20100520

Yamaha 4OP Comparison

Here's a quick comparison of the classic Yamaha 4 operator FM synths. Created as inexpensive alternatives to the 6 OP DX7, when handled with care these synths are capable of some great sounds. First off, naturally, they all have four operators; one being the carrier frequency and the other three the modulating frequencies. Another thing is they all seem to have 8 note polyphony. So looking around a bit, the consensus on the difference in sound between these comes down to; multitimbrality or how many instruments it can play at once, the waveshape of the operators and the bit depth of the digital to analog converter (DAC).

DX-21

Multitimbral: Mono
Waveshape: Sine
DAC Bit Depth: 10 bit

DX-27 / DX-100

Multitimbral: Mono
Waveshape: Sine
DAC Bit Depth: 10 bit

FB-01

Multitimbral: 8 part
Waveshape: Sine and Noise
DAC Bit Depth: 10 bit

TX81Z / DX-11

Multitimbral: 8 part
Waveshape: 8 shapes
DAC Bit Depth: 12 bit

TQ-5

Multitimbral: 8 part
Waveshape: 8 shapes
DAC Bit Depth: 12 bit

So looking at this as a timeline we can see some changes occurred with the FB-01 with the addition of 8 part multitimbrality and the ability to set one of the operators as a noise generator. Then the TX81Z added the ability to change the waveshape outright. The other important thing folks seem to point out on this is the 12 bit DAC on the TX81Z. You might wonder how the bit depth at the input side of the DAC could really make any difference, and honestly I'm not so sure myself, but people claim that it gives the signal more breathing room resulting in a fuller sound. The 10 bit DACs seem to result in a more lo-fi gritty sound (and a really good bass sound). There are camps that swear by either side so it's definitely a personal preference thing.

Addendum: So the FB-01 actually is unique in the way it handles programming the voices. The volume of the operators is controlled on a 0 - 127 scale, all the other 4OP synths use a 0 - 99 scale. I think this is the main reason the stock voices are not up to par with the rest of the Yamaha 4OP synths. Basically the engineers at Yamaha had to relearn the whole system to program the voices; probably not worth it for a product that was supposed to be the cheapest of their synth range. So they probably pushed it through quickly to get something on the market. There are a handful of decent patches on there but most people prefer to program their own for this unit.

For an extremely in depth analysis, check this guy's page: http://www.angelfire.com/in2/yala/2fmsynth.htm

He refers to the FB-01 range as the "CX". I assume what he actually means is the "SFG-01 / SFG-05" which were the sound modules on the CX5M computer. I intend to follow this post up with a more in depth post on the FB-01 explaining the relationship of these units.

4 comments:

Daniel said...

"Waveshape: Sine and Noise"

"So looking at this as a timeline we can see some changes occurred with the FB-01 with the addition of 8 part multitimbrality and the ability to set one of the operators as a noise generator."

No, the FB-01 does not have a noise generator.

"DAC Bit Depth: 10 bit"

It uses the YM3012, which takes an input signal comprising 10 bits of mantissa and 3 bits of exponent, for a total of 13 bits.

"Addendum: So the FB-01 actually is unique in the way it handles programming the voices. The volume of the operators is controlled on a 0 - 127 scale, all the other 4OP synths use a 0 - 99 scale. I think this is the main reason the stock voices are not up to par with the rest of the Yamaha 4OP synths. Basically the engineers at Yamaha had to relearn the whole system to program the voices; probably not worth it for a product that was supposed to be the cheapest of their synth range."

Firstly, nothing needs to be re-learned assuming one knows some fundamentals about how Yamaha's FM synths work. The maximal value is always 0 dB, and every 8 steps below that (or on the SY77/TG77/SY99 only, 10 steps, seemingly) cause an attenuation of -6 dB. On synths that needlessly clamp their ranges to the ugly and illogical scale of 0-99 (why not 0-100?), values below 20 are mapped non-linearly to an internal scale that also runs from 0-127. All other parameters - EG, feedback, etc. - use the same scales and are almost identical, barring small differences in clock rates, etc.

So, one would hope the engineers of the FB-01 would have known this and could simply have added 28 to all volume levels (above 19) accordingly. If anything, the lack of quality, if it exists - which I won't confirm or deny because I find the voices OK on average - is more likely to be due to the suspicion that the FB-01 was developed by Yamaha's hi-fi division rather than their usual synth engineers.

Daniel said...

Having said that about the output levels above, it does start to matter when one is analysing modulators.

The FB-01's modulation index at full volume (of the modulator) is ~12.566, whereas other 4-op synths (except for the pointless DX9) go all the way to just over 25.

In fact, as I have discovered in the course of writing a converter, one can compensate for this simply by checking whether an op is acting as a modulator, and adding 8 to its level if so (which actually = subtracting 8 from its TL). But of course, if the necessary headroom is not available, the modulation will be less than on the other 4-op synth, and therefore the timbre might change a bit. How many voices really use modulation indices that high, though? An open question...

The strange thing is that the YM2151 itself should support the max. index of 25, if similar chips and VOPM (an emulator of the YM2151) are indications. I have to assume that the makers of the FB-01 chose the re-map its modulation for some reason. Either they felt 12.566 was enough, or maybe they wanted to bring it closer to the 13.123 used by the DX7/9.

Daniel said...

I would love to disassemble the ROM of the FB-01 and check whether the modulation index is indeed being clamped to half of its maximal value (8 steps less) internally... something to add to my vague plans to mod its electronics to increase its gain and tame its overly aggressive lowpass filters, I guess. :P

Anyhow, enough from me! Good blog.

Mages said...

Hey, thanks for your comments, some great information you've shared!

What kind of converter are you working on?

And I agree the more I used the FB-01 the more I found the sounds to be perfectly adequate. It's all about how you use them.