even used techniques which were not commonly known back in the middle 1960's.

    John Dyson
     
    kmetal and pcrecord like this.
  13. Davedog

    Davedog Distinguished Member

    Joined:
    Dec 10, 2001
    Location:
    Pacific NW
    The history of the VCA as a modulation device is quite interesting and a brain-load for us normal persons. The DBX approach has always been one of the basic food groups (if you will) for a compressor circuit. Taken with the opto-style comps as well as the FET's of the 1176 you have the basic sounds of music over the last 50 years.
     
    Kurt Foster likes this.
  14. Kurt Foster

    Kurt Foster Distinguished Member

    Joined:
    Jul 2, 2002
    Location:
    77 Sunset Lane.
    kmetal likes this.
  15. Davedog

    Davedog Distinguished Member

    Joined:
    Dec 10, 2001
    Location:
    Pacific NW
    Yeah. My Bad. I completely left out the Vari-Mu styled comps which are, of course, the oldest form used for compression. The bridge diode style was an early Neve and EMI developed compression circuit but has now come back into style with new companies reintroducing these to the market. Buzz Audio and IGS both offer recent examples.
     
  16. John S Dyson

    John S Dyson Active Member

    Joined:
    Feb 1, 2017
    Location:
    Fishers, Indiana (near Indy)
    I never quite understood the love of the past like that. One can say that the Audimax and Volumax also used a diode scheme (who wants to use those?), but there is really little benefit over the THATCORP stuff. If one is looking for distortion (the good old 25mv diode rule), then do the distortion AND use new technology that allows for easier balance/etc. (The diode method comes from the 1960's or maybe even before.) The variable mu scheme is mostly due to the captive audience back then *(not very many choices), except for the beam deflection tubes (otherwise mostly used for RF mixers and probably 100% unavailable nowadays.)

    When I think of the older schemes, I think of distortion and SNR challenges with the diodes, matching issues for FETS, heat/availability/stability for tubes, distortion and SNR for SIMPLE balanced transistor schemes. To me, for real audio quality, there are a few compelling SUPER good choices nowadays -- DSP/software (of course), OPTO (it is just so very cool), and THATCORP or similar. *JEFT schemes (Sony even made a very special MOSFET that was meant for audio-type AGC) can work super well and can have characteristics that challenge THATCORP type stuff -- but are simply more tweaky. (THE ONLY REASON WHY I USE THE NAME 'THATCORP' is that I forget the circuit name -- there are several variations of it - is the name 'Balmer' or somesuch?...)

    For a casual build, it seems like the DSP/software or OPTO choices might be the best. THATCORP or similar are practically perfect, but to do a superior compressor/expander it takes more than just an RMS circuit, a 'nonlinear capacitor' circuit, and the gain chip. OPTO does some things to help make it sound good, while the fancy bipolar design from the likes of THATCORP require the designer to do all of the tricks themselves. This is probably why so many 'compressors/expanders' have too much 'personality.' ;-).

    Anyway -- I DO NOT MEAN TO CRITICIZE, because people do have their reasons to do things in a way that I might not understand. There is a definite art to all of this, and I certainly know that I don't know everything (far from it!!!)

    John
     
  17. Davedog

    Davedog Distinguished Member

    Joined:
    Dec 10, 2001
    Location:
    Pacific NW
    And then for all the true DIYers theres the Gyraf site. Personally I would go for the SSL 4000 clone.
     
  18. pcrecord

    pcrecord Quality recording seeker ! Distinguished Member

    Joined:
    Feb 21, 2013
    Location:
    Quebec, Canada
    Home Page:
    I did check gyraf site. They don't seem to sell kits. I wouldn't start from scratch with schematics... I'm not that good :unsure:
     
  19. John S Dyson

    John S Dyson Active Member

    Joined:
    Feb 1, 2017
    Location:
    Fishers, Indiana (near Indy)
    Never heard of the Gyraf site -- it looks very nice. They even seem to have examples about the two compressors that I know most about -- DBX/THAT stuff, and the FET design. There are some really interesting papers on the THAT site (AES documents) that talk in detail about attack/release times for feedforward AND feedback compressor designs. The math isn't difficult, but can be quite subtle. Those docs were the first time that I saw the formal derivation of the way that log gain chips (like THAT/DBX) map their attack/release times to dB/sec. Lots of good techie reading on www.thatcorp.com.

    John Dyson
     
  20. John S Dyson

    John S Dyson Active Member

    Joined:
    Feb 1, 2017
    Location:
    Fishers, Indiana (near Indy)
    I have been thinking about the strong interest in a FET compressor, and how to do it more easily/repeatably (and easier to match to tandem channels) -- the easiest way to make a compressor nowadays (that is very repeatable) is a THATCORP (or other DBX style chip.) If you want the sound of a 'linear' compressor instead of true RMS while using a THATCORP and associated RMS detector, you can do so by using a standard peak/average full wave detector, then use that output directly to produce your attack/release times. Then, use the RMS chip as mostly a linear to log converter. The sound of direct peak/average detector vs. signal^2 vs. LOG averaging are quite different. The tightest aggressive sound seems to come from the linear/peak detector. The natural sound comes from the RMS-linear and RMS-log approaches. The good thing about the log approaches is that the attack/release can be made to be consistent on a dB basis, and therefore the apparent attack/release can be constant or consistent vs. the depth of compression. Using the RMS-linear approach (as you might use with a FET), will have a distinctive character, where deeper compression will elicit more aggressive sound. When using RMS-log, the deeper compression will maintain a similar sound. When working on the linear domain instead of the squared domain, the sound will be more aggressive given the same attack/release and whatever compression ratio that you can define. I have also played with the cubed domain, and I that that also has a useful purpose.

    My favorite sound is an upfront RMS-log type scheme driven by an RMS-linear detector (quick attack/release to prefilter before the log domain) (with the RMS log havingdynamic attack/release times), and for the limiter and/or fast-post compressor, I like the good old fasioned RMS-linear method. I tend not to use the average/peak detect version because of my own bias, but I do have that capability in my toolkit because it is pretty good at making the sound more dense when needed.

    John Dyson
     
  • AT5047

    The New AT5047 Premier Studio Microphone Purity Transformed

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice