Subharmonic Synthesis

Update

  • The bottom of the page now includes an upgrade to an old (80’s or early 90’s) DBX 120X-DS to include modern op-amps and an upgraded power supply.  “DS” stood for “Digital Series” which was fashionable marketing nonsense of the day.  There was nothing more “digital” about it than any other model, but it is the best 120 made to date.  By comparison, the 120A could be seen as a trimmed-down el cheapo version – maybe good enough for DJs. 🙂

Abstract

This isn’t a “build” page, but rather a meticulous application of a “professional” product to a home audio system.  And as usual in my approach, good things can’t be rushed.  There’s plenty of preparation to be done before simply plugging something in and expecting miracles.  So this page may seem tedious with few people expected to read let alone do anything on it.  The page discusses the application of subharmonic synthesis to two or four low frequency-capable sealed speakers¹ in a room without any kind of dynamic range compression or other shenanigans.

  1. I would not recommend that this be tried by audiophiles with their “open baffle subs”².
  2. Which is an oxymoron³. 😆
  3. Yes – an oldie but a goodie.

Background

Multiple large surface area subwoofers as part of a home hi-fi system when used sensibly can provide very low distortion bass reproduction at moderate to average listening levels.  Rarely however does the opportunity to seriously get them going present itself as the repertoire is very limited.  Of the thousands of albums in my collection, only a handful have very deep and solid bass in the very lowest registers.  Outside of such albums, cranking the subs is an exercise in futility.

Subharmonic synthesis was developed in the late ’70s by DBX Inc. to create deep bass an octave below the deepest notes recorded on the source material.  See this US Patent.  I never knew about this until recently, but that’s how they did it in discos and other venues.  It was developed in the twilight days of LP which as a medium, was (and remains) incapable of the required fidelity.  That awesome deep bass in venues was not present on the tracks and apart from the fact that most home playback systems had inadequate extension to cope with it anyway – even with digital sources, that’s why “nobody” (except the lucky owners of a DBX 120X-DS) reproduced it at home.

Aside

It’s really just an effects unit for a bit of fun.  It’s unsuited to modern recordings with existing deep bass content, so it should be left in bypass mode most of the time.  If “smart” audiophiles object to it however, I’d recommend escorting them to the door before returning to decide for yourself whether to press the bypass button again! 😆

Playback software plug-ins

VST plug-ins for mixing and mastering software suites are available which apparently can simulate all sorts of things including subharmonic synthesis.

But VST effects plug-ins are a PITA unless you’re a recording engineer, and they don’t lend themselves to casual user installation for use with playback software given the ridiculous licencing rigmarole!  Anybody who can get a Plug-in Alliance VST plug-in for example to load in JRiver, please let me know.  Then again don’t bother!  Twiddling the settings of a perplexing user interface reminds me of Lee Meriwether at the Time Tunnel’s control console forever “losing the time coordinates” of their lost time travellers – hardly conducive to distraction-free enjoyment of the music!

My set-up

I’m running 4 channels through a multi-channel DAC to facilitate an independent 7ms delay to the sealed coffee table subs which are situated about 2.5m closer to the sofa than the front speakers (which are capable of even lower extension).  This delay actually aligns the peaks of the waveforms in the upper bass (around 90Hz) from the front speakers with those from the subs – no kidding.  At the lower end the wavelengths are too large for the delay to be useful.

There are two main options:

  1. Apply the subharmonics only to the subs – leaving the main front speakers to do the hi-fi part, or
  2. Apply subharmonics to all 4 speakers.

As a first approach I’m going with Option 1 and know of no software capable of achieving anything so complex.  Also quite frankly while I trust Harman to have perfected the DBX 120A after so many years, I have no such confidence in the ability of random software writers to mimic the amazing old analogue technology.  And they’re asking hundreds of dollars for one-trick pony plug-ins among huge catalogues when you can buy a used DBX unit for less! Also the general “effects” provided as standard in JRiver and the like are IMO unconvincing at best, so if that’s any indication, then hardware it is.  🙂

Hardware options

Various hardware options are out there including “professional” 1RU chassis from Peavey, Behringer, DBX and others – all with ¼” sockets or XLRs, and some home hi-fi units like the long obsolete DBX 120X-DS with RCAs only.  There is also the modern DBX 510 module (same basic circuit as 120A) for professional “plug-in” use and various “Driverack” models that seem to be Linux-based software implementations of the 120’s functionality throwing in a whole lot of other stuff as well.  The internet provides little in the way of any meaningful information to distinguish the various models.  The forums are just the usual rabble.  Someone has gone to the trouble of uploading FFTs to YouTube showing the output of some of the models, but the operator basically stuffed up the settings and the videos are not very useful.  What I can say is that all of the hardware options throw in “kick” harmonics as well as the basic subharmonic, but these are at reduced levels.  These are usually at odd orders of the subharmonic.  For example, from say a 40Hz fundamental, not only do they produce the 20Hz subharmonic, but a 60Hz (3rd) and 100Hz (5th) at progressively reduced levels.

