Benedikt wrote:

I have added an additional resistor pack to the FTL Sound Adapter replica for short circuit prevention. […]
Show full quote

I have added an additional resistor pack to the FTL Sound Adapter replica for short circuit prevention.

FTL_Sound_Adapter_replica_3d.png

Without these additional resistors that are connected in series with the control lines, using the joystick port could potentially damage CMOS-based parallel ports. (Or anything else that uses push-pull outputs.)
Older TTL-based ports should be safe, either way, because they use pull-up resistors to create the high-level.

Does the joystick port actually work with Dungeon Master? The game supports it, and that is how the port was implemented (absent extra resistors), but does it actually allow the cursor to be controlled in the game? I would definitely buy one if it did.

i soldered a db9 connector to my covox some months ago and it works, but i did not add Resistors.

matze79 wrote:

i soldered a db9 connector to my covox some months ago and it works, but i did not add Resistors.

It is good to hear that it does indeed work. Most parallel ports will not need the additional resistors, but any port that connects a control line directly to +5V when it is switched to 1 might get damaged in their absence.
I believe that FTL renamed it to "future expansion port" for liability reasons. If someone fries their parallel port by plugging a joystick into a random "future expansion port", they cannot sue the company.
My intention is to write some sort of TSR driver, as well. That driver could then e.g. emulate cursor keys and space bar or something.

Change of scene: The SoundJr replica. It looks like I'm stuck again, because I cannot see through the resister packs. Everything else is complete.

Those three traces that go beneath the resistor pack on the front layer have to go somewhere and I'm not quite sure where that is.
They start at pin 1 of Q2, Q3 and Q4, respectively.

So, what you're saying is, we have to disassemble one, slightly, to complete a proper reconstruction?

Not further than has already been done. One would have to probe around with a multimeter and find connections with near-zero resistance. So there's no need to get out the soldering iron.
Once that is sorted out, only the value of the filter capacitors remains unknown, but that can be determined via trial and error, i.e. by comparing the frequency response.

Ah, ok. I should be able to do that, once I locate mine.

Alright. There is some progress. The traces that go below the resistor array on the front side of the board have been figured out. (Thanks, Cloudschatze!)

But there is one such trace on the back side, as well. Where could that go? Does it really connect to DAC_OUT?
And normally, one would connect one side of an R-2R ladder to GND, but I cannot find any such trace.
It should also be noted that the wiring between RN1 and RN2 is entirely speculative.
I connected the upper row of pins on RN1 to the upper row of pins on RN2, because it would make sense, but I cannot verify that, either. Maybe it is shifted on the original board.

Anyway, we are getting closer.

And we are almost there!
Yes, it was shifted a bit and no, this particular R-2R circuit apparently does not connect one end to GND.

Only one question remains: Are the resistors next to RN2 connected to DAC_OUT or to the bottom row pin next to it, i.e. DAC_OUT via 10kΩ resistor?

EDIT:

Cloudschatze wrote:

The circuit containing the resistors connects to the corner Pin 1, per your latest diagram.

Thanks! That means it is DAC_OUT via 10kΩ resistor.
(And it looks like my initial theory was mostly correct and we are indeed dealing with an 8+3 bit DAC. So forget my rambling about implicit volume control.)

And this is the 98mm×37mm PCB for production:

Now we'll have to wait.

So, this is what you get when you put them together? Interesting.

I'm very sorry I couldn't be more help with working out the design for the SoundJr. In the end the only thing I could do is get the original software. I'm glad you were able to determine it is an 8+3 system. I didn't think the original designer was stretching the truth. In the end it seems obvious that the SoundJr. would be a far better design to use for anyone that wants to build a covox/covox-like device since it can do everything a speech thing can do, plus more, as I originally surmised.

Paralel wrote:

In the end it seems obvious that the SoundJr. would be a far better design to use for anyone that wants to build a covox/covox-like device since it can do everything a speech thing can do, plus more, as I originally surmised.

It is indeed clearly superior. Unfortunately, it will also have between three and four times the BOM cost and it will take about ten times as long to assemble it by hand, at least without special equipment.

I do however plan to do some component substitution to reduce the cost:

• The Bourns 4816P-1-103LF / 4816P-1-203LF resistor packs are electrically and mechanically equivalent to those in the SoundJr and only cost half as much
• The M1MA152WKT1G diodes are practically identical to the DAN202K, except for a lower current rating which does not matter here
• A BCX51-10 transistor by Diodes Incorporated should be just as good as any other
• ON Semi MSA1162GT1G (PNP) and MSD601-RT1G (NPN) should be acceptable substitutes for ROHM Semi MMST2907A and MMST2222A, but I'd get more than twice as many for less than half the price and six out of eight are only for switching, anyway.
• Tantalum capacitors in a slightly smaller C package are cheaper than the ones in a D package, but should still fit and have suitable voltage ratings, nowadays

Combined with the savings from squeezing all three designs on one board, that should enable a reasonably appealing sub-€10 price point for a SoundJr DIY kit, provided that everything works out as expected.

Paralel wrote:

So, this is what you get when you put them together? Interesting.

