Author Topic: Spin semiconductor fv-1  (Read 50907 times)

jmasciswannabe

Spin semiconductor fv-1
« on: October 24, 2006, 12:57:41 PM »
I don't know much of anyhting concerning this stuff, but ran across an add for it in the new products page in Music Inc magazine. It looked pretty rad especially with the analog potentiometer inputs. Searched for it on this board, but didn't find anything. Maybe a good chip for a project? Here's the datasheet!

http://www.spinsemi.com/get_datasheet.php?prodnum=SPN1001&pdf=FV-1.pdf
....the staircase had one too many steps

puretube

Re: Spin semiconductor fv-1
« Reply #1 on: October 24, 2006, 01:26:10 PM »
very interesting, THANX a LOT!

 :icon_cool:

Peter Snowberg

Re: Spin semiconductor fv-1
« Reply #2 on: October 24, 2006, 04:37:25 PM »
Eschew paradigm obfuscation

jmasciswannabe

Re: Spin semiconductor fv-1
« Reply #3 on: October 24, 2006, 05:55:56 PM »
My pleasure, glad I could be of some service!! I have been getting ready to make the jump over to the digital/dsp realm and have read about the femtoverb....thats why this clicked when I read over it.
....the staircase had one too many steps

SeanCostello

Re: Spin semiconductor fv-1
« Reply #4 on: October 25, 2006, 12:41:01 AM »
For anyone on the board who is familiar with the Alesis AL3201, this chip seems very similar in design and architecture. It is a reverb/delay/modulation/dynamics processor, and probably should not be thought of as a full-fledged DSP. Built-in LFOs, single cycle LOG and EXP conversion, etc.

Unlike the Alesis chips, it has ADC and DAC on board. Basically, all you need to build a circuit is this chip, the pots needed for the parameters, yer usual box/switch/etc, and a 32.768 KHz watch crystal for the clock. And an EEPROM, if you design your own algorithms. It is surface mount, but only 28 pins.

Sean Costello

Arno van der Heijden

Re: Spin semiconductor fv-1
« Reply #5 on: October 25, 2006, 06:04:25 PM »
Very interesting indeed!! :icon_mrgreen: :icon_mrgreen:

This looks like a perfect opportunity for me to get started with DSP.
Does anybody have an idea how much such a chip (and eventually the developers kit) will cost?

This brings me to something I'm curious about: Assembly vs. C++.
My C++ knowledge is very limited, but my knowledge of assembly is non-existent. As I understand it programs for this chip have to be written in assembly and there's no compiler available that will compile C++ code into the right format.

Is assembly normally to be preferred above C++? How hard is it to learn assembly?

SeanCostello

Re: Spin semiconductor fv-1
« Reply #6 on: October 26, 2006, 01:44:04 AM »
About the cost: Contact the distributor on the page. It seems fairly reasonable - more than some of the DSP solutions, but probably cheaper when you add in the built-in ADC/DAC, delay memory, and controller DACs. The development kit is also reasonable.

The FV-1 can only run 128 instructions per cycle. C++ just would not compile to efficient code (unless you create inline functions for C++ that are simply macros that map to the assembly instructions). The assembly syntax for the FV-1 is compact and VERY powerful: you can code an allpass delay in 2 instructions!!!

One caveat: the FV-1 does not look like the type of chip you can program an unlimited variety of DSP algorithms on. It does not have the horsepower for FFTs, bandlimited distortions are probably out of the question, the coefficient quantization imposes limits on what you can do, etc. On the other hand, delay buffer management is taken care of for you, the built-in LFOs are very powerful, and the EXP and LOG units are a dream for those of us who have had to come up with C and assembly approximations of these functions (did you ever use SCALB in C++?).

Sean

Peter Snowberg

Re: Spin semiconductor fv-1
« Reply #7 on: October 26, 2006, 09:54:37 AM »
I spoke with their distributor a couple days ago and from what I remember the price breaks are as follows:

1-99: $10 ea.
100-999: $8.50 ea.
1000-4999: $7.50 ea.

Please don't quote me on that pricing schedule.

Once you factor in the other parts required for a design based around the Alesis Semi/Wavefront Semi Digital Reverb Engine, the FV-1 becomes a very complete and less expensive solution. Other little DSPs require an additional microcontroller to provide the kind of flexability and controlability that the FV-1 provides without outside help.

The FV-1 is limited in scope of the algorithms that it can handle, however it is still VERY powerful and it should do what it does rather well.

On programming: Assembly is the only language that the chip actually speaks and for me that makes it much easier to relate to that an abstract C++ program. I've been programming in assembly for over 20 years but I still don't do C++ and don't have any desire to. I just can't relate to C++.

As Sean mentioned above, the built-in EXP & LOG instructions and the addition of sine LFOs are very welcome features! 8)
Eschew paradigm obfuscation

SeanCostello

Re: Spin semiconductor fv-1
« Reply #8 on: October 26, 2006, 04:23:52 PM »
I personally like programming DSPs in C. However, you need a good compiler to do so ($$$$), and you may have to use a lot of "intrinsic" operations, which are basically compiler hints to tell it what assembly operations to use. For the FV-1, you would have so many specialized instructions you would need to use, that it makes sense just to dive in and learn the assembly.

i wonder how quickly programs could be ported from the Alesis AL3201 to the FV-1? My concern with the FV-1 is the opposite idea: that an algorithm I create could easily be copied and ported over to the AL3201, or the Coolaudio V1000 if you want to get more precise about my fears. The FV-1 does not have a method of running encrypted code, so it would be easy to disassemble any code written on the external EEPROM. This would not be a concern for people who are using the built-in programs, or the (excellent) example code available on the FV-1 website.

