Soundcard for Toshiba Satellite T1960CT

[SleepySheep]

Hey y'all. I've got this T1960CT working pretty flawlessly. It's a 486x with 4mb of ram running Windows 95 version 'a'. I'd really like to get some sound out of it (aside from the PC speaker which I am too hard of hearing to perceive anyway). It does not have a built in sound card but it does have a PCMCIA slot.

Are there any sound options for me? I was looking at the Digigram VXpocket 440 PCMCIA sound card on ebay, but I can't find anything in the way of system requirements for it. The other card that pops up a bunch is the sound blaster Audigy 2, but that requires 256mb ram and windows xp, so that seems like a no-go.

I'm out of my element looking around for this kind of hardware so any help is appreciated. Thanks!

 

[eswan]

Might look at the parallel port soundcards, like the opl3lpt. I remember the pcmcia adapters as working well in windows, but next to impossible to use with dos software.

 

[Zare]

Both. PCMCIA to have sounds in Windows if you wish, LPT soundcard for DOS games.

Also be sure to check driver support for PCMCIA, not what it claims or what you seen online to work, but an actual driver download.

 

[GiGaBiTe]

If you want sound in Windows 95, you can try the Windows 3.1 PC Speaker sound driver. It uses the 8253 programmable interval timer to bit bang the PC Speaker using PWM to play PCM audio. It'll give you 8/16 bit audio at a variable sample rate, usually between 11-16 kHz on the 486. The driver has two modes, the first where it locks up the system while it plays a sound sample uninterrupted. This isn't very useful. The second is allow the system to run while the sample is played, but you'll have pauses in playback when other system interrupts are in use, like the mouse. The other limitation is that no MIDI will work

It's better than nothing. I used it in the 90s when I couldn't afford a sound card.

Linux actually has a PC Speaker driver still (snd_pcsp) and it's way better than the old Windows 3.1 driver, if you have a decent quality motherboard. It allows for much higher sample rates, and it can play sounds uninterrupted. My X670E motherboard can do close to 30 kHz. It's great if you have a system on a test bench and can't be bothered messing with speakers.

 

[Svenska]

Are there any sound options for me?

There are four audio options:

• PCMCIA sound adapter
  • best choice for Windows applications
  • supports music and digital audio
  • limited compatibility with DOS games
• Parallel port DAC (Covox Speech Thing)
  • very easy to build yourself, just a few passive components
  • digital audio only (no music)
  • well-supported by some DOS games and MOD players
  • Windows drivers exist, but very high CPU load
• Parallel port OPL
  • somewhat easy to build, kits available
  • music only (no digital audio)
  • new design, so no out-of-the-box software compatibility
    • many DOS games can be tricked or patched
  • no Windows support?
• internal PC speaker
  • already available, very low quality
  • digital audio only (no music)
  • well-supported by many DOS games and MOD players
  • Windows drivers exist, but very high CPU load

 

[CommodoreZ]

• internal PC speaker
  • already available, very low quality
  • digital audio only (no music)
  • well-supported by many DOS games and MOD players
  • Windows drivers exist, but very high CPU load

I do this with a PC Speaker driver for Windows 95 on my T1960CT called SPEAKR.DRV, and it chugs through CPU cycles. It's supposedly better than the official MS one. Best you can get is the Microsoft Sound on startup, but beyond that, prepare for suffering.

I genuinely would love to find a proper PCMCIA option with Windows drivers but prepare to shell out some cash when they show up.

 

[GiGaBiTe]

Technically, both the OPL2 and OPL3 could output crude PCM audio by abusing the FM channels, but it was virtually never done because it required tons of CPU time to micro manage the audio chips. Game consoles could get away with stuff like that because they usually had a dedicated CPU like a Z80 for sound.

This is why the PC Speaker driver uses 100% of the CPU time, because it has to bit bang the program interval timer, which fires millions of system interrupts, one for every bit of audio having to be played. The system literally can't do anything else while this is happening, or the audio will cut out.

 

[Zare]

The Win3.1 driver was nearly useless, used as a curiosity to hear the chimes or play a clip from Sound Recorder. "That's a WAV file coming out from PC speaker!". I mean it can probably perform the task if you run it on something faster than 486, but that's what was out there at the time.
We also had a thread recently about playing PCM over speaker on 8088, seeking limits of what's possible with the finest timing and all CPU power.

This is why the PC Speaker driver uses 100% of the CPU time, because it has to bit bang the program interval timer, which fires millions of system interrupts, one for every bit of audio having to be played. The system literally can't do anything else while this is happening, or the audio will cut out.

Compaq SystemPro with 2x386 was available in 1989. In an alternate history where this was not rare and done couple of years before, maybe there would be software that would use one CPU exclusively as PCM generator.
Just a passing thought; but in googling I found AMD employee on Quora with 3.4 million answer points that claims no dual 386 existed. Again, beware of Quora, Stack Overflow, Reddit and consequently any LLM for these topics.

 

[GiGaBiTe]

The Win3.1 driver was nearly useless, used as a curiosity to hear the chimes or play a clip from Sound Recorder. "That's a WAV file coming out from PC speaker!". I mean it can probably perform the task if you run it on something faster than 486, but that's what was out there at the time.

On my 486 DX/2 66 (and later an Am5x86), the PC speaker driver was tolerable when used in interrupt mode. I used to play Windows 3.1 games with sound, and besides the regular tiny pauses in audio when the mouse was being moved around, the audio was fine. I much preferred it over having no sound at all.

All of this talk about PC Speaker sound makes me want to try it on my newly acquired PS/1. I need to fix the bricked BIOS first though, still waiting on an EPROM UV eraser so I can reprogram it.

Compaq SystemPro with 2x386 was available in 1989. In an alternate history where this was not rare and done couple of years before, maybe there would be software that would use one CPU exclusively as PCM generator.
Just a passing thought; but in googling I found AMD employee on Quora with 3.4 million answer points that claims no dual 386 existed. Again, beware of Quora, Stack Overflow, Reddit and consequently any LLM for these topics.

Not surprising the nonsense being spread, it is the internet after all.

Compaq I also believe had an AMP server that had a 486 for the main CPU and a 386 that was used as a slave I/O processor to speed up the system. It required some weird flavor of Unix or Windows NT 3.11/3.5x with a custom HAL driver to work properly.

As for using a second CPU on a SMP system for the PC Speaker driver, it could work if the slave CPU could service interrupts without pausing the master CPU. I don't know how the modern Linux PC Speaker driver works, but it can play sound without the hitching the Windows 3.1 driver did.

 

[Svenska]

The Win3.1 driver was nearly useless, used as a curiosity to hear the chimes or play a clip from Sound Recorder.

I've also had some success with SpkQQ, which did slightly better than Microsoft's driver. Old systems often had decent case speakers, so both volume and audio quality were not abysmal. It was definitely good enough for system sounds, even on slower systems.

However, this thread is about a notebook - and the speakers used in those are only good for beeps. PCM audio will be barely audible with any driver. Not usable.

Compaq SystemPro with 2x386 was available in 1989.

Was that a standard SMP system, or something custom? Because even an 8086 can share the bus with other masters, but it took until the 486 instruction set to get the necessary instructions for efficient multi-threading in SMP configurations (such as lockless data structures). Windows 95 / NT3 don't support full atomics for that reason (Windows 98 / NT4 officially required a 486).

Linux removed the 386 support mainly because its 80386 support was fundamentally incompatible with its SMP support. The user-space side was 486 anyway (the kernel pasted over the few differences in its "undefined instruction" handler), but the kernel-side was a different story and had bitrotted without testers.

I need to fix the bricked BIOS first though, still waiting on an EPROM UV eraser so I can reprogram it.

I've had some success with LEDs used to harden gel nails. Took about three days to erase an EPROM, but I had no issues reprogramming and using them since.

 

[Zare]

On my 486 DX/2 66 (and later an Am5x86), the PC speaker driver was tolerable when used in interrupt mode. I used to play Windows 3.1 games with sound, and besides the regular tiny pauses in audio when the mouse was being moved around, the audio was fine. I much preferred it over having no sound at all.

All of this talk about PC Speaker sound makes me want to try it on my newly acquired PS/1. I need to fix the bricked BIOS first though, still waiting on an EPROM UV eraser so I can reprogram it.

I can't remember running much games in 3.1, the only "normal" game I had was Pitfall The Mayan Adventure. I was quite hooked on Civ2 but I played it later, on 95.
The Olivetti PCS5130 of mine has the loudest most clear PC speaker I've ever heard. It has 5 volume levels in BIOS and 'normal' volume is somewhere between 1 and 2. On 5 it's not a PC speaker it's a PC blaster.

Gonna try some PIT stuff on it too, monotone tracker is a nice test. FWIW the entire machine is about sound, and I did not know it has an weird electrical quirk in that area, but it fits - the computer activity bleeds to the PC speaker. I can hear the HDD running but there is no HDD, only CF. It is all very quiet but clearly heard in a silent environment. A very low volume hum appears if a certain SVGA mode is engaged. The volume of these artifacts is not impacted by changing the master volume in BIOS.

Compaq I also believe had an AMP server that had a 486 for the main CPU and a 386 that was used as a slave I/O processor to speed up the system. It required some weird flavor of Unix or Windows NT 3.11/3.5x with a custom HAL driver to work properly.

As for using a second CPU on a SMP system for the PC Speaker driver, it could work if the slave CPU could service interrupts without pausing the master CPU. I don't know how the modern Linux PC Speaker driver works, but it can play sound without the hitching the Windows 3.1 driver did.

Was that a standard SMP system, or something custom? Because even an 8086 can share the bus with other masters, but it took until the 486 instruction set to get the necessary instructions for efficient multi-threading in SMP configurations (such as lockless data structures). Windows 95 / NT3 don't support full atomics for that reason (Windows 98 / NT4 officially required a 486).

GiGaBiTe I believe is referring to the same system, it is AMP and it was advertised for Windows NT.

Technically in an AMP system the second CPU for sound task can be treated as dedicated sound hardware. Any kind of MP platform starts with giving at least one shared resource to all CPUs, in a simplest form the code CPU can orchestrate the sound CPU which has its local data and is being serviced by PIT.

Btw in normal single CPU case, how about a PWM generator coroutine from a TSC based program loop on the Pentium? I can't get results for rdtsc overheads on the ancient P5 arch, but should be more efficient than interrupt based loop.

However, this thread is about a notebook - and the speakers used in those are only good for beeps. PCM audio will be barely audible with any driver. Not usable.

On-topic this is the right answer, OP should know that PC speaker digital audio is programatical movement of membrane thus you need something OK to begin with.

 

[Svenska]

I don't know how the modern Linux PC Speaker driver works, but it can play sound without the hitching the Windows 3.1 driver did.

Linux has a more efficient, lower-latency interrupt handling system than Windows 3.1. Interrupts are serviced partially in interrupt context (top-half), partially in kernel mode (bottom-half), allowing interrupts to be handled while other parts of the handler are still executing (within reason). This allows interrupts to fire more often, increasing the sample rate.

Additionally, the PC speaker driver simply runs in kernel mode, does not require additional context switching during playback, and has a high dynamic priority.

(Although Linux will start polling devices if their interrupt rate is too high for extended amounts of time. Trying to push 8+ Mbps through a 3c589 PCMCIA card will briefly freeze the system, at which point Linux uses a timer to get rid of the interrupt storm.)

 

[SleepySheep]

Well I wound up buying a Digigram VXpocket 440 PCMCIA in really good condition because I saw the driver has Win95 support, but I don't think the drivers can properly load with only 4mb of RAM even though they seem to install okay. Is there any PCMCIA sound card that can work on such a meager amount of memory?

I really did want a good Windows solution but I think this computer is specced more for the 3.1 era. Maybe I'll check out the OPL3LPT card and hopefully at least get some sound out of DOS.

 

[Svenska]

Maybe I'll check out the OPL3LPT card and hopefully at least get some sound out of DOS.

A new/updated Windows 3.1 AdLib driver for OPL2LPT/OPL3LPT is available at [https://github.com/andreiw/adlib21].
Windows 95 tends to support Windows 3.1 drivers, so maybe you can use it.

Is there any PCMCIA sound card that can work on such a meager amount of memory?

Most cards will work just fine. But user experience is bad (sound or no sound), no vendor wants to take the blame.

You shouldn't run Windows 95 with only 4 MB RAM. Try to find a memory expansion card for your system, or downgrade to Windows 3.1.