BBD and me, and Ali Express fakery

In my previous post about the Roland stringers, I mentioned that they all used the Reticon SAD512D chip, but the second iterations of all these designs (circa 1980) switched over to Panasonic chipsets. I thought I would mention a bit about why this happened, and why a dead SAD chip make you, well, sad.

Reticon was actually an early pioneer in the use of CCDs, both in BBD (Bucket Brigade)delays and video sensors. Although co-founder Gene Weckler wasn't part of the Traitorous Eight who left Shockley and co-founded Fairchild, he followed a similar path, working for Shockley straight out of university, then Fairchild, then onto his own startup in 1969.

Their products included the SAD (Sampled Analog Delay) chips, including the SAD1024 with two independent 512 stage delay units, and the SAD512 which was basically a SAD1024 where one half of the chip had failed testing (a 1000-transistor chip in the early 70s would have been a biggish deal to produce in quantity and yields would not have been great). To understand how these worked, one cannot think of them like a digital memory chip, or even a digital delay line like the Princeton PT2399 (their distant descendant), where every location can be in use simultaneously. 

The reason these were called Bucket Brigade was that they consisted of a line of capacitors and electronic switches connected to each other. As a voltage came into the chip at one end it was progressively passed through the entire chip, but instead of being done so in a ripple arrangement, where one voltage is passed out and each other voltage shuffles forward one by one, half the storage locations would be empty at any given time, and each next cycle would either sample an incoming signal and shuffle existing locations forward by one, or output a delayed voltage and shuffle existing locations forward. This may seem inefficient, but it removed the need for complex circuitry to address each delay location individually, as well as ensuring that all locations moved forward simultaneously within a single clock cycle. Indeed, all one had to do was connect the gates of half the bucket transistors to a clock input, and the gates of half the bucket switches to another, and send alternating signals to each clock input. 

What the SAD512D did was no longer make the 512-stage version of the delay line an afterthought, but shrink it down into a compact 8-pin package. It also included some circuitry to generate the alternating clocks from a single square-wave input, thus making the chip very guitar pedal-friendly. 

So sunshine and roses, right? Well, in 1977 EG&G bought Reticon, and it appears they really wanted them for their optoelectronic tech as all further refinement of the SAD BBDs seems to have ceased. Enter Panasonic, who soon developed rival products and continued refining their performance and voltage requirements, and by 1980 had convinced Roland to switch over to their products, even though the Panasonic MN series BBDs always required two external alternating clocks. Before too long both the SAD and Panasonic MN chips were discontinued in the face of digital tech, and became very expensive on the aftermarket. Eventually Uli Behringer's CoolAudio reissued the MN series chips, leaving the SADs high and dry and consigned to history's dustbin. What did not help either is that the SAD chips have a reputation for dying (ie my RS09) and this has further constrained supply and made second-hand parts suspect.

However.....

It appears that SAD512Ds are available again-not officially of course. The Chinese semiconductor market can be a wildwest of recycled scrap parts, sometimes labelled to be that what they are not, or even non-working chips which have been labelled to con desperate and gullible Westerners. Worse still are US Ebay sellers who import this junk and resell it offering it an air of legitimacy at a higher price. Still...when used SAD512D chips that are probably on their way out fetch triple-figure prices, it's tempting...

 

And indeed I was tempted. So after spending $A15 on AliExpress, this arrived...

The total lack of antistatic packaging was not promising.

And no mistaking the engraved printing for a new-old-stock 1980s manufacture (ink printed) chip. Still, testing time!!!! First up the testing rig:

Followed by removing the precious original from the pedal board for safe storage (after disconnecting the battery)

Then the unknown chip in circuit.

Results: both chips were non-functional. As static damage to both chips in transit is unlikely, one can only assume they were duds to begin with. Lesson learned. On to the repair!

Oh, your hype is beautiful, oh the price...I C O N I C

 Well, in a coda to my Tandy Tampura article of last year, and echoing my sentiments in the article, it appears that the home of hype and greed in the service of the price inflation of music gear has decreed that this rhythm box is now "iconic", with "sonic territories" waiting to be explored (and presumably prices to be gouged) by issuing a sample pack. It's hilarious that they try to make you believe this was ever worth 24 bucks, when it takes longer for you to read the legal terms and conditions PDF than it probably did for a person to assemble this entire "extensive" sample package, but here it is.

You honestly can't make this stuff up. Bonus points for recording a piece of 40-year old consumer-grade ephemera with a recording setup you could probably use for classical music recording. Says it all about the vintage gear hype cycle, really!

A very SAD Panda (plus MXR shMXR)

So...as we already saw, my RS09 needs a bit of open-heart surgery to get back to full health, with the transplant required being the most costly one possible after the master octave generator,  an SAD-512 BBD chip.

Here is a close up of the deceased piece of metal-oxide coated silicon in question.

So...despite being out of production for decades, this chip appears to be for sale in bulk lots from China, at high prices, as "new". Normally I wouldn't go near the Chinese aftermarket for silicon, as chips pulled from electronics recycling are often sold as new, and sometimes chips are even relabelled to look like more expensive and desirable models-which doesn't make them work like the more expensive brethren!

Therefore, I did what anyone would do, find a crummy piece of gear that used the chip that I would have no qualms about sacrificing. Picking up courage, I waded into that wretched hive of scum and villainy known as Reverb. Hmmmmmmmmmm.....

As crummy but useful things go, an MXR "Commande" chorus pedal from the early 80s is about the ideal, containing a SAD512D BBD and costing about $100 shipped. This was an abortive attempt to compete with the Japanese onslaught from Boss and Maxon/Ibanez, which saw affordable, well-made pedals in rugged die cast enclosures flood the US market. Well, at least MXR tried to match them on price, pity about the rest.

This pedal even looks cheap, from the poor screen-printing on the enclosure, to the enclosure made out of two bits of plastic held together with a single screw, with even the battery door on a plastic hinge...ugh

The innards sort of match the outside, with a cheapo fibre PCB. Surprising use of metal film resistors for the time though, even if they are 5%. The transistors appear to be Japanese types with flat chamfered cases, which may be a giveaway as to where this was actually made.

Note the (single) socketed BBD inner the switch-this unit generates a chorus by modulating a single BBD with a single LFO against the source signal, which is a far cry from the RS09's ensemble. Also note the (thankfully obsolete) American-style minijack power connector.

The back of the PCB-it appears someone has tried to repair the battery snap, and ripped traces off the board, requiring link wires to keep it working. Blechhhh....

At first sight I thought the socketed SAD512D and the appearance of rubbing on it meant that the unit had been repaired and a dodgy "reconditioned" part had been substituted. Looking at other units, it appears that this is standard ex-factory on both counts - as is the messy soldering of the pots.

Well, it works OK, but I would never use this seriously in music, so the chip is a good candidate for the RS09. That leaves the chorus pedal as a perfect testbed for other chips, even cheapo Chinese clones....hmmmmm

(to be continued)

Zing went the strings

As mentioned before, I have a Roland RS09. Recently I took advantage of an online sale to add a Behringer VC340 to my collection. It seems to be the universal opinion that the VC340 *is* the reincarnation of the Roland Vocoder Plus version 2, and I'm very happy with it. I realised I now own two of the "big three" 1978-9 stringer designs by Roland, the other being the now-ridiculously-overpriced Paraphonic 505. These broke the mould of triple chorus "Solina wannabes" set bv the earlier RS202 stringer, introducing Roland's own four-chorus take on the ensemble effect. So putting aside the considerable differences in other aspects of their designs, what are their differences as stringers? No-one seems to have done a direct comparison of all three, there is misinformation around, and some comparisons have an almost comical confirmation bias towards expensive purchases, so here goes.

