Wish New Product: Just the 2 Amps+Cabs of the Axe III

[Dr. Dipwad]

I think there's a market for a small FAS product focusing only on the 2 Amp Blocks of the Axe III, plus 2 Cab Blocks.

It could be 1 rack space, possibly with removable ears and a tiltable front face so you can mount it on your pedalboard;
or,
It could be a pure-electronics package small enough to fit into 1 rack space, but less than that in width, with a detachable set of footswitches;
or,
It could be a stompbox sized in such a way that it isn't awkward to rack, and the controls can still be used from rack-front. (Probably that last is the most-popular option, if I'm honest about it, even though I myself prefer to keep everything in the rack.)

There would be a small screen (as large as would fit in 1 rack space) plus several knobs and a power switch on the front, plus a couple of buttons for navigation.

The knobs would be soft-knobs (rotary encoders) to allow quick access to the usual Amp Block settings (Treble, Mid, Bass, Presence, Gain, Volume, etc.), but, using the buttons, one could select a different page on the screen, and those knobs would instead modify different Amp settings, or Cab settings, as desired.

On the back there would be MIDI ports, an Expression Pedal jack, a Footswitch jack, 2 inputs, 2 pre-Cab-Block outputs, and 2 post-Cab-Block outputs.

The simplified "grid" in this unit would allow some basic routing and mixing options, plus the usual full-featured Continuous Controller/Expression/Modifier capabilities of the Axe FX III. The routing and mixing would allow you to run the signal chains as entirely independent, or to blend them, as desired.

Usage Example #1:
You plug your pedalboard into only 1 of the input jacks; you send a pre-Cab output to your on-stage power amp and cabinet, and you send a post-Cab output to the house mix. You use the expression pedal to "morph" from one Amp/Cab signal chain to the other.

Usage Example #2:
The guitarist uses one of the parallel signal paths; the bassist uses the other. Or, you use the two paths for two different guitarists.

Usage Example #3:
By plugging this unit into the FX Loop send and returns of your Axe FX III, you suddenly have a rig with 4 Amps and Cabs able to run simultaneously.

Usage Example #4:
By plugging this unit into the FX Loop send and returns of your FM3, you suddenly have a rig with 3 Amps and Cabs able to run simultaneously.

...and obviously there might be more use-cases I haven't thought of.

 

[Dr. Dipwad]

Nobody?

Thought for sure someone would be interested in having FAS-quality Amps and Cabs, for cheaper than the FM3, that could be added either to pedalboards or racks.

 

[edo]

It would be even better if it were in a pedal form, think of the two note torpedo c.a.b. m.. just place it at the end of the pedalboard chain and use your old school pedalboard as if it were an ax8 / fm3 between guitar and pa!

 

[Dr. Dipwad]

Ah, thanks @SML59 for the Like!

For what it's worth, here's another reason I'm pursuing a product like this as a WISH:

Various companies (Strymon, TwoNotes) have been releasing products allowing us to have just the Amp section (and optionally, the Cab section) of our signal chain on our pedalboard, or on one shelf of our racks.

BUT, it's always one sound at a time (unless you buy multiple pedals).

AND, I don't think the quality of the modeling rises to the level of excellence that we see from FAS products.

AND it's certain (indeed, widely-acknowledged) that other companies don't provide firmware updates constantly improving on the sounds, the way FAS does.

So it seems to me that a 2-Amps-At-Once product from FAS has great advantages:
1. It allows all the usage examples given in my original post, above;
2. It allows FAS to play to its strengths in an area where FAS already outshines the competitors;
3. It allows FAS to keep existing FAS fans (I count myself one!) "in the fold" by always using FAS products for the Amp/Cab part of their signal chain, rather than having to go elsewhere.

 

[Dr. Dipwad]

It would be even better if it were in a pedal form, think of the two note torpedo c.a.b. m.. just place it at the end of the pedalboard chain and use your old school pedalboard as if it were an ax8 / fm3 between guitar and pa!

