That’s the thing
I don’t know, I have an opinion and that’s all it is.
Thats precisely why I am on here questioning it to begin with
Why do you think that?
I’m a professional keyboardist and I have owned several Elektron items.
Anyhow, it’s not asking a lot to be able to have a 2018 digital synth able to have significant polyphony+multitimbrality… which is the whole point of this thread.
You can’t even have 2 different parts playing a few notes, not to mention voice stealing totally changes a performance even for non chords…
As I said - just a wild guess.
Analog polys get very fat, it’s true. I still don’t think it’s a lot to ask to be able to have a few multi-note parts coming out of a digital synth at this point in time. Any way you slice it, the Digitone maxes out very quickly.
Dunno why Digitone has only 8 voices.
Why not pick up a Kronos or Montage? Those have all the poly you need.
3 Likes
Fair enough. I’m not trying to be like my forum name either. 
I just like talking about this stuff. I’m not an expert though on the digital side. Just read and talk a lot. 
The Coldfire processors in the Digitone aren’t really DSPs like say the 56000 series (like in the Waldorfs for example) or SHARCs etc. either if I’m reading the correct datasheets anyway. They seem more suited toward more general applications, though they’re pretty fast, and have a lot of IO capabilities.
I’m actually not sure which would lend itself better to the types of things the Digitone is doing as an example. It does seem to do a very good job sound-quality-wise though, and 8 voices between two of these actually seems pretty good to me.
I would think the Elektron devices would benefit more from a general purpose processor because of all of the other functions they perform. Sequencing for example. I’m not sure how well that kind of thing translates to a DSP. There are quite a few MCUs that include DSP functionality too. Some of the new STMs I think, the DSPic (which I think is used on some of the Synthesis Technology modular modules) etc. The more I think about it, something like that or maybe a combination of something like that and a full DSP would be perfect with all the functions these boxes have to run. I don’t actually know what kinds of general purpose processing are used in things like the Blofeld to run the OS, or if the 56000 based DSPs can also do some general purpose stuff. It’s definitely worth looking into though.
The Modors appear to use a PIC + 56725.
I would guess that the more DSP power being used the faster your other processors would need to be as well to coordinate all of the additional functions, synchronization, etc. as well.
I do know someone that would be able to answer just about all of this in quantifiable terms. Maybe I’ll shoot him an email later. 
1 Like
The more I look at the synths that represent the current pinnacle, the more the design trade-offs are apparent. The Virus TI probably has the highest poly count while using DSPs for everything. It has two of them, but I’m not sure which ones they use from first glance. (Edit: It uses two 56321s it looks like. The Blofeld uses one 56371 in comparison.) While it’s a REALLY fully featured synth, and high quality overall, it’s not my favorite sound. It sounds good, that’s pretty undeniable, but not beyond incredibly amazing to my ears.
The Solaris is ten voices. It uses CEMs for VCOs, but also has the “Rotors”. It uses SHARC DSPs (no surprise since John was a Scope developer) for a lot of the other functions. They run at 96KHz to avoid aliasing. I had the pleasure of chatting with John at a meet a couple of years ago. I’d love to have one of these.
The Waldorf Quantum has 8 voices. I have no idea what it’s running on though.
The Roland System 8 uses 3 custom DSPs (multi-core) two for synth voices and one for effects. The Jupiter 8 model on the System 8 uses almost 50% of an Intel I7 CPU on the VST version which runs the same code. So there is a good frame of reference.
(I’m speaking with the person I mentioned earlier, and he’s filling in some of these gaps.)
Really it all comes down to design trade-offs. Less high quality voices, more lower quality voices. Or TONS of processing power or analog circuitry (if you want to go that route). DSPs are getting more powerful, (and other processors) but there’s no magic to this. Higher quality higher voice count equals higher processing requirements and higher cost. Real DSPs like the SHARCs aren’t cheap either. I remember my 6-DSP Scope card being pretty expensive, and that was basically just a PCI card with 6 SHARCs on it. Software to actually use them was separate and also expensive.
As far as multi-purpose ICs go, I was just told about a new SHARC that has dual DSP cores, a filter, FFT co-processor, and an ARM CPU on one die. That would probably get one some decent performance in a small space, but I imagine it’s also pretty expensive. Plus you need someone who knows how to code on all of that just to use the one chip.
Factor all this in, and it’s pretty expensive to make relatively small batches of a hardware synth, let alone something unique that sounds good with a lot of features in a small box that’s still pretty fun to use. Industrial design, PCB layout, metal work, hardware costs, component costs, tooling, software development, etc. etc. It’s not just the ICs that sit on the board.
I know that it’s pretty much not worth it for me to design and build the TINY batches of modular modules that I build. I barely make money at all off of it. Just do it because I love doing it. Most of the money goes back into coming up with something new to build And the cycle continues.
Anyway, some of this was on topic, some maybe not, but I thought I’d post while I was having the conversation with my friend. (who does in fact develop all sorts of high end equipment that uses this sort of technology)
Oh, just got another little tidbit. Apparently the Bricasti reverb runs 6 SHARC DSPs and a Blackfin CPU. I’m told the algorithms that run on this couldn’t even run real-time as a VST period. They put all this DSP power into just making a better sounding algorithmic reverb. I’m pretty much just quoting here as I have no experience with this reverb unit.
Edit: Just took a look around for some numbers for single DSP ICs. Because, why not? This is QTY 1, so I’m sure if you buy a thousand of them, you’re getting a decent break.
The 56000 series that are currently sold sit around $15-30 each.
The SHARC range is wider, some are older, some less available, so I’m not sure why some are as expensive as they are (scarcity or complexity) but these range from around $20 to $930. Average for the ones I recognize is around $45 per chip.
Designing a system, PCB, OS, selecting CPU/MCU, memory, data conversion ICs, etc. around this can be pretty costly. Then having someone to program it all. Then housing it in a pretty case with nice controls, IO, etc. I think we’re pretty lucky to have some of the prices we can get our beloved gear for actually. These companies also need to profit, and stay in business. These little boxes can get pricey, but there doesn’t seem to be much in the way of gouging either. The smaller the company, the more boutique the instruments are, the fewer sold, the trickier this gets too.
Edit: Quick update on those super-high priced SHARCs. Seems that they are likely legacy units that are for applications like military. I’ve seen this with analog multiplier chips from ADI too with similar applications. So the average price on the new/modern SHARC parts appears to be right around $45-50 for a single chip.
9 Likes
yes… probably easier to program than dsp chips and now also powerful enough for crunching numbers in most applications.
The teensy 3.2 comes to mind… the one used in the RadioMusic module… this thing doesn’t even have FPU, but still can do audio synthesis.
1 Like
Yes, I see the Teensy a lot, and also the STM32 variations. The FM Ogre Phase Modulation Oscillator runs on the STM. I believe it’s one that has some DSP function, but would have to check. It may have originally been a DSPic then ported or something. I spoke with Bill/WSY before about it, but can’t remember the details. It’s basically like one fully featured DX-style Operator. Side note: My analog VCO that does the same thing (called the Operator) sits on a 4" x 6" PCB. Uses around 20-30 op amps, and a ton of passive components. Nice what a little digital IC can do in comparison. heheh
The Cylonix Cyclebox actually uses an Altera Cyclone evaluation board. I thought that was pretty cool. It’s an FPGA on a little board with some IO headers, and maybe some conversion. Then the front panel IO and controls are on a separate PCB that attaches to the eval board. It’s surprisingly clean, and is a pretty advanced VCO.
1 Like
That is how we hoomans work. Restrictions, limitations, it enables creativity. It has to. Animals work much the same way. It’s a fundamental function, inside most of us, that makes us curious, makes us move forward and in turn reward us. Some people landed on the Moon, because it couldn’t be done. Some people make complete tracks using 1 single (monophonic) track in a tracker, because: how? Beavers loves pools, but they don’t got moniez so they build their own. I’ve never been as productive as when I sat in front of FastTracker2, somewhere in the 90s, granted it was quantity over quality when it came to the nr of tracks I made in that tracker (a hard drive crash took care of that problem though!:).
1 Like
Most people don’t want all the extra crap the workstation brings, and with their generally unintuitive UI.
I had a Kronos, sounded great but absolute pig to program.
Been saying it for years, If either Korg or Yamaha extracted their incredible FM engines from the Kronos and Montage into knobby standalone synths with the immense power they possess, they will clean up.
Only a matter of time I think before we see a real powerhouse FM synth released, hopefully with a few other digital bells and whistles also.
2 Likes
I actually agree with you completely here. If either company, both, or someone else entirely did this, they would sell tons of them. I’d probably buy one.
What I would say though… Digitone related… is that the simplified engine shouldn’t be discounted as not being powerful. Because it is powerful, and capable of making some exquisite sounds.
Sure, if something comes along with a ton of knobs, more algorithms, more operators, and high quality audio, I’d probably want that too, but it would still be a different instrument altogether. In fact, it might be fun to sequence and supplement such an instrument WITH the Digitone.
2 Likes
Yeah as it stands the Digitone is probably the best option for a hardware FM synth (apart from my Nord G2 that is 
) but as I only work alongside a DAW it doesn’t offer me enough to warrant purchasing it.
If working outside the box it’s a no brainer, and as mentioned if one of the big guys release their awesome FM synths I’ll be much more interested for sure.
It seems odd they haven’t, as the hardwork has been done, the code is sitting there ready and waiting to be unleashed on the world
1 Like
That’s a big one for me. (working away from the PC) I didn’t used to care at all. I’d use all software for one thing, all hardware for something else, mix and match, etc. These days though I work more and more in front of computers, work-benches, etc. and if I get an hour or two to be creative, I grab one of these little boxes (maybe two or three) and go somewhere comfortable to create. Then I bring it all back to the PC to record and polish a bit. Since I started making instruments though, I kind of use them less, so I’ve been less serious about creating final products, and more about just exploring, noodling, and occasionally record the ones I like. So the DT+DN+Blofeld combo is precisely what I was looking for and didn’t know it until just these last few weeks.
1 Like
That would be an expensive beast then, and probably wouldn’t touch a very large user base… why do you think Roland sells containers of Boutique synths while there is no equivalent of, say, a MC909 these days ?
Yes you are sadly correct unfortunately.
Unfortunately the big guns think releasing dispensable toys rather than real instruments is where the money lies these days, hence my general frustration.
As I said they have done the hard work already though, so just extracting one of the engines from their big boxes could be an affordable option.
I know I would probably pay for it.
1 Like
I do think more people now want a few (or many) boxes that do a lot of things. Rather than a single box.
Personally I enjoy minimal setups, even like working with one little box at a time, but kind of like having a few to choose from.
Roland has a suite of little slabs now, which I think is closer to the way people work now.
I do think a modernized slightly larger MonoMachine with a bit more manual control might be kind of cool in the “All in one” department though. Or maybe a DSI box that combines the Tempest (more or less) with a Pro-6 desktop. It would be SUPER expensive, but also super nice. Actually maybe a Pro-2 in that box would be more realistic.
1 Like
And it would be fraught with bugs, never to be fixed. Dave Smith has proven, on many occasions, that he simply does not possess the objectivity nor the sense of commitment required to manage larger projects.
Roger Linn could do it; but I doubt another MPC style device is on his radar.
Elektron, on the other hand, have proven themselves in this arena many times (for the most part).
Cheers!
I operate the opposite
It’s the reason I love the Nord G2. There are virtually no limitations and restrictions. It’s why I love working in a DAW as well.
Anything imaginable can virtually be done.
I have a Novation Peak, great synth but the ‘limitation’ of only 2 lfo’s definitely doesn’t spark my creativity, rather just gives me the shits 