Aside

  • There’s heaps of hardware out there for DJs.  These include dynamic range compressors and the like.  IMO there is absolutely no place whatsoever for dynamic range compression in hi-fi, so any models that even offered the option were dismissed.

Anyway it seems that the DBX 120A is a good option with subharmonics generated gradually from program about 200Hz down.  They don’t make it any more, but dead stock was still available.  Interestingly, since voice is generally higher than 200Hz, no strange “monster voices” ought to be generated.  And even if a skerric gets through, it’ll be masked by the fundamental.  Voice range in blue. →

Back home

The DAC at the front of the room sends two channels (3 and 4) of full frequency spectrum to the “stereo” subwoofer controller box → and Class-D amp at the sofa.  The controller box includes a pair of ESP P48As, a volume control and bandpass filtering to roughly 20 to 90Hz with very steep slopes at both ends (36dB/octave).  With the DBX unit placed in front of the controller box, “magic” subharmonics will appear from fundamentals not present in the passband of the subs.  In other words, when the subs would otherwise do nothing, they’ll do something.  Then with fundamentals within the passband, the subharmonics will accompany them. 😮  And if it sounds silly with certain music, the bypass switch can be pressed, or the knobs twiddled without going (or remoting) to the computer! 🙂

Aside

  • The 120A (and others I have seen) mixes left and right channels and generates subharmonics in mono.  The subharmonics are then divided and mixed back into the individual stereo pass-through channels.  That is the original signals pass through in stereo, with monaural subhamonics mixed in.

Option 1 first

I decided to try Option 1 – to apply the effect to just the two subs.

Possible advantages (later found to be irrelevant):

  • Retains the independent delay to the subs.
  • Leaves the DBX unit out of circuit for the front hi-fi speakers.  With so many TL074 op-amps in it there may be some perceptible noise.

Possible disadvantages:

  • As the modal region EQ settings for the room are based on measurements taken with all 4 speakers driven simultaneously, the generated harmonic frequencies emanating from the subs alone might not be optimally EQed.
  • Overall reduced SPL of the synthesized output because the four 12″ Morel bass drives of the front speakers are not involved.  I’m no SPL freak, but this suggests that at moderate listening levels, distortion of the synthesised effect will be slightly higher as maximum available driver surface area is not exploited.
  • Possible “localisation” of the synthesized effect to the subs.

A little preparation while waiting for the 120A to arrive – what about bass EQ?

IMO bass EQ (from about 150Hz down) is essential in any enclosed small listening space, so I had a few parametric EQ filters set up in JRiver to apply the same EQ to all four channels.  The Owner’s Manual for the DBX 120A says “When using a chain of processors, the 120A should generally be placed as far down the chain as possible”, but I say this is incorrect where subwoofer equalisation is involved and the owner’s manual for the older home audio “X-DS” version of the DBX 120 actually confirms this.

The equaliser must follow the DBX generator so that whatever comes out of the generator is EQed correctly to the room.

Accordingly the up-front LF parametric EQ filters as previously applied by JRiver’s DSP Studio to all four channels cannot be presented to the DBX unit.  Instead a hardware substitute for those EQ settings must be inserted downstream.

Good thing I have a two-channel P84 subwoofer equaliser set aside. →  Also a good thing that one of its fixed bands is centred right where it needs to be (at 40Hz) to “de-boom” the room.

Here is the EQ trace of the two channel P84 unit retrieved from retirement and with a few preliminary adjustments made:

The “all pots centred” trace was carefully set at unity gain by matching to a loop-back sweep.  That way, if the bypass feature is used, the bass output transitions level with the midrange.

The EQ is compensating for an almighty 40Hz “boom” at the sofa.  Imagine an 80Hz signal through a system with these EQ settings (with SFA adjustment at 80Hz) applied ahead of the DBX unit.  What would happen?  The 120A would generate the 40Hz subharmonic at “full” level, but that subharmonic would receive no EQ whatsoever because it’s too late – the EQ happened “back there” already!  Moreover, an 80Hz signal would “boom” at the sofa at 40Hz!  The other generated harmonics would also be off-whack.  No thanks! 😕

  • Just as an aside, here is the awesome quality typical of ESP projects:

That’s a line level measurement of P84 and the distortion is too low for anyone to hear.  All eight bands work exactly to design.