My feeling was that it should have a form factor which is able to compete with TwoNotes, but with better MIDI and ContinuousControl/Expression capabilities (nobody does those as well as Fractal!) and that isn't awkward to put in a rack.

It seems to me that if the electronics can fit within a single rack-space, or even a half rack-space, then one can add a cheap 3-button footswitch for control (like the add-on one sees with Strymon reverb and delay pedals, or like one of the commonplace Digitech 3-button switches).

But the important thing is the functionality and usability.

 

[edo]

Personally, if I had to carry a rack and a pedalboard, I would just use the axe fx and the midi controller. If I could place a fas amp modeler in a stompbox size and put it in a small pedalboard with a couple pedals, I would buy it immediately

 

[lqdsnddist]

I don’t think it would be really that much smaller/cheaper etc just if it couldn’t do effects. The Fx8 wasn’t smaller than the ax8 after all, so it’s not like each possible effects block adds bulk

 

[IronSean]

@lqdsnddist Remember that the form factors of the FX8 and AX8 were at least partially dictated by the requirement for the floorboard form factor with multiple switches. A form factor like the FM3 is a lot more manageable if it's only amp blocks.

Initially with asks like this I usually think they sound superfluous. Helix offers lots of form factors, and although there's value it also adds a lot of overhead to their software releases making sure all the hardware is kept in sync.

There is a world where this could make sense. The Axe FX III has two high powered DSPs (one for guitar, and one for effects), the FM3 has one processor with 2 DSP cores (assuming again one for amp modelling and one for effects) and one ARM core for system software (USB, ui, etc). If they could just make a single DSP core plus ARM chip they could just run AMP modelling and nothing else. But, that one core is not significantly expensive, and taking the chip from 3 cores to 2 wouldn't be that much of a heat and power difference for smaller form factors. And the software would be significantly different because there would be no handling of routing, effects on other chips, loops, etc.

I think ultimately though it might be cool, removing that one core has diminishing returns for saving size and cost. The rest of the unit, board, hardware etc all still has to exist. And if it was smaller it breaks the uniform form factor they've developed. And if it's not much smaller, and not much cheaper, most people would opt to pay a little extra for the FM3.

 

[Rex]

If they could just make a single DSP core plus ARM chip they could just run AMP modelling and nothing else.

But if you're running an amp block, you'll probably want to run a cab block, too. There's your second core.

 

[IronSean]

But if you're running an amp block, you'll probably want to run a cab block, too. There's your second core.

Well now it's got all of the horsepower of an FM3, has the same heat dissipation needs, the same form factor... and all they took out was the effects from the software which doesn't really save on costs

 

[Rex]

Well now it's got all of the horsepower of an FM3, has the same heat dissipation needs, the same form factor... and all they took out was the effects from the software which doesn't really save on costs

This.

 

[Dr. Dipwad]

Okay, you fellows seem to understand how which (and how many) chips are required for different uses, but I'm struggling to catch up.

For example, @IronSean says:

The Axe FX III has two high-powered DSPs (one for guitar, and one for effects).

...and I'm assuming that when he said "guitar" he meant "amp modeling"; but I'm unsure whether that also includes cab modeling. Then he says:

[T]he FM3 has one processor with 2 DSP cores (assuming again one for amp modelling and one for effects) and one ARM core for system software (USB, ui, etc).

This confuses me because it makes me think that, since Axe FX III obviously has system software, shouldn't it require an ARM core just like the FM3 has? Doesn't each "high-powered DSP" have at least two cores? What's the difference between an ARM core and some other kind of core? Sean seems to know a lot about this but I find I'm unable, from what he's saying, to guess what the proper "recipe" of chips/cores would be for such-and-such set of functions. (Sorry if I'm being dense.)

So, let me ask 5 numbered questions:
Q1. What kind of chip or core would it take to do Amp modeling for 1 Amp block?
Q2. What kind of chip or core would it take to do Amp modeling for 2 Amp blocks?
Q3. What kind of chip or core would it take to do Cab modeling for 1 Cab block?
Q4. What kind of chip or core would it take to do Cab modeling for 2 Cab blocks?
Q5. What kind of chip or core would it take to do system software that's rather less complex than that of the FM3?

It seems to me that straightforward answers to questions 1-5 would allow me to know what the chips/cores "recipe" would be for each of the following Product Ideas (labelled A-D):

Product Idea A: 1 Amp, 0 Cabs, no FX
Product Idea B: 1 Amp, 1 Cab, no FX
Product Idea C: 2 Amps, 0 Cabs, no FX
Product Idea D: 2 Amps, 2 Cabs, no FX <--- my original product idea

(I'm assuming that A-D would have a simplified UI because the grid/routing options are fixed and minimal, but the sound quality and expression/continuous-control flexibility are up to the usual FAS standards.)

I imagine that Product Idea B is roughly equivalent to the Strymon Iridium (which I haven't tried) but with more Amp/Cab options. But perhaps FAS quality/processing-load is so much higher that an FAS-equivalent product can't be put into a similarly small enclosure?

 

[Rex]

Okay, you fellows seem to understand how which (and how many) chips are required for different uses, but I'm struggling to catch up.

For example, @IronSean says:

...and I'm assuming that when he said "guitar" he meant "amp modeling"; but I'm unsure whether that also includes cab modeling. Then he says:

This confuses me because it makes me think that, since Axe FX III obviously has system software, shouldn't it require an ARM core just like the FM3 has? Doesn't each "high-powered DSP" have at least two cores? What's the difference between an ARM core and some other kind of core? Sean seems to know a lot about this but I find I'm unable, from what he's saying, to guess what the proper "recipe" of chips/cores would be for such-and-such set of functions. (Sorry if I'm being dense.)

So, let me ask 5 numbered questions:
1. What kind of chip or core would it take to do Amp modeling for 1 Amp block?
2. What kind of chip or core would it take to do Amp modeling for 2 Amp blocks?
3. What kind of chip or core would it take to do Cab modeling for 1 Cab block?
4. What kind of chip or core would it take to do Cab modeling for 2 Cab blocks?
5. What kind of chip or core would it take to do system software that's rather less complex than that of the FM3?

It seems to me that straightforward answers to questions 1-5 would allow me to know what the chips/cores "recipe" would be for each of the following products (labelled A-D):

Product A: 1 Amp, 0 Cabs, no FX
Product B: 1 Amp, 1 Cab, no FX
Product C: 2 Amps, 0 Cabs, no FX
Product D: 2 Amps, 2 Cabs, no FX <--- my original product idea

(I'm assuming that A-D would have a simplified UI because the grid/routing options are fixed and minimal, but the sound quality and expression/continuous-control flexibility are up to the usual FAS standards.)

No one outside of a select few designers at Fractal has enough information to answer your questions.

 

[Dr. Dipwad]

No one outside of a select few designers at Fractal has enough information to answer your questions.

Seems I run into that problem often. Has to do with the picky analytical kinds of questions I ask, I guess!

Still, I wouldn't mind someone taking an educated guess, provided it was labelled as such.

 

[Rex]

Seems I run into that problem often. Has to do with the picky analytical kinds of questions I ask, I guess!

Still, I wouldn't mind someone taking an educated guess, provided it was labelled as such.

I wish I could. But I didn't write the code, so the best I could offer is a random guess that would serve no one.

 

[IronSean]

For example, @IronSean says:

...and I'm assuming that when he said "guitar" he meant "amp modeling"; but I'm unsure whether that also includes cab modeling.

You're right, I meant one runs the amp modelling and the other all the other effects.

This confuses me because it makes me think that, since Axe FX III obviously has system software, shouldn't it require an ARM core just like the FM3 has? Doesn't each "high-powered DSP" have at least two cores? What's the difference between an ARM core and some other kind of core?