Sean Costello




puretube

Re: Spin semiconductor fv-1
« Reply #9 on: October 26, 2006, 04:32:00 PM »
have no fear, Sean: I`m just an "as-is"-user...  :icon_wink:
(with analogue surrounding ideas...)


puretube

Re: Spin semiconductor fv-1
« Reply #10 on: October 27, 2006, 03:07:58 AM »
sorry for the confusion, but hey now, what is this?
(if link won`t work, go here,
and search for part# " RA-FX3C " ).

"resolute audio"  :icon_question:

just got this offer from the distributor, that contacted me after having filled in Spin`s
contact-form...


first look seems larger S/N ratio than the Spin`s...

[EDIT:]
OK - I`ve searched after having had a closer look at the pdf picture:
"Nemesis"...

whaddaya guys think?

[Edit 2:] another one... (inferior ?...)
« Last Edit: October 27, 2006, 03:28:52 AM by puretube »

Arno van der Heijden

Re: Spin semiconductor fv-1
« Reply #11 on: October 27, 2006, 04:29:01 AM »
The aforementioned Cooladio V1000 can be found here:
http://www.coolaudio-semicon.com/files/public/V1000_DATASHEET.pdf

How does the Spin FV-1 compare to this and the products mentioned by Puretube?

Peter Snowberg

Re: Spin semiconductor fv-1
« Reply #12 on: October 27, 2006, 11:41:21 AM »
It looks like those boards could easily be based on the AL3201B/V1000. 
Eschew paradigm obfuscation

puretube

Re: Spin semiconductor fv-1
« Reply #13 on: October 27, 2006, 11:59:41 AM »
I`ll give it a closer look with a magnifying glass, once they arrived...

SeanCostello

Re: Spin semiconductor fv-1
« Reply #14 on: October 27, 2006, 03:10:44 PM »
How does the Spin FV-1 compare to this and the products mentioned by Puretube?

I have not worked with either the FV-1 or the Alesis chips. However, the FV-1 was designed by Keith Barr, founder of Alesis, and the architecture of the DSP cores of both chips seem rather similar. I am not sure if the Alesis chips use the floating point representation for the delay lines.

Some of the differences between the FV-1 and the Alesis chip, from my admittedly inexpert position:

- FV-1 has built-in stereo ADC and DAC, while the Alesis requires external ADC/DAC
- Alesis requires high frequency crystal (in MHz), while FV-1 only requires a 32.768 KHz crystal, which should help FV-1 designs in RF testing
- FV-1 has really nice example algorithms, while Alesis has crappy example reverb code (you probably could port the FV-1 examples to Alesis, but that might violate terms of code licence)
- FV-1 seems like the assembly code is more concise. On the Alesis, you need to spend some cycles refreshing memory locations, and I haven't seen this in the FV-1 code.

Sean Costello

MetalGuy

Re: Spin semiconductor fv-1
« Reply #15 on: October 28, 2006, 08:54:44 AM »
Quote
WWW
   
   Re: Spin semiconductor fv-1
Reply #10 on: October 27, 2006, 02:07:58 AM
   
sorry for the confusion, but hey now, what is this?
(if link won`t work, go here,
and search for part# " RA-FX3C " ).

"resolute audio"  icon_question

just got this offer from the distributor, that contacted me after having filled in Spin`s
contact-form...


first look seems larger S/N ratio than the Spin`s...

[EDIT:]
OK - I`ve searched after having had a closer look at the pdf picture:
"Nemesis"...

whaddaya guys think?

[Edit 2:] another one... (inferior ?...)

I'm happy we already have a wider choice for such chips/products without the hassle of PCB making and SMD soldering. Now after the reverb "problem" is solved I think these are /at least for what I need/ still "semi-solutions". I would like to see for example a chip/board which will allow for at least a chorus-delay-reverb FX chain with the option of individual /per effect/ parameters /all of them/ control and presets memorizing.
FV-1 comes close to that but still the parameters control is not flexible enough.

amz-fx

Re: Spin semiconductor fv-1
« Reply #16 on: October 31, 2006, 08:19:34 AM »
I have some chips out of the first batch made, and one of the developers kits/board.  Seems like it cost $130, if I remember right.

It is not easy to port from the Alesis to this chip...  in fact, it is not easy to understand how this chip works.  I can program in several different assembly languages and the instructions for the FV1 are just not intuitive...  definitely NOT recommended for the beginner.  Its instructions are DSP specific and there's not much in the way of basic register instructions....  however, even though this makes it difficult, it also makes the chip quite powerful.

Because many basic instructions are missing, some types of DSP manipulations will be difficult or maybe impossible... the chip is really optimized for reverb/echo/chorus.

You'll need a programmer to load the ram if you intend to write your own code...

regards, Jack

Peter Snowberg

Re: Spin semiconductor fv-1
« Reply #17 on: October 31, 2006, 10:52:49 AM »
The FV-1 is really more of a MAC (Multiply ACcumulate) unit with an instruction sequencer rather than a full featured DSP or microcontroller. It does what it does very well, but it's made for a specific purpose.

It seems a little funny, but the first way to make presets that comes to me would be to use a trio of e-pots with a microcontroller driving them. Analog->digital->analog->digital. :D
Eschew paradigm obfuscation

puretube

Re: Spin semiconductor fv-1
« Reply #18 on: October 31, 2006, 12:41:41 PM »
over here, the DEV-board is more like the "3" as first digit...  :icon_eek:, and that`s in ...

amz-fx

Re: Spin semiconductor fv-1
« Reply #19 on: October 31, 2006, 10:13:36 PM »
over here, the DEV-board is more like the "3" as first digit...  :icon_eek:, and that`s in ...

As in 330?  If so, that's steep!  Do you want me to order you one and mail it to you?

-Jack