EQ adjustment

Waiting for the 120A to arrive I manually reset P84 to a target curve determined some time ago with both the coffee table subs and front speakers driven simultaneously for the measurement/averaging around the sofa.  The fixed Q=4 bands of P84 are a compromise compared to the parametric filters determined previously (in particular a broader Q=3 filter at 40Hz), but it worked out pretty well.  Moreover, a little adjustment of the 32Hz and 50Hz ⅓ octave band centres at either side produces the following response:

The above traces show a couple of twiddles at 100 and 125Hz.  These were determined by manual adjustment in REW to get as close as possible to the target curves like the more ideal parametric filters determined previously (including one at 110Hz with a Q of 3) for JRiver.  They’ll have little effect on the coffee table subs because they’re slightly out of their passband anyway, but including them is no drama.

So that will do.  Notice there are two traces there – almost identical.  That’s the left and right channels!  Again a testament to ESP (and a little capacitor matching on my part).

These exact same filters then replace the previous ones in JRiver (front channels only) so that the front speakers receive identical bass EQ.

Here are the settings:

← P84 for subs; JRiver for front speakers →

 

The JRiver DSP Studio settings are processed in order.  Duplicating the Front Left and Front Right channels to SL and SR prior to the application of EQ filters to the Front Left and Front Right channels sends unequalised signals to “rear” channels 3 and 4 of the USB DAC for the DBX unit, then P84 applies identical EQ filters.  Set up as a 4-channel “quadraphonic” system, I designate channels 3 and 4 as “Sub Left” and “Sub Right” ¹.  The cut filter at 3kHz is unrelated – it addresses a front speaker analogue crossover/phase shifter resonance only (it tames an odd summing of the tweeter and midrange).

  1. A conventional 5.1 or 7.1 surround set-up would designate channel 3 as Centre, and 4 as LFE.  SL and SR would designate Surround Left and Surround Right, but it doesn’t matter that I have departed from this convention here.  A channel is a channel.

And here are the REW target and prediction curves with these filters applied.

There’s a slight droop around 60-70 Hz there.  The 120A’s “LF Boost” is a low Q rise centred around 60Hz and it might fill that a bit.

So that’s the EQ sorted.

Back to the 120A

Which has arrived from the USA.  It’s marked on the back as “100V~50/60Hz  120V~60Hz”.  I’d been hoping it would have a dual primary transformer so I could simply rewire it for my 230V~50Hz supply, but no such luck.  It has a crummy chassis-mounted single primary EI transformer.

Not even a mains fuse – just a thermal fuse in that crap transformer.  At least it’s well away from things which is probably a good thing.  Nup.  Will have to fix it. Cheap bastards! 😆

Addendum:

A proper Aussie transformer (cheap NOS oversized Harbuch of the correct voltage) and a fuse were added after initially posting this page. 🙂

OK…  It appears to be a condensed SMD version of the 120XP circuit using quad (TL074) op-amps where the XP was all through-hole and had dual (TL072) types and it also had its Earthed chassis tied to audio ground by its metal-bodied sockets ↓.  Even the output op-amp is the same NE5532 type.  120XP inside ↓ (not mine – just grabbed from the ‘net).

The block diagram of the 120A shows that the “crossover” switch at the back just adds a (shallow) high pass filter for the stereo outputs 1 and 2 and brings the corner frequency selector switch on the front panel into circuit.  I switched that off because stereo bandpass filters are already provided in the “controller box” at the sofa.  It doesn’t matter what is presented at the “Subwoofer” output socket as it isn’t used.  Suffice to say that the “crossover” switches do not affect its output.  The front panel “Synthesis Bypass” switch basically turns the unit into a stereo buffer stage when depressed to illuminate its indicator.  Hey Harman – please get someone as smart as the circuit designer to write your next manual.  “Pressing the SYNTHESIS BYPASS will engage (when lit) or bypass the Synthesis effect of the 120A” is the opposite of what happens! 😛

Measurements

Here are some FFTs that I took.  They’re from steady sine waves input to one channel only.  This was a mistake on my part as the unit sums left and right inputs for generating subharmonics.  In other words the generated subharmonics shown here are low.  Not to worry…   All measurements were taken with the front panel controls of the 120A centred and at no time during these measurements did any of the front panel synthesis indicator LEDs illuminate.  I used REW and a Focusrite 2i2 USB sound card to do these.  I set REW’s generator to a very high level of just -5dB and the 2i2’s monitor (output level) control set to maximum with the input level control set to just below clipping.  On music program with significantly lower input and the same settings, the LEDs flash away merrily.

