Separating high pass and low pass TIA output bands. - Page 2

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
Re: Separating high pass and low pass TIA output bands.
On Wednesday, August 16, 2017 at 7:42:42 PM UTC-4, Jim Thompson wrote:

Quoted text here. Click to load it


Yes. The pad to the plane and then back to the other pad can have lower a impedance than the resistor. It also acts like a capacitor to ground. I often cut holes in the plane around TIAs and I never worked at frequencies as high as 100MHz.

Re: Separating high pass and low pass TIA output bands.
Quoted text here. Click to load it




Quoted text here. Click to load it

It's super educational to try putting realistic strays in the circuit simul
ation and seeing what they do. For a 4 layer board with ground on L2, figur
e 0.15 to 0.25 pF per pad. Adjacent pads don't couple as much as you'd thin
k because there's air on one side and the ground plane on the other.  

Resistors have about 0.05 to 0.1 pF in parallel, depending on the ground co
nfiguration.  

I generally put a bootstrapped pour under the summing node.  

Cheers

Phil Hobbs

Re: Separating high pass and low pass TIA output bands.
On 8/16/2017 4:19 PM, Wanderer wrote:
Quoted text here. Click to load it

If I were to place a cutout beneath high frequency parts wouldn't this  
cause a disturbance in the image plane effect? Ref Section 17.2.2.3,  
page 7, of: https://focus.ti.com/lit/ml/sloa089/sloa089.pdf

Or would this not matter when the components are small enough, and close  
together enough, to be considered lumped?

--  
To respond to me directly remove sj. from the my email address's domain  
name. This is a spam jammer.

Re: Separating high pass and low pass TIA output bands.
Quoted text here. Click to load it

I've seen a 100MHz receiver built with 3/4" parts between tag strip  
and vacuum tubes.



--  
This email has not been checked by half-arsed antivirus software  

Re: Separating high pass and low pass TIA output bands.
On 7/11/2017 1:09 PM, snipped-for-privacy@gmail.com wrote:
Quoted text here. Click to load it

The current plan is to use your front end topology you show in Figure 9,  
Page 4 at: http://electrooptical.net/www/frontends/frontends.pdf with an  
error integrator in the feedback path as shown in Figure 12, page 11, at  
http://www.ti.com/lit/ds/sbbs001/sbbs001.pdf.

The question is where is the best place to connect the output resistor  
of the error integrator (R3 Figure 12)? In your front end schematic is  
it best connected to A1's summing junction (inverting input)? Or the  
emitter of Q1?

I can see how the R3 connection will sacrifice bandwidth either way.

The schematic in your book is slightly different from the one you give  
in the PDF you linked to. In your book you have in Figure 18.12 a  
transistor with its collector and base connected to the +15V supply and  
2Mohm resistor. Is that there to compensate for the base emitter voltage  
drop on Q1? If so wouldn't be best to make that transistor the same as Q1?

--  
To respond to me directly remove sj. from the my email address's domain  
name. This is a spam jammer.

Re: Separating high pass and low pass TIA output bands.
On 09/01/2017 07:35 PM, Artist wrote:
 > On 7/11/2017 1:09 PM, snipped-for-privacy@gmail.com wrote:
 >> Good TIAs at that bandwidth are not that easy to design. I suggest  
spending most of your time on a fast, quiet DC-coupled TIA and filter  
afterwards.
 >>
 >> You might find this article helpful.
 >> <http://electrooptical.net/www/frontends/frontends.pdf
 >
 > The current plan is to use your front end topology you show in Figure  
9, Page 4 at: http://electrooptical.net/www/frontends/frontends.pdf with  
an error integrator in the feedback path as shown in Figure 12, page 11,  
at http://www.ti.com/lit/ds/sbbs001/sbbs001.pdf.

Why bother?  Are you working at very low supply voltages or super wide  
ranges of photocurrent, or something like that?  By the time the  
photocurrent gets as far as dropping 50 mV across the feedback resistor,  
you're already in the shot noise limit.

 >
 > The question is where is the best place to connect the output  
resistor of the error integrator (R3 Figure 12)? In your front end  
schematic is it best connected to A1's summing junction (inverting  
input)? Or the emitter of Q1?
 >
 > I can see how the R3 connection will sacrifice bandwidth either way.

Need more info.  If you want to reject DC, why do that instead of using  
a series capacitor between the first and second stages?  There are  
sometimes reasons, of course, but a series cap is pretty simple and  
trouble-free.

 >
 > The schematic in your book is slightly different from the one you  
give in the PDF you linked to. In your book you have in Figure 18.12 a  
transistor with its collector and base connected to the +15V supply and  
2Mohm resistor. Is that there to compensate for the base emitter voltage  
drop on Q1? If so wouldn't be best to make that transistor the same as Q1?