I'm very sorry I couldn't be more help with working out the design for the SoundJr. In the end the only thing I could do is get the original software. I'm glad you were able to determine it is an 8+3 system. I didn't think the original designer was stretching the truth. In the end it seems obvious that the SoundJr. would be a far better design to use for anyone that wants to build a covox/covox-like device since it can do everything a speech thing can do, plus more, as I originally surmised.

Hm i don`t see any "+". except for using a Headphone without Amplifier.

matze79 wrote:

Hm i don`t see any "+". except for using a Headphone without Amplifier.

There's also the 8-level volume control, which can also be used to increase the dynamic range. (With an appropriate driver or player)

Benedikt wrote:

I believe that FTL renamed it to "future expansion port" for liability reasons. If someone fries their parallel port by plugging a joystick into a random "future expansion port", they cannot sue the company.
My intention is to write some sort of TSR driver, as well. That driver could then e.g. emulate cursor keys and space bar or something.

That definitely has a convincing ring to it.

matze79 wrote:

Paralel wrote:

So, this is what you get when you put them together? Interesting.

I'm very sorry I couldn't be more help with working out the design for the SoundJr. In the end the only thing I could do is get the original software. I'm glad you were able to determine it is an 8+3 system. I didn't think the original designer was stretching the truth. In the end it seems obvious that the SoundJr. would be a far better design to use for anyone that wants to build a covox/covox-like device since it can do everything a speech thing can do, plus more, as I originally surmised.

Hm i don`t see any "+". except for using a Headphone without Amplifier.

I would say being able to use headphones without an amplifier is a nice +, and double the frequency range of the Speech Thing, along with volume control. However, as Benedikt said, those features markedly increases the cost and complexity of the device. When the devices in question were contemporary, that would obviously be a considerable factor. These days, considering the price of the hardware, the difference is much smaller, so the benefit of using a clone of the Speech Thing, compared to a clone of the SoundJr. becomes much more debatable.

Benedikt wrote:

It is indeed clearly superior. Unfortunately, it will also have between three and four times the BOM cost and it will take abou […]
Show full quote

Paralel wrote:

In the end it seems obvious that the SoundJr. would be a far better design to use for anyone that wants to build a covox/covox-like device since it can do everything a speech thing can do, plus more, as I originally surmised.

It is indeed clearly superior. Unfortunately, it will also have between three and four times the BOM cost and it will take about ten times as long to assemble it by hand, at least without special equipment.

I do however plan to do some component substitution to reduce the cost:

• The Bourns 4816P-1-103LF / 4816P-1-203LF resistor packs are electrically and mechanically equivalent to those in the SoundJr and only cost half as much
• The M1MA152WKT1G diodes are practically identical to the DAN202K, except for a lower current rating which does not matter here
• A BCX51-10 transistor by Diodes Incorporated should be just as good as any other
• ON Semi MSA1162GT1G (PNP) and MSD601-RT1G (NPN) should be acceptable substitutes for ROHM Semi MMST2907A and MMST2222A, but I'd get more than twice as many for less than half the price and six out of eight are only for switching, anyway.
• Tantalum capacitors in a slightly smaller C package are cheaper than the ones in a D package, but should still fit and have suitable voltage ratings, nowadays

Combined with the savings from squeezing all three designs on one board, that should enable a reasonably appealing sub-€10 price point for a SoundJr DIY kit, provided that everything works out as expected.

Subbing in parts makes sense to me. I am sure, at the time, the given parts used made sense. However, parts price and availability change, especially over such a large span of time. As long as there is no material difference, I'd say go for whatever makes the most sense from a pricing perspective. Anyone expecting to get a parts perfect replica would be crazy. None of the Speech Thing clones are anywhere close to the originals as far as the parts used.

Headphone amps ? I once built one using a pair of EF95 electron tubes. 😁
http://www.elexs.de/ef954.htm (English Translation)

For mono, one single tube is enough.
Just note that the amp produces DC, so better add a transformer (or two, another one for the source).

The PCBs are now on their way.

Jo22 wrote:

Headphone amps ? I once built one using a pair of EF95 electron tubes. :D http://www.elexs.de/ef954.htm (English Translation) […]
Show full quote

Headphone amps ? I once built one using a pair of EF95 electron tubes. 😁
http://www.elexs.de/ef954.htm (English Translation)

For mono, one single tube is enough.
Just note that the amp produces DC, so better add a transformer (or two, another one for the source).

Mmh... I have a feeling that the tube amplifier would cost more than all three devices combined.

Benedikt wrote:

Mmh... I have a feeling that the tube amplifier would cost more than all three devices combined.

Hi, yes, commercial/professional tube amps are quire pricey, I believe.
For this homebrew project, all you basically need is a a pair of EF95 (aka 6AK5/5654/6SH1P/6J1/M8100).
These tubes cost around 5€ each and were originally made for space and aeronautics usage, afaik.
Or was it russian tanks (in case of 6Sh1P, a military tube) ? 😉

The boards have arrived:

The three parts were relatively easy to separate, because I put the separating line on the mask layer, this time, rather than on the silk-screen.
This creates an ever so shallow groove that is apparently sufficient to nicely guide a blade. The individual parts can then be broken apart with tongs.

Looks good to me. I look forward to hearing how it goes with the prototype.