All any of us outside Factal can due is speculate based on what info is available. From the shop page it lists for the Axe FX III " Equipped with four processors, the Axe-Fx III contains more raw processing power than any guitar product ever made, by far. " and on the Wiki:

Processors:

• Two 1.0 GHz floating-point “Keystone” DSPs (2.8 times faster than the TigerSHARC DSPs in the Axe-Fx II)
• Video display processor
• 500 MHz 16-core XMOS USB microcontroller:
  • Supports 8x8 audio at 48kHz, 24-bits
  • MIDI-Over-USB is about 10x faster than an Axe-Fx II

My understanding is that there are 2 DSP processors which may be two chips or two cores on one chip, one GPU for running the screen, and the USB Microcrontroller appears to run the USB interface. Exactly how much of the system software runs on the DSPs vs the other microcontroller isn't clear, but when your CPU usage gets above 80% on the DSPs the slow UI response and audio dropouts indicate that some amount of the system hits a wall when there's too much running on the DSPs.

My understanding is that in the Axe FX II, one core would also work for amp modelling and the other for effects. It used to be that One amp model would use most of a core, and running two would run both at half the oversampling rate, so you get two amps at half quality so to speak. On the Axe FX III I believe both amps can run at full quality on the one DSP core. I'm also not as familiar with the low level architecture of DSP chips as I am CPUs/GPUs/etc. But since analog signal processing is largely serial data (ie, one continuous string not a bunch of chunks you can process separately) I don't think they benefit as much from multiple cores the way some CPU laods can, or tasks like video encoding really can by splitting the image up to different chunks and processing them each in parallel. That said, there's some level of parallel-ish work involved in things like delays and reverbs, and since stereo data is so common there may be some parallel lines within the chip.

And ARM core is a CPU processor that runs the ARM instruction set, the same as cell phones tablets and some smaller laptops do. These are general purpose computing processors and I imagine well suited for things like managing the system itself, USB communication, sending and receiving data to the FCs, etc.

I'm not sure exactly how this work is divvied up, among the cores in the III. I thought I had read somewhere that the ARM core on the FM3 was able to do more of the work and free up the DSPs to focus more completely on audio processing, but I can't find it now. I also know the CPU usage on the FM3 has been normalized so that 100% isn't really 100%, but it's more like 80% on the Axe FX III. ie. the point where you are doing to much and things will stop working as well. The FM3 cheap is also a lower powered one as well, hence only one amp block and lower CPU overhead.

So, let me ask 5 numbered questions:
Q1. What kind of chip or core would it take to do Amp modeling for 1 Amp block?
Q2. What kind of chip or core would it take to do Amp modeling for 2 Amp blocks?
Q3. What kind of chip or core would it take to do Cab modeling for 1 Cab block?
Q4. What kind of chip or core would it take to do Cab modeling for 2 Cab blocks?
Q5. What kind of chip or core would it take to do system software that's rather less complex than that of the FM3?

I'm not sure if the CAB blocks run entirely on the effects cores or with the amp block. But I think relative to the amp modelling, they'd be pretty low CPU cost. So for Q1 and Q3 I'd say "A DSP core at least as powerful as one of the 2 cores in the FM3", and for Q2 and Q4 I'd say the same, but for one of the Axe FX III DSPs.

As for Q5, that really depends. The system software might seem like it's less complex, but you might still want USB audio functionality, external controller functionality, modifiers, the ability to talk to an editor like Axe Edit, controlling and running the UI, etc.

It seems to me that straightforward answers to questions 1-5 would allow me to know what the chips/cores "recipe" would be for each of the following Product Ideas (labelled A-D):

Product Idea A: 1 Amp, 0 Cabs, no FX
Product Idea B: 1 Amp, 1 Cab, no FX
Product Idea C: 2 Amps, 0 Cabs, no FX
Product Idea D: 2 Amps, 2 Cabs, no FX <--- my original product idea

(I'm assuming that A-D would have a simplified UI because the grid/routing options are fixed and minimal, but the sound quality and expression/continuous-control flexibility are up to the usual FAS standards.)

I imagine that Product Idea B is roughly equivalent to the Strymon Iridium (which I haven't tried) but with more Amp/Cab options. But perhaps FAS quality/processing-load is so much higher that an FAS-equivalent product can't be put into a similarly small enclosure?

One other thing to note is the more different this is from the standard existing products, the more work it takes to actually build and start selling one. As you can see from the FM3 being announced a year before it started shipping, and even some of the early growing pains in hardware and software that making new hardware devices is hard, even for the skilled people at FAS. The FM3 is on a different DSP, has a different main processor architecture with the ARM core, different overall hardware design and circuit board layout I'm sure. Different compile and build processes were needed to get the model code that could be shared built and running on it, there were compiler bugs that made development hit snags, etc. It's an expensive and complex process to bring a new device to market.

I feel like it just starts becoming a trade-off: To keep things as simple in terms of development effort and just get a product out there, maybe you use the exact same FM3 hardware, but use the 2 DSP cores for an amp+cab block each and nothing else. The physical device is the same and costs the same to produce, so who would buy it for 2 amp blocks instead of an FM3 with one amp and all the effects for the same $999?

It you still try and keep it similar with limited additional development costs and the most shared software you end up with a version of the FM3 chip with one DSP core instead of 2 for lets arbitrarily say $120/chip instead of $150. It has similar heat and power needs, a similar enclosure, and similar other features around USB and display, and costs $799 instead if $999. Probably won't be a hit when people could just spend a little more and get all the effects too.

And as you start stripping away more and more to save costs. Maybe you remove USB interface functionality, build a new pared down system interface. Remove the FC controller ports and the ability to use them, use an even lower powered chip. You could get a physical device that costs maybe half as much to produce. But would you sell it for half? You've also had to write a lot of custom system software, deal with development challenges for these new chips, this new UI, and all the removed functionality, and deal with the headache of keeping this third separate system synchronized with your software updates. So although the BOM (Bill of Materials) is lower, the cost to make it might not be that much less after all that development work, so maybe you can't afford to sell it for much less than the FM3 which causes the same comparison as above.

And the other question is are any of those ideas worth committing a couple years of work to build, or would it be smarter to spend that time on something else like a successor to the FX III, or whatever new idea Cliff said he patented that would be a game changer as a new product, etc.

Would it be cool to have a box the size of a medium Pedal that has the FAS amp modelling in it? Sure! But how cheap would it have to be to sell enough to make up for the costs of product development? And what about the Opportunity cost of spending the time on this product instead of something else? The more expensive it gets, the more someone can look at an FM3 and go "well, I get a lot more for the money this way".

Or maybe they'll come out with a HX Stomp sized amp only product in a year and I'll have to eat my hat But I'm more than happy to take the opposite path of @Rex and make all sorts of random guesses. Mostly because I'm sure FAS has already thought long and hard about which form factors and price points they think makes sense, and if they think this one does I'm sure it's already being worked on,

Basically I had a fun roller coaster of thinking "This probably doesn't make sense" then "Wait, well if they removed one of the two DSP cores maybe it would be simple and would make sense" then "No, on second thought I'm not sure they could remove enough stuff for it to really still make sense". But I'm mostly just making semi-educated guesses and thinking out loud. Who knows what will happen. I can say for sure what won't happen is me going back and editing out half of the unnecessary text in this post

 

[Dr. Dipwad]

Wow. Thanks for the substantial effort in that response, @IronSean!

Lots to think about.

I guess what I ultimately wish I could get is something that's an FAS equivalent of an Iridium, or of one of the Mooer Preamp minis: Something small and inexpensive and integratable into a larger signal chain, yet with the benefits of FAS (sound quality, feel, low latency, continuous/expression control, and lots of different amp and cab models to choose from).

How such a product would be constructed, and whether it could be made cost-effective and targeted at a good market, is apparently beyond what we can discern on the basis of our limited information. I guess I can always hope!