baktra

The CAS2Audio 1.0.3 has been released. A new option to invert polarity of the pulses was introduced List of 12 most recent tape images allows easier re-selection UI has been slightly refined It is a minor update at best. Since the app is distributed as a loose apk package, do not forget to read the installation instructions. If you prefer simplicity, the CAS2Audio can be a viable Atari-only alternative to the TapDancer app. I will have to make a decision on how to continue with this app. If it is to prosper, then I would need to review the Android app development best practices and implement them in the app. Not sure if it is something I am interested in at the moment.

Time to tickle the CAS2Audio android app. Though its architecture is dated, I am adding an option to reverse polarity of the signal, and most importantly, list of 12 recently selected tape images for better usability. This will be in 1.0.3 soon. Then a well-deserved refactoring, perhaps.

The 9.2.3 is out. Bugs in the Turbo 2600 support have been fixed (lack of automatic program title generation when converting files using Wizard/Batch/CLI, too many transfer speeds offered by the Wizard) The LBE format provided by the Standard plugin was refined as described in the previous post The Report tab was removed from the Tape Image Extractor There is a new "International" flag. The change is purely esthetical, not driven by any politics.

For the next adventure, I've chosen some leftovers. When you create a Turbo 2600 playlist item with Wizard/Batch/GENCAS, you do not get any automatically generated program title. This needs fixing. The LBE file format of the Standard Plugin. I've decided to streamline the file format. Instead of having the file metadata (program title, colors, soundr, crsinh and others) in the first 40 bytes of the first long block, the metadata will be in the LBE loader itself. That makes the long blocks unified and streamlined. And as a bonus, the program title is displayed right when the LBE loader is loaded, and lasts for at least 5 seconds. This is also more friendly to dual track tapes. The Report tab of the Tape Image Extractor tool is useless and will be removed.

It seems it was so awesome, that it had to be replaced quickly. It remains just a curiosity, nothing else. Thanks for the drill-down, though. An interesting reading.

It is more complicated than that. The Atari computer expects and requires a digital signal from the data recorder. The Atari data recorder (which is a specialized device) outputs a digital signal, because it demodulates the analog signal recorded on the cassette and converts it to a digital signal. The conversion is performed by a specialized circuit called FSK demodulator, present in the data recorder. The ordinary cassette recorders (like those sold by Radio Shack) do not have the FSK demodulator. These output is an analog signal to their output jack. So a simple connector adjustment wouldn't suffice. To connect an ordinary cassette recorder, there are two options (none of them easy) An external FSK demodulator. Basically a box connected between the ordinary cassette recorder and the Atari computer. Schematics can be found. I believe one Polish Atari enthusiast nicknamed Zaxon is or was selling some. Hardware hack of the Atari data recorder. Adding an extra JACK connector to the Atari data recorder and using its own FSK demodulator to connect to an ordinary data recorder. Soldering and knowledge of electronics is required. If we forget the ordinary cassette recorder at all, then at least rudimentary data recorder emulation is provided by devices like SIO2PC, SDrive Max and others.

Not sure what you mean. The .cas file holds a hybrid binary load file that is heavily compressed. The file is prefixed with a tiny hybrid binary file loader. Therefore, you have a lot of 'turbo' already included. If I understand the video, it is demonstrating FujiNet emulating a data recorder, booting from tape.

In the end, I changed the codename to the name of the Polish city, where the Turbo 2600 was born.

TURGEN 9.2.2 - "Swiebodzin" has been released. The Polish Turbo 2600 is now supported by the Standard Plus plugin. When using Turbo 2600, remember the following: With non-modified data recorder, you can achieve around 900 bps. For higher transfer speeds, you would need to have a special interface. The XL/XE computer with 64 KB RAM is required. The software resides in the RAM under ROM; binary files using the addressing range during the loading process might destroy the loader Binary files with INIT segments are supported. However, the motor is not stopped. Use the silence lists to mitigate the problem. Binary files tampering with MMU through PORTB can accidentally ruin the loading, if the tampering enables the OS ROM.

With the latest commits, TURGEN converts binary files withe the same efficiency as the Trans File DB utiility for Turbo 2600. So what is left? Support for the Wizard and documentation updates. Then 9.2.2 can be released. The code name of the release will be "Shock", tribute to the authors of Turbo 2600 I shamelesly stolen the loader from. To redeem myself, I will make a pligrimage to Swiebodzin, the town where the company resided.

Finally, some success with the Turbo 2600. Conversion of segmented binary files can be done. You can try the attached .cas file created by TURGEN. It is self-contained, just boot with START+OPTION. It works with a non-modified data recorder. Two more quirks of the system: 1. Doesn't seem to be stopping the motor when an INIT segment is encountered. 2. The code of the loader lives partially in the RAM under ROM, so an XL/XE machine is required. The Turbo 2600 doesn't work with the pre-XL machines. Also binary load files using RAM under ROM during loading can fail to load. I will have to work on conversion efficiency. Currently, each binary file segment has its own data block (or more). The file format allows multiple segments in one block, provided a synchronization sequence separates them. battybuilders_t2k6.cas

The Turbo 2600 is offering some resistance. The overall format (generated by the Trans File DB utility, which transforms binary load files to the Turbo 2600 format) is the following: Stage 0, boot file, two blocks Stage 1, long, non-standard block Stage 2, long, non-standard block in the Turbo 2600 format. Stage 3, non-standard block in the Turbo 2600 format. Meta-data, such as file name, RUN address and possibly others (?) Binary load file in the Turbo 2600 format. The Stage 3 is the remaining a mystery, because it needs to be zapped for each binary load file individually, and I still don't know how exactly. But I will.

As the support for Polish Turbo 2600 is coming, the Standard Plus is becoming the first “multinational” plugin. Now the code supports up to three flags for a plugin.

Development continues with the ver_9.2.2 branch. The Wizard can now play Tennis.

How unfortunate 🙂900 bps is anyway at the limits of practicality, with recorders like 1010 or XL11. A backlog item for 9.2.2, then. Together with making the loader available for GENCAS in full mode, batch processing, and optionally Wizard for files.

The next system on the list is Turbo 2600. It gets to the top of the backlog, because it brings value to the users with non-modified data recorders.

You can also place __asm__ (" BRK ") right before the offending statement, setup Altirra to invoke the debugger whenever BRK is encountered and trace instruction by instruction to see what is different.

Tennis in version 9.2.1 https://sourceforge.net/p/turgen/blog/2023/08/turgen-921/ Some final words about the Tennis (Turbo SOFTWARE) loader. Very well executed idea of entertaining the impatient user waiting for a game to load. The loader doesn't seem to support binary files with INIT segments, however it is a generally accepted limitation in the world of Atari cassettes. I would be delighted to see a streamlined version of the loader without obfuscation and copy protection removed. Perhaps one day, dogdark and Willysoft will make it possible.

Al diablo con "las trampas"... I cannot ZAP the loader, so that the program name is displayed, instead of the "TURBO SOFTWARE" string. I've found two places where the string is. Zapping one has no effect (that is not unexpected), zapping the other one (obfuscated by EORing with $46) results in a crash. I know it was meant to be like this, but why don't give the user something extra... In any case, new update version is getting closer...

I believe I got it. The attached .cas file was generated by a development version of TURGEN. If someone can test it with real hardware? A few more details on the file format: 1. Stage 0 loader - tricky tape boot file, one block 2. Stage 1 loader - two standard blocks, one of them is EOF. 3. Protective raw FSK signal (mandatory) 4. Stage 2 loader - Non-standard, long block #1 5. Stage 3 loader - Non-standard, long block #2. This block holds a three-byte length-code at offset 0x0231. Three encoded decimal digits. 6. Stage 4 loader - Non-standard 134-byte blocks, numbered from 0x0000 7. Binary load file - Non-standard 134-byte blocks, numbered from 0x000C 8. Protective raw FSK signal (mandatory) Layout of the block (134 bytes in total): 1. 0x55,0x55 (calibration sequence) 2. 0xFC|0xFA|0xFE (block type - full, partial, eof) 3. Data bytes (user data, 128 bytes. If fewer valid bytes, then the last byte is the number of valid bytes) 4. 2 bytes block number (for re-reading/progress indicator. Numbering begins with 0x0C for the blocks of the binary load file) 5. SIO Checksum (standard SIO checksum) I've enhanced the Tape image extractor so that if more tape image chunks are selected, the data of all selected chunks is displayed, not just the data of the first one. This helped, when I needed to compare .cas files side by side. The tape image extractor also displays the AUX bytes of the "fsk" chunks, something that was missing. scorpions_tt.cas

Some progress to report on the Turbo Tennis loader support. TURGEN can generate the loader, so it boots and makes it to the stage where the pong game begins. Now I will move on to converting the binary load file to the 134-byte blocks. That shouldn't be too challenging.

I see some instances of LDA SKSTAT followed by AND #16, clearly direct access to the signal at DATA IN.

Some raw information about the "Turbo Tenis" loader. The general file structure is the following: Stage 0: Cassette boot file, using the partial block trick; 1 block Stage 1: Cassette file, 1 data block, 1 empty EOF block. These two are followed by a short raw fsk signal that is not a block, but without it, the loader will not start. Copy protection, calibration? Stage 2: Non-standard block, 588 bytes, non-standard block, 1019 bytes Stage 3: 11 non-standard, 134-byte blocks Then the binary load file follows, stored using non-standard 134-byte blocks. The structure of the blocks is almost the same as standard "C:" records. The two additional bytes before the checksum are used to store block number. The binary load file is not obfuscated (almost too good to be true). Also, the IRGs are rather short, less than 100 ms. An interesting beast, for sure. I have converted two different files and compared the resulting .cas files to see how much one has to .ZAP the loader. I see only two bytes are different in one of the blocks that belong to the loader. Perhaps the total length of the file(?).

Yes, that one https://github.com/a8dogdark/atari_turbo_software

There is at least some progress with TURGEN. I've updated the GENCAS command line tool so that it supports the -o:audio option. As the option suggests, you can use GENCAS to send the signal to the sound card. Added mostly for orthogonality and completeness. I've made some progress deciphering the Tennis loader, so I might fulfill the plan to support it in the Standard Plus plugin.