200Hz input:

Anything higher than 200Hz produces even less in the way of harmonics.

100Hz input:

A 3rd harmonic of the 50Hz subharmonic appears.

50Hz input:

A 5th harmonic of the 25Hz subharmonic appears.

25Hz input:

As well as a 3rd and 5th harmonics of the 12.5Hz subharmonic, tiny 2nd and 3rd harmonics of the fundamental appear.  These would be inaudible.

Again – that’s with all dials centred.  Turning the effect up raises the primary subharmonic toward the level of the input signal.  This also brings the other harmonics up proportionately.

Observations

  1. If that’s what it does, then all the dire warnings about over-driving speakers and amplifier clipping don’t concern me at all because there are steep high pass filters to prevent unwanted infrasonics reaching all power amps.  People who don’t do this are simply asking for trouble.
  2. This is an interesting thought that has me a little confused.  An equal loudness curve chart like the one on the right → shows that below 100Hz the 80 Phon line (80dB at 1kHz and a fairly loud listening level) rises sharply to the left at about 10dB/octave.  This suggests that a 50Hz subharmonic would be swamped by auditory masking of the its 100Hz fundamental.  And with the pots-centred settings as graphed above, the subharmonic is about 25dB down to begin with anyway.  How could anybody even hear it?  More evidence that the steady state measurements are not much to go by?

A note on dynamic performance

The above FFTs have very little to do with how the original units actually performed with dynamic music signals.  Things are very different as per the patent document, a couple of extracts of which appear below:

I do not know whether the 120A has this non-linear capacitor.

Set up and ready to play

That’s the 120A set upon the Focusrite Scarlett 18i20 multi-channel DAC and with a white power cable to distinguish it from the others as USA voltage.  The interconnects are coax with TS jacks because it’s unnecessary to use balanced cables and TRS jacks over such short distances.  Indeed the stereo cables from P84 to the back of the room are single ended too – with no ill-effects.  The USB cable is HP and only cost a few dollars.  Also the PC is an HP and has a cooling fan.  Shock, horror and shame – NOT! 🙂

Initial listening test

I left it with the controls centred as a first test.  The “LF Boost” dial is a bit of an odd one.  Apparently its effect is centred at around 60Hz and it mildly boosts both the original content and the sub-harmonics and might take some tweaking down from the centre position.  Anyway, as a first “experiment” I tried an amazing new Camerata Nordica album appropriately titled “Tales of Sound and Fury” on the BIS label. Track 11 is sensational!  →

And it sounds astonishing!  Not “fake” or bloated in the bottom end at all, but rock solid with a crazy sensation of a huge listening space with bass that kind of envelops the room, but it remains very punchy and clean.  Maybe I’ll see what happens if it’s cranked a bit.

More to come.  🙂

Update

An extended listening and adjusting session with various genres of music has the sub-harmonics dial back to their centred positions.  The “LF boost” dial seems dumb and does SFA and I figured it could only ruin the EQ (as its mild low Q filter is applied to just two of the four channels) so I turned it right down.  The two Subharmonic band trimmers if turned beyond half way see more of the effects LEDs flashing, but the bass can be too heavy, then if the “Subharmonics” (master) dial is turned back below half way it gets the sound back to where it was, but with the LEDs still flashing.  It was kinda fun watching that for a while, but since it sounds exactly the same, I figured Harman has probably played with this for decades and the mid positions are just fine.  The trimmers are probably just to protect systems that are easily over-driven.

Conclusion

The dire warnings in the owner’s manual can pretty much be ignored in a system naturally capable of powerful relaxed bass to 20Hz with infrasonic filtering.  No matter how stupidly I set the controls (even with high overall volume) there was never any sign of amplifier clipping, voice coil former “poling”, or other stress – just deep flowing and visceral bass that could easily be considered absurd in the context of the music from which it was generated.  As I’m so happy with it, I feel no need to try Option 2 (involving the front speakers).  At the mid-point settings in a properly EQed and calibrated system, it’s just wonderful.  Whether it’s “hi-fi” is a question that can be mooted at the press of a bypass button. 🙂

Not quite …

Well as usual I can’t help myself – and just as well!  I’d forgotten I’d ordered these “Monster indeed” 😉 RCA splitters which arrived today.  So Option 2 could be tried very easily.

