10 MHz discrete ADC for measuring oscillatory signal

Primoz,
Are you looking for an ADC function or a DAC function? Your message indicates you are looking for an ADC. It may be more appropriate to address this in the MCU micro controller forum if you need an ARM processor along with an ADC.

The High Speed Converters forum mostly covers discrete ADC and DAC products ranging from 10Msps to 4Gsps in 10-16 bit flavors.

I will forward your request to the MCU forum.

Ken

For TivaC there's the launchpads. There's the TM4C123 launchpad with 80Mhz max clock MCU and the TM4C1294 launchpad that has a MCU with a maximum clock of 120Mhz.
Note that I referred to the maximum clock speed, you can of course run the processor at lower speeds!

Both boards are pretty cheap with a price in the Estore of 12,99$ for the TM4C123 and about 20$ for the TM4C1294 (can't quite remember on the top of my head).
The TM4C123 launchpad has a ADC rated to a maximum of one million samples per second and the TM4C1294 launchpad has the ADC rated at two million samples per second. Note that getting these sample rates is not that easy, it involves using the DMA for data fetching among other considerations

These items - w/in your post - confuse/confound:

a) 10MHz DAC for "lock-in" operation - I've been in this field beyond 25 years - enjoyed some success - don't believe "lock-in" is (yet) a universal term - as your usage suggests

b) you list "parallel communication" as "best" - but provide no clue as to the recipient of such communication.  Parallel data transfer is (usually) fastest - but speed is only one measure of "merit."

c) you note, "price is "somehow crucial" - my experience is that price mates often with degree of features, functions & performance - required/demanded/delivered.  Luis suggests a good "initial building block" approach - your job is to read, read, read - and determine a far more detailed "feature set" prior to developing any meaningful price data.

Your (unexplained) use of DAC in post's title suggests that "proofread" may require some tweaking...

Hi CB1, yes and forever missing details to what is application...
This forum is a disaster, TIVA unrelated and unfiltered post...
If someone filter initial request on new user then can do some refinement then move to proper area... All forum can get more readability and information at tip.
10MHz Dac or ADC too I fear if some real time filtering has to be applied is out of target for TIVA. I read datasheet and discovered TIVA has DSP like instruction but a filter @10 or more MHz hogs processor on few taps.
Can be managed by some C2000 with CLA but forever IMHO domain of High speed DSP or OMAP or high end ARM / FPGA pre post processing.

Yes Roberto - forum management "clings" to hallowed "status quo" - despite cascading evidence that an, "Unguided, undirected (i.e. unmanaged) forum meets beginning posters' (always apt) issue description, "Does not Work!"  That's poetic - is it not?

Forum bosses claim there is no "status quo" - yet numerous suggestions for improvement sit untested - tossed aside - without any attempt at evaluation. 

Beginning poster seeking a, "10MHz, Locked In DAC" may best make that request of, "locked-in" forum management.  (seemingly entrenched in "locked-in, status quo" despite cascading evidence that, "Do nothing" surely leads to posters' (oft repeated) and building cry, "Does not Work!")

cb1_mobile said:

Beginning poster seeking a, "10MHz, Locked In DAC" may best make that request of, "locked-in" forum management.  (seemingly entrenched in "locked-in, status quo" despite cascading evidence that, "Do nothing" surely leads to posters' (oft repeated) and building cry, "Does not Work!"

 This is an unpleasant consideration but also from professional work forever this "doesn't work" came forever from habit of same country... They are not able to do what they claim and cry doesn't work, example doesn't work for our project, we have to sell to our client, your example doesn't work...

 Then a lot of consultng  firm closed by starvation or bankrupcy....

 Doesn't work... Tell to commercial say some eng... from some country are genius just why pay'd nothing.. and you have the unsustanaible forum too.

 This remember also another {Kid [beginner guru] aspirant} cry asking for examples exactly close to what it need to sell his product... Cry and cry, and offend who know and refuse to solve for nothing, this is just change two basic HTML or C functions,.. CB1, I think I , you and Amit too have to quit here and leave this asylum at his destiny then producing again  a valuable thing to sell to a professional market... Single unit Arduino like to sell to hobby people DOESN'T WORK, leave them sell 20 year old technology for a so high price than waste away latest one or remove from professional style or support on separate avoid poisoning all other one.. but we leave to commercial foolish of all profit from debris.... I fear bad feel for TIVA and my project too have another possible NRND failure stop. .:(

Dear Ken and All,

Thank you for all your answers and please excuse me for incomprehensibility of my initial post. Proofreading of it by myself before posting it would certainly be necessary. In my defense, I did it, but this was after I was cruising ADCs for more than 3 hours on Farnell, Ti, AD pages...
Anyhow to start from the beginning. The title should read

10 MHz ADC FOR MEASURING SINUSOIDAL SIGNALS FASTER THAN 1MHz

Using sinusoidal signals at input to analyze the response of a system is widely used, since by simply measuring the amplitude and the phase of the response one can obtain system characteristic response time. There is commercially available equipment that measure exactly this - Lock-in Amplifier.
I want to measure a response that can vary from approximately 0.5 us to 5 us (micro seconds) in real time - fluorescence lifetime response. The measured photo diode signal is quite noisy (>10MOhm amplification @ bandwidth > 1MHz), but can be filtered by analog filter / ADC buffer. This response is quite easily measured either by commercial Lock-in amplifiers or oscilloscopes, but I want to implement it in affordable handheld equipment, therefore hoping to get the processor + ADC part for less than 200US$.

From the theory (Nyquist) the digitization rate should be at least 2x the measured bandwidth. With nice design, knowledge of the driving frequency and fast(er) sample and hold part this can be even surpassed, (EG&G 7265 for example).

But "life" and data analysis (FT, or fitting) is much easier if ADC rate is a bit higher. From above I came to a conclusion I wish a digitization rate of ~10 MHz. The ADC resolution can be low, since the Lock-in technique is based on averaging.

I would like to use a "simple" ARM processor - Tiva C Launchpads (80-120 MHz) for data acquisition and analysis. They have implemented ADCs with a rate up to 2 MHz, but I would prefer faster for ease of use and future compatibility (for response faster than 0.5 us).
So I am basically looking for a sub 100US$ evaluation board I can use for my purpose (or an ADC I can solder myself). I will perform data analysis after the acquisition, so processor only needs to be fast enough to collect the data in real time (10MHz x 10bits serial is probably to fast, therefore I suppose parallel output is needed on ADC).

A note on a "driving" signal. It would be great to have a > 10 MHz DAC for sinusoidal waveform generation, but a simple ON/OFF driving will be used to drive the system. Anyway theoretically - with fast enough digitization - there are a lot of sinusoidal functions in the square wave :)

If someone can point me to a specific part (ADC or preferably evaluation board) it would be highly appreciated.

Primoz

Hello Primoz

Based on your inputs I would suggest going through the following page for the EVM's

www.ti.com/.../high-speed-adc-greater-10msps-tools-software.page

Phase-2 of the project would be to see if one of the EVM's mentioned have a serial interface to transfer the data. Please note that even @ 10MSPS would amount to 100Mbits per second. So the controlling uC has to have an interface which can support such data rates,

Regards
Amit

Second post displays far more thought, effort than initial (mis-titled) one.  Lock-In minus "Amplifier" may not be universally understood.

KISS seems missing in the suggested (necessity) to "rush the acquired data" from the MCU.

Earlier you mentioned "parallel" and MCU's ports enable 8 bit data - w/2nd port providing needed strobes & command signals.   More expensive MCUs here employ EPI - automating, extending & (likely) speeding the external data transfer process.

Again - vastly improved 2nd effort - project is of interest - wish you well...

Primoz Kusar said:

If someone can point me to a specific part (ADC or preferably evaluation board) it would be highly appreciated.

 Hi Primoz, before to drive you somewhere to hell.. then some question about:

 Why a lock in amplifier cost so much?

https://orders.thinksrs.com/OnlineOrders2003/html/AddToCart.asp?EPrdGroup=SR810

 This is first point to think about, ok this is a not so widely diffused system, again some more clean has to be applied..