Note 1: all schematics are the 1979 "rocker tab" versions which all used SAD512 BBDs in the ensemble circuit. These chips are now very expensive to replace if they die!

Note 2: Sustain in the schematics is actually "release" on the panels and in modern parlance. 

Mode I means that pressing a key while another is held will not cut off any note tails that are still sounding, but pressing a key after all others have been released will cancel all sounding notes. This is useful for vocoders or paraphony where you do not want note smearing when the VCF envelope (or the speech signal) re-articulates. 

Mode II means that note tails will not get interrupted by new notes, and is best for pretty much anything else.

Note 3: Ensemble Mode I: animated from slow triangle LFOs and fast sine LFOs both being applied (think classic stringer)

Ensemble Mode II: slow sweeps from just the slow triangle LFOs (think goth icescapes like the Cure's Funeral Party)

Ensemble Mode III: Mode I but with feedback loops on the right and left chorus groups, meant to emulate rotary speakers but sounds more flanger-ish

Parameter	RS09 Organ/Strings mk1	Vocoder Plus mk1	Paraphonic 505
Schematic date	Dec 78	Sep 79	Sep 78
US Price 1979	$795	$2695	$1695
Keys	44	49	49
Keysplit	none	upper/lower	upper/lower
Master Oscillator transpose	Octave down switch, fixed /2 divider	variable bend slider, range adjustable up to an octave down	variable bend slider, range adjustable up to an octave down
Master Oscillator divider ranges	8',4',2',1'	8',4'	8',4' 16' (bass only)
Ensemble modes	I,II	I	I,II,III
Ensemble LFO rates	6.7 sec, 5.5 sec 210ms, 170ms	3.9 sec, 4.6 sec 150ms, 175ms	6 sec, 7 sec 160ms, 180ms
Ensemble external inputs	Yes	No, vocoder carrier only	Yes
String assign to Ensemble	switchable on/off	fixed on	fixed on
String footages and waveshapes	4'- 3-bit saw 8'- 4-bit saw summed from divided square waves, "log curve" global waveshaping	4' - spaced sharktooth circuit on each key	4', 8' - spaced sharktooth circuit on each key
String filters	high on 4' tabswitch low on 8' tabswitch	global	high on upper tabswitch, low on lower tabswitch
String footage selection	two tabswitches	none	4'<-> 8' cross-fader into each filter
String LFO vibrato	yes	no, vocoder and voice vibrato modulated via SAD1024 BBD delay	yes
Release modes	I,II (panel switch)	I (fixed)	I,II (Mode II only via footswitch)
Polyphonic Release control	yes, release start level, foot switch control	yes, release start level, no foot switch	yes, true variable release time, foot switch control
Paraphonic attack control	yes, strings only	yes, independent for strings and voice	yes, strings only
Tone control	yes, pre-main output, common with organ	yes, individual pre-chorus	yes, post-chorus, on back panel
Synth connections	gatebus out, "raw organ" (unfiltered) out	No	gatebus out, trigger out (retriggers on additional keypresses)
Phones out	yes, stereo, master fade	yes, mono, separate fade, transformer isolated	none

As can be seen, there is a lot of similarity between the RS09 and its (then and now) expensive siblings, even surpassing them in some respects. All three can effectively layer other sections to "thicken up" the strings' low end-the RS09 has organ, the Vocoder has 8' human voice, and the RS505 has an actual mono bass section. Interesting also to note that the prices are for sure a reflection of a time when electronics cost far more than the metal-and-woodwork that housed it, as the lightweight, utilitarian RS09 was not that much cheaper than the deluxe wood-cased RS505.

Also interesting to note that both the most expensive and cheapest of these three survived into the 80s in a revised version with touch switch controls and updated Panasonic MN BBD chips, and the RS505 didn't. Maybe on the eve of affordable true polysynths (the Jupiter 4 was out at the same time for $2795), the RS505, with its "synth" section with just a single VCF that didn't track the keyboard and a single VCA, simply turned out to be neither fish nor fowl. 

