Author Topic: clipping a fast negative pulse  (Read 5040 times)

0 Members and 1 Guest are viewing this topic.

Offline capt bullshot

  • Super Contributor
  • ***
  • Posts: 3033
  • Country: de
    • Mostly useless stuff, but nice to have: wunderkis.de
Re: clipping a fast negative pulse
« Reply #25 on: May 09, 2019, 09:57:51 pm »

Quote
See that quick and dirty schematic, hope you get the point. Component choice is up to your requirements, choose R2 to limit the current into the OpAmp input to a level typically lower than 5mA, better 1mA (check the datasheet and appnotes).
Thank you for the suggested schematic, kapitän, but would that circuit be any more accurate than the clamping diodes? Their forward voltage varies quite a lot!

Yes, the "precision" clipping isn't done by the diodes but the OpAmp. The negative supply is e.g. -0.8V, if you use an RRIO amp, the output voltage follows the input as long as it is above -0.8V, but can't go below. The diodes clip to -(0.8V + diode drop voltage) to protect the amp input.
Safety devices hinder evolution
 

Offline OwO

  • Super Contributor
  • ***
  • Posts: 1250
  • Country: cn
  • RF Engineer.
Re: clipping a fast negative pulse
« Reply #26 on: May 10, 2019, 03:48:51 am »
I don't understand why you can't just clip the signal inaccurately first and then clip it again in software after it is digitized.
Email: OwOwOwOwO123@outlook.com
 

Offline brumbarchrisTopic starter

  • Regular Contributor
  • *
  • Posts: 223
  • Country: ro
Re: clipping a fast negative pulse
« Reply #27 on: May 10, 2019, 05:48:52 am »
Quote
What's there to ADC if your clipping circuit works?
The remaining slow decaying waveform between -0.8V and GND. That is where the actual information in this signal is.

Quote
Quote from: brumbarchris on Today at 06:24:29 am


    Quote

        See that quick and dirty schematic, hope you get the point. Component choice is up to your requirements, choose R2 to limit the current into the OpAmp input to a level typically lower than 5mA, better 1mA (check the datasheet and appnotes).

    Thank you for the suggested schematic, kapitän, but would that circuit be any more accurate than the clamping diodes? Their forward voltage varies quite a lot!

Yes, the "precision" clipping isn't done by the diodes but the OpAmp. The negative supply is e.g. -0.8V, if you use an RRIO amp, the output voltage follows the input as long as it is above -0.8V, but can't go below. The diodes clip to -(0.8V + diode drop voltage) to protect the amp input.

I'll have to look at that in more detail then, could be a winner!

Quote
I don't understand why you can't just clip the signal inaccurately first and then clip it again in software after it is digitized.
We could probably do so, but as the whole system is a bit unblurry now, and we do not know yet what we'll use down the line exactly, we want to make it as easy as possible for it and take the difficulties in the portions of circuits that we are designing now. It might be that later, when we will ourselves know more about these (what type of ADC we'll use, what analog signal processing we will employ before it, how the signal slope varies from product to product etc; we first need to learn some things and perform additional measurements), we could also do what you suggest here.

Best regards,
Cristian
 

Offline OM222O

  • Frequent Contributor
  • **
  • Posts: 768
  • Country: gb
Re: clipping a fast negative pulse
« Reply #28 on: May 10, 2019, 06:07:28 am »
well then let me tell you that you're building your circuit backwards. the saying would be "making a coat for a button that you found" ... when nothing about your circuit is determined (something as basic as the ADC!) why do you even bother creating analog section for a very specific type of signal? what kind of information are you even trying to look at that can't be gathered from simulations? LTSpice is very accurate in that regard.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: clipping a fast negative pulse
« Reply #29 on: May 10, 2019, 06:19:18 am »
This sounds very much like an eddy current sensor, for whatever purpose: material resistivity, thickness, metal detection or identification, etc.  You don't need to be cryptic about it, there's nothing special about this.

The impulse method never impressed me.  In particular, why discard whole microseconds of the strongest part of your precious signal, when you can get the long tail (low frequency) components directly, with a sine wave impedance bridge method instead?

A frequency sweep with a DDS and ADC, under MCU control, would be very easy, require no fiddly current source or beyond-the-rails voltages, and as it has no dead bands, and integrates the signal continuously, offers an arbitrarily high SNR.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline brumbarchrisTopic starter

  • Regular Contributor
  • *
  • Posts: 223
  • Country: ro
Re: clipping a fast negative pulse
« Reply #30 on: May 10, 2019, 08:17:10 am »
Quote
This sounds very much like an eddy current sensor
Ah, professionals can read between the lines, good catch!
I do not want to withhold information, particularly when asking for help. However, as I am myself not very knowledgeable in this field, I would want to refrain from putting forward the wrong information or hypotheses.
Besides, theoretically, it is much probable to get pertinent and focused and usable answers when you split the big problem in smaller, simpler chunks (simpler than "please design this eddy current detector for me".)

