Yamaha TX16W—owners please help test—odd pitch/timbre issue?

[db0451]

Can anyone with a TX16W confirm, deny, or explain this? Before I invest money upgrading it, I want to be sure that it does not have a weird hardware fault.

The official OS might do this too, but I have no way of writing a 720 kB disk at the moment and only the much more popular Typhoon 2000 OS. Most other users probably have that, anyway! So, the easiest way to replicate this – or not… – is and to follow the instructions in its manual PDF as follows: Load the OS. Press SYSTEM SETUP. Press UTILITY. Choose to Load the Setup 01: DEMO. Set PERFORMANCE SELECT to 01: MULTI. Now, follow their advice of trying out the Typhoon LPF by going to VOICE EDIT / voice 1: ANA_STRS / parameter 5: filter / table 17: LOWPASS. However, the official LPF at table 01 should also produce almost as noticeable results.

Now play any E note from MIDI channel 1 of your controller, thus playing the ANA_STRS voice. Do you hear repeated square-wave-y alterations in timbre, increasing and decreasing (not only pitch) high-frequency content – which almost sound like the filter being opened and closed periodically? I do! Yet there is no modulation programmed from any LFO to the cut-off frequency, and the same strange effect does not occur on any notes other than E There is, however, a modulation set from LFO to pitch, and if you really ramp up that PMD, D# and F notes start to be affected…

Having fooled around with various waveforms, filters, and modulations, I can only assume this pattern of sudden stepping between timbral variations is due to some strange interaction between—

• the filters and their quantisation to 10 steps on each axis and/or
• some sort of octave-based scaling for waveforms’ pitch, which results in particular notes becoming ‘transition points’ for timbral artefacts

—causing sudden step-wise changes in timbre to be heard when the pitch moves above and below these ‘pivotal’ frequencies, whether that change is caused by the keyboard or pitch modulation. I have also sort of noticed similar but much less obvious changes without filters, but those were not consistent and only occurred on some keypresses.

but my aforementioned technical theories could be totally wrong, and maybe I just got unlucky with an isolated glitchy TX16W… hence this thread needing your feedback!

All feedback and testing is welcomed. I’m keen to know whether this is ‘normal’, a.k.a. just bad programming/design – or whether I have yet another ‘quirky’ module… There was, at least, a glimmer of hope on a page written by one of the regulars here, @madtheory: http://madtheory.com/site/?p=340

Other things were impossible to fix- the digital filters are still very limited, and
modulation can cause glitches with certain pitch and sample rate combinations.

So, hopefully I can be reassured this is a usual thing – then I can get to making the most out of this daft old beast and its silly glitchiness.

[madtheory]

As I described in the link above, this is normal. There is an interaction with the transposing algorithm and sampling rate. The workaround is to resample the offending wave to a different sampling rate. This will move the glitch to a different pitch, hopefully one you're not using in the piece of music you're playing!

I never looked in to exactly why this happened, so your guess is as good as mine. I just used the quick and dirty fix. IIRC it also happened in the Yamaha OS, but it's been a very long time since I used that. And I never will- I would like to retain the will to live!

To be honest, it didn't happen very often in an actual track I was recording/ performing. Although when programming new sounds it can be annoying.

If you're into it, there is at least one decent reverse engineered analysis of the hardware out there on the tinterwebs. Or you could DIY- the whole machine is laid out inside with off the shelf chips, no proprietary Yamaha ones. So it is possible to figure the thing out with data sheets. You could also ask Magnus at Sonic Charge, who made Typhoon.

[db0451]

Hey, thanks for the confirmation, which suggests what I described is a known issue. I just wanted to be sure the weird glitches were normal glitches, so to speak I was encouraged when I read your line quoted above, but I couldn’t be sure it was referring to the same thing. It sounds like my unit is not abnormal compared to others.

