At one time there was Tritronics which I thought was spin off from Vinci Labs or vise versa.

Today I was told that Jade Design helped engineer and build products from Krell, Sunfire and Emotiva [ direct sales division of JD ].

Just a question I've always pondered. I also wonder how many firms are writing the DSP coding or is it just a "hired gun". It was once thought that Meridian did the best DSP programming and it was all in house.

Is it like the auto biz such as a GMC Yukon is a Chevy Surburban.

In short who builds and designs the products we all sell and or crave?

Neal Feay inc. does alot of industrial design

Cullen Circuits is an amplifier build-house

ATI oems quite a bit of amplifiers for several companies

I'm sure that others can chime in as well.

Dan

Dan thanks for throwing me a bone here!

Cullen Circuit is AKA Wyred4Sound made many PS AUDIO items.

I have been told that Jade Design [ tried googling but not many audio hit ] is the parent company of Emotiva and makes products for Krell and Sunfire. Some say the Emotiva prepro is a Sunfire Grand pre pro.

At one time I think Vinci Labs made the platform for Parasound and NHT. It would be interesting to get some information about this.

A few years ago one of the founding DSP engineers of Tritronics/Vinci Labs showed me and an associate a prototype pre pro platform. Very interesting meeting we actually were giving some input on what features it should have. The next day at CES he was going door to door trying to get someone to bite. As I recall he designed the Parasound C1/2. He told us that it would take $1,000,000.00 to bring a product to market. This was around 2003-2004.

I said how much is design and DSP work. He claimed more than 50%. I said OK if you are in for that then my associate and I might come up the other 50% [ $250K each ]. Well he declined of course.

I just find this interesting to see what is really under the hood.

A McIntosh high up told me 3 yrs ago when Mac was coming out with the two piece rig that never made it to market that there were only 5 top end DSP developers for the Motorola and TI DSP chip sets. They sunk $250K in engineering, design and software and had nothing to show for it...

A McIntosh high up told me 3 yrs ago when Mac was coming out with the two piece rig that never made it to market that there were only 5 top end DSP developers for the Motorola and TI DSP chip sets. They sunk $250K in engineering, design and software and had nothing to show for it...
In my old job I had a team that ported and optimized signal processing algorithms in DSPs and other platforms like it. And we received many design wins for them in CE products.

It is true that few people can do a good job here. The problem is not lack of DSP programming skills. There are thousands of people with skills in the above platforms. The problem is finding someone that knows programming plus knowledge of audio/video algorithms. The latter is super important. For example, when we first tried to port WMA audio codec to an embedded platform, we were give 1/10th the memory space and 1/4 of the CPU performance of our most optimized implementation!

We got the CPU load there quickly but how on earth do you get 1/10th the memory usage? I remember we needed more than that just to perform the basic transform. Well, if you understand signal processing, then you can also figure out equiv. ways of making the same operation happen. That a months of burning the midnight oil got us there eventually :).

As to your original question, these are the scenarios in play:

1. The chip company wants to differentiate its product so goes out and gets as many algorithms ported to their chip as possible and offers them (usually) free of charge to their customers. This is the most common way people get MP3, DD, etc these days. Essentially all mass market products fall in this category. They are simply repackaging what is shipped to them from the core platform provider.

2. The technology company (who invented something new) goes and pays either the chip company or a third-party (e.g. Berkeley Designs) to port the algorithm to target DSPs used in A/V product.

3. A big customer (top 5 CE company) wants certain new feature. That gets the chip company interested to do the port for them and then the rest of the industry gets a free ride (usually after some exclusivity). This is a good scenario as then the chip company's resources are applied with excellent knowledge and ability to reuse existing building blocks.

4. In rare situations, the CE company itself develops the algorithm. This rarely happens as the above measures are usually in play. The work is done as to create additional differentiation. Example is the one you gave: Merdian. As you noted, this is expensive, and risky as there is no assurance that your folks know more that say, the technology creators.

In our case, we did some of the ports, and others did the rest. I don't recall any of our OEM customers doing any of the work though. All in all, there are hundreds of millions of devices with our technology in it (excluding the billion+ PCs). Sitting here, it is hard to imagine that outcome, sitting through our first chip company meeting with the guy telling me no one would have any interest in our technology in CE world :). To give you some context, I had 20 to 25 people doing this work full time. So it cost quite a bit to get there...

Amir as always a very detailed post.

I still think it would be interesting [ if not condeming in some circles ] if there was list something like this...

MIKE'S AV DESIGN INTERNATIONAL built and designed the following platforms:


ABC Company Model 123 --- XYZ Company Model 999 [ if listed on the same line it indicates units are similar if not identical "internally" ]

ZULA DELTA Model XRAY

BRAVO AUDIO Model TACT Killer


etc. etc. etc.

Again I have more time on my hands than I should.

Your information about Jade Labs, Sunfire, Emotive, etc is correct.

Vinci Labs made surround processors for a number of companies. Parasound, Classe, NHT, Halcro, and others. The vast majority of surround processors made by third parties, all come from a total of about four companies. In other words, very few of the small high end companies make their own surround processors. Almost all of them, buy them from elsewhere.

With regards to DSP platforms in surround processors. I think the main reason that there aren't that many people specializing in this, is that there is no money to be made. The entire yearly sales volume of every single surround processor sold by a mid or high end company, in the world, is probably about the same number as ATI sells of one video card model in the PC world. If you are a DSP programmer, where are you going to spend your time working?

If you are a DSP programmer, where are you going to spend your time working?Probably cell phones.