Why Adding a Voltage Amplifier after a Transimpedance Amplifier reduce BW? Book Claim by 0.64

Makes no sense to me. How does the TIA know that there is an amplifier downstream? How would the amplifier know that it's being driven by a TIA?

Can you post your schematic?

I would expect more noise and increased BW if you cut the classic TIA gain by 3 and add a wideband gain of 3 afterwards.

A little spurious feedback, through power supply rails, could affect bandwidth, but not by a scientific factor of 0.64.

If you attached a page from the book, we can't see it here.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

With all due respect to Jerry Graeme, his book has confused a lot of people. He has an eccentric nomenclature, for one thing. He also seems to think that op amp TIAs are the be all and end all. If you follow that advice, it will lose you a good 30 dB of SNR in many cases. A much better book at the same level is Mark Johnson's "Photodetection and Measurement", and of course there's no substitute for doing your own algebra.

What Graeme seems to be trying to do here is to split the gain between two identical amplifiers such that the resulting circuit bandwidth is maximized. The choice of R_f is based on the amplifier's GBW product. Since you can get gigahertz op amps with good DC and noise properties for a few dollars, there's no reason in the world to worry about doing this.

And since this discussion appears to ignore photodiode capacitance, it isn't worth much anyway.

Just calculate the bandwidth the normal way, and it'll work fine.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

The 0.64 (actually 0.643594... :-) comes from the bandwidth shrinkage equation from cascading _two_ identical bandwidth-limited stages.

General equation: ShrinkFactor = sqrt(2^(1/N)-1)

where N is the number of cascaded identical bandwidth-limited stages. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

f days, I hope you guys don't mind if I paste my email that I just sent to Dr. Hobbs seeking advice on my issue. I found this group on Dr. Hobbs websi te, so I hope you guys can offer any insight as to what's going. I understa nd that I'm about to post a wall of text, so here's a summary:

on-inverting amplifier to a transimpedance amplifier, the BW is reduced by

0.64. Gain of this amplifier is 3. I'm reducing my transimpedance gain by 3 so the overall gain remains the same. Where does the 0.64 reduction come f rom? Why? When we use an inverting amplifier, we do not see this reduction in BW. Here's my reddit post showing pictures of the book and equation in q uestion:

n_question/

I've been seeking answers to my issue for several days now and an individua l in one of the forums that I posted on recommended that I contact you to s ee if are available to help.

mpedance amplifier, why is there a bandwidth reduction when using a non-inv erting amplifier? Assuming you're not limited by the bandwidth of the non-i nverting amplifier.

guy is going off of your bootstrap amplifier discussed here:

d we have learned a lot from your papers and your book, but we're just conf used as hell about our issue.

er and added a second stage for voltage amplification. After prototyping it , he noticed a reduction in bandwidth when using a non-inverting amplifier with a gain of 3. With the fT of the amplifier (non-inverting amplifier) be ing quite high for his application, I wouldn't suspect that to limit his ba ndwidth (his signal bandwidth is 50kHz). However, when he connects the non- inverting amplifier to his transimpedance amplifier, he sees a significant reduction in bandwidth (reduced down from 50kHz to 15kHz or so). We have bo th looked at your Building Electro-Optical Systems book as well as the 3rd edition of the Art of Electronics, along with the papers referenced above. Unless we are missing something, neither of these books address this issue.

Amp Solutions by Jerald Graeme) that the bandwidth circuit gets reduced by

0.64 when adding a non-inverting amplifier. I attached a picture of the pag e showing the schematic and equation. Nowhere in this book does the author address or detail this result. And no one in the office expects and or can explain this.

if you change the non-inverting amplifier to a inverting amplifier then th e BW is not reduced.

Not used to posting on Google groups and I don't see any places for attachm ents, so I'll assume there are none. With that said, here's a link to the s chematic we're working with and a picture of the equation from the book: ht tp://imgur.com/a/bbu4N.

