Author Topic: Slew Rate, Bandwidth, Settling time, rise time for high speed transipmedance amp  (Read 1538 times)

0 Members and 1 Guest are viewing this topic.

Offline arivalagan13Topic starter

  • Contributor
  • Posts: 23
  • Country: in
Hi all
I'm in the process of designing a Transimpedance amplifier for an analytical instrument. The sample waveform (hand-drawn) is attached in the attachments. The requirement is to amplify a pulse kind of signal whose maximum amplitude is around 700 nA and the width of the pulse is around 1 ns. So, I need a high-speed transimpedance amplifier.
My question is what are the opamp parameters that decide the speed of the amplifier. I need a concrete relationship between slew rate, settling time, rise time, bandwidth/gain-bandwidth. And particular to the signal which parameter is most important so that I can choose the right opamp for the design.
Regards
M Arivalagan
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
What's the source capacitance? TI has a reference design for a 200 MHz TIA with a link to a reference how to calculate the necessary GBP.

Personally though I'd try to dig up some old stock for MAX3266CSA, one of the few GHz range TIAs available in non-die form.
 

Offline arivalagan13Topic starter

  • Contributor
  • Posts: 23
  • Country: in
The source capacitance is around 20pF
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Really? That's kinda huge for something with a GHz worth of bandwidth and will present an almost insurmountable challenge to any opamp and TIA you can get in a solderable package.
 

Offline awallin

  • Frequent Contributor
  • **
  • Posts: 694
that one seems close(ish) to 1GHz:
http://www.ti.com/lit/ug/tidud08/tidud08.pdf
it might use this photodiode, specified as 2 pF (with some reverse bias)
https://acphotonics.com/downloads/High_Speed_800nm_900nm_PIN_Photodiode.pdf
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Do an online search for the Burr-Brown application notes which discuss wide bandwidth transimpedance amplifiers.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Even better question: how large, physically, is that 20pF?  If the answer is more than a few cm, end to end, then it's not going to behave as a capacitor, but a transmission line stub of some sort!

Note that 1ns * 700nA = 700aC and 700aC / 20pF = 35uV.  You're looking to amplify and differentiate a step of this magnitude.  Even if you don't need much signal level (low mV?), you're asking for a tremendous amount of gain at high frequency to turn that into a pulse versus a step.

There is also a noise associated with capacitance, independent of system resistance, or bandwidth (they cancel out).  For 20pF at room temperature, this is 28uV.

In general, you are asking for the impossible.

If the detector and amplifier are cryogenically cooled, it may still be possible, but you may find it is more productive to fix the detector.

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

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Note that 1ns * 700nA = 700aC and 700aC / 20pF = 35uV.

It's not supposed to charge the capacitance, the input resistance of the TIA is supposed to be far lower than the impedance of the capacitance. So in this case a couple ohms.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Yes, I was implying a conceptual split between the source and the amplifier: the signal and impedance, looking into the source, will be as described.  The impedance looking into the amplifier is supposed to be low, and this tells you how low it must be.

Which, good luck achieving that at the PCB level -- a few ohms [equivalent inductance] at 1GHz is a trace length of maybe 0.5mm.

Also, even on a thin substrate, a trace impedance of 10 ohms is much wider than 0.5mm, so we can't really do anything meaningful at all, at this speed and impedance.  Even on HDI where you can stack vias into alternating traces in parallel, you're limited by the amplifier's pins/bond wires.

Respectively, 20pF resonates at 1GHz with 1.3nH.  That detector better be right on top of the amp!

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

Offline StillTrying

  • Super Contributor
  • ***
  • Posts: 2850
  • Country: se
  • Country: Broken Britain
Don't start any more threads on this subject, 4 is more than enough, :) vital bits of information on the problem always get left behind.
You can always wake-up a previous thread with a new post updating with a bit of new info.

https://www.eevblog.com/forum/projects/high-speed-transimpedance-amplifier/msg2489382/#msg2489382
.  That took much longer than I thought it would.
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
AFAICS a microchannel plate has 50 Ohm output. You don't need a TIA, you need a RF gain block.

If you don't care too much about gain flatness just pop a SPF5189 amplifier in there, you can get modules really cheap. Or for a bit better build quality this module which I assume is based on TQP3M9036 doesn't seem a bad deal.
« Last Edit: August 25, 2019, 07:12:58 pm by Marco »
 

Offline Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 2077
  • Country: us
    • The Messy Basement
....There is also a noise associated with capacitance, independent of system resistance, or bandwidth (they cancel out).  For 20pF at room temperature, this is 28uV....

Tim

I always thought pure capacitive or inductive reactances had no noise, though it seems like there should be at least some. How is this calculated?

Conrad
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
....There is also a noise associated with capacitance, independent of system resistance, or bandwidth (they cancel out).  For 20pF at room temperature, this is 28uV....

Tim

I always thought pure capacitive or inductive reactances had no noise, though it seems like there should be at least some. How is this calculated?

There must necessarily be a resistance somewhere -- otherwise you can't measure any signal at all (though you might be able to sense it parametrically, with other limitations).  That resistance gives the RC time constant and therefore the cutoff frequency, but it's also the noise source and therefore the noise level.  Noise over the total bandwidth happens to cancel out, giving the simple result. :)
https://en.wikipedia.org/wiki/Johnson–Nyquist_noise#Thermal_noise_on_capacitors

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

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
AFAICS a microchannel plate has 50 Ohm output. You don't need a TIA, you need a RF gain block.

If you don't care too much about gain flatness just pop a SPF5189 amplifier in there, you can get modules really cheap. Or for a bit better build quality this module which I assume is based on TQP3M9036 doesn't seem a bad deal.

And for that matter, hopefully the MCP is built in such a way that it has a transmission line structure.

For example, it might be a sinuous path over a ground plane, which has the bonus of allowing you to correlate propagation delay to impact location.

You're still very limited by noise, and may need to stack another MCP to get higher particle amplification, or something like that.

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

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
In theory you could segment the anode for a MCP in say cm^2 pieces and bring multiple connections outside, that should allow you to use multiple very low impedance TIAs.

But given that in the other thread he mentioned connecting it to a Phillips lab amplifier he almost certainly just has coax coming from the device and he needs a RF amplifier. The TQP3M9036 has an almost unbelievably low noise which should bring the noise down by about 2x relative to the Phillips amplifier (for the same bandwidth, but it has far more bandwidth than the Phillips).
« Last Edit: August 26, 2019, 04:30:05 pm by Marco »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf