Hardware handholding

On Nov 3, 11:19 am, "mitakeet" <mitak...@sol-biotech.com> wrote:
> >No difference with taking a lead from a uC or, parallel pin or serial
> >control pin.  Talking FPGA is just irrelevant at best.
>
> Since I have many (minimum 10, ideally at least 100) independent
> conductors to measure simultaneously,

You can easily find uC with 50 I/Os.  Two of them is far cheaper and
easier to deal with then FPGA.  Alternatively, a uC with a simple
CPLD.

But, first of all, you need 100 amplifiers to bring the nA and nV up
to 1 to 2V.  You are really asking for a custom analog chip.

> I thought the ability of FPGAs to be
> programmable to operate in parallel would be better than having a single
> processor sequentially check each conductor.  Even presuming there aren=
=92t
> problems with propagation delays (I have been warned that such delays wil=
l
> probably put a ceiling on the speed of the monitoring processor), if the
> duration between each iteration of the serial checking is too large then
> there exists the possibility to miss a signal.  I figure to have a reliab=
le
> chance to detect the passing of the molecule I need to check at least 10
> MHz, probably 100 MHz.

You can sample pins at MHzs, but always get zero if the voltage level
is too low anyway.

>You can easily find uC with 50 I/Os.  Two of them is far cheaper and
>easier to deal with then FPGA.  Alternatively, a uC with a simple
>CPLD.

I thought that you could get FPGAs for not much more than twice that of a
CPLD and that it was reasonable to get a 100 pin FPGA for $100 or less. 
Did I do a poor job of research or is $100 reasonable?  At present, I only
need a single prototype of the hardware and presuming the chip idea works
the relative volume of the embedded hardware will be several orders of
magnitude less than that of the chip, so $100 ea. for several FPGAs is not
a significant issue.  

>But, first of all, you need 100 amplifiers to bring the nA and nV up
>to 1 to 2V.  

Can the amplifiers be something as simple as some transistors on a PCB?

>You are really asking for a custom analog chip.

I know that ASIC is the way to go, but prototyping those is definitely out
of my budget.  For a successful commercial product I expect to embed most
of the control circuitry on the same chip, which should mitigate much of
the concern.

Thanks again for your help!

mitakeet wrote:
>> Your biggest challenge is to build the semiconductor sensor to detect
>> pico amps.  Forget the FPGA, it's not likely critical enough for your
>> project.
>>
> 
> Despite your sarcasm, at least you provide something more useful than
> complaining about what a stupid idea my post is.

If you're referring to Vladimir's and my comments, I had no argument 
with your original post, only suggestions on what you may have to do to 
get a professional to help you (hint: professional == gets paid).

> As I mentioned, I have no clue how practical even detecting the change in
> current is.  I will have a 10 nm or so channel (3-5 nm deep) with a fluid
> that is mostly water with some number of charged molecules flowing through
> it.  The charged molecules will be around 1 nm in diameter.  My goal is to
> detect the change in current flow between two conductors that are broken by
> the channel.  I know this sort of thing can easily be done on the macro
> level (conductivity meter), but I also know that things get strange when
> you get down to the nm and single molecule level.  I also know that
> researchers have been able to put 'patch clamps' on the surface of a cell
> membrane and measure the movement of charged molecules through pores in the
> membrane.  I have read anecdotal reports that some researchers have been
> able to isolate single channels in membranes, hence measure the charged
> molecules moving through a single pore (which, btw, would be smaller than 1
> nm), but haven't been able to put my hands on the primary literature which
> might lead me to the technical means they used to detect the changes.
> 
> What I have read makes me think that if I take a lead from an FPGA and
> connect it with my chip that the level of current might not be so high that
> it causes the fluid to boil.  If it is that simple then doing what I need
> should be straightforward.  If getting that to work is highly unlikely then
> it indicates that the initial prototype might need embedded circuitry just
> to know if it has any chance of success, greatly complicating prototyping
> within my budget.

What you want will take sensitive analog circuitry, which is _not_ what 
you get inside of an FPGA.  Your overall product may end up with an 
FPGA, but that's a tactical, not a strategic decision.

> So, is there a chance for helpful advice from what I thought was a forum
> devoted to embedded hardware or just more unhelpful comments about how
> asking for help and being willing to pay for it is a bad thing and subject
> to ridicule?
> 

You said:

-> While this is very low budget, it does have huge economic
-> potential so the usual NDA and consulting restrictions will
-> apply.  If you are curious but not so

----> willing to work for peanuts <----

-> you might still be interested in a bit of a gamble:  if the
-> prototyping is successful the revenue potential within the
-> next 12-18 months is such that full-time employment and/or
-> extensive consulting agreements are quite possible as
-> undoubtedly several versions of the embedded hardware will
-> need to be made and the commercial product will require
-> embedded hardware as well as ASIC design.

Later you insist:

-> Yes, I know all about equity.  Enough to know it is
-> nonsensical to

----> pay someone equity potentially worth millions <----

So you want some poor sap to sign away their future in some all-binding 
NDA, in return for absolutely no promises at all, _and_ you promise to 
pay them peanuts, _and_ you aren't willing to share any equity.

Can I go halves with you on a lottery ticket, scratch it and look, then 
decide if I want to give you your $0.50 back or the winnings?  That's 
what you're asking everyone in this group, and when it's pointed out you 
call it "ridicule".

If you showed any indication of wanting to pay a fair wage for fair 
work, either in dollars or in trade, I'd be on your side.  But you 
don't.  Your posts assert positively that you want someone who will work 
for less than the going rate on something that has a huge market 
potential, in return for your _maybe_ hiring them later for what you 
should have in the first place -- but in return for their charity you 
get to keep all the chances at getting the big bucks.

You didn't think this was a forum dedicated to embedded hardware, you 
thought this was a forum full of financial ignoramuses!

But here's how you can show me up:  Go to some of your profs in that MBA 
program that you're so proud of, and offer to pay _them_ 1/2 their going 
rate in return for no equity what so ever in this venture that may be 
worth millions.  Be sure to tell them that you'll put their name on the 
top of the hiring list for some vaguely worded job if they go for it.

Report back here with their response.

-- 

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

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html

On Nov 3, 9:19=A0pm, "mitakeet" <mitak...@sol-biotech.com> wrote:
> ...=A0I figure to have a reliable
> chance to detect the passing of the molecule I need to check at least 10
> MHz, probably 100 MHz.
> ...

100 MHz bandwidth in pA range (precision is TBD, I assume something
modest like 8 bits) is dreamland given your level of expertise as you
describe it.
It is probably dreamland for anyones level of expertise today, for
that,
(but I am not sure - I have not looked deeply into it).

As for finding someone to hold your hand... most if not all of what
Tim said is correct. There are loads of freelancer cites, I am sure
you will locate loads of helpful timewasters there. For a real job
like that to be done, you need either to pay its real value or do
it yourself.

Didi

------------------------------------------------------
Dimiter Popoff               Transgalactic Instruments

http://www.tgi-sci.com
------------------------------------------------------
http://www.flickr.com/photos/didi_tgi/sets/72157600228621276/

Original message: http://groups.google.com/group/comp.arch.embedded/msg/0f4=
c26371a2efbbd?dmode=3Dsource

mitakeet wrote:

> I am working on a prototype chip using electron beam lithography near
> the technique's limits (10 nm wires).  I need to bridge from the chip
> to the PC world and need some embedded hardware.  I am a biochemist and
> programmer (with an MBA, so I am a 'suite' who wants to grab a
> soldering iron) and while I have studied IC fabrication for a number of
> years, much of the stuff is still pure voodoo to me.  
	(snip)

Ludicrous restriction of trade agreements and laughable remuneration
offers aside, what bothers me most about this post is the seemingly
incongruous level of expertise claimed versus the complete failure to
identify the core technical challenges of this project.

The hand-waving over the whole problem of measuring the current in
question and the focusing on what is - relatively - a trivial aspect of
the solution begs the question as to the validity of the whole project.

"I've got a theory on how to turn lead into gold, but I can't start on it
until someone can help me work out how to melt it down into ingots".

Anyway, it was good for a laugh - I especially liked the offer of possible
full-time employment if he made it big!!!

Regards,

-- 
Mark McDougall, Engineer
Virtual Logic Pty Ltd, <http://www.vl.com.au>
21-25 King St, Rockdale, 2216
Ph: +612-9599-3255 Fax: +612-9599-3266

In article <ZqadnUe7ZdsD7ZLUnZ2dnUVZ_uudnZ2d@giganews.com>, mitakeet@sol-
biotech.com says...
> >You can easily find uC with 50 I/Os.  Two of them is far cheaper and
> >easier to deal with then FPGA.  Alternatively, a uC with a simple
> >CPLD.
> 
> I thought that you could get FPGAs for not much more than twice that of a
> CPLD and that it was reasonable to get a 100 pin FPGA for $100 or less. 
> Did I do a poor job of research or is $100 reasonable?  At present, I only
> need a single prototype of the hardware and presuming the chip idea works
> the relative volume of the embedded hardware will be several orders of
> magnitude less than that of the chip, so $100 ea. for several FPGAs is not
> a significant issue.  

I think you need more details by first checking out some ball park 
pricing and working out some rough complexities required first. Actually
get some real figures to get an idea of what you want.

You do not seem to have a handle of the costs of making your 
prototypes and incorporating FPGAs, soft cores and the like.
 
> >But, first of all, you need 100 amplifiers to bring the nA and nV up
> >to 1 to 2V.  
> 
> Can the amplifiers be something as simple as some transistors on a PCB?

They could be, if you want something potentially drifting with 
temperature, age, and potentially with low voltage and current input
picking up all sorts of stray interference as well.

This looks like a lot of work without a lot more details to work on.

As others have said the FPGA is not your weak point the analog 
front end is. Which if that is not right means the rest of it might as 
well be made of chocolate
 
> >You are really asking for a custom analog chip.
> 
> I know that ASIC is the way to go, but prototyping those is definitely out
> of my budget.  For a successful commercial product I expect to embed most
> of the control circuitry on the same chip, which should mitigate much of
> the concern.

A lot of ASICs start off as prototype circuits, then go through various 
design phases to become an ASIC.

One of my customers is a design house and manufacturing for ASICs
with customers all over the world and all sorts of requirements.
Some of these have 10-20 year lifespans which involve things like
respinning the ASIC every 8-10 years as manufacturing processes
become obsolete.
 
> Thanks again for your help!

-- 
Paul Carpenter          | paul@pcserviceselectronics.co.uk
<http://www.pcserviceselectronics.co.uk/>    PC Services
<http://www.pcserviceselectronics.co.uk/fonts/> Timing Diagram Font
<http://www.gnuh8.org.uk/>  GNU H8 - compiler & Renesas H8/H8S/H8 Tiny
<http://www.badweb.org.uk/> For those web sites you hate

Thank those of you who saw fit to provide some helpful advice.

Mark McDougall wrote:

> The hand-waving over the whole problem of measuring the current in
> question and the focusing on what is - relatively - a trivial aspect of
> the solution begs the question as to the validity of the whole project.
> 
> "I've got a theory on how to turn lead into gold, but I can't start on it
> until someone can help me work out how to melt it down into ingots".
> 
> Anyway, it was good for a laugh - I especially liked the offer of possible
> full-time employment if he made it big!!!

I think we call it "playing fort" over here in
rainy California.

Didi wrote:

[...]

>100 MHz bandwidth in pA range (precision is TBD, I assume something
>modest like 8 bits) is dreamland given your level of expertise as you
>describe it.
>It is probably dreamland for anyones level of expertise today, for
>that,
>(but I am not sure - I have not looked deeply into it).

1pA / e = 6E10/s

Oliver
-- 
Oliver Betz, Munich
despammed.com might be broken, use Reply-To:

On Nov 4, 3:58=A0am, Jim Stewart <jstew...@jkmicro.com> wrote:
mployment if he made it big!!!
>
> I think we call it "playing fort" over here in
> rainy California.

And I thought it never rained there (at least in southern
California)... :-)

Didi