Eventide Harmonizer H969

h969 front panel
h969 overview

Last Update 01-10-2015

Harmonizers were always a specialty from Eventide; few effects manufacturers have ever succeeded in getting respectable pitch shifting or harmonizing, and Eventide was a pioneer in that field.  The H969 is the last Eventide harmonizer design without a microprocessor system; while it is a mature design in that the infamous "glitch" has at last been tamed (but not quite eliminated), it is not capable of diatonic pitch that is a feature of later models.  Diatonic pitch is the ability to set musical harmonic intervals and then by defining a key the intervals follow the key scale as you play.  Also the later models feature two harmonizers so you can create two harmonies and assign them in the stereo field.  So if you're a guitar player looking to use a harmonizer for doubling or tripling with intervals, this is not your box.  Also the user patch storage is very limited so if you want instant changes from harmonizing to chorus to other FX, look elsewhere.  The manual even warns that this box is not designed to withstand drops so it is definitely not rugged enough for touring or gigging.  I wanted a high quality studio device for vocal processing so the lack of these features was not a concern.

I got outbid on an H949 and shortly afterward this H969 came up.  In hindsight it was a better unit for my needs.  I got a good score with this ebay unit on several counts:

  1. Recently overhauled by the vintage Eventide restoration expert - David Kulka of Studio Electronics (not the retro synthesizer guys)
  2. Unlike the H949, the H969 features variable delay time, longer delay times, and buttons for immediate pitch ratios.
  3. Original owner had the MIX slidepot removed and the unit wired 100% wet - no problem for a sidebuss effect in a studio system
  4. Because of item (3), no guitar player would have any use for a 100% wet only device.  That eliminated a lot of ebay bidding competition.
  5. Because of item (4), I was the sole bidder and got the unit for a really decent price.  It can't be used inline with a guitar, sorry guys (gosh darn it).

The fact that the H969 was designed sans microprocessor is an impressive feat.  A pop under the hood reveals a LOT of circuitry.  I counted at least eight digital converters, heaping amounts of TTL latches, decoders, flip-flops, adders, counters, discrete OR/AND/inverter gates, tri-state transceivers, some custom PALs, as well as a few exotic opamps and ICs whose function was unknown to me (and I've seen just about everything in my career).  No EPROMS, no full-fledged sytem cpu.  The manual makes mention of "software" but no EPROM is found anywhere in this thing (the "software" IS the TTL gates - no OS upgrades for this baby!!!).  There are sixteen DRAMs dedicated to the 16 bit PCM system.  The closest thing to a microprocessor is the crude AMD 2901 bit slice processor and it is not the system controller.

UPDATE A fellow EE on Facebook has found that the H969 does indeed have a microprocessor.  The MC68705U3S is a 40 pin microcontroller with integrated EPROM.  There are many unusual ICs in the H969 and this was one I did not recognize.  Unfortunately the contents of the EPROM cannot be read out of the device so no backup is possible.  But it certainly protects their IP, which companies like Eventide and Lexicon were paranoid about (and for good reason).

I did spot an NE570 compandor so this appears to be the first Eventide design to omit the dreaded DBX compandor boards that have rendered many an H910/H949 inoperable when they went dead (and until Kulka's restoration business surfaced they stayed dead as no spares or detailed service docs were available).  I've done troubleshooting on some complex systems but this thing is one item that is beyond my capabilities - and few in the industry understand what is going on under that hood. 

The heart of the Eventide harmonizing effect is the phase locked loop (PLL) configured as a voltage controlled frequency converter which is capable of dividing (pitch shift down) or multiplying (pitch shift up) the frequency of input signals.  Division using a PLL is a simple, but designing a reliable PLL multiplying circuit is very tricky.  This not only requires expert circuit design but expert system design and PC board design in that there are plenty of evils that can trip the best PLL system - power buss noise, EMI from external high frequency sources, you name it.  Careful PC Board design is critical to reliable PLL operation - local power regulators, power supply bypass, low impedance ground busses, guard traces around critical paths - the works!  Take a look at the PLL system comprising three circuit boards.  An excellent PLL system separates the men from the boys.  The same is true for pro audio digital converters - a flawed PLL design will create jitter in the clock signal which will cause irreversible distortion in the digital conversion of an analog signal (not clipping but conversion error which results in loss of stereo imaging and depth).   Few hobbyist recording musicians have any concept of jitter and its foibles - digital was supposed to make our lives better but it does have its achilles heels.