Quote
The impulse method never impressed me.
Well, we have to build and extend whatever experience already exists within our company (not much, admittedly, but more than nothing with regards to the pulse induction principle).

Best regards,
Cristian

 

Offline Kalvin

  • Super Contributor
  • ***
  • Posts: 2145
  • Country: fi
  • Embedded SW/HW.
Re: clipping a fast negative pulse
« Reply #31 on: May 10, 2019, 08:55:13 am »
It is funny how these toy metal-detectors are still developed and manufactured. Is this due to the lack of competition or what?
 

Offline StillTrying

  • Super Contributor
  • ***
  • Posts: 2850
  • Country: se
  • Country: Broken Britain
Re: clipping a fast negative pulse
« Reply #32 on: May 10, 2019, 02:36:00 pm »
.  That took much longer than I thought it would.
 

Offline OwO

  • Super Contributor
  • ***
  • Posts: 1250
  • Country: cn
  • RF Engineer.
Re: clipping a fast negative pulse
« Reply #33 on: May 10, 2019, 02:47:33 pm »
I second the suggestion to use frequency domain methods to make impedance and transfer function measurements. This is very common at RF where network analyzers are used to measure impulse response which gives better results than a TDR.
Email: OwOwOwOwO123@outlook.com
 

Offline AndyC_772

  • Super Contributor
  • ***
  • Posts: 4315
  • Country: gb
  • Professional design engineer
    • Cawte Engineering | Reliable Electronics
Re: clipping a fast negative pulse
« Reply #34 on: May 10, 2019, 04:01:29 pm »
Quote
Wait.  Why are you clipping the pulse you fought so hard to perfect?

In order to further signal process it and feed it to an ADC eventually.

Hang on... what?? You want to precisely clip a signal which is going to be digitised?

In that case, your clipping 'circuit' is two lines of code:

Code: [Select]
for (i = 0; i < SAMPLE_BLOCK_SIZE; i++)
  sample_data[i] = (sample_data[i] < CLIP_LEVEL) ? CLIP_LEVEL : sample_data[i];

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: clipping a fast negative pulse
« Reply #35 on: May 10, 2019, 07:56:39 pm »
Quote
Wait.  Why are you clipping the pulse you fought so hard to perfect?

In order to further signal process it and feed it to an ADC eventually.

Hang on... what?? You want to precisely clip a signal which is going to be digitised?

In that case, your clipping 'circuit' is two lines of code:

Code: [Select]
for (i = 0; i < SAMPLE_BLOCK_SIZE; i++)
  sample_data[i] = (sample_data[i] < CLIP_LEVEL) ? CLIP_LEVEL : sample_data[i];

Analog clipping is probably still required to protect voltage ranges, but in the end, digital will be far more accurate than anything else.  ADCs don't usually suffer from recovery effects either, so the saturation (to 0x000 / 0xfff) can be used implicitly as well (but, that may be something to test; ADCs fail in all sorts of fascinating ways).


It bears noting that the abstraction process only works if the one doing the abstraction, fully understands the specifications of the systems above and below the abstraction interface.  This exercise strikes me as an overzealous (and perhaps inexperienced?) abstraction, that would benefit more from a holistic design process instead.

That just leaves one thing --

Well, we have to build and extend whatever experience already exists within our company (not much, admittedly, but more than nothing with regards to the pulse induction principle).

-- the force of "not invented here".  Even when the existing background is sparse, there is a strong tendency to stick to what is "known", even if very little is actually known.

It is funny how these toy metal-detectors are still developed and manufactured. Is this due to the lack of competition or what?

Beats me -- they're definitely out there.  Last time I saw an eddy current analyzer at work, it was being used to check the martensite fraction (and to what depth) of an induction-hardened steel workpiece.  A subtle application, but apparently representative and repeatable enough to work.  Not that I can recommend anyone just go out and buy an industrial eddy current analyzer for any generic purpose (guessing they're in the $10k price range, as much industrial kit is), but the basic idea in any case is quite tractable, and simplified versions are available for peanuts (like metal detectors) that arguably could be used for the core functionality, and hacked to the OP's goal, without having to invent anything new. :)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf