Opinions, on this newfangled thing, please

I just ran across this:

http://www.eetimes.com/author.asp?
section_id=36&doc_id=1328618&_mc=sm_eet_editor_maxmaxfield&hootPostID=09e55671236236acbe4121d86cb78c72

http://tinyurl.com/zhdcerx

It looks like it could be a nifty thing to use in certain circumstances, 
particularly where one needs a complicated analog block in little space 
with fairly high bandwidth (yes, I know -- it's digital inside, but I 
don't care about that if it's analog outside).

Anyone have any mileage with the company or any predecessor products?  Do 
you know of any competing products out there?  It looks like a good arrow 
to have in my quiver, if it can meet the expectations they're trying to 
set.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

On 07/01/16 19:14, Tim Wescott wrote:
> I just ran across this:
> 
> http://www.eetimes.com/author.asp?
> section_id=36&doc_id=1328618&_mc=sm_eet_editor_maxmaxfield&hootPostID=09e55671236236acbe4121d86cb78c72
> 

(If you put < > around your pasted urls, they will work fine in
newsreaders without wrapping issues - there is no need for tinyurl's.)

<http://www.eetimes.com/author.asp?section_id=36&doc_id=1328618>

> http://tinyurl.com/zhdcerx
> 
> It looks like it could be a nifty thing to use in certain circumstances, 
> particularly where one needs a complicated analog block in little space 
> with fairly high bandwidth (yes, I know -- it's digital inside, but I 
> don't care about that if it's analog outside).
> 
> Anyone have any mileage with the company or any predecessor products?  Do 
> you know of any competing products out there?  It looks like a good arrow 
> to have in my quiver, if it can meet the expectations they're trying to 
> set.
> 

I have been trying to find an excuse to use one of these things in a
product - they look so fun!  This new device, with in-system
programmability by I&sup2;C (I'm guessing that you can burn the NVM via the
I&sup2;C, as well as just change the current RAM-based configuration) opens
up new possibilities.  For the previous generations, you had to either
order a gazillion factory-programmed devices or program them by hand
using the development kit.  That's great for people making hundreds of
thousands of boards, and for people making tens of boards, but not for
those making hundreds and thousands.  The I&sup2;C programmability and
reconfigurability makes them interesting for a wider range of uses.

On Fri, 08 Jan 2016 10:10:52 +0100, David Brown wrote:

> On 07/01/16 19:14, Tim Wescott wrote:
>> I just ran across this:

>> http://tinyurl.com/zhdcerx

>> It looks like it could be a nifty thing to use in certain
>> circumstances,
>> particularly where one needs a complicated analog block in little space
>> with fairly high bandwidth (yes, I know -- it's digital inside, but I
>> don't care about that if it's analog outside).

> I have been trying to find an excuse to use one of these things in a
> product - they look so fun!  This new device, with in-system
> programmability by I²C (I'm guessing that you can burn the NVM via the
> I²C, as well as just change the current RAM-based configuration) opens
> up new possibilities.

All of this sounds like sex, but what would I use it for? Maybe I need to 
do more analog (or at least asynchronous) to be able to think of 
something.

On Fri, 08 Jan 2016 15:38:26 +0000, Aleksandar Kuktin wrote:

> On Fri, 08 Jan 2016 10:10:52 +0100, David Brown wrote:
> 
>> On 07/01/16 19:14, Tim Wescott wrote:
>>> I just ran across this:
> 
>>> http://tinyurl.com/zhdcerx
> 
>>> It looks like it could be a nifty thing to use in certain
>>> circumstances,
>>> particularly where one needs a complicated analog block in little
>>> space with fairly high bandwidth (yes, I know -- it's digital inside,
>>> but I don't care about that if it's analog outside).
> 
>> I have been trying to find an excuse to use one of these things in a
>> product - they look so fun!  This new device, with in-system
>> programmability by I²C (I'm guessing that you can burn the NVM via the
>> I²C, as well as just change the current RAM-based configuration) opens
>> up new possibilities.
> 
> All of this sounds like sex, but what would I use it for? Maybe I need
> to do more analog (or at least asynchronous) to be able to think of
> something.

I tend to close a lot of control loops in software; there's a definite 
ceiling in what you can use a microcontroller with because of code 
execution speed.

Something that I can think of off the top of my head for a chip like this 
would be as a controller in a snazzy switching supply, assuming that 
there's enough resources on board to implement a controller as well as a 
proper PWM generator.  At today's switching speeds one can't really do 
cycle-by-cycle control with a microcontroller -- one should be able to 
with an FPGA or CPLD.

-- 
www.wescottdesign.com

On 08/01/16 20:52, Tim Wescott wrote:
> I tend to close a lot of control loops in software; there's a definite
> ceiling in what you can use a microcontroller with because of code
> execution speed.

Have a look at the XMOS processors and boards. Cheap,
many cores (i.e. >10) and execution time for each code
block is /guaranteed/ by the compiler. Low jitter and
guaranteed timing helps in control loops!

Many of the architectural concepts are by Prof David May,
who had a large hand in inventing the Transputer and Occam.

On 1/8/2016 3:52 PM, Tim Wescott wrote:
> On Fri, 08 Jan 2016 15:38:26 +0000, Aleksandar Kuktin wrote:
>
>> On Fri, 08 Jan 2016 10:10:52 +0100, David Brown wrote:
>>
>>> On 07/01/16 19:14, Tim Wescott wrote:
>>>> I just ran across this:
>>
>>>> http://tinyurl.com/zhdcerx
>>
>>>> It looks like it could be a nifty thing to use in certain
>>>> circumstances,
>>>> particularly where one needs a complicated analog block in little
>>>> space with fairly high bandwidth (yes, I know -- it's digital inside,
>>>> but I don't care about that if it's analog outside).
>>
>>> I have been trying to find an excuse to use one of these things in a
>>> product - they look so fun!  This new device, with in-system
>>> programmability by I²C (I'm guessing that you can burn the NVM via the
>>> I²C, as well as just change the current RAM-based configuration) opens
>>> up new possibilities.
>>
>> All of this sounds like sex, but what would I use it for? Maybe I need
>> to do more analog (or at least asynchronous) to be able to think of
>> something.
>
> I tend to close a lot of control loops in software; there's a definite
> ceiling in what you can use a microcontroller with because of code
> execution speed.
>
> Something that I can think of off the top of my head for a chip like this
> would be as a controller in a snazzy switching supply, assuming that
> there's enough resources on board to implement a controller as well as a
> proper PWM generator.  At today's switching speeds one can't really do
> cycle-by-cycle control with a microcontroller -- one should be able to
> with an FPGA or CPLD.

Microsemi bought Actel some time back and their mixed signal FPGA.  I 
have never done anything with it because the price is a bit too high for 
some apps I've had and the analog performance was too low for others. 
Mixed signal can be hard to do well.  Two companies who could do it well 
are Analog devices and Silicon Labs.  They both have produced mixed 
signal MCUs with great analog.  They just need to branch out into FPGAs, 
lol.

-- 

Rick

On Fri, 08 Jan 2016 14:52:13 -0600, Tim Wescott wrote:

> On Fri, 08 Jan 2016 15:38:26 +0000, Aleksandar Kuktin wrote:

>> All of this sounds like sex, but what would I use it for? Maybe I need
>> to do more analog (or at least asynchronous) to be able to think of
>> something.
> 
> I tend to close a lot of control loops in software; there's a definite
> ceiling in what you can use a microcontroller with because of code
> execution speed.
> 
> Something that I can think of off the top of my head for a chip like
> this would be as a controller in a snazzy switching supply, assuming
> that there's enough resources on board to implement a controller as well
> as a proper PWM generator.  At today's switching speeds one can't really
> do cycle-by-cycle control with a microcontroller -- one should be able
> to with an FPGA or CPLD.

Hmm... an idea. :)

Yesterday, I also thought of a thread, but I think it was on 
sci.electronics.design, where it was asked if such-and-such circuit 
exists, and one of the answers being to implement the required device in 
digital.