Early harmonizers like the H910 and (early) H949 suffered from glitch defects during pitch shift up - the PLL in early designs didn't accurately create a contiguous waveform and this flaw became more apparent at higher pitch shifts.  Some would call this a "feature" and the early units are sought out for this (?!?!?).  Later H949s had a "deglitch" daughterboard that cured this problem, which was available as an option for early units.  The H969 is a mature design and has deglitching over the complete bandwidth.  This was called the Lupin deglitch circuit and was the first of its kind in the world.  You'll still get some glitching at extreme pitch shifts but this unit offers three algorithms to reduce glitching.

According to Eventide ad copy, the H969 incorporated 16 bit linear PCM.  There's a 1975-era AMD 2901 bit slice processor in the H969 that appears to be processing the PCM converted signal - the 2901 is a crude 4-bit arithmetic IC with very limited instruction set, certainly not enough to support an entire system like this.  This is one of those applications that is easier done in microprocessors but remember back in the early 1980s there were few microprocessors with enough horsepower for pitch shifting (those microprocessors were custom developed for large mainframes (IBM) and supercomputers (Cray) and not available to the public).

The H969 has no MIDI interface and very limited user patch storage.  What it has over the H910/H949 is a set of preset pitch shifts that are immediately accessable at a button.  On older harmonizers you had to dial in the pitch interval in abstract ratios (was a minor third up equal to 1.188 or 1.260?).  The H969 has buttons marked "minor third", "major third", "fifth", "seventh", and "octave".  You could get manual diatonic pitch shifting by pushing buttons on the fly or using the CV input.  Working in tandem is a button that toggles between "sharp" and "flat" for pitch shift up and down.  Perfect for musicians that speak in musical intervals (something that is sadly rare these days).  There's also a micropitch button for doubling and other effects.  A "unison" button returns to normal pitch.  The display reports the pitch ratio for the interval selected, and you can tweak the interval using the adjustment control in the center of the unit.  Three pitch shift algorithms are available to minimize glitching or for applicable effect - while one may sound better, there's no reason you can't use another for a "retro" effect.  The third algorithm has the option of user defineable splice points for manual deglitching.  The severity of glitching and effectiveness of each algorithm is dependent the type of program material.  Unfortunately there's no user storage for these parameters.

Also a delay unit is implemented with 16Khz bandwidth all the way to 1530 millisecond delay (you can double the delay length to 3060 milliseconds with the "DOUBLE" button but at the expense of half the bandwidth).  The delay time is adjustable in 6ms increments, not as fine as the 0.1ms resolution of my Korg SDD-2000.  Herein lies the sole patch storage in that five user defined delay times can be stored and recalled at the touch of a button.  Considering the limited patch storage, this unit packs a big ass lithium backup battery which I am relieved is showing no signs of leaking (pretty good for a unit made in 1984!!!).  The same adjustment control tweaks the delay time and a button directly underneath this control toggles between delay or pitch ratio tweaking.  The audio chain is delay then harmonizer.  When combined with the harmonizer you can bypass the delay unit by setting the delay time to zero but there is always a 12ms delay inherent in the harmonizing processing system.  On the left side of the panel is a feedback control which gets interesting.  Not only can you create multiple delays, but each delay is pitch shifted in succession.  Unfortunately this control is not programmable.

The adjustment control begs for explanation.  It is a concentric control with four different adjustment modes (the mode is stored in memory when power is off).  In the first two modes, the knobs are coarse/fine adjustment and only delay or pitch can be tweaked at a time - the ADJUST button selects the parameter under control and a large backlit arrow points to the delay or pitch unit under control.  In delay or pitch shift mode the concentric knobs function similar to "pass through" mode on some synthesizers - the parameter doesn't change until the knob position matches the setting.  In this mode a red backlit arrow lights up above the control.  This red arrow flashes as you move the knob, and the flashing gets faster as you approach the setting.  The manual calls this "Positional I" mode.  While this isn't entirely intuitive to a novice, combined with the preset pitch interval buttons it is ideal for small deviations from a preset interval or delay preset.  If you switch to flange/doppler mode and hold the ADJUST button for three seconds, the concentric knobs now change to "Positional II" mode and this makes knob changes instantaneous.  While this is intuitive like a volume control, it does have the disadvantage of abrupt jumps in tweaking when the knob(s) are moved.   On my unit I cannot seem to get the "Positional I" mode to work.

Hold the ADJUST button for three seconds in delay or pitch change mode and the adjustment mode changes to increment/decrement (the manual calls it "counting" mode); each knob can now adjust delay and pitch independently.  In this mode a red backlit arrow lights up above the control to tell you where the off position is for the knobs.  If you line the pointer of both knobs to the arrow this is the OFF position.  Turning them clockwise increments the delay/pitch, couterclockwise decrements it.  This mode gives you quick access to both delay and pitch shift at the same time.  The fourth mode is only applicable to the flanger/doppler which will be explained below.

To round out the system at the right side of the panel are buttons that select delay only, pitch shift, flange/doppler, reverse, repeat, and CV select (rear or front panel input jack).  The flange/doppler button toggles between flange or doppler mode, with a backlit indicator appearing near the adjustment control.  Here is where the fourth adjustment mode is applicable only in flange/doppler mode.  Flanging and doppler can be automatically (via LFO) or manually swept.  The large knob now controls the sweep frequency and the small knob sets the point of manual sweep.  The algorithm button in the pitch shift section allows offers some control here.  In flange mode, "A" algorithm enables automatic sweep while "B" algorithm will stop the sweep and enable manual sweep using the small knob; pressing "A" algorithm again resumes automatic sweeping at the setting of the manual sweep knob.  In doppler mode the operation is similar with the exception that the "B" algorithm enables randomized sweeping with the small knob setting the random "probability" setting.  You cannot tweak the delay time when in flange/doppler mode, but you CAN recall delay presets from the buttons.  For the pro audio reputation that Eventide enjoys, I am appalled that they did not offer negative feedback which gets that "jet flange" effect.  None of the flange/doppler settings can be stored.

You can find your way around most of this box without a manual but automatic sweep of flange and doppler and the different adjustment modes wasn't obvious until I had the manual in my hands, which is available from the Eventide Store.

Reverse plays the delayed signal backwards and can be really weird, I'll have to experiment to find a musical use for that. It only works in flange/doppler or pitch shift mode.

When repeat is enabled the delay unit signal acquisition stops and the entire delay memory is cycled repeatedly.  It would have been nice if the H969 had implemented manual trigger, turning it into a rudimentary sample playback system like the Korg SDD-2000.  The delay and pitch shift can be changed in repeat mode.

And here is a feature unique only to the H969 - voltage control of pitch shifting.  It is designed to use a passive attenuator such as a passive volume pedal or control pedal.  The CV jack is a TRS with ring carrying +5vdc to a potentiometer and tip returning attenuated CV from the pot.  The currently selected delay or pitch unit will be modulated continuously by the CV.  For a synthesizer enthusiast like me, that's a cool feature.  While any CV source such as a keyboard, LFO, mod wheel, S&H, envelope follower, you name it can be used- the manual warns that exceeding the 0-5 volt range will destroy the unit.  There are a lot of 25-year-old exotic components no longer available in this unit so heed this warning.  The H910/H949 offered the optional HK940 keyboard for CV control of pitch shift but it was a rare option and it used a proprietary input jack.

There are 1/4" and XLR inputs and outputs which can operate at -10dBV or +4dBu.  Be warned that the H969 XLR jacks have pin 3 "hot" instead of the industry standard pin 2 "hot" - I had to build adapters with other "non-standard" gear like this one to make it play nice with my studio system.  The audio at the 1/4" and XLR inputs are summed and can be used simultaneously although you may have to use lower levels to avoid clipping.  When the Line button is pushed and the IN indicator is off, the unit is hardware bypassed.  The 1/4" input accomodates instruments such as guitar and the gain control provides sufficient gain for low level devices, although my unit has a defective gain slidepot that cuts out when I use it in the +4dBu range.  There's a ten segment input level indicator with 6dB per segment - the manual says that the first segment should light at -30dB so quick math tells me that center segment is 0db and clipping is +24dB.  The +4dBu output was pretty damn hot and I had to keep the input gain down to maintain proper gain staging with my studio system, and the input level indicator was riding about halfway.

contact info