Moreover, for about the same price people could have had an RS09 and an SH09 synth combo, which for roughly the same price could either provide strings and a mono bass/lead with filter tracking and actual portamento, or when patched together, paraphonic filter and VCA articulation. 

The RS09's low price earned it thousands of sales, a place on all three of the Cure's "dark trilogy" of 17 Seconds, Faith and Pornography, and even a failed, market-misreading spinoff in the Saturn 09 organ (ah Roland, you've done it again!). The fact that layering the organ onto the strings along with Mode II ensemble gives a passable synth-like pad sound would not have hurt the RS09's sales, either.

As for the Vocoder Plus, the string section was obviously a cheap and easy add-on to the insanely complex vocoder that, along with its unique analogue voice synth (hey, why not add another 7 bandpass filters while we're at it?), gave it the ability to replace the Mellotrons of the rich and famous on tour, and earning it a place in the rigs of people who never even used the vocoder, like Tony Banks. The vocoder does feature on 80s hits both groundbreaking (O Superman and Behind the Mask) and timelessly cheesy (Mr Roboto).

To close with a bit of subjectivity, if you can get a good deal on it, the Behringer VC340 is well worth the asking price, and better value new than any of the above after 40 years of use, as well as adding DIN and USB midi control. It sounds "off" through a mono signal chain, but godlike if you can keep in in stereo and keep levels well down to avoid overloading artifacts.

Maxon Payne 2

Well, I tested my original hypothesis of how to fix the Maxon Autofilter tonight, with a roll of duct tape from a $2 shop. Would layers of duct tape on the overly-soft sponge buffer between the miniature switch and the pedal movement get the Maxon Autofilter switching again? Time to cut them up and see!

I first tried applying a 6-layer thick square onto the pedal foam to add enough contact pressure to switch the pedal. No dice. Then I added a further 4 layers of duct tape.

Success!! The pedal switched over - just- at the bottom of its travel. Screwed the switch in, clicked it a few times-then it stopped clicking. Time for more tape. I added another 9 layers, when it started working, then another three more for good measure. It worked, but again, it eventually stopped changing over the switch as the foam block "adapted". Argh.

Another 8 layers of duct tape later, the pedal is now changing over nicely. As I predicted in my earlier post, the foot pedal movement barely moves far enough to disengage the pedal (the deepest travel of the switch) so alignment is critical. I think I will be keeping the pedal in the engaged mode when not in use to minimise switch pressure on the foam.

And of course with all of this insertion and removal of the switch, one of the flimsy wires to the switch carrier board snapped! Murphy's Law in action, and time for more soldering.....

The Joy of Forgetting

About a year ago, I was browsing a pawn shop when I saw a guitar pedal that immediately drew my attention for a few reasons. It was big, it was colourful, it was made by Electroharmonix (who have a legendary rep in their own right and also employed former EMS VCS3 designer David Cockerell in the 1970s), and it had the word synth in it!!!

Although it appeared a bit beaten-up and well-used, it seemed to be in one piece, so I asked the staff what the price was. $29. Instant buy.

The original pedal was indeed designed by David Cockerell in 1978, and was very interesting. It employed an obscure RCA transconducting op-amp (CA3094) extensively, and had some innovative ideas. For example, the "subharmonic" generator was actually a chopper timed by what the pedal's circuitry saw as the fundamental harmonic waveform. The chopper signal silenced the original signal during every second cycle for a subharmonic with traces of the original harmonic structure. Its square wave output (a pre-filtered saturation distortion) had the input signal's amplitude envelope re-imposed on it. The octave output was not time-based, but actually a full-wave rectifier with high-pass filtering. There was a threshold-triggered variable attack generator controlling a VCA.