 Why lock to so high frequency and why use a lock in amplifier to your analysis of fluorescence "lifetime" response?

 Again what mean for you lifetime lifetime?

 ( see also second post after this, I post separate due to policy of forum avoiding delay on post.

 Why use a lock in amplifier I don't see how to lock to reference of what?

 Again

Primoz Kusar said:

A note on a "driving" signal. It would be great to have a > 10 MHz DAC for sinusoidal waveform generation, but a simple ON/OFF driving will be used to drive the system. Anyway theoretically - with fast enough digitization - there are a lot of sinusoidal functions in the square wave :)

 Sinusoidal reference are the lowest jitter, using a square wave reference is worstened, are you meaning about reconstructed signal phase reference only?

 Precision of lock in amplifier come from time-frequency indetermination, from poster word i see this:

Primoz Kusar said:

I want to measure a response that can vary from approximately 0.5 us to 5 us (micro seconds) in real time - fluorescence lifetime response.

 SO is appropriate use a similar instrument to analyse FLIM?

Primoz Kusar said:

I want to measure a response that can vary from approximately 0.5 us to 5 us (micro seconds) in real time - fluorescence lifetime response. The measured photo diode signal is quite noisy (>10MOhm amplification @ bandwidth > 1MHz), but can be filtered by analog filter / ADC buffer. This response is quite easily measured either by commercial Lock-in amplifiers or oscilloscopes, but I want to implement it in affordable handheld equipment, therefore hoping to get the processor + ADC part for less than 200US$.

 Second section analysis, here I suppose a FLIM is from idea...

http://en.wikipedia.org/wiki/Fluorescence-lifetime_imaging_microscopy

 Is your goal this? So your choice to hell is this room? Good we can get you very hot in few TIVA clock cycle:

 This one if you try obtain from pulsed light and light is from which one of method?

 You have to deal with a convolutional integral and if this is your goal then I can focus on light phase detection as from optical reader...

 So IMHO TIVA is out of range or data presentation/transfer secondary rule...

 So I think from lock in the Hypothesis of frequency modulate pulse light? If so scattering require not just  bandwidth of analog probably got from .5uS therm instead of what is information content of spectral of noise to signal not monochrome.

 Again time correlation and convolutional IMHO drive this application to multicore DSP arena or FPGA, DSP can be TI FPGA from Xilinx or Altera or close competitor of High end.

 Multicore DSP can be the cheap solution, not the simplest.

 Noise from photodetector, light source, pulse shaping etc... require resolution in the pS order, this is a non trivial task and can be your definitive hell.

 This bias forever resolution of ADC sample rate and noise floor too, I think no less than 1Ghz or half of it ADC sampling rate can be applied nor less than 12 bit depending from s/n and pulse shape...

 If this is third post with photodetector instead of ADC and DAC but again I fear target price can be matched....

Primoz Kusar said:

I am trying to measure sinusoidal signals up to 1 Mhz to obtain the phase and the amplitude of the signal. Since the signal is noisy high resolution is not needed.  I would also like to use a simple ARM processor (Tiva  C) for data acquisition so therefore parallel communication should be the best? (FPGA solutions would probably be too time consuming at this moment).

 So if precediing two chunk fosused then do last think:

http://www.hamamatsu.com/jp/en/C11367-14.html

 this instrument is out of your target so:

Primoz Kusar said:

already exist, but with a price tag not much above 100US$

has to be read as 200K$... ok we know we can do, avoid Arduino, PIC like all can be done with these low computing power,....

 You are in the VERY HIGH SPEED Arena and CB1 can confirm you a well shaped sharp (close to soliton) is not so simple to generate. Don't be fooled by processor speed and toggle rate, it doesn't work!!!!

 So instead of digging low speed ADC (10Mhz is very less two order magnitude) think in therm of better insrtumentation and ask what primary sensor cost to you.

 Yamamay site can drive you in the right reasoning way, if this is the best answer we can move on appropriate area, forget TIVA as primary number cruncher, it cannot do nothing, high end are to be applied OMAP + MultiDSP, are you prepared for?