This makes no sense to me either. Even if Jim Thompson is correct in that t he author assumes identical bandwidth amplifiers and then using the shrinka ge formula to get 0.64 then why even bother with the second stage?

gn_question/

is.

When you say, just calculate the bandwidth the normal way and it will work fine...Are you saying f = 1/(2pi*Rf*(Cd+Cin)) and then f3db,ckt = sqrt( f*fT)? Just like eqn 18.10 in your book?

If that's true then that's what we have done. We calculated the bandwidth o f the TIA using eqn 18.10. Built the circuit. Tested it. Saw that it worked fine. Then wanted more gain so we added an amplifier with a non-inverting gain of 3. That's when we noticed the reduction and subsequently, found tha t 0.64 formula in Graeme's book.

With all due respect to Jerry Graeme, his book has confused a lot of

You're using an AD8066, which has a GBW of ~ 70 MHz, and running the second stage it at a noise gain of 3 or 4. The 10k/30k resistors are too large for the speed of the amp, but they're OK below a megahertz or so. The op amp by itself _cannot_ be causing gain funnies down in the kilohertz. Something's oscillating, or the supply bypass is inadequate, or there's a measurement problem, or the circuit you built is not the circuit you drew.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

This is officially a non-binary newsgroup.

The schematic doesn't show the entire picture.

Does the photodiode anode literally connect to the -10 supply? Are the opamps powered by -10 too?

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

e of days, I hope you guys don't mind if I paste my email that I just sent to Dr. Hobbs seeking advice on my issue. I found this group on Dr. Hobbs we bsite, so I hope you guys can offer any insight as to what's going. I under stand that I'm about to post a wall of text, so here's a summary:

a non-inverting amplifier to a transimpedance amplifier, the BW is reduced by 0.64. Gain of this amplifier is 3. I'm reducing my transimpedance gain b y 3 so the overall gain remains the same. Where does the 0.64 reduction com e from? Why? When we use an inverting amplifier, we do not see this reducti on in BW. Here's my reddit post showing pictures of the book and equation i n question:

sign_question/

n. I've been seeking answers to my issue for several days now and an indivi dual in one of the forums that I posted on recommended that I contact you t o see if are available to help.

nsimpedance amplifier, why is there a bandwidth reduction when using a non- inverting amplifier? Assuming you're not limited by the bandwidth of the no n-inverting amplifier.

ne guy is going off of your bootstrap amplifier discussed here:

and we have learned a lot from your papers and your book, but we're just c onfused as hell about our issue.

ifier and added a second stage for voltage amplification. After prototyping it, he noticed a reduction in bandwidth when using a non-inverting amplifi er with a gain of 3. With the fT of the amplifier (non-inverting amplifier) being quite high for his application, I wouldn't suspect that to limit his bandwidth (his signal bandwidth is 50kHz). However, when he connects the n on-inverting amplifier to his transimpedance amplifier, he sees a significa nt reduction in bandwidth (reduced down from 50kHz to 15kHz or so). We have both looked at your Building Electro-Optical Systems book as well as the 3 rd edition of the Art of Electronics, along with the papers referenced abov e. Unless we are missing something, neither of these books address this iss ue.

Op Amp Solutions by Jerald Graeme) that the bandwidth circuit gets reduced by 0.64 when adding a non-inverting amplifier. I attached a picture of the page showing the schematic and equation. Nowhere in this book does the auth or address or detail this result. And no one in the office expects and or c an explain this.

hat if you change the non-inverting amplifier to a inverting amplifier then the BW is not reduced.

chments, so I'll assume there are none. With that said, here's a link to th e schematic we're working with and a picture of the equation from the book:

t the author assumes identical bandwidth amplifiers and then using the shri nkage formula to get 0.64 then why even bother with the second stage?

Hi John,

Now I know what my older buddy at works feels like when a phrase is tossed out and not knowing what it means. What's "This is officially a non-binary newsgroup"

As per your questions, here is a picture of the complete schematic. I'm at home at the moment so it's drawn from memory and I don't know the RC values for the power supply filters, but other than that this is it.

PD is connected to -10V. OpAmps connected to +/-10V. +/-10V measured. No os cillation on supply lines. LED connected to a function generator is the lig ht source. a 50+kHz sine wave is applied to the LED. The output is connecte d to an SMA which then connects to a high impedance scope input.

Please note that the upper left circuit is the one with the reduced BW whil e the upper right circuit has no BW reduction.

I really appreciate your help with all of this and I hope I'm not being too much of a pain.

Hi Dr. Hobbs,

Here's the complete schematic of the circuit:

I'm at home so I don't know the RC values for the power supply filters, but they were measured with a DMM and with a scope. +/-10V measured with no oscillations. The output signal is connected to an SMA which then connects to a high impedance scope.

Please note that the top left circuit is the one with the BW reduction while the top right circuit has no BW reduction. The changes are: 20k changed to 30k and opamp changed from non-inverting to inverting.

And I really appreciate the help from both you and Mr. Larkin. Thank you guys. My apologies for being a pain in the ass.

-Felipe

Oh and I'm fairly certain the non-inverting amplifier is wired up correctly. Between switching back and forth from the non-inverting to the inverting configuration several times, I am fairly certain that both configurations were wired up properly.

You can't in theory add a binary (like a jpeg file) to a post here. Some news servers allow it, some don't.

No, this stuff is interesting.

The photodiode is connected to the same -10 rail as powers the opamps. That could allow the second opamp to feed spikes back into the front end.

Try adding an RC lowpass filter, or another C-multiplier, to provide a dedicated, clean bias voltage to the photodiode.

The AD8066 is a dual opamp, which presents other paths for the stages to interact. Single amps, some small distance apart, with individual power supply decoupling, would be safer.

Was this built on one of those dreadful plastic breadboards? A really tiny amount of capacitance from the second amp output, back to the input, will wreck things.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Hmm... "several times" does make it sound like a solderless breadboard, as John suggested. Those things are slabs of misery, for sure.

The AD8066 doesn't come in a DIP package, so I assumed it wasn't one of those things, but having bad layout strays is one of the ways to not build what you think you're building.

Oh, one other thing--did you get your parts from an authorized distributor? Otherwise it's possible the op amp is counterfeit.

(And do knock off the 'Dr.' bit--that's in my sig for SEO reasons. It's all free-men-and-equals round here.)

Cheers

Phil Hobbs

Good point. Or load capacitance and really poor supply bypassing.

To the original poster (OP): is that schematic complete? If it is, add some nice big bypass caps on the supply leads.

Cheers

Phil Hobbs

Ah, you did post a complete schematic.

Your negative cap multiplier is wrong. You need to wire it like the + one, but with a PNP.

Cheers

Phil Hobbs

Yeah, or buy Phil's book.

Felipe, If you give us your specs, light level (and wavelength), spot size, needed BW, then I'm guessing we can make something better... Lots of fast opamp's.

Phil, the one gripe I have with your book...(well gripe is perhaps too strong. ).. is that you don't really do the classic PD TIA opamp analysis. Not a problem for me, since by the time I found your book I'd read numerous app notes and worked through the math several times. But maybe an appendix where you "show your work". (I've only got the 1st ed.)

George H.

That was on purpose.

Because of the ridiculously broad coverage, I had to be selective about where to go deep, and since everybody building circuits has AoE, I tried to dovetail with Win and Paul.

Cheers

Phil Hobbs

That feedback path explains the change in bw as a function of the sign of the gain of the output amp.

One polarity decreases BW; the other increases it!

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

And in this case, I actually did go through it completely, P 626-631 of the first edition.

Cheers

Phil Hobbs

Huh, Well there it is. My bad, I hope I haven't besmirched your wonderful text.

George H.