Measuring Frequency

[ATu]

Hi Dear Digilent Team,

A question concerning the measurement capabilities ot the Analog Discovery 2 and WaveForm software:

Is it possible to measure continously and "without losses" a frequency of an analog or digital entry?
Goal is to log a frequency over a long period of time between EACH following flanks
Optionnaly also measure times between flanks of different channels

Examples:
- Measure the freq/time between each following rising flanks of a channel
- Measure the freq/time between each following falling flanks of a channel
- Measure the freq/time between each following flanks (rising and falling) of a channel

...

Most scopes measure frequency only "around" the trigger flank

Thanks for each feedback

Regards

[jpeyron]

Hi @ATu,

I am sorry it appears your post got overlooked. I have reached out to a more experienced engineer about your question. 

thank you,

Jon

[attila]

Hello,

The "without losses" depends on you trigger rate, the needed sample rate, long time meaning..

There are several millisecond gaps between repeated acquisitions, depending on computer performance and opened measurements, views....
With record mode you can capture up to 10M samples at up to 1-2MHz (without gaps).
You can also create custom script or dedicated application for better performance and "infinite" length processing.

Here for instance Math1 is C1 as long C2 is above 1V, recording for 10sec at 1MHz and freq/time shown in spectrogram view.

Similar posts:

 

 

[ATu]

Thanks attila.

It is still unclear to me how to do this, so I'll try to explain the problem in a better way.

I use a frequency modulated digital signal as input (digital is easier for the description) between 100Hz and 100kHz.

[ATu]

Sorry for split,

I want to measure real time and without the loss of pulses the instantaneous frequency of this digital signal.

At each flank I want to extract, plot and save the frequency or the  time between this flank and the previous flank of same type (rise-rise or fall-fall)

Ideally this extraction should be done on rise-rise and fall-fall flanks of a signal.

Ideally time resolution should be 10nS (0.1% of a 100kHz signal)

How can this be done?
 

[attila]

Hi @ATu

According your requirements it has to save "continuously" the samples at 100MHz.
For 1 bit digital data this means 12.5 MBps. For 16bit analog sample on one channel 200MBps, requiring large device memory or faster connection than USB 2.

With Analog Discovery you can get 100MHz acquisitions of up to 16k samples. The acquisition rate is around 10 ms depending on system performance and processing done.
The Digital Discovery has 256MB memory and lets you acquire with 8 digital inputs 256M samples at up to 800MHz or for 2 seconds at 100MHz , or with 32 inputs 64M samples at up to 64M samples...

[ATu]

Hi attila,

Ideally i want to acquire at 10nS, but it could be done at lower freq.

My first question is to verify if and how I can measure a frequency of a digital or analog signal,
meaning I'd like to plot on the scope the instantaneous frequency of a signal which is FM modulated.
  On each rising or falling flank it plots 1/time separating this flank to the previous flank of same type.
  For an analog signal the flank will be the threshold crossing.

Practically it means that for each flank or threshold cross a time stamp must be recorded, and
the difference to the last flank of the same type must be subtracted from it.
1 over this value must then be plotted at this timestamp (X-axis) on the scope Y axis

Do you understand my question?
 

Regards

[gasstationwithoutpumps]

For logging continuous frequency measurements of low-frequency signals (under 1MHz), I use a different instrument:  PteroDAQ on a Teensy LC or Teensy 3.2 board.  The board costs $12–20 and the software is free at https://bitbucket.org/abe_k/pterodaq/

The measurement uses a fixed sampling rate and counts pulses within the sample time to determine frequency (so is not very precise for low frequencies, where it would be better to measure the period).

Alternatively, I've used PteroDAQ to record precise times for edges on a digital signal, with accuracy to about 100ns (limited by variable latency of interrupts on the Teensy boards).  I can then post-process the record to get frequency.  As an example, see https://gasstationwithoutpumps.wordpress.com/2017/06/20/fidget-spinners/ using a photogate to measure the speed of fidget spinners.

[ATu]

Hi

Thanks for the reply.
In fact I already implemented such an analysis into a STM32 and into an FPGA and a Labview Acq unit.
and I am not looking for a way to do this on other hardware.

I find the Analog discovery 2 very nice and versatile tool and want to use it more for measurement applications.
And as it includes an FPGA the type of measurements I'd like to do can probably be implemented easily.

Does a design kit exists to directly access and reconfigure the HWD/FPGA at low-level?
Or does someone has an idea on how to implement the required function?

Regards

[attila]

Hi @ATu

Please take a look at the following post:

 

[ATu]

Hi attila,

Tks for the reply.
Is there another way to implement this real-time frequency measurement?
Could you implement such a function into it?

Regards
 

 

[ATu]

Hi attila,

Do you have a suggestion how to measure a frequency and a duty-cycle in real time with an analog discovery 2 ?
 

 

[attila]

Hi @ATu

I don't think such "gapless" measurement will be added to AD FPGA any time soon.

In case you can't find any device with such feature, it can be implemented relatively simple with a microcontroller or for higher resolution and better precision with an FPGA development board.

[ATu]

Hi attila,

A pitty, i wanted to use this with all other measurement capabilities of this tool.
I have something similar already implemented in a uC but AD converter and other stuff
are far from the one integrated in AD2.

Maybe the implementation of a simple counter triggered on a digital channel could do the job,
I'll look if a can create a soft directly for it.

Regards