 So its poster diligence to get a way and choose if we got the underway to his hell.

 After too many children if this I am interested too. Also in nuclear mass spectrography, can be done at low cost but nos so low as 200$, raise at almost one order magnitude but .. no real limit.. if you perform better and have correct biasing about we can proudly help.. CB1 or poster turn.

 --- P.S. I am not so healty so please check if something sound strange or unreadable too... I promise read back and if inappropriate or wrong fire over my writing too but not soon..

Primoz Kusar said:

ertainly be necessary. In my defense, I did it, but this was after I was cruising ADCs for more than 3 hours on Farnell, Ti, AD pages...
Anyhow to start from the beginning. The title should read

10 MHz ADC FOR MEASURING SINUSOIDAL SIGNALS FASTER THAN 1MHz

 Forgot...

 if preceding brainstorming of possible application of your text failed...

 please can we have idea from where these sine signal come and if not foolished again by spectral content how many sine and changing at what frequency or time?

 If sine is FM or PM some communication analysis can also help isolate spectral line (if not dense spectral area) by evaluation of modulation bandwith and or phase speed as time derivative of frequency or spectral density by probability of...

 This is another source..

  Is TIVA for this? I think no without an FPGA front end for non real time analysis.

Friend Roberto has (clearly) gone, "above/beyond" all prior posts "pale" in comparison.

May I second Roberto's guidance - most always it makes great sense to study the methods & implementation of industry leaders - and then "hope" that your cleverness & use of newer/faster/superior devices - leads to some "notable" advantage.  That "notable" is critical - you must be able to demonstrate marked superiority (usually price, performance, features) - to make serious inroad upon established/entrenched competitors... (decade ago my small tech firm did just that w/LCDs/Plasma displays - sold so much - so fast - that we took firm public.  And firm continues today - although all founders have "walked plank" {departed}...)

Many firms succeed by operating "behind the curve."   You hint (or that's my biased read) at a lesser performing product - but hugely price advantaged - and with performance, "Good for G'ovt Work!"   Portability, battery power has great appeal - and (of course) NO (ok few) serious electronic products exist today w/out a good display.  Simple, eased User Interface is most necessary - if you are, "Lone Ranger" it's unlikely you'll succeed.  (unusual for one person to harbor all of the required areas of excellence...)

An interesting project - but one which may require "l'assistance" beyond "Tiva only/always - as universal Kitchen sink."

cb1- said:

An interesting project - but one which may require "l'assistance" beyond "Tiva only/always - as universal Kitchen sink."

 Kitchen sink where we cook or sunk too, as reference this is another overcook of the past:

http://e2e.ti.com/support/microcontrollers/tiva_arm/f/908/p/344459/1204066#

  I like do this, I like a lot than solve simple C error or two HTM lines... ;)

 But ok kitchen sink has to be wash out from some blood of beginner not believing on us then "peeling"  his finger with knife too.

 Nor knowledge nor skill can bet know how cook of both.

Dear Everyone,

thank you all for your effort. I understand there are a lot of problems coming my way, but I can deal with mathematics and physics of the problem and the subsequent analysis (programming). I can have a system quite nicely running with an oscilloscope and PC. Tiva with 120 MHz and floating point capability should therefore be quite OK, since data analysis can be performed after the data acquisition and not necessary simultaneously.

Therefore I was hoping for a response like this:

"There is a nice ADC from Ti capable of 10 MHz conversion. Like ADS802, ADS804, ADS850. Ti team has gone into a great effort of preparing an evaluation board for one of these with direct compatibility to Tiva C boards. You can order them at Ti shop. There is also some software available..."

Since this is not gonna happen, I was at least hoping for.:

"There are these two nice Ti's ADC ADS802-ssop, ADS804-soic capable of 10 MHz digitization (40 US$) you can soldier yourself.  I am afraid there is no evaluation board readily available at his time, so you will need to make your prototype board yourself, but with 40US$ chip, 15US$ front end, 20 US$ board, 10 US$ clock plus some change you might be able to end up below 100 US$ mark."