Option 2

This is to feed the 120A’s signal to all four speakers.  I had some reservations about the possibility of audible noise from all those TL074 op-amps in the 120A.  Also despite the great measurements of the P84 above (filled with OPA2134s), I still didn’t like the idea of having “all that” up front of the “awesome” LME49720 filled active crossover for the front speakers.  However one must be open-minded, so I thought that maybe this experiment could prove that there’s nothing wrong with TL072(4)s and self-debunk some lingering audiofool notions that maybe I hadn’t quite shaken.

So these splitters were plugged into the P84’s outputs to feed not only the long cables to the subwoofer controller at the back of the room (P84 has 100Ω series resistors at its outputs, so capacitance of the long cables won’t destabilise the OPA2134 output buffers), but also the active crossover inputs.  All those EQ filters were un-checked in JRiver first of course.

Result

Well as if it wasn’t astonishing enough!  This is probably the final straw in my decades long slow shedding of residual audiofoolery!

With the system changed over to Option 2 there is no turning back.  There is no perceptible hiss, hum or other noise attributable to the 120A from the front speakers even with an ear at a midrange or tweeter driver.  Also no apparent change to the feasome sound quality that I’ve tweaked into those speakers over the years either.

With the subharmonics no longer isolated to the coffee table subs, there is a multi-fold improvement in the “punch” from the overall system and the subwoofers again totally “vanish” into the room.  And I’ve never seen the cones of the four up-front Morel 12″ woofers move quite like that!

I played a lot of music and in particular REO Speedwagon’s “Take it on the Run” and a few tracks from the INXS “X” album – as examples of the type of recording that I thought might benefit the most.  They look like this in DSP Studio → which is obviously no big deal at all in the bass department (typical of that style of music from the early 80s), and the cones moved like you wouldn’t believe.  An incredible punchy new bass line that you’d never suspect to have come from something that’s basically an “effects generator”.

It does not seem to matter that there is no longer a digital delay to the subs either.  Indeed I reset JRiver’s output format from 4-channel to stereo and removed a now superfluous pair of analogue output cables from Channels 3 and 4 of the DAC.

A note about a multi-sub expert’s comment found on the ‘net

Yeah – the guy who looks and sounds exactly the same as the original Uncle Fester and who advocates the overlapping of multiple subs with front “mains” (no crossover between them).  That part I don’t mind and it’s what I’m doing here, but IMO he misses the boat where he claims that “at no time should one ever EQ the mains based on room data.”  Unless he means something very specific by “room data” that’s utter nonsense!  Of course you wouldn’t EQ mid and high frequencies based on measurements taken at the seat.  But if I leave the coffee table subs switched off and flick the bypass switch on the equaliser so that the mains receive no EQ (as he advocates), then the enormous 40Hz room mode immediately hits you at the seat – and it’s not a speaker resonance!  It was determined by in-room measurement data!  The fronts must be EQed exactly the same as the subs for optimisation of the response at the sweet spot!  Someone has written software based on Fester’s “theories” to help people to EQ each sub independently of all others.  It cannot work.  Stick to the so-called “global EQ” for a sweet spot alone and be done with it.  Any attempt to “spread the goodness” around the room can only compromise the sweet spot and therefore fail!  Stick to light bulb tricks – seriously.

Additional Conclusions and observations

  • The 120A’s LF Boost control might now potentially be more useful since it’s mild EQ is delivered to all of the speakers.  I ended up turning it fully anti-clockwise (off) however.
  • If you do it at all, then just like applying the exact same EQ to multiple subs, apply the synthesis to all transducers in the room that are capable of the required extension.
  • Don’t try it without infrasonic filtering ahead of the power amps unless you really enjoy trouble.
  • For those rare albums with very deep bass content, the bypass switch should be used.  For example the best recording of Tchaikovsky’s 1812 Overture is by far the Pletnev version with The Russian National Ochestra on DG.  The 120A muddles the cannon fire.  In bypass it kicks some serious arse.

A side lesson from all of this is kind of a confirmation of what I’ve tried to accept for a long time – that audible distortions from a solid state sound system are predominantly speaker-related and when it comes to LF performance, that pretty much comes down to maximising surface area with multiple subs.  The use of very expensive and exotic op-amps and the like pales into insignificance.  The insertion of a plethora of lowly TL074s (available for less than 50¢) up front of the system is completely irrelevant to the overall sound quality and only goes to validate the often repeated observation that “there were probably heaps of them used to produce the album anyway”.