Yes.  But then it's only a first-order compensation, so it doesn't  
matter very much which transistor you use.

Cheers

Phil Hobbs


--  
Dr Philip C D Hobbs
Principal Consultant
We've slightly trimmed the long signature. Click to see the full one.
Re: Separating high pass and low pass TIA output bands.
On 9/1/2017 5:18 PM, Phil Hobbs wrote:
Quoted text here. Click to load it


Quoted text here. Click to load it


Quoted text here. Click to load it

Quoted text here. Click to load it

Because the AC modulation is only 1% of the DC level. If I can optimize  
the TIA feedback resistor for only this AC there is a potential 100 to 1  
noise reduction to be gained over a design where the feedback resistor  
has to be optimized for the AC plus the DC.

--  
To respond to me directly remove sj. from the my email address's domain  
name. This is a spam jammer.

Re: Separating high pass and low pass TIA output bands.
On 09/02/2017 04:39 PM, Artist wrote:
Quoted text here. Click to load it

If Iphoto * Rf > 200 mV, there's only 1 dB to be gained by going to the  
limit Rf -> infinity, because the shot noise power is already 4 times  
the Johnson noise power.  You can't separate the signal from the shot  
noise unless you go to spooky quantum measurements, and even there it  
doesn't usually get you that much. (*)

If you have a very wide range of photocurrents, so that you can't keep  
within the range 200 mV < (Rf * Iphoto) < VDD, you just have to pay the  
price in voltage swing to avoid dying of Johnson noise at the lower  
limit.  It's not unheard-of to run a current-nulling servo off a 250V  
supply.

Cheers

Phil Hobbs

(*)  (There's a scheme called 'quantum illumination' that I don't  
understand well, but which bids fair to help quite a lot in special  
situations.)

--  
Dr Philip C D Hobbs
Principal Consultant
We've slightly trimmed the long signature. Click to see the full one.
Re: Separating high pass and low pass TIA output bands.

Quoted text here. Click to load it

Another approach is to segment the resistor in the nulling loop and use a d
iode in parallel with all but one section, forming a breakpoint servo amp.  
  The good news is that you don't need a 250V supply. The bad news is that  
the servo bandwidth depends on the signal level.  

Cheers

Phil Hobbs

Re: Separating high pass and low pass TIA output bands.
On 09/02/2017 02:04 PM, Phil Hobbs wrote:
Quoted text here. Click to load it

Rf * Iphoto for the AC component is expected to be roughly 25mV if the  
Rf were to be optimized for AC +  DC.

In an effort to reduce lost bandwidth due to R3 from the error  
integrator to the virtual ground suppose I do any of: substitute R3 for  
a current source such, as the Howland Current source:  
http://www.falstad.com/circuit/e-howland.html , invert output of the  
error integrator and connect R3 to a two transistor current mirror on  
the positive supply, or use that current mirror with a current sink such  
as one  
https://www.electronicsweekly.com/blogs/engineer-in-wonderland/current-sink-stability-2015-10/
?

--  
To email me directly remove sj. from my email address's domain name.  
This is a spam jammer.

Re: Separating high pass and low pass TIA output bands.
Quoted text here. Click to load it


That's not the right way of thinking about it--the shot noise is white, and
 depends only on the DC. If I_DC R_L > 200 mV, you've only got 1 dB of room
 to improve, independent of frequency, regardless of how fancy a circuit yo
u use.  

Cheers

Phil Hobbs

Re: Separating high pass and low pass TIA output bands.
On 9/4/2017 2:56 PM, snipped-for-privacy@gmail.com wrote:
Quoted text here. Click to load it

Quoted text here. Click to load it

Today I found out that the TIA is to be used over a wide variety of  
illumination levels on the photodiode. I have to design so as not to  
saturate at the highest expected power level. For the lowest  
illumination level it very well could be that the DC level of Rf *  
Iphoto will be under 200mV.

--  
To respond to me directly remove sj. from the my email address's domain  
name. This is a spam jammer.

Re: Separating high pass and low pass TIA output bands.
Quoted text here. Click to load it

That makes sense. In that case, the two normal ways of proceeding are:
(1) use a current feedback loop running off a much higher supply voltage, or
(2) Use a CFB loop at normal voltages but with a nonlinear feedback resistance cobbled together from diodes or diodes plus resistors.  

Option 1 requires a separate supply, and the range may be impractically large. OTOH the loop BW stays constant.  

Option 2 relaxes the supply voltage, but the loop bandwidth changes with signal level.  

Cheers

Phil Hobbs

Re: Separating high pass and low pass TIA output bands.
On 9/6/2017 4:02 PM, snipped-for-privacy@gmail.com wrote:
Quoted text here. Click to load it

The output has to be linear so option 2 is out.

Why would a much higher power supply be required for option 1? Are there  
any options for current sources other than than the ones I mentioned?

It is permissible for there to be a second power supply in this design.

--  
To respond to me directly remove sj. from the my email address's domain  
name. This is a spam jammer.

Re: Separating high pass and low pass TIA output bands.
Quoted text here. Click to load it
  
Quoted text here. Click to load it

As I said, it doesn't matter how complicated you make the circuit. The issu
e is that for a DC-coupled amplifier there is no escape the thermal noise o
f whatever resistor you use to remove the DC, whether it's the feedback res
istor in a TIA or the emitter resistor in a current source. Do the algebra  
and you'll see.  

In an AC-coupled amp, you can use transformers and/or capacitors for feedba
ck instead of resistors, but not at baseband.  

You can also use range switching.  

Cheers

Phil Hobbs

Re: Separating high pass and low pass TIA output bands.
On 9/6/2017 4:02 PM, snipped-for-privacy@gmail.com wrote:
Quoted text here. Click to load it

Today I got more specific specifications.

The minimum photodetector size is .3 mm. Larger would be preferred as  
this would provide a larger target for the laser beam making easier  
alignment.

The laser's the total irradiant power on the photodetector will range  
from 10 uW to 5 mW. Linearity must be maintained over the entire range.  
Of course they would like to be shot noise limited over the entire  
range, but they recognize this might not be feasible. They are open to  
having two or more units, each optimized for a narrower range of input  
power levels. At 40 MHz how practical is it to switch feedback resistors?

The modulation will be roughly 1% of the DC level.

The DC output band will be from 0 Hz to 50 kHz
The AC output band will be from 20 kHz to 40 MHz. They do not want  
bandwidth any higher that 40 MHz for noise suppression reasons. The  
signal of interest is 40 MHz and below.

I see your point that any source that puts current into the virtual  
ground to tune out the DC offset will introduce its own noise.

--  
To respond to me directly remove sj. from the my email address's domain  
name. This is a spam jammer.

Re: Separating high pass and low pass TIA output bands.
On 09/12/2017 06:21 PM, Artist wrote:
Quoted text here. Click to load it

That's 500:1, which is not horrible. What's the responsivity of the PD?

Assuming it's around 0.25 A/W, which is typical for silicon in the  
visible, that would be 2.5 uA to 1.25 mA.  If you pick a 20k resistor,  
that gets you to the shot noise + 3 dB at the low end, and 25V at the  
high end, which you can do with a reasonably normal amplifier if you run  
it off asymmetric supplies like +28/-4.

At 40 MHz you'll probably need an AC tweak to keep the response flat,  
and that would differ between ranges.

If you do decide to use range switching, use a relay to switch the whole  
feedback resistor + AC tweak in and out.

Cheers

Phil Hobbs

--  
Dr Philip C D Hobbs
Principal Consultant
We've slightly trimmed the long signature. Click to see the full one.
Re: Separating high pass and low pass TIA output bands.
On 9/13/2017 1:26 PM, Phil Hobbs wrote:
Quoted text here. Click to load it

Quoted text here. Click to load it

What low noise op amp has sufficient GBW and can operate off a 32V power  
supply difference?

By tweak do you referring to a subsequent stage that is a filter with a  
zero in it, or a multipole filter that has some overshoot at the high  
end of its frequency response?

The photodiode and its TIA is for a laser ultrasonics system's detection  
laser.

--  
To respond to me directly remove sj. from the my email address's domain  
name. This is a spam jammer.

Re: Separating high pass and low pass TIA output bands.
Quoted text here. Click to load it

There are more nowadays. You might look at the ADA4898, which may not be quite fast enough, or (with a sufficiently efficient bootstrap) an LM6171.  

Cheers

Phil Hobbs

Re: Separating high pass and low pass TIA output bands.
On 2017-07-11 12:13, Artist wrote:
Quoted text here. Click to load it

I have done this once. The bands did not overlap (probably could have)  
and there was one TIA optimized for RF, low noise and such but that had  
high offset and drift. The other was optimized for low drift and low  
offset errors because that was used to servo something. I can't go into  
details but it all worked fine. RF guys call this sort of splitter a  
diplexer.

As Phil said in your case this might not make much sense because 20kHz  
is already audio range and 40MHz is RF but not very high. In my case it  
was different, an audio range servo loop and then the RF was a certain  
band in the low VHF range. If you need professional support on this Phil  
is the guru in that field.

--  
Regards, Joerg

http://www.analogconsultants.com/

Site Timeline