So if someone can add to this "last response", either knowing for a better chip (ease of soldering, at lower price, same speed, parallel com.), some low price evaluation board, or has experience using either of the two, layout or software, or an idea why one is more suitable than the other, the response will be appreciated.

Best regards

Primoz

Primoz Kusar said:

I was at least hoping for.:

Mon ami - how are your (many) helpers here able to "detect" your hopes?   Have you reviewed your initial post - not much usable data lurked therein - I'd say.  Does "hoping" harvest the same (effective) response as, "clearly framed/detailed, direct questions?"   Appears that you "share some blame" for your dissatisfaction...

This vendor - thru past acquisitions of Burr-Brown, National Semi - is rich in powerful & specialized ADCs.  Should your interest "peak" (as it has in your most recent posting) upon ADCs and the ready availability of proper, high frequency ADC Eval boards - should not you have posted in an "ADC-centric" forum?   (that's our "hope.")

May I note that you (still) have not provided any real, "in depth" spec requirement for your "hoped for" ADC.

Is it not true that you are attempting to "off-load" your serious investigation efforts onto others?   And - that much of this "off-load" falls far from the MCU-centric nature of this MCU forum?   Suspect that's the best explanation for your request's full/total satisfaction, "Not gonna happen!"

Primoz Kusar said:

hank you all for your effort. I understand there are a lot of problems coming my way, but I can deal with mathematics and physics of the problem and the subsequent analysis (programming). I can have a system quite nicely running with an oscilloscope and PC. Tiva with 120 MHz and floating point capability should therefore be quite OK, since data analysis can be performed after the data acquisition and not necessary simultaneously.

 Ok I feel some kidding are alone, so its time to shut down..

 What is running on a scope? what type of scope? What type of pc is involved and what software?

 So after cloud came out of, sun shine over wet soil and fog appear with iridescence of light spectral scattering....

Primoz Kusar said:

Therefore I was hoping for a response like this:

"There is a nice ADC from Ti capable of 10 MHz conversion. Like ADS802, ADS804, ADS850. Ti team has gone into a great effort of preparing an evaluation board for one of these with direct compatibility to Tiva C boards. You can order them at Ti shop. There is also some software available..."

 Is this a calling for free software? Ok we are at usual request same as from past and forever...

 You can make and build your own board, usually an ADC board RF grade doesn't cost less than 100$ probably 99$ so you can make. You can buy a better ADC for same price too.

Primoz Kusar said:

Since this is not gonna happen, I was at least hoping for.:

 Usual wording again appear...

Primoz Kusar said:

So if someone can add to this "last response", either knowing for a better chip (ease of soldering, at lower price, same speed, parallel com.), some low price evaluation board, or has experience using either of the two, layout or software, or an idea why one is more suitable than the other, the response will be appreciated.

 At last we can say this application is out of our control and drop in unclear tech capability of performer...

 No data, no details, no ability to sustain high tech level forever adhere to Einstein rule:

 http://www.amazon.com/Albert-Einstein-understand-Photograph-Photography/dp/B00J0D299S

 I wanna one with instruction on how to stick on this forum!!!

Primoz Kusar said:

evaluation board readily available at his time, so you will need to make your prototype board yourself, but with 40US$ chip, 15US$ front end, 20 US$ board, 10 US$ clock plus some change you might be able to end up below 100 US$ mark."

 Cost of this board is similar to one you choosed youst some better performance:

http://www.ti.com/tool/ads6125evm

 So you are able to build an impedance controlled board for so less than 20US$?? Please can you tell me where to  build some for me too I never found impedance controlled for less than some hundreds uSD?

Roberto Romano said:

 I wanna one with instruction on how to stick on this forum!!!

NO rules, regulations, nor "template-based, poster guidance" seems to work...

Nice to (always) add, "I wanna", "I need", and newest, "unexplained hope" - minus any usable detail - to guide hapless responders.

Poster's reach may exceed his grasp - his "thanks" is severely clouded by his statement that all here have (somehow) "missed" his unexpressed hope!

Several here may qualify as "savants" not psychics - despite poster's great hope...