Icing on the cake, it had an 18dB lowpass VCF with resonance whose control source defied both received guitar pedal wisdom (from the 70s until now, preferred logic is that filters should follow the input signal amplitude envelope ie Moogerfooger LPF and Maxon Autofilter) and received synth-head wisdom (filters should be controlled by an envelope generator with defined rise then fall times ie Moogerfooger MuRF). Instead, input signal threshold triggers a linear ramp generator with arbitrary start and stop levels, and a time control.

So, I got it home and turned it on. All seemed to work. The slider travel seemed a little rough but there was no noise or dropouts when I moved them. However, there was no filter motion at all-no matter what the settings, the filter was stuck tracking the stop level slider. Oh well, maybe it broke and that's why it was hocked for so little? Let's take a look!

The board inside is (nearly) all surface mount, with the prominent exception of tantalum caps (Why tho????), and looked very different from the 1978 schematic-and not just due to the absence of CA3094s. The board did not seem to be very logically laid out, either. I put it aside to work on other things.

Yesterday I found that a brave soul (Bernard d'Uur) had actually made a trace of the schematic for this revised modern reissue. Yay! A few things were apparent. Not just the OTAs switching to 13700s, but much of the circuit had been revised. Gone were the nice dual 12V power supplies, now there was a cheap and nasty op-amp rail-split of the 9V signal to give virtual ground at 4.5V (c'mon EHX, even hobbyists know to use the LT1054 charge pump IC to get dual 9V supplies!), and the circuits appeared to have been cheapened a lot-gone were the zener references and a lot of the OTAs, replaced by generic transistors, generic signal diodes and resistors in paralleled configurations. In short it looked like a design that had the hallmarks of a synth designer (whose design language was typical late-1970s) had been re-imagined by guitar pedal guys (whose design language often seems stuck in 1967!). Still, the filter circuit seemed intact, despite the current design cheaping out on ceramic caps rather than film.

So, I got to work. Still working on the assumption that the pedal had broken after purchase, I suspected that one of the many signal vias (courtesy of the haphazard-seeming layout) was to blame. With expansion of the many fibreglass layers in a PCB, signal via traces running between board layers can be vulnerable to failure, which is why they they are often filled with epoxy, solder (better), or soldered solid wire contacting the top and bottom pads of the board (NASA standard!) to guarantee longevity.

I noticed that there were empty part positions, largely to accommodate through-hole versions of parts where no SMT equivalent was available at time of design, such as JFETs and some capacitors, and set to getting the board out of the case.

At this point it was clear that it was not designed for field servicing, as the jack sockets would not allow the board to be easily removed. I eventually pried each side by increments with a small screwdriver, and it was free.

Looking at the front, it was clear how dirty the board was from use, probably due to a total absence of dust filters on the slider runs (cmon, EHX!).

I returned to Bernard's schematic, and decided to start looking around the ramp generator, starting from the timing capacitor C6. Ummm, what timing capacitor????

The filter ramp timing cap had never been fitted at the factory!!! This explained why the filter was static alright, and it must have been the case that the guitarist who bought this shrugged their shoulders and just used it as a glorified fuzzbox. Still, it begs the question of what sort of quality control something that costs $A600 new deserves-there were no quality control stickers on the board or inside the unit.....

The solution was a nice fat 10u tantalum cap scrounged from a prototype I had to hand and soldered in seconds. First switch on-yay!

The 18dB filter squelched beautifully when animated by the ramp, and with ye old Concertmate drum machine I was in heaven.

Conclusions: Quality control matters. Having any parts such as tantalum caps that can be assembled by machines, assembled by machines, is A Good Idea. Not leaving multiple empty part positions on your boards so that hand assemblers don't get confused when visually checking the parts that should be there, is A Good Idea. Designing your PCBs so the layout is logical and aids service, like Roland analogue synths, is A Good Idea. And when troubleshooting, don't ignore the obvious, assume cost equals quality, or discount human stupidity :)

Hey Mr. Tampurine Man

You never know what you find in an "uncurated" antiques shop. In this case, something that looked like a vintage answering machine that had been hacked to be an analogue preset rhythm box-for about 50 bucks after negotiation with the seller. Woo-hoo!!!