And who marked this? 😆 😆 😆

Update September 2018

Still using the 120A and loving the solid oomph that it delivers, but it’s front panel looks a bit silly (buzz words all over it) so I “needed” to replace it. 🙂  Just kidding.  Anyway, with a view to doing a little project, I finally found an old 120X-DS model in perfect condition in Japan.  Many months of searching found only damaged units that were either over-priced or previously rack-mounted, scratched and otherwise beaten-up by DJs.  The “DS” was for home use without balanced in/outputs (that nobody needs), and had four separately adjustable synthesis bands.  But also – its PCB is through-hole which lends itself to a good butchering.  Here it is:

And it has a covered switch to change the primary winding configuration to local voltage:

Note May 2020: The price of these things has shot through the roof since I published this information.  I didn’t pay anything remotely close to what they’re asking today!

Inside:

This is way better in construction and attention to detail is superior to the 120A, but nonetheless everything is/was built to a price.  It has a some marginally voltage-rated electrolytic capacitors and very low-grade op-amps throughout – mostly low slew rate RC4558Ds.  Even the ones noted as “Selected” in the service manual and maked “S” on the PCB are lowly LF353s.  Part way through production they must have identified a problem.  Additional parts and coloured jumper wires were added (bottom centre in above picture). I have no idea what the problem might have been, but am glad nonetheless to have lucked on a late production one!

That auxiliary board is for the front panel LEDs and the 4 band trimmer pots.  Could have used a few cable ties!

I won’t bother posting images of the FFTs, but it seems to perfrorm pretty much the same as the 120A.

A little project is underway and it involves replacing all of those op-amps with non-exotic, but modern ones (with 100nF ceramic bypass caps under the PCB for each) and replacing the electrolytic caps at the PSU with larger ones.  More to come…

DBX120X-DS rebuild (Sept 2018)

I pulled it apart and did what I said above.  All 22 of the op-amps were removed and gold flash low profile machined pin sockets installed with bypass caps across supply rail pins 4 and 8 of each – shown here:

The rail tracks were cut between the main PSU caps and 0.56Ω ½W resistors installed (seen there at the left) to form π filters to reduce high order switching harmonics from the bridge rectifier.  Actually I cut two of them off after that photo was taken since only one was required on each rail. Pre-regulated Vdrop attributed to these is only about 0.2V.  Roughly ±24V unregulated DC is required by parts of the circuit, so it was important to minimise the drop.  Dissipation in each resistor is only around 0.07W.  The 470μF main filter caps were replaced with 35V/2200μF 105°C:

The heat sink is large because the regulators are dropping around 12 Volts – and just as well since the new op-amps will draw more current.

I know I said there was no perceptible noise, but this seemed like a fun project, so I did it anyway:

A bunch of left-over OPA2134PA op-amps (JFET input) replaced all LF353s (like TL072s).  Special low noise NE5532APs (bi-polar input) replaced all of the RC4558D op-amps.  These are the Japanese JRC equivalent of crappy μA741s which should not be used for audio!  Now they were really bad opamps.  THD+N and slew rates are a lot better in the replacement devices.  In “bypass” mode just one OPA2134 buffer is in circuit.

  • Note: some of the op-amps are not for “audio” anyway, but rather perform supporting functions around the flip-flop and comparator chips.  It was too hard to determine which were which, so I simply replaced the lot.

Although the PCB had provision for fuse holders, none were factory-installed.  Instead jumpers were soldered across.  These were removed to install fuse holders – top centre in above photo.  200mA fuses were installed for certain markets, but not for this version with the switchable primary voltage.  As the modern high performance op-amps draw about 150mA more current in total than those originally fitted, 400mA slow blow fuses were installed.  And the regulators were mounted on sil-pads, but without spring washers!  When will these tight-wad manufacturers learn that sil-pads creep?  Naturally the nuts were both loose so the heatsink might as well have not been there!  So new insulators, bolts, Nylocks and spring washers were installed.  Really!  How much does a spring washer cost?

Inter-stage coupling caps were 4.7μF BP electros which I’d prefer not to keep being around 30 years old, so all 14 were replaced.  I used PETs for longevity:

They’re the white ones.  To clear neighbouring resistors, some were perched up on header pin plastic with pins removed. →  All other electrolytics were also replaced.  The 16 Volt ones in particular were replaced with 25s.  None of the original film or ceramic caps were touched.  It would be pointless anyway but I was also confused about a “non-linear capacitor” required for enhancement of transient attack signals.  No kidding – the patent says. 🙂  A misunderstanding here – see comment below from “Fomoco94” who is correct. The patent shows an opamp circuit that forms a “non-linear capacitor 48”.  Here it is:

It’s actually three caps, two diodes and a resistor in the feedback loop of an opamp.  Although I cannot see that replicated exactly in the schematic in 120X-DS service manual, I’m not sufficiently expert to say whether it’s implemented or not.  There certainly are caps, diodes, transistors and resistors in feedback loops all over the place however.