Dear Poirot,

yes my first post was quite poorly written. But again in my defense I did post it on the ADC-centric forum. As the history goes it was instantly moved to this one (see the first response of this thread). My second mistake I did not start a new thread at the ADC section.

My understanding of the forum is the following: Someone asks a question, than someone else who already has the knowledge - if there is such a person and is willing to answer it - answers it. Resulting in much help for the first person without much work for the second. So it was never my attempt to off-load anything.

Just in case someone fitting the second person description of my understanding of a forum comes by (for my particular problem anyway) - you know the title of that Bond movie...:

I really do not have any "in depth" spec requirements: ~10 MHz, >=10 bits, parallel output (with levels somehow compatible with Tiva c processor - we are in the TM4C Microcontrollers Forum :) ), hopefully nicely packed in evaluation board, sub 100 US$ for complete ADC package (it feels like I am repeating myself). I can manipulate the input signal, I can do the programming for data analysis, I can take care of the power supply.

LP Primoz

Primoz Kusar said:

Dear Poirot,

See - when (properly) motivated you do have (some) investigational capabilities.  (and - my time @ engineering school & then law-school may justify the name...)

Surely it's ok to "fish" for the 1 in 100 (or 10,000) who may have (overlapping) recent experience - and be willing to "donate" much time/effort - your behalf.

But - when thanks is "backhanded" - and over-weighed by condemnation for failure to "tease out" your "hope" - motivation to assist disappears as fast as that you seek to measure...

cb1- said:

Several here may qualify as "savants" not psychics - despite poster's great hope.

 Uh sir, sorry I was some way kidding too much but I know

 <<UBI MAIOR>>

<< Semper minor cessat!>>

Ancient past quote, forever to your kind pleasant mastering :

<<scire quod sciendum>>

Dear Roberto,

thank you for your reply. Really a nice card, I am looking for something like this, but this one is a bit of an overkill for me (the speed - I doubt TM4C could handle it - and the price - I need a cheaper solution). Anyway as far as my understanding goes 10 MHz does not really call for an impedance controlled board - certainly would not make any damage, except to the valet.

I am pretty sure the suggested board also comes with some "free software" :)

Regards,
Primoz

Maybe some VHDL are with the one I pointed to, Altera has a board fit there, I am sorry say board can cost ten times the ADC ...

http://www.terasic.com.tw/cgi-bin/page/archive.pl?Language=English&CategoryNo=13&List=Simple

 if you wish free some shelf space I am proud sell you an old Stratix main board for your target price excluded shipping and handling ( may be export rules apply), that one securely fit seamless ADC board and securely solve DSP power need..

 Ah.. post was no FPGA, so why am I using FPGA on EPI s front end of TIVA I don't remember, it was an 6Mhz only....

 At least you can also fit fast DAC to TIVA EPI port, still we don't know how deep is memory sample nor what is sampled, so initial post is fighting to other...

 Not so high resolution... SO are 6 bit ok? Then appear a 12 bit with worst S/N, for less price you can get the one I suggested, better speed Floor Noise and 12Bit too.

 What can I suggest to mystery project?

 Sunk it as Iceberg sunk'd Titanic too?

 Your pointed ADC has some 65 s/n from 6.02dB bit we can see it is just 10.79 effective bit or less, the one I suggested is 12 bit before to see =>10 bit.. I don't like changing target, I can shot moving but not evanescent one!

 Is application FLIR? What  method apply and what you are expecting from FLIR result ?

 Evil forever lurk but we have some working detector now ;)

You guys should really point out to me there is this More button bellow every post, that enables one to edit his old posts. I slightly changed my first post for better clarity. Admittedly it makes other posts less clear, but for the sake of future readers it might be of some benefit.
And yes the way I see it I am fishing for 1 in1....., sort of like hitchhiking. With a little help I might get somewhere.

Primoz Kusar said:

You guys should really point out to me there is this More button bellow every post

You have (apparently) very high expectation of others - do you not?   Resourcefulness seems not (too high) your list.  (note: no one told Roberto, myself, many others of the "more" button - it is useful)

Returning to high (and now extremely unrealistic) expectations of others - I've never (successfully) had the first car passing, brake to a stop!  (then again - I'm not young, very pretty, roadside (would be) passenger.)   And - should we not note - first forum "disgorged you" - their first opportunity!  1 in 1 - fast/free - not a chance!