I’ll experiment with resampling and whatnot. and with trying the official OS, which – let me say something shocking here – I’m looking forward to trying! It has a few features Typhoon never quite managed to replicate, so I wonder whether those outweigh its oft-supposed painfulness… Although I doubt the original OS is ideal, knowing Yamaha – I suspect its terrible reputation is partly urban legend. I have a TX802, which looks nearly identical both physically and in menu structure, and I have no problems with that. However, the first modules I programmed were an FB-01 and a TX81Z, so maybe I have an unfair advantage! Anyway, I can’t even try any other disks until I find a FDD that can write 720 kB/DD types…

Yeah, I’ve found lots of interesting info on the hardware and possibilities for reprogramming. This is the first module I have that will boot directly from any old thing on a disk, so that’s quite tempting. I’ve previously had ideas about modding other OSs, but all of those would require an EPROM burner, and my cheap one is hopeless.

[edit] although not all the chips are generic: certainly, the Wave GEneratorand filters are proprietary LSIs by Yamaha. For info, the GEW was also used in the RX11, albeit filter-less, and possibly in some other contemporary products. [/edit]

thanks again!

[madtheory]

What facilities did the Yamaha OS have that Typhoon doesn't? Typhoon 2000 is not short of anything AFAIK, it even maps samples for you once they're tuned and the modulation is pretty extensive. They improved the resampling from the original Typhoon so it didn't alias as much, which was a pity IMO Didn't know about the filter LSI, I thought I remembered reading that they ran on the 6800?

Anyways, let us know if the resampling trick works for you.

[db0451]

Yup, I will do some resampling once the holiday rolls around! and yeah, the controlling CPUs are 6x0x architecture, but they just tell the GEW and ADF chips what to do – with all the actual processing being carried out within those proprietary LSIs. This pattern was totally standard for Yamaha throughout the 80s. I really must learn 6x0x!

For sure, I know Typhoon added a tonne of fantastic things – including some very cool modulation opportunities and other abilities that haven’t been achieved in sampling elsewhere/since – and I intend to make the most of those.

That said, and for posterity, here is the info you wanted on what was not inherited from the official OS. I’ll even add my own idle commentary in exciting technicolour! Most of these centre around converting original, official-OS-targeted voices, so admittedly that is of less concern for creating new sounds – but I’ll be wanting to explore all the original factory disks and whatnot as a ‘training course’ and plundering for sounds to use.

So, firstly, notes on the automated conversion from Typhoon’s own release notes:

The import routine will do its best at translating all the parameters from the Yamaha format, but some things are diIIicult or simply impossible to convert. Pay special attention to the following cases:

• Alternating voices. Alternating voices cannot be supported because there is no easy way to recreate this effect in Typhoon. I don’t know about you, but I’m really looking forward to using this feature to implement ‘round robin’ on drum samples to mitigate the ‘machine gun’ effect of static samples hit in succession.
• Pitch and Iilter envelopes. LFOs and other modulations. These modulations could be emulated by defining entries in the modulation table. However, since at most eight of them can be defined this way we would run into trouble deciding which modulations to include and which to ignore. End of story: the only modulation currently converted by the import routine is the pitch bender. For using voices that were originally written for the official OS, this sounds like a lot of work would be required to convert them. In these cases, I would boot into the official OS (unless I really need to combine such voices with native Typhoon ones).
• Pitches oI waves. Typhoon will try to figure out proper pitch settings for the waves by looking at how they are used in the Yamaha timbres. If a timbre places a wave at a single key, the wave is considered to be unpitched. Otherwise the pitch is found using a combination of the different pitch and tune parameters for the voice and timbre. Most often this is correct but sometimes it can be way out of line.
• Stereo perIormances. Performances constructed for stereo playback are not treated differently from other performances. The voices and waves of these performances will not be converted to true stereo voices and waves (i.e. some voices play the left channel and others the right channel). Thus, perIect stereo phase-synchronization is not achieved unless you reconstruct these perIormances Irom scratch. (Hint: use the wave edit function Join to create a stereo wave from two mono waves.)
• Velocity response curves. The response curve parameters in Typhoon are totally unlike the ones in the Yamaha OS. Some curves, albeit really weird ones, cannot be recreated in Typhoon at all. Typhoon will try to convert the velocity curves that can be recreated in Typhoon. If a curve cannot be translated, a default velocity setting will be used. as above for modulations

Secondly, from the unofficial FAQ – written for Typhoon v 1, so some of these might have been fixed in 2000, but I’ll catch the ones I know about so far:

'Missing features'
Features that existed in Yamaha OS 2.x, but not present in Typhoon v1.0:

• You can't trigger notes from a footswitch as in the Yamaha OS 2.x. meh
• It's not possible to play the (newly sampled) wave from the keypad (with "-") as you could in Yamaha OS 2.x. sounds kinda useful
• There is no way to change the parameters of all the Groups (splits) in a Voice like you could in Yamaha 2.x, you have to change them all by hand. ditto!
• There are no mix or reverse utilities for waves. added in 2000 alongside various new editing processes
• There is no support for making a backup copy of the system disk MEH. Just copy that floppy! It’s a standard DOS format like any bootable disk for the TX16W.
• There is no support for importing Yamaha OS setups, performances or voices added in 2000 but with limitations as noted earlier

a mixed bag, really. Typhoon wins on most points, but I can see myself wanting to boot into the official OS from time to time, especially for voices that were originally programmed for it (and have not had all modulations including ENVs painstakingly converted to Typhoon – preferably by someone else )

[madtheory]

That's really interesting. It would be so cool if you could run some other programs in this machine. Mr. RIchard D James would be an interested person!

But I'm pretty sure that because the Yamaha OS was so slow and difficult, and the filters so weak, that no one really dug in to it very deeply. For example I have the Japanese and UK factory library. AFAIK none of it has any evidence of major modulation and definitely no round robin. I was not impressed when I tried the UK library in Yamaha OS- I initially converted them (in the original Typhoon) by writing down the zones and reprogramming.

The UK library is superior- one of the programmers was Paul Wiffen who among other things did the Synthex Laser Harp, as well as a ton of work as an EII programmer. If anyone was gonna make the TX sing, it's him. And he did- the actual samples are great, especially the strings. But there's not much use of modulation. I might be wrong though. I could ask him to chime in on the thread if he's into it.

I think Magnus and co were perhaps being pedantic in detailing the differences with the Yamaha OS. Pitching is 99% perfect IME. The Yamaha library performances are mostly meh. I think the velocity response curves are heavily dependent on the keyboard you're using, and your playing style. I'd always adjust them anyway, on any sampler/ synth. But that's just my opinion.

[db0451]

Yeah, I would love to program support for some weird and wonderful other sample formats, plus any other ideas that came to mind! But I find it hard to justify the time that would be needed. Which makes me appreciate Typoon a lot!

Are these libraries archived anywhere? I might have them, but most of what I found has been scattered and often unlabelled files on FTPs, etc. I am keen to try the official set once I have a way to load them in. It would be really interesting to hear from Paul! I’ve seen his name appearing all over the place recently, for various companies.

[madtheory]

Some library posted in the Cyclone thread at Sonic Charge site.

[PaulWiffen]

Thanks to Tomas Mulcahy for drawing my attention to this thread.

With the TX16W library, I was asked to quote a fixed price for producing the two sets of sounds. It took so long to sample, loop and key assign the sounds because of the rather long-winded operating system that I overran the delivery schedule by a week. They were screaming for the disks to launch the product at a NAMM show so I never had time to do any mod settings. The urgent need was for good authentic sounding library, it was assumed users could add any required mod settings themselves as chances were they were already Yamaha owners and used to the obtuse user interface. I was not, having previously worked for Sequential, Emu & Akai so we concentrated on my sampling and looping skills which no one in-house at Yamaha had back then!

I did eventually acquire a SY99 to be able to use the sound sets I developed live.

[madtheory]

Thanks Paul. I always felt the so called UK library was far better than the previous US/ International/ Japanese one. As you know I'm a big fan of the Cello Trio

[db0451]

Great to hear the story! And I can't wait to hear the sounds, once I get a drive to write them from my PC.

chances were they were already Yamaha owners and used to the obtuse user interface

Yeah, that's me.

[madtheory]

Are these libraries archived anywhere?

I've uploaded the complete Muki Pakesch archive here:
http://madtheory.com/site/?p=1028