Getting it home, it sounded very much like a poor man's DR-55-claves, open and closed high hats, snare and kick and that's it, but all voices were nice and punchy.

On the back was a label which made it appear that it had been part of the holdings of an educational institution-presumably in the US given the date format from when it entered inventory in 1980 -  T. something College of the Arts?

The label is interesting for a couple of reasons-firstly, it dates it from the brief period when analogue synths were moving into the consumer space before the Casiotone onslaught, and indeed the quirky design brings to mind Paul Schreiber's recollections of how he was told to design the Moog Concertmate for Radio Shack around the same time. For example, they required all line outputs be on phono jacks for connection to domestic hifis-as is the case here. It has a 2.5mm start/stop footswitch input that screams "dictation machine"-and indeed it was sold bundled with a plain-looking footswitch that just reeked of typing pool anti-chic, and was probably excess stock from a Realistic dictaphone.

Secondly is what appears to be the college's intended use of it-a "tamura". In South Asian music, an electronic tampura or tanpura box is an electro-acoustic accompanist for live acoustic musicians (it was originally invented in India in 1979). When I got the Realistic home and turned it on I was surprised-the built-in speaker is loud-really loud, easily sufficient to accompany a small acoustic performance. In these days, when we are conditioned to expect inbuilt speakers in portable instruments are usually toys or at best sufficient for personal practice, this came as quite a surprise.

There was also something else that was really loud-a continuous tone that would not stop unless the input power voltage was dropped to 4.5V. Fortunately the schematic was not hard to find online. It appeared that the "tuned" voices (bass drum, claves and snare) were produced by the "pinging" of four pole single-transistor hipass filters generating tones from decaying feedback from the collector, and the emitter of each transistor is connected to a trimmer which sets the gain on the collector resistor and therefore controls feedback. Could it be that one of the voice trimmers was out of alignment and feeding back too much so that the sound never decayed?

Time to open it up!

The biggest surprise is that the speaker takes up almost half the space inside the instrument. The rhythms are driven by a Sanyo preset rhythm generator chip called the LM8972 (bottom left of the main board). Many other devices used this, and many of them used a remarkably similar circuit to this one, but in stomp box format (such as the Sound Master Rhythm 1). Transistors were all the bog-standard Japanese NPN 2SC1815 except for the reverse-biased noise generator, which was a 2SC1923, presumably because it was noisier! Note the big red inductor-a single transistor LC cymbal noise filter is a big part of the DR55's sound, and a similar circuit technique is present here too.

Any way, to work. After all the effort it was almost anticlimactic. A slight twist on one of the trimmers and the droning stopped, with the snare now decaying nicely!

I tried turning the noise trimmer, and it did vary the level of the noise source pretty effectively.

So to wrap up, it's a fun instrument with a powerful voice that is hamstrung by its limited rhythm selection. Also its primitive RC power bypassing puts the sensitive analogue circuitry at the mercy of interference from the digital circuitry, so that even when stopped a very faint ticking is audible from clock interference "pinging" the clave circuit. Nonetheless, it has great potential as a sound module. Adding trigger inputs would be easy as it already has pulse extractors on all voices, likewise adding individual outs, with plenty of room in that huge case.

Another thought-remember those pedals based on the same circuit we mentioned earlier?  Thanks to prominent Youtubers featuring them, they are now quite expensive, although they offer very little of the customisation potential of this unit due to their cramped size. This one on the other hand seems to have slipped under the collectors' radar courtesy of its odd appearance and the Realistic brand name and thus is undervalued accordingly. But then again, so was the Moog Concertmate synth once upon a time, for exactly the same reasons...