Unrealistic expectations - from bitter experience (myself, other small biz, VCs) will NOT prove your friend - in fact are likely to "repel" those who may be best able to assist.  Bit more tact, diplomacy - real "thanks" (not masked as being outside of unexpressed "hopes") should serve you much better...

cb1- said:

real "thanks" (not masked as being outside of unexpressed "hopes") should serve you much better.

 Hi CB1, I just think this in response of new title...

 Is this appropriate answer to topic? So I get the right way to home...

There is no secret to my project, I simply do not see how what I am gonna write in this post is gonna help:

I am making a hand held equipment to measure a fluorescence lifetime. Lifetime of this particular fluorophore is in the range 0.5 - 6 micro seconds. Let me point out state of the art lifetime measurements techniques measure lifetimes shorter than 100 ps (TCSPC), there equipment is in the expensive range. 20kUS$ is needed (even for 1 ns lifetimes). This technique includes pulsed lasers, and single photon detectors with response faster than 1 ns. Plus a card similar to what you suggested (speed, price etc...).

Frequency domain lifetime measurements is another technique used - One simply measures how fluorescence light is shifted (phase) to the sinusoidal excitation light. So the system as such can measure lifetimes faster than the bandwidth of the system. State of the art of this technique is in low tens of nanoseconds. The serious disadvantage of this technique is much more light is needed. Photo-diodes tend to be much faster if enough light is available.

The third technique is simply to observe the temporal behavior of fluorescence light after a short light pulse. This one is least suitable since one needs everything (detector, pulse, and acquisition) faster than the lifetime. Would not know what state of the art is. And even more light is needed.

But let me point out: my goal is to measure something way out of state of the art! 1 us >> 10 ns or even 50 ps.

My application is - lifetime changes with change of environment - therefore we have a sensor.

Currently I have a two stage PD amplification 500000 V/Ohm OPA transimpendance amplifier plus 20V/V OPA which gives a bandwidth close to 1 MHz and enough signal even to measure the lifetime (1 us) directly. Rigol for 800Eur (100MHz) can easily do the measurement.

And for excitation a simple medium current LED is used pumped with OPA - 100 mA current with better than 10MHz response.

Rigol as such is not really an option for hand held equipment for obvious reasons, but any digitization with 10 MHz should do the trick. I will use a square wave to for excitation, since linearity of LED with OPA (current control) is quite bad, and I do not have the signal generator anyway (except for the square - TM4C). This does not really matter since square can be described with few sinusoidal signals at different frequencies pretty well giving us even additional information. The fluorescence can be described with a linear differential equation - this is why the additional information is possible.

In theory one can measure phase and amplitude of the signal all the way to the digitization rate of the used ADC, but extreme care needs to be taken for a correct timing.

Many reasons for the high price of the Lock-in are: Scientific equipment, very nicely made, phase/frequency lock is extremely nicely done, sensitivity is just beautiful. But what they sell for 5000US$ is actually 20 years old SR810 or LG&G 7650 (different name now but still the same model - I've been using it in 1999 already). Now imagine any other equipment so old and so expensive - a Ferrari maybe? Anyway ADC in 7650 runs at the 180 kHz, but it can measure all the way to 250 kHz. So some nice knowledge is implemented.

But I do not need any of the above, since I can lock my excitation modulation directly to the ADC clock - for this somehow higher ADC rate is needed than signal modulation - therefore >5 MHz, TM4C at 100 Mhz should also be able to read a parallel port at 10 Mhz, I can fill half of its memory with the measurements. And use the rest of it for analysis later. The signal should be to some degree constant so my front end can be quite simple.

Since many repetitions can be performed - only 10 points per oscillation compared to the MCU memory... and the signal is quite noisy even the most significant 8 bits should be sufficient to get the phase (with proper signal amplification) - the advantage of the technique. I have no Idea what S/N is nedded - a single scan (non-averaged directly from detector) gives S/N ~5 (linear), but with 100 averages it is 10x better (experimentally).

I have never used a FPGA - I do not wanna go there right now. We will not sell the equipment if it is gonna be above 1000 US$, so here I am...
... hoping for a suitable ADC solution I can use.

Regards
Primoz

Primoz Kusar said:

here I am...
... hoping for a suitable ADC solution I can use.   

You've provided much, most useful data - your past, several writings.

Yet - quite clearly - the MCU's impact/effect upon this project is "dwarfed" by the analog requirements.

Vendor's "Data Converter" forum seems so much "more appropriate" - might you have, "given up" (there) too soon?   Only when you can narrow your selection of the Analog sub-system would MCU "mating" seem fully appropriate.

Suggest you "repost" there - especially these past 2,3 data-laden, analog-centric posts.

Your "hitching" attempt - targeting a "near abandoned/improper" county road - seems less effective than one along major interstate (i.e. Data Converter) forum.  Your (past) "miscasting" of the post's title may have emboldened forum crue (there) to cast you away - their hallowed turf.  Clearly that's your best bet...

<Hi Primoz, from some your word I inferred you are in some part of Euro area, so from your other word you are using from 1999, you are in some area of research or manufacturing, but I cannot focus what are you targeting measure nor what are your goal...

 FLIR, ok Phase, Amplitude,

Light excitation can be

Dirac delta like (soliton) very sharp pulse then convolutional filter...

Single Sine Pulse

Single sinx/x pulse (Six)

Sinusoidal modulation single spectral line... I don't see how you can get useful data

Sinusoidal modulation FM or PM modulated too, this case syncronous IQ demodulator analog or digital apply to demodulate response and further processingon decoded data.

  About ADC if you require as I read evaluate 50pS this is related to noise floor in first, so a bigger oversamplig must apply..

 You state using a RIGOL 100MHz, samping I suppose is at least @400Mhz if chepest but 100Mhz generally require 500Mhz-1GHz for better sampling window

 And finally....

Primoz Kusar said:

Since many repetitions can be performed - only 10 points per oscillation compared to the MCU memory... and the signal is quite noisy even the most significant 8 bits should be sufficient to get the phase (with proper signal amplification) - the advantage of the technique. I have no Idea what S/N is nedded - a single scan (non-averaged directly from detector) gives S/N ~5 (linear), but with 100 averages it is 10x better (experimentally).

 this sound me as worst measure theory i am not sure you Have real Idea of what is Jiitter nor noise figure nor Noise floor nor quantization.

 This assertion stride too much...

Hi Roberto,

Yes I am using a Gs/s oscilloscope, gives me great insight into my detection system, but in no way am I needing this bandwidth to extract the phase. As is somewhat schematically presented here: 

http://www.lambertinstruments.com/technologies-1/2014/12/4/frequency-domain-flim-for-beginners

One can measure this even at a sampling rate at twice the modulation frequency, or even less - as is the case with 7265 model (sample and hold bandwidth is much higher I guess). In this case Jitter and all the nasty stuff really needs to be taken with special care, but having 10 times the frequency - life should be much easier then.

I am from Slovenia, shifting from research to "sort of" industry.

Regards Primoz

Primoz Kusar said:

Yes I am using a Gs/s oscilloscope, gives me great insight into my detection system, but in no way am I needing this bandwidth to extract the phase. As

 If your instrument support it, try reduce sampling rate to see what happen on low bandwidth and if possible use an external clock as you wrote to sampling. I don't own Rigol I see is quite diffused due to low cost and acceptable performance, both my Agilent and Tek have this feature.

Primoz Kusar said:

http://www.lambertinstruments.com/technologies-1/2014/12/4/frequency-domain-flim-for-beginners

 This is a microscope stereo FLIM imaging, is your goal close to this?

 leave me some time to think about and what problem can come from then I got here again.

Primoz Kusar said:

I am from Slovenia, shifting from research to "sort of" industry

 I am from worst close to you country Italy, I suppose you wish start some marketing of instrument, we can also  hear from CB1 maybe we can do something to help you better focus target.