A Word of Caution about the LAPC-I

[Great Hierophant]

While obtaining an Roland LAPC-I Sound Card may seem like the greatest thing this side of the Adlib, the use of it can pose some challenges and problems:

Usage
The LAPC-I is an 8-bit ISA card.  ISA slots are not to be found in modern computers and are seldom seen in motherboards more advanced than the coppermine Pentium III boards.  The LAPC-I is best suited for a retro system, and using it is harder than simply connecting the proper voltages to the proper pins.  Using the card remotely (putting it in another PC to power it up), requires its MCB-1 breakout box (no ordinary midi cables work) and software to allow the MPU-401 controller to accept sysex commands through the external MCB-1 breakout box. 

Length
The LAPC-I is a 13" long (full length) ISA card.  It will not fit into every case, and many modern cases often have hard drive cages partially or fully obstructing the ability to insert the card.  While it is not as wide as the full-length AWE32s, it can be a very tough fit even with cases that do have the room.  Also, the card is only an 8-bit card and really should have some support on the back end, especially if the case is vertical.  You don't want the PCB to crack under the strain. 

Power Requirements
The LAPC-I is among the very few known ISA cards of any importance that utilizes the -5V signal on the ISA bus. The -5V line was originally put on the 8-bit ISA bus by IBM for its diskette drive adapter.  I do not know of any graphics or sound/midi interface ISA card other than the LAPC-I that uses this line.  Some implementations of the ISA bus, like the Tandy 1000 series, leave out the -5V line.  Some modern PSUs do not have a line for the -5V, because modern systems do not use it.  No interface beyond the 8-bit ISA bus ever used this line either.

Noise and Volume (not unique to the LAPC-I)
The external version of the LAPC-I's LA synthesis, the CM-32L and similar external modules, is shielded in metal to protect the analog circuity from electromagnetic interference.  The LAPC is connected inside a PC, which is a very electrically noisy environment and has no shielding.  Also unlike the external modules, the LAPC-I has no mechanism to control the volume, so you must rely on the volume control if any, from the speakers. 

Service (not unique to the LAPC-I)
If your LAPC-I breaks down and needs repair, you have few options.  The card is stuffed with components (resistors, capacitators, diodes and transistors) which are easy to break off if you handle the card roughly.  Roland will not service ISA devices because they got rid of their ISA testing hardware.  You may have to pay a pretty penny to find another working LAPC-I.  If the EPROM goes bad, then at least it can be replaced (if not from Roland.) 

Resources (not unique to the LAPC-I)
The canonical approach to the MPU-401 portion of the interface puts it at I/O 330, IRQ2.  This is necessary because most software expects the MPU-401 interface to be at that address and use that IRQ.  Other midi devices, including Sound Blasters, also prefer that address, if not the IRQ.  It can be very difficult to have two MPU-401 interfaces inside a PC and use them without having to fiddle around with the settings all the time.  Windows 9x doesn't work well with more than one Roland MPU-401 interface.  Also, it usually sets IRQ2 to the APCI interface, which can be difficult to disable.  I way to get around this is necessary if you wish to use the LAPC-I in Windows. 

Compatibility (not unique to the LAPC-I)
The LAPC is an 8-bit ISA sound card an expects to be run at the 8-bit bus speed, 4.77MHz.  16-bit ISA bus cards expect an 8.33MHz speed or something very close to it.  More modern motherboards do not have 8-bit ISA slots, the presence of which would guarantee that the card would be run at the slower bus speed.  Undoubtedly, some later 8-bit cards can run at either 4.77MHz or 8.33MHz, but the LAPC is not one of them.  Modern motherboards should be able to detect the presence of an 8-bit card in a 16-bit slot and adjust the speed for that slot accordingly.  If they don't, expect problems.

[HondaSiR]

Hmm...so to summarize everything...don't get an LAPC-I over a similar external module. Makes me all the more loving to my CM-500 now...LA/GM/GS/extra 33 sound FX all in just one wonderful box. Life surely is sweet. 

[Ari]

Well, I don't know if the title: "A word of Caution about the LAPC-I" is the most appropriate one.

The first LA synth I ever got was the LAPC-I. To get it to work for me was as simple as plug&play. I bought it about 6 years ago when all I had was a Pentium II 266, and it has been sitting there since when and operating without a hitch. One of the advantages of having an LAPC-I, is that you don't need to bother yourself with an interface card, you'll never get the "buffer overflow", you don't have to worry about power adapters if you buy the unit from the States, and, of course, it saves up space.

Sure, External modules are cool, and the MT-32 has a lot of nice buttons to play with and a nice display window (The lack of volume control like the volume knob on the CM modules really isn't that big a deal) , but personally, I've had more problems with external modules in the last 2 years, than with my LAPC-I in since I bought it.

[Rhizome]

I'm glad I went for the external MT-32 instead of a LAPC-I.

Reason?

My Pentium II has both of it's ISA slots used up by 2 other sound cards

[Tom]

I've never owned the LAPC-I.  I've had a CM-500, though.  The overbearing "vibrato" issue inherent to its L/A section was too much for me, so I just learned to deal with the noisier MT-32.  Does the LAPC-I have the vibrato characteristics of the CM-500's L/A synth, or is it more like the original MT-32?  Did later revisions of the MT-32 also have that overwhelming vibrato sound with some of its patches?

[Ari]

I haven't really noticed. Do you have a specific MIDI file that I can use to compare the two synths?

[Tom]

I no longer have a CM-500.  But I remember Eric commenting on it, too.  The L/A part of the CM-500 had a very unnatural vibrato added to some of the sounds.  I think it was the oboe, bassoon, or some other wind instruments. It was VERY noticeable, so I guess the LAPC-I doesn't share that characteristic; you would have noticed.

[HondaSiR]

Based on past threads regarding the vibrato issue, I think it is inherent only to the CM-500. No big problem for me since I don't notice it when playing games or LA MIDI files (I also have a 1st gen MT-32). It's (CM-500) LA synthesis is still way way ahead when compared to Munt or any software emulation.

[Great Hierophant]

No other card quite matches the LAPC-I + MCB-1 capabilities.  The MPU-401 + MIF-IPC(-A) or MPU-IPC lacks the ability to switch settings, the MPU-IPC-T cannot select certain I/O addresses that the LAPC-I can and lacks the sync output.  The SCC-1 and MPU-401AT lack the sync, metronome, tape and 2nd midi out and use custom cables.  If you had one, you really had it all, and for that reason alone I would keep it, regardless of difficulties. 

[Ari]

I actually do have one. Got it off Ebay for 20$ or so a few months after buying the LAPC-I. It was brand new and came with the original package. I certainly consider myself lucky. 

[TuomasT]

I owned a LAPC-I once. It worked in a modernish '01 PC with just one ISA slot fine.

I didn't notice any extra noises over MT-32. It was actually more quiet and had those extra percussion instruments. On the other hand SCC-1 in that same ISA slot makes all sort of electrical whine and buzzing in FFT spectrum. Fortunately it can't be heard even at loud listening volumes, but it is certainly something which must not be mixed in when recording eg. MT-32.  Perhaps LAPC-I had better circuitry in that respect.

What is the real annoyance of LAPC-I though is the inability to change the midi master output volume. But as far as I know, CM-32L can't do that either. The same thing is a bit of a problem in SCC-1 too. The MT-32 volume knob does that.

But I like the way the SCC-1 is always on when my multimedia station is on and doesn't require extra power converters.

And just to add: the actual sound produced by LAPC-I compared to MT-32 was 1:1 as far as my ears can tell and I did compare them quite a bit. They played some different notes though at places.. in Ultima Underworld 1 "Armed" piece LAPC-I always played one extra horn note, which was very noticeable, that MT-32 never played. I guess it is because of the faster CPU in LAPC-I. That's the only place I remember different output though.. the LAPC-I had buffer overflows just like MT-32.

And here's a quite detailed picture of that LAPC-I

[NewRisingSUn]

What is the real annoyance of LAPC-I though is the inability to change the midi master output volume. But as far as I know, CM-32L can't do that either.

It can with a system exclusive message.

In Ultima Underworld 1 "Armed" piece LAPC-I always played one extra horn note, which was very noticeable, that MT-32 never played

First generation MT-32 units with the slow C8095-90 CPU are known to behave somewhat unpredictably in sophisticated multitimbral arrangements. Second-generation units as well as the CM modules don't have this problem, as they are based on the D-110, which has the newer P8098 CPU. Any compatible module will have that advantage.

Basically, the original MT-32 is a castrated D-550 --- noticable from the fact that the it also recognizes some D-50 system exclusive messages, according to its MIDI implementation booklet.

Whereas the second-generation MT-32 and the CM modules are castrated D-110s, noticable from the fact that they recognize RQ1/DT1 system exclusive messages in the 06 00 00 range, which is unused in the MT-32, but used in the D-110 spec --- a leftover from the D-110 system code basically.

the LAPC-I had buffer overflows just like MT-32

How would you know that without the display? And which games? I have not encountered any buffer overflow problems on the CM-32L at all.

[TuomasT]

The need to change the digital master output volume for a midi file or a game arises only after the playback has started. For example I enjoy Master of Orion now and then, and it sets MT-32 at 50%, and I always crank it up to 100% in game. That's a lot of hassle to do without the MT-32 knob. The same goes for SCC-I, sometimes I would like to lower the digital volume to get rid of digital clipping in some midi files.

The guy who I exchanged LAPC-I to SCC-1 for reported having buffer overflows in Thexder II until he slower his computer's speed and I also had some, but can't remember any examples. One can tell of course from the sound output.

[NewRisingSUn]

Thexder II and other Sierra games are not a buffer overflow issue, but a checksum issue --- sysex data bytes get lost because Sierra's MT32.DRV checks the MPU-401's "data receive ready" flag only 256 times before sending a MIDI data byte.

Which is enough if the computer's mainboard runs the 8-bit LAPC-I at 4.77 (like it should), but not enough if it runs 8-bit cards at 8.33 MHz, which it should only for 16-bit cards.

If you try those games on that computer with a second-generation MT-32, it would display "Excl. checksum error", not "Excl. buffer overflow".

the actual sound produced by LAPC-I compared to MT-32 was 1:1 as far as my ears can tell and I did compare them quite a bit.

Second-generation MT-32s sound exactly like the LAPC-I indeed. There are noticably differences between first-generation MT-32s though and LAPC-I:

SQ1 on first-generation MT-32
SQ1 on CM-32L

Notice how the squeaky sound effect is almost inaudible on the CM-32L. Other games sound right on the CM-32L, but not on the first-generation MT-32.

[Rhizome]

I've got a first generation MT-32 (ROM version is 1.06).

[Great Hierophant]

Thexder II and other Sierra games are not a buffer overflow issue, but a checksum issue --- sysex data bytes get lost because Sierra's MT32.DRV checks the MPU-401's "data receive ready" flag only 256 times before sending a MIDI data byte.

Which is enough if the computer's mainboard runs the 8-bit LAPC-I at 4.77 (like it should), but not enough if it runs 8-bit cards at 8.33 MHz, which it should only for 16-bit cards.

If you try those games on that computer with a second-generation MT-32, it would display "Excl. checksum error", not "Excl. buffer overflow".

I have updated my original post to reflect this issue.  I would guess that certain 8-bit cards can run at the higher speed, simply run at their own speed using oscilators on the board, or use the oscillator signal of 14.1318MHz.  Can motherboard detect this by determining that the 16-bit portion of the ISA slot is unconnected and adjust the speed?  Certain cards, like the Sound Blaster Pro, can work in either an 8-bit or a 16-bit slot.  Does this mean the card runs faster in the 16-bit slots?

[NewRisingSUn]

It's not that the LAPC can't run at 8.33 MHz, most likely, it doesn't care about bus speed at all. The issue is not what the LAPC runs at in and of itself, but how much delay a single write to its ports will enforced by the ISA bus hardware, and at 8.33 MHz, a single port read/write will cause a delay of about 0.12 µs, whereas at 4.77 Mhz, a single port write will cause a delay of about 0.21 µs.

MIDI has a baud rate of 31250, or 3125 bytes (10 bits per byte) per second, thus requiring a delay of 320 µs between bytes. Sierra's MT32.DRV waits at most 256 port reads, or 256*0.21 µs ~ 55 µs, probably more since I'm not counting the delay caused by the 8088 LOOP instruction, but even when taking that into account, it's obvious that Sierra walked a *very* tight line with that driver, probably relying heavily on the MPU-401's FIFO buffer, too. I noticed if I reduce the wait count in the driver from 256 to 200 port reads, it already fails.

To summarize: the LAPC will happily work in all ISA slots; it's just a few really crappy sound drivers like Sierra's that require some ultra-precise bus timing. Other software is more defensive in their programming, and thus doesn't have that problem. The 8-bitness of the LAPC only comes into play to cause the 4.77 MHz bus timing required for Sierra's tight timing loop.

Can motherboard detect this by determining that the 16-bit portion of the ISA slot is unconnected and adjust the speed?

I think if the MT-32 can detect if only the left audio jack is connected and thus force mono output, a high-tech computer mainboard can detect if additional pins are unconnected too.
I actually remember my old PC tuning guide recommending against using 8-bit VGA cards, because that "causes the mainboard to believe that the card could only stomach a bus speed of 4.77 MHz" and thus slow down the graphics.

[Cloudschatze]

Basically, the original MT-32 is a castrated D-550

Hey now, they did add multi-timbrality after all.

[Alistair]

And so we come back to an age-old topic- which would be the best MT-32 compatible device to use for hearing the music, and on a personal note, for recording it?

- Alistair

[NewRisingSUn]

Since the first-generation MT-32 occasionally sounds different from later modules, and since a composer may have composed with either sound in mind, the proper solution is to have two modules:

• a first-generation MT-32 with a modification (i.e. the Beaverton modification) that improves the signal-to-noise ratio while retaining all the sound-generating characteristics;
• a CM-32L, which has the best signal-to-noise ratio of all the second/third generation sound modules.

Which one to use for recording/listening to a specific game?
Assume a CM-32L if:

• the game uses CM-32L sound effects, or
• the game causes "buffer overflows" (*not* "checksum errors"!) on a first-generation MT-32 even on a slowed-down computer.

Assume a first-generation MT-32 if

• The game came out before 1990. The LAPC-I's MIDI implementation is dated March 30, 1989; the CM-32L's is dated April 18, 1989; but of course they won't have become widespread until 1990.

If these analytical solutions don't apply, find out empirically by just comparing them side-by-side and seeing if you can determine that one module sounds "obviously" wrong, i.e. missing or hanging notes on one but not the other module (so the case with Lemmings 2, The Lost Vikings, Space Quest 1), etc.

How about we start a list about cases where we know which module was used? Maybe it can be incorporated into the Wikipedia list, indicated by small numbers or something.

It is known that Sierra used a first-generation MT-32 for all games except LSL5 (which wasn't composed in-house and uses CM-32L SFX), likewise, the Fat Man wrote that he used an early MT-32 for Wing Commander and others.
(This makes Ultima Underworld difficult to decide, as the music was composed by the Fat Man (and so you'd assume first-gen MT-32), but the game uses CM-32L SFX for in-game sound effects.)
LucasArts on the other hand always used CM-32L, and is able to cause the early MT-32's firmware to "crash" at times.