VOGONS


First post, by auron

User metadata
Rank Oldbie
Rank
Oldbie

since there is very little information on this module out there i thought i'd do a little writeup from the standpoint of GM playback. first, some basic info: https://soundprogramming.net/synthesizers/kawai/kawai-gmega/

the GMega shares its hardware with the K11 keyboard and was likely released in early 1993, though interestingly some of the chips can have 1992 and even 1991 date codes. 32-part multitimbrality and 32-voice polyphony was certainly top of the line for a 1993 GM unit, as was the 6 MB sized ROM, which AFAIK was only matched by the korg 05R/W at the time. there is also the possibility to link two GMegas together, with a dedicated option in the menu for this. an 18-bit DAC is claimed in the manual and demo song text, unlike the 16-bit stated in the link above, and the 44.1 khz sample rate is also better on paper than the sound canvas series' 32 khz. a mac-only serial port is present. apart from the standard GM patches, extra drumkits are also onboard and are compatible with the SC-55.

in terms of effects, reverb is there but chorus is missing. like on the SC-55 a delay can be applied to the reverb, which is globally activated on the GMega by default, though it's a fairly light delay effect and can be turned off. the reverb intensity setup by standard MIDI controls (CC#91) is rather odd: instead of being adjustable from 0-127, every channel has a an "effect level" with hi or lo setting, with the lower half of CC#91 being mapped to lo and the upper half to hi. it's not well explained in the manual what this exactly means, but from what i can tell lo simply disables reverb altogether for that channel, while hi uses the (roland sysex incompatible) settings in the menu. factory settings for all channels are "hi", so if CC#91 isn't sent there always will be quite a bit of reverb by default. this is in contrast to some other modules like the SC-55 that can actually set reverb to 0 in certain circumstances.

regarding front panel controls, this has to be one of the worst designs of the day. there is a lack of dedicated controls so the cursor needs to be switched around, and then inputs have to be made with the big increment dial, even confirming certain options which is awkward, and the dial seems to be prone to wearing out over time and working less reliably. in terms of sounds, they generally are of a higher quality than the SC-55 and especially the TG100, however DOS game music in particular will not necessarily sound better than on the SC-55, mainly due to instrument volume balance being often rather different. the standard drumkit partly takes after the room kit, which may be appreciated. main strong points in my book are the guitars, both acoustic and electric, which can make certain tracks sound very good. synth basses have a very unique sound as well but are also quite loud. conversely, some of the synth pads can be bit quiet. the synth drum patch has to get a special mention for being all but incompatible with what roland did for the SC-55.

now, for the problems with the GMega: for one, the unit does not react to GM reset sysex whatsoever - instead two buttons are marked on the front panel that have to be pressed simultaneously. since the TG100 came out earlier and does process GM reset sysex, this is quite an annoying oversight. apparently with the later GMega LX they fixed this issue (or rather they had to, due to a lack of front panel controls). as expected, GM resets do not affect parameters that are specific to the GMega, like the reverb settings; these can be restored by performing a factory reset instead.

the bigger issue though is that with certain MIDIs the GMega does not seem to perform like other 32-voice capable hardware. compare heretic's E1M1, which can be hardly considered a heavy MIDI track, on the TG100 and the GMega:

https://www.wavetable.nl/files/comparison/gam … %20TG100%5D.mp3
https://www.wavetable.nl/files/comparison/gam … %20GMega%5D.mp3

from 0:54 onward, drums and strings quite noticeably cut out on the GMega, while the TG100 doesn't really break a sweat, despite releasing earlier and only having 28-voice polyphony. unfortunately, the GMega has problems like this with quite a few MIDI files. somewhat related, triggering too many notes at once, sometimes even as few as 4 or 5, tends to cause timing issues in the playback that i never noticed on any other synth. i was not able to pinpoint what exactly is the cause of these issues; looking through the manual, some patches do use two voices, but there isn't much difference to other GM hardware there. it does seem that turning off the percussion channel helps more often than not. perhaps either voice resources are spread around internally in a weird way or, more likely, the on-board CPU can't keep up with too rapid MIDI messages; for instance page 93 in the manual asks to not switch effect types during song playback, and the menu can in fact lag heavily during MIDI file playback. unlike other devices that used off-the-shelf parts, the CPU here is either integrated with the main chip or another part with custom markings.

overall, it's an interesting unit with some unique characteristics for the time, but also has problems that aren't usually present on other GM hardware.

Reply 1 of 4, by RichB93

User metadata
Rank Oldbie
Rank
Oldbie

Good write up - I recently got ahold of old of these units and it is indeed an interesting beast.

It’s worth mentioning that the GMega LX is a totally different beast and is actually based on the KC20 - it doesn’t even use a different ROM! So presumably if one had the know how, they add some controls and display to it.

Last edited by RichB93 on 2024-10-09, 23:43. Edited 1 time in total.

auq80m-99.png

Reply 2 of 4, by auron

User metadata
Rank Oldbie
Rank
Oldbie

some GMega XC-3 LX differences:

- supposedly came out in 1993 as well, with MSRP cut down from 795 USD to 399 USD (source)
- no display/controls besides volume (obviously), headphone jack moved to back
- 160 patches instead of 128
- multitimbrality reduced from 32 to 16 parts
- polyphony reduced from 32 voices to 20 instrument+8 drum voices
- 3 demos instead of 1, demo button added to back, can be used as a GM reset button as well
- serial support expanded from just macintosh to PC/AT and PC98, switchable on back

all other info taken from the manual. so, overall, that unit seems like a weird sidegrade of sorts - the reduction in parts is not an issue at all and the 32-part support on the original was a strange thing to focus on in my opinion, but the reduction in polyphony is surprising to see from a supposedly later generation based unit. needless to say, an sc-55mkii with same total voices but flexible distribution already suffers in many MIDIs that were made post-90s, so i would expect problems there with the LX. but the GM patches seem to be considered higher quality by some people, and it supports GM reset sysex, so if the performance issues were actually fixed, that might make up for it. of course, given the aforementioned issues with the original GMega, i still have to wonder whether that one may also have some hidden limitations in regards to polyphony.

by the way, this source suggests a 3 MB sample ROM for the original GMega instead of the 6 MB i mentioned before. that number is from the manual, mentioned as 48 Mbit of wave data there. but the manual also mentions two ROMs, one for GM and one for single patches, so that could mean 3 MB for GM.

Reply 3 of 4, by RichB93

User metadata
Rank Oldbie
Rank
Oldbie
auron wrote on 2024-10-09, 03:50:
some GMega XC-3 LX differences: […]
Show full quote

some GMega XC-3 LX differences:

- supposedly came out in 1993 as well, with MSRP cut down from 795 USD to 399 USD (source)
- no display/controls besides volume (obviously), headphone jack moved to back
- 160 patches instead of 128
- multitimbrality reduced from 32 to 16 parts
- polyphony reduced from 32 voices to 20 instrument+8 drum voices
- 3 demos instead of 1, demo button added to back, can be used as a GM reset button as well
- serial support expanded from just macintosh to PC/AT and PC98, switchable on back

all other info taken from the manual. so, overall, that unit seems like a weird sidegrade of sorts - the reduction in parts is not an issue at all and the 32-part support on the original was a strange thing to focus on in my opinion, but the reduction in polyphony is surprising to see from a supposedly later generation based unit. needless to say, an sc-55mkii with same total voices but flexible distribution already suffers in many MIDIs that were made post-90s, so i would expect problems there with the LX. but the GM patches seem to be considered higher quality by some people, and it supports GM reset sysex, so if the performance issues were actually fixed, that might make up for it. of course, given the aforementioned issues with the original GMega, i still have to wonder whether that one may also have some hidden limitations in regards to polyphony.

by the way, this source suggests a 3 MB sample ROM for the original GMega instead of the 6 MB i mentioned before. that number is from the manual, mentioned as 48 Mbit of wave data there. but the manual also mentions two ROMs, one for GM and one for single patches, so that could mean 3 MB for GM.

Both the GMega and GMega LX are basically KAWAI keyboards stuffed into a DTM module. The GMega is based on the K11 keyboard and the GMega LX is based on the KC20. They are totally different devices, despite appearing as siblings.

auq80m-99.png

Reply 4 of 4, by Spikey

User metadata
Rank Oldbie
Rank
Oldbie
auron wrote on 2024-10-09, 03:50:
some GMega XC-3 LX differences: […]
Show full quote

some GMega XC-3 LX differences:

- supposedly came out in 1993 as well, with MSRP cut down from 795 USD to 399 USD (source)
- no display/controls besides volume (obviously), headphone jack moved to back
- 160 patches instead of 128
- multitimbrality reduced from 32 to 16 parts
- polyphony reduced from 32 voices to 20 instrument+8 drum voices
- 3 demos instead of 1, demo button added to back, can be used as a GM reset button as well
- serial support expanded from just macintosh to PC/AT and PC98, switchable on back

all other info taken from the manual. so, overall, that unit seems like a weird sidegrade of sorts - the reduction in parts is not an issue at all and the 32-part support on the original was a strange thing to focus on in my opinion, but the reduction in polyphony is surprising to see from a supposedly later generation based unit. needless to say, an sc-55mkii with same total voices but flexible distribution already suffers in many MIDIs that were made post-90s, so i would expect problems there with the LX. but the GM patches seem to be considered higher quality by some people, and it supports GM reset sysex, so if the performance issues were actually fixed, that might make up for it. of course, given the aforementioned issues with the original GMega, i still have to wonder whether that one may also have some hidden limitations in regards to polyphony.

by the way, this source suggests a 3 MB sample ROM for the original GMega instead of the 6 MB i mentioned before. that number is from the manual, mentioned as 48 Mbit of wave data there. but the manual also mentions two ROMs, one for GM and one for single patches, so that could mean 3 MB for GM.

For reference, 8Mbit = 1 MB, so 48Mbit = 6MB. That could certainly be 3MB samples + 3MB GM samples, however - this is true for other devices like the Alesis S4 Plus, where GM is 4MB (I think?) in addition to the other 16-20MB of samples. In the original SC-55 2MB is GM, and the other 1MB are MT-32 and GS samples.