And after all that (over 340 resolderings!) the thing still actually works, and well!  I certainly couldn’t hear any noise from it before, but it’s measured noise performance is now sensational.  The scale in the next picture goes way down to -50dB below zero and the primary subharmonic at 50Hz is up at around +95dB:

That’s with all subharmonic adjustments maxed out and with a high level 100Hz input to both channels.  REW thinks the subharmonics are “noise” but check out the actual harmonic distortion (those calculated off the fundamental with numbers shown above the peaks).  The general noise floor is 130dB below the primary 50Hz subharmonic!  Amazing.

And just as a little fuss, I didn’t like that the chassis was not Earthed.  It’s an old Japanese item and is typical in being neither Earthed nor double-insulated. Not safe!  Also the audio ground was tied to the chassis with a strap off the zero volt line at the transformer connection to the PCB.  My whole audio system comprises Earthed equipment with floating audio grounds all tied together via the coaxial interconnect shields.  The 120X-DS in effect added its chassis as an antenna to audio ground!  Fortunately the RCA inputs/outputs are insulated from the chassis, so I could fix it!

Just some pics showing the installation of a 3-conductor power lead with it’s heavy shield used for Earthing, strapped to the chassis where the old PCB ground strap was secured:

That could not have been done so easily with a 120XP for example, in which the input/output sockets are continuous to chassis.

Listening

It’s actually better than the 120A.  More punchy/snappier (perhaps since the 120A deletes many circuit features such as the non-linear capacitors), less bloated (not that the 120 was bloated) and completely upper-frequency transparent in both bypass and “synthesis in” modes.  The 4 individually adjustable bands also make for a more integrated and natural sound when tweaked.  It has therefore replaced the 120A permanently in the upstairs system.

11 thoughts on “Subharmonic Synthesis

  1. Thank you so much. This is a very comprehensive review of these units and i think you hinted at what I have in mind to modify my own. Would it be possible to drop the subs final filtering an additional octave like some people do with guitar and bass amps? You mentioned swapping the Inter-stage coupling caps and i believe you did this for noise reduction but this can also have an effect on the units overall subsonic filtering correct?

    I’m handy enough with mathematics and a soldering iron; I would like to modify this unit to generate and output frequencies as low as 16 hz to transform the deeper 56-32 hz material in modern music and want to consult you before i tear into this beautiful unit with my limited knowledge (Engineering Background). I recently acquired a DBX 120X-DS. It is in great condition and I use it in oversized DIY 2.1 setup with a large low tuned sonotube subwoofer and bi-amped control over a pair of 3 way sealed mains using an l-pad for additional control over the midrange. I listen to mostly music and the sub is built to quickly swap out the 6″ port and normally gets tuned “high” to 27 hz for a big spike in efficiency right where the the synth drops off; the system is nearly bombproof and even stands up to extra eq to bring things to ridiculous levels without breaking a sweat.

    The 120X-DS is amazing for classical and older rock music but this system has the low end guts to go deeper with the right signal. Without the 120X-DS it responds well to sine waves and organ tones nearly down to 16 hz which was the ultimate goal. It seems like the unit will synthesize tones in the low 20 hz range but they seem filtered out which would be appropriate for off the shelf or equipment from the 80s. Would it be possible to drop the subs final filtering an additional octave like some people do with guitar and bass amps? I’m also interested in scaling up or down the center frequency on the units 4 built in eq pots but I don’t even know if this is possible. Otherwize would be keeping all pots flat or finding some additional EQ for the real subsonics.

    Once i finally achieve this tremendous effect I have plans on scaling the entire system up to foundation shaking levels. I am dead set on having the capability of a full scale pipe organ at the flick of a wrist.

    Thank you,
    Dave

  2. Hi Dave. Sorry – Akismet treated your post as spam perhaps due to its length and it sat in the spam folder for a while. You raise many points – a most interesting one being the lowering of the bottom end filter cut-off. No doubt that could be done, but I’d have to study the schematic in more detail. My system is -3dB at 12Hz, but somehow I doubt that such a tweak would achieve much since the filters are all of low order anyway. As to the inter-stage coupling caps changing the sound – no I don’t think so. I only did it for stability/reliability. The film caps are of the same value as the original BP electros and the only measured difference was a much lower ESR. There may be some merit in bypassing the rail pins at the CMOS chips with 100nF ceramics too, but I didn’t think of it at the time. This was an enormous overhaul and I wouldn’t recommend trying it unless you’re really well experienced. For example the 120X-DS main PCB uses very reliable NASA style bent leg soldering to nearly all of the components including the corner legs of all op amps. A lot of patience is required with over 300 of these and it would be pretty easy to accidentally lift tracks.

