Sunday, January 8, 2012

Realizing the Realizer

This post was inspired by Big City Music's announcement this week that they are offering a PPG Realizer for sale. The Realizer, as many synth enthusiasts know, was the legendary attempt by Wolfgang Palm's PPG company to pioneer the soft synth concept using 1980s technology. However, the development costs became the final straw for the financially struggling PPG, and led to the company liquidating itself in 1988. The Realizer never went into production; rumor has it that two prototypes were built. (If this is true, I don't know where the other one is; I've checked the listings of the Audities Foundation, the New England Synthesizer Museum, and the Eboard Museum. None of them list a Realizer, but that doesn't mean they don't have one in their collection somewhere.


Realizer control desk -- from ThePPGs.com

So what was the Realizer actually? Modern legend has it that it was the first virtual analog synthesizer. Actually, however, it was both more than that, and possibly less than that. According to Palm, the Realizer was an early attempt to build what we now call a "workstation"; it would have been capable of synthesis, multi-track recording, processing, and mixing. The unit you usually see in pictures, which PPG called the "control desk", is not the whole system; it's only the control unit and user interface. The control desk could interface with a combination of sound modules that actually did the audio processing, and hard disk units (HDUs) which provided audio and data storage, up to 8 units total. The sound modules and HDUs were intended to be somewhat mix-and-match with other PPG products, such as the Waveterm B.



The sound module was the piece of the Realizer that did most of the work. Like many of today's digital synths, it contained a bank of digital signal processor ICs -- in this case, eight of the TMS 32010 -- and a Motorola 68010 that managed everything. (The HDUs also contained a pair of 32010s, but it appears that they were not used in the Realizer configuration.) PPG developed three synthesis packages that ran on the sound modules. The one that everyone talks about in regard to the Realizer was the "Minimoog" virtual analog software that reproduced not only the features of the Mini, but also its panel layout. However, there was also application software for additive synthesis and for the sampling and wave scanning method of synthesis that PPG was know for. Further, according to Palm, the software allowed individual processing functions to be mixed and combined in the style of a modular synth. (You could say that PPG invented Reaktor in 1986!)


LinkPPG HDU with stand-alone control unit (in foreground). From Synthmuseum.com

Further, the Realizer also incorporated the functions of what we now call a digital audio workstation. It was capable of multi-track recording and editing using the HDUs as storage. It had "plug-ins" for adding common studio effects (although, according to Palm, the reverb software was troublesome and was never completed to a satisfactory degree). And it could do mixdowns, producing a digital master that could presumably be transferred directly to CD mastering systems, although it is not clear how that would have actually worked.

The control desk is the part of the Realizer that everyone has seen. It contains a monitor (which was to have been color in the production version, but the prototypes had green-phosphor monochrome), 31 knobs, 6 faders, a data entry wheel, a keypad with numeric and function keys, and a graphics tablet. Each of the knobs and faders had line graphics drawn on the panel from the control to the edge of the screen opening; the software generated lines on the screen that led from the screen edge to the parameters on the screen, and by that method, the user got a visual indication of which knob or fader controlled what on a given screen. The graphics pen would have been used for drawing, and probably also for selecting commands from menus, in the fashion of the Waveterm. The best photo I've been able to find that shows a screen and illustrates the layout and the knob graphics is the following, taken from Palm's old blog on Myspace. Palm describes this photo as being a photo of an early mock-up, which accounts for the crude-looking panel.




So what made the Realizer so expensive to develop, enough that it killed the company? One thing that Palm mentions is that in the timeframe when the Realizer (and presumably the Waveterm B also) was being designed, PPG had decided that laying out the circuit boards by hand, which was how they had done all of their previous products, wasn't going to work given the density of the boards they wanted to design. So they purchased an electrical CAD software package (which would have been very expensive in 1986), and then leased a computer system to run it. Unfortunately, apparently the leased computer didn't have sufficient performance, and the system ran very slowly, harming productivity. There probably would also have been a learning curve for the engineers. Palm also mentions, in his account of the Waveterm B development, that the Waveterm B and Realizer were PPG's first products to use 16-bit sampling, and that they had a hard time getting their 16-bit analog-to-digital converters working properly. (In order to build the initial sample library to be shipped with the Waveterm B, they hacked a Sony F1 digital tape system and used its converters.)

However, I'm guessing that the real killer was the software. Even though there was probably some commonality with PPG's other products, there would have been a ton of new software to be written. They had to write a lot of new software for the control desk displays and user interfaces; the Moog emulation and the additive synthesis was new, and they had implemented a lot of improvements to the sampling which required new software. (Plus, knowing how things were done back in the day, the 12-bit sample handling software used in the Wave keyboards probably did a lot of "tricks" with unused data bits, which would have had to be re-written to handle 16-bit samples.) The sound modules required a bunch of new software to manage all of the DSPs, not to mention the actual DSP processing code. Also: they were writing all of this in assembly language. Palm states that a C compiler was not available for their systems at the time -- a statement which puzzles me, since C compilers were readily available for most processor families by 1986, and Sun, to name one, had one for the 68000-series CPUs.

The final factor was that, due to a combination of circumstances, PPG found themselves without a lot of money to spend on R&D. In 1986 they had invested money in moving production to a larger facility, in part due to robust sales of the Wave, but at about the time the new factory opened, Wave sales began to decline. The Wave, especially by the time of the 2.3 revision, had very sophisticated capabilities for wave scanning and manipulation of samples, but a lot of users didn't care about that -- they only wanted basic sampling and playback, or just playback of canned sample libraries, and so they gravitated towards less expensive samplers like the Emulator or the Ensoniq Mirage. The Wave was PPG's bread-and-butter product, so when sales declined, the company's revenues suffered. A few Waveterm B's were sold, and a few HDUs were sold as stand-alone products, but the Realizer never reached production before Palm and the other founders realized that they were going to run out of money. They liquidated rather than continue and be forced into bankruptcy.

Going back to the sale by Big City Music, I don't know exactly what it is that they are selling. The ad shows only the control desk. Although that item no doubt has significant collection value by itself, the point remains that if the goal of a buyer is to get the system actually running again, it won't do anything without at least one sound module and one HDU. Perhaps Big City has these items and is including them in the sale; the ad copy doesn't say.

Below is a link to a 1987 demo, from Palm's Myspace page. The Realizer's control desk can be seen on the left for much of the video. (The device that the demonstrator is holding appears to be a stand-alone control for an HDU, and not part of the Realizer configuration.) Note that the demonstrator never touches the Realizer control desk, which suggests that the software was still not stable at this point.

http://mediaservices.myspace.com/services/media/embed.aspx/m=,mr=60628615,t=1,mt=video

Finishing up with a historical curiosity: The astute observer may have noticed that in the photo of the control desk, following the first paragraph of this post, the desk does not have the data entry wheel. I don't know if this implies that prototypes were built both with and without the data wheel, or if it was added to the pictured unit after the photo was taken.

No comments: