Turns out my soldering iron cleaner/holder is useful for more than one thing:

#lifehacks 😎

Good stuff 😀

By the way: All design files are now online!

• Speech Thing replica
• FTL Sound Adapter replica
• SoundJr replica

The community really owes you a major thanks for your work on reversing the SoundJr. Despite not being as well known, it's capabilities are significant. It is something that a lot more people would use if they were aware of its capabilities and it was easier to acquire (at least I'd like to think so).

One of these days I hope to get up the gumption to do a little disassembly on mine and nail down the values for those couple little bits were there was a little guess work.

Cloudschatze's photos and measurements were also very helpful. The SoundJr reverse engineering certainly would not have been possible without them.

The only components of which the value still remains unknown are the two 0805 ceramic capacitors on the back side of the board.
Unless you want to desolder and measure them, comparing the frequency response of the original device and the replica should be the most pragmatic approach.
The effect of the transistor substitution on distortion etc. would be interesting, as well.

But first things first: The next logical step is to add the proper short circuit protection resistors to the FTL Sound Adapter replicas.
Unfortunately, I lifted a pad here and there when I removed the Sn99Cu1 solder joints at 400 °C. It should all be repairable, though.

Follow-up post:

The replacement resistor packs went in in mid-November and by the looks of it I was able to fix the only seriously damaged pad, but final measurements and tests have been pending, since.
While I was at it, I also made another two Speech Things from surplus PCBs.
There are PCBs for another two FTL Sound Adapter replicas, as well, but I didn't have enough DE9 connectors.

The evaluation of the joystick ports on the FTL Sound Adapter replicas has been complicated further by the recent realization that my only suitable joystick appears to be broken.
If the appropriate pins on the adapter's DE9-connector are bridged with jumper wires, the software on the PC-side will however detect the change, accordingly, which means that the adapter itself is now indeed fine.

Evaluation of my only readily assembled SoundJr replica turned out to be surprisingly easy, though, because Modmaster in Covox Speech Thing mode does not touch the parallel port control port that the SoundJr uses for volume control.
You simply write a value to the control port register and start the player afterwards.

The following table lists all possible control port values and the respective volume levels:

1Ctrl 3..0 | Volume level
2----------|-------------
30000      | undefined (weaker 8)
40001      | undefined (weaker 6)
50010      | undefined (weaker 7)
60011      | undefined (weaker 5)
70100      | 8
80101      | 6
90110      | 7
100111      | 5
111000      | undefined (weaker 4)
121001      | undefined (weaker 2)
131010      | undefined (weaker 3)
141011      | muted, if data 1..0 = 0
151100      | 4
161101      | 2
171110      | 3
181111      | 1

It is derived from the schematic and appears to be accurate.

Thanks for the update, Benedikt.

Hello,

Benedikt wrote on 2019-10-06, 14:34:

By the way: All design files are now online! […]
Show full quote

By the way: All design files are now online!

• Speech Thing replica
• FTL Sound Adapter replica
• SoundJr replica

i've got some questions about the FTL:

- if it's a passive DAC, can it just work with the same drivers & softwares as the covox speech thing? Otherwise, which drivers are there & how can we make it work with DOS software ?
- does the serial port have full functionality ? can we put a mouse on it for example? If i put a midi interface on the serial port of the PC, i won't be able to use a mouse then, so this other serial port would come in handy 😀
- Do you have some left for sell?

In another thread, @dreamblaster said he was going to work on a CVX5 and asked for suggestions on what improvements to make over the earlier CVX4 design.

Adding support for these variants would be an interesting suggestion indeed. I also asked for Disney Sound Source and Stereo-on-One compatibility.

Would it be possible to combine support for all these (relatively) simple LPT DAC variants into a single device, ideally in a software-switchable way?

ultramars wrote on 2022-02-12, 17:16:

if it's a passive DAC, can it just work with the same drivers & softwares as the covox speech thing?

It is fully Speech Thing compatible. The only difference is the additional functionality, namely that it can be detected by software and that it has a joystick port.

ultramars wrote on 2022-02-12, 17:16:

does the serial port have full functionality ?

As mentioned above, there is no serial port. Assuming that DE9 equals RS232 serial can be dangerous with these old machines.
Serial mice and Amiga mice, for instance, both use the same DE9 plugs and sockets, but mixing them up can cause hardware damage.
The expansion port on the FTL Sound Adapter (and my clone) is for Atari-style joysticks, only.
Such a joystick can then control the mouse cursor in Dungeon Master, the game that originally included this adapter.

digger wrote on 2022-02-12, 21:46:

In another thread, @dreamblaster said he was going to work on a CVX5 and asked for suggestions on what improvements to make over the earlier CVX4 design.

Adding support for these variants would be an interesting suggestion indeed. I also asked for Disney Sound Source and Stereo-on-One compatibility.

Would it be possible to combine support for all these (relatively) simple LPT DAC variants into a single device, ideally in a software-switchable way?

The economically most sensible solution would be a small STM32 µC in a parallel port dongle with external power supply.
It could be programmed to imitate virtually all the other parallel port dongles.

Benedikt wrote on 2022-02-21, 18:45:

As mentioned above, there is no serial port. Assuming that DE9 equals RS232 serial can be dangerous with these old machines. Ser […]
Show full quote

As mentioned above, there is no serial port. Assuming that DE9 equals RS232 serial can be dangerous with these old machines.
Serial mice and Amiga mice, for instance, both use the same DE9 plugs and sockets, but mixing them up can cause hardware damage.
The expansion port on the FTL Sound Adapter (and my clone) is for Atari-style joysticks, only.
Such a joystick can then control the mouse cursor in Dungeon Master, the game that originally included this adapter.

Ah ok, i didn't get that.
thanx for this precision 😀

Benedikt wrote on 2022-02-21, 18:45:

The economically most sensible solution would be a small STM32 µC in a parallel port dongle with external power supply.
It could be programmed to imitate virtually all the other parallel port dongles.

Makes sense.

If combined with an OPL3 FM chip, perhaps the STM32 would even be powerful enough to emulate both OPL3LPT and Digispeech DS3xx, allowing the Sound Blaster emulator and Windows 3.x drivers for the Digispeech parallel port products to be used with this hypothetical universal LPT sound device. It would require some reverse-engineering of the bitstreams going between the drivers and an actual Digispeech device on the parallel port, but that seems feasible, right? 🙂

However, would it be smart to bypass the microcontroller logic when such a device would be running in unbuffered "dumb LPT DAC" mode? Otherwise perhaps some unnecessarily latency would be introduced?

digger wrote on 2022-02-22, 13:27:

If combined with an OPL3 FM chip, perhaps the STM32 would even be powerful enough to emulate both OPL3LPT and Digispeech DS3xx, allowing the Sound Blaster emulator and Windows 3.x drivers for the Digispeech parallel port products to be used with this hypothetical universal LPT sound device. It would require some reverse-engineering of the bitstreams going between the drivers and an actual Digispeech device on the parallel port, but that seems feasible, right? 🙂

However, would it be smart to bypass the microcontroller logic when such a device would be running in unbuffered "dumb LPT DAC" mode? Otherwise perhaps some unnecessarily latency would be introduced?

If you want to reverse engineer the Digispeech, I am not going to stop you.

I do not see any need to bypass the microcontroller when our hypothetical dongle is acting as a simple DAC, because the average STM32, as well as similar microcontrollers, runs at at least several 10 MHz and even has DMA functionality, whereas the audio signal has a sample rate of at most several 10 kHz.