  3. I have a little project studio, bought a 120XP and found it priceless for mixing my stuff. Recently i both got a 120X-DS and learn a little bit on how to solder, so i was thinking in recapping the unit as soon as i receive it.

    Just read your awesome review (thank you very much for doing it), and after seeing the mods you applied, I was wondering… would you modify mine?

    And if not… what kind of mod would you recommend me to try, given I am basically a noob-but-willing beginner in mods / upgrades ? Perhaps sticks to just recapping? Or I should dare to change opamps?

    So far I just practiced desoldering on junk, and did a simple “mod” putting some solder bridges on my mixer (Soundtracs Topaz Project 8) to select some configurable options.

    Sorry if my questions sound silly, i just got started in this world.

  4. This is definitely not a beginner’s upgrade! LOL. And I’m not a commercial set-up – just a hobbyist, so I couldn’t take on a job for others – especially since I’m in Australia. Your IP address suggests you’re in Germany! LOL If you want to try it as an easy start, then maybe just do the electrolytic caps – being sure to replace all the non-polarised ones with film caps. See how confident you are after that. Cheers.

  5. Thank you. You are correct. The non-linear capacitor is shown in the DBX patent at 48 to be quite similar to what’s shown in your reference. Will amend. Cheers.

  6. Hello,

    I have a 120X-DS as well. I really would like to do the same upgrades as you did, but I am not tech savvy, I can solder pretty good but I am not into adding pcb’s or upgrading etc…. So replacing the PS caps with ones you put in and those bypass caps as well? It would be nice if you had a walk thru step by step of what you did? the Op Amps as well but then the newer ones will draw more current so what else needs to be upgraded?

    Cheers!

  7. The main thing to do re higher current draw of the modern opamps is to make sure the two regulators are attached firmly to the aluminium heat sink. Mine were loose. Add spring washers, replace standard nuts with nylocks and tighten up. It will be fine since the heat sink is really oversized. Mine has been reliable and you can feel a bit of warmth in the chassis lid over the heat sink but nothing serious at all. But be cautious. It was brazen of me to take on this overhaul. If the thing didn’t work afterwards, it would have been near-impossible to trace a fault. It could have been an expensive brick.

    Don’t go soldering bypass caps to the underside of in situ opamps! It’s not warranted anyway as the originals are slow devices and do not oscillate in circuit. I only did it because the new opamps had much higher bandwidth and had an increased potential to oscillate. It’s a fiddly job requiring excessive soldering time and could easily overheat the opamps. I did it with just sockets installed and inserted the opamps afterwards.

  8. Hi Ian,
    Dave here again from Nov 2018, slowly continuing on my quest for full subharmonic synthesis with modern recordings. The only thing slowing me down is fear of getting evicted. I just wanted to thank you again for this amazing write up. I never did tamper with the 120X-DS unit and I am still beyond happy with it today.
    In my quest for perfection I have also tried out a Behringer Ultrabass EX1200; I can say the DBX unit is far superior for home Hi-Fi. I’m going to try out an Audio Control Epicenter next which is from the 12V car audio world (My class A/B mains amp and 31 ch EQ are 12V and run off a big computer power supply so this is an easy inclusion, audiofools may laugh at this but I have no issues, the 12V PSU even trips out and acts as an E-Stop cutting the entire signal chain when things get out of hand as it is sized just below the max draw of the power hungry class A/B mains amp). Note that the reason the system is built using a car amplifier is to simplify bi-amping of the mains with the individually adjustable “front” and “rear” channels, but I digress. The thing about the Epicenter is that in factory form it has a nasty highpass at 33 hz and has far less functionality as far as tweaking is concerned. However, the mfg sells chips to adjust this highpass as low as 15 hz. I will report back with the results and confirm if the unit is worthy of Hi-Fi. I remember laughing at this unit when shopping around for the DBX120X-DS but here I am today.

    Thank you,
    Dave

  9. Thanks Dave. Look forward to the update. A friend offered me one of those Behringer units (might have been the earlier German Ultrabass EX1000 come to think of it) but my general aversion to the brand and the fact that DBX was the originator and “them” the mere copycats put me off. Glad to hear that you have used both and prefer the DBX – just confirms my suspicion. So what was actually “wrong” with it? My hunch is that it had sub-sub-harmonics which would make it all a bit woolly. The punch of the 120X-DS is just insane. Your audiofool comment is spot on! I have numerous audiofool friends and not one of them likes sub-synth despite not ever experiencing it. I find that it works best on older pop recordings like ABBA (blush). LOL 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *