Bielefeld devblog@posteo.de

Select your language

Startrack - what is it about?

tl;dr

Startrack was a hardware for Atari computers that I developed in the 90s. It was a professional audio card for the VME bus of these computers. There was software for recording audio, editing it, arranging it into tracks and then outputting it in very high quality via the audio card. The audio card worked on the Atari TT, but only more powerful hardware such as a Hades 060 or a Milan had the power required to work smoothly with many tracks simultaneously. This article documents the development of this hardware and software, prototypes, the variants in hardware and software, as well as the expansion with additional interfaces.

 

Startrack is a professional audio card for the VME bus of the Atari TT, as well as fast, compatible computers such as Hades 060 and Milan.


Back in 1994 I would have liked to have had an audio card for my newly acquired Medusa 040 computer. The Medusa 040 was a fast Atari ST clone with Motorola MC68040 CPU.
Unfortunately, I quickly found that such hardware did not exist in the quality that I imagined. There were only Soundblaster-like sound cards that couldn't do much and didn't sound particularly good.
So I started to construct and build such an audio card personally for me. The assembly was performed on a breadboard. The first question for me was which audio converter circuits i could use, which can also offer an excellent sound. So I screwed up my CD player and my DAT recorder, and looked at what was used in these devices. Crystal circuits were used in both HI-FI components. So I got the component catalog for Crystal audio converters, and then selected the right ADC and DAC circuits for my project.
At that time I had already gained some experience with the corresponding tools, which you needed for such a project. For measurements I had a 3 channel oscilloscope from Hameg, a (naturally) selfmade soldering station of ELV, as well as a multimeter. I knew PALs and GALs from Lattice Semi and had the appropriate, still relatively cheap hardware and software to program them with my own logic.
It quickly turned out that my Atari Mega STE had problems pushing the data quickly enough to the audio converter. Data packages were lost again and again. This happened especially when the computer had to do other tasks. Therefore, I used large, fast FIFO buffers, which could temporarily store the data for a short time. This solved the problem.

The 1994 Atari trade fair in Bonn was coming up. I took the prototype with me to the trade fair to find out, among other things, whether there was already software that could possibly be adapted for the audio card, and whether there was any interest in it among the players in the Atari scene. I had noticed a great interest at the MW Computer exhibition stand there, and the conversation was very good. In the course of the conversation, it quickly became clear that the audio card would be much more interesting if it had a DSP and compatibility with Atari Falcon. Here, a collaboration with MW computer was scheduled, which later also took over the sales of the audio card. In the period that followed, I started to extend the audio card with a DSP accordingly. For this, the Atari Falcon documentation, which I was provided with by MW Computer, was very helpful.

However, there was still a lot of work ahead of me: it should be the DSP56002 from Motorola, but it was difficult to get. I therefore built the prototype with an existing DSP56001. I bought this a year or two earlier at the 'Hobby and Electronics' trade fair in Stuttgart. At that time that was an excellent point of contact for exotic components. This was also available in a pin grid version, so it could be built on a breadboard. This would have been much more difficult as a SMD construction element. I also had to familiarize myself with the special assembler language. The DSP can execute several commands at the same time, but not in any combination. These commands had to be coordinated to get an optimal execution speed. Motorola had a small and cheap development system where the hardware is already fully built up and the hardware could be connected to a PC. This was very helpful to be able to test some small effect programs in advance.
All in all, it was a lot of work. When the hardware was as good as finished, the decision was made to turn the project into a 'product'.

A circuit board had to be developed, or actually there were 2 circuit boards. The DSP part was plugged onto the audio card. It was also fully functional without the DSP. However, it was not compatible with the Atari Falcon without the DSP, which required a DSP56001. Many programs for the Atari Falcon used the DSP, so that the audio card possibilities to run software was limited without DSP.
Therefore, the audio card was ultimately always sold with DSP. Due to the modular concept, other, more powerful DSPs could have been used later (that was the plan).

The DSP made the audio card much more expensive. But hey, you wanted a high-end product. You got that too. Some Falcon programs also ran on it, which would not have been possible without DSP.
The 4-layer boards were completely developed with the Atari ST software Platon, and also produced via Gerber export and control of the pick-and-place machines.
The program was really first class, there were no crashes or serious bugs. Only the production of the layers for the power supply was problematic. The insulation rings for the plated-through holes of the layers were missing, these were not created. Unfortunately, that was not noticeable in time. This made the first test production of 4 boards directly for the bin.
The second attempt was fine. I soldered the first board manually by hand. After all, I wanted to make sure that there were no hidden mistakes here.
There were also a few minor wiring errors, but, with a little copper lacquer wire, were easy to correct in a few places. Overall, however, it was very satisfying.
There were no serious mistakes here, even the DSP card worked perfectly. In view of the complexity of the circuit, this was an excellent result in my opinion.
The multiple testing of the circuit board layout clearly paid off here.

Board layout with Platon

 

Front side of the boards

 

Back side of the boards

 

Now the audio card in Hades 060 computer could be tested for the first time.

 

The Hades 060 Tower with the audio card (above), including the optional XLR connections (prototype)

 

Connection of the audio card with ribbon cable to the VME bus of the Hades 060 (here without DSP extension)

 

You can find all data on the project, including circuit board layouts and software, on my Github repository.

 

 

 

 

Log in to comment
Stephan Wilhelm replied the topic:
7 months 2 weeks ago
Das ist ein Test