Taming the ADA4817

Do you mean cap values across the feedback resistor?

If both inputs are grounded, the output should be on one of the rails; is it? (Open-loop gain is low, so maybe it isn't. You could force in some DC offset to make it rail.)

If the amp is railed and you still see the 3.5 MHz, the oscillation is probably coming from somewhere else.

The low oscillation amplitude is another hint that the opamp may not be oscillating.

That is a pretty scary part. Layout and bypassing and such will really matter.

Can you post the circuit and a pic of the board/layout?

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

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

What's PD?

--

Rick C

as it continues to oscillate at 3.5MHz with about 20mVpp-40mVpp in a TIA c onfiguration.

yout has all copper removed from all signal pins on all layers. I have disc onnected the TIA from the following stages, so I'm only looking at the outp ut of the TIA.

r as the TIA gain and I also put a 20Meg ohm resistor from the inverting in put to ground. Changed different cap values and nothing. Dropped the feedba ck resistor down to 5Meg, 1Meg, 100k ohm while trying various values and no thing. Same oscillation.

t the output. Same thing, but with more of an offset.

Yes I was changing the feedback capacitor for various feedback resistors.

Here's a link to the layout pic. U1 is the TIA. In the picture below, I had both the inverting and non-inverting inputs grounded and no feedback compo nents.

As for the schematic, just a standard TIA configuration. non-inverting grou nded and inverting input is where the PD connects to with the feedback resi stor and cap on that pin. 0.1uF caps on the supply pins and nothing else

mp as it continues to oscillate at 3.5MHz with about 20mVpp-40mVpp in a TIA configuration.

layout has all copper removed from all signal pins on all layers. I have di sconnected the TIA from the following stages, so I'm only looking at the ou tput of the TIA.

tor as the TIA gain and I also put a 20Meg ohm resistor from the inverting input to ground. Changed different cap values and nothing. Dropped the feed back resistor down to 5Meg, 1Meg, 100k ohm while trying various values and nothing. Same oscillation.

at the output. Same thing, but with more of an offset.

ad both the inverting and non-inverting inputs grounded and no feedback com ponents.

ounded and inverting input is where the PD connects to with the feedback re sistor and cap on that pin. 0.1uF caps on the supply pins and nothing else

R1 and C1 is the feedback. C16 and C15 is the supply 0.1uF caps

as it continues to oscillate at 3.5MHz with about 20mVpp-40mVpp in a TIA c onfiguration.

yout has all copper removed from all signal pins on all layers. I have disc onnected the TIA from the following stages, so I'm only looking at the outp ut of the TIA.

r as the TIA gain and I also put a 20Meg ohm resistor from the inverting in put to ground. Changed different cap values and nothing. Dropped the feedba ck resistor down to 5Meg, 1Meg, 100k ohm while trying various values and no thing. Same oscillation.

t the output. Same thing, but with more of an offset.

Yeah if he's changed the feedback resistor (by factors of ten or more) and the oscillation is still at the same frequency, then something else is wrong. (speaking to OP now) I assume power supplies are bypassed and (maybe) filtered.

What about if you hook it up as a unity gain buffer? (Maybe you fried the opamp... ?)

George H.

it seems you are picking up 3.5MHz from something other than the opamp?

Your power supply?

a local Tx?

scope problem?

m

mp as it continues to oscillate at 3.5MHz with about 20mVpp-40mVpp in a TIA configuration.

layout has all copper removed from all signal pins on all layers. I have di sconnected the TIA from the following stages, so I'm only looking at the ou tput of the TIA.

tor as the TIA gain and I also put a 20Meg ohm resistor from the inverting input to ground. Changed different cap values and nothing. Dropped the feed back resistor down to 5Meg, 1Meg, 100k ohm while trying various values and nothing. Same oscillation.

at the output. Same thing, but with more of an offset.

s wrong.

+5V voltage in goes through a capacitive multiplier then goes to an inverti ng charge pump to generate the negative supply voltage. These voltages supp ly the op amp. The inverting charge pump has a 10uF cap at the output and t he TIA opamp has 0.1uF on the top layer near its pins.

I will try a unity gain configuration to see if that changes anything. But I will have to cut a trace so maybe a inverting gain of 1 will be okay to t ry instead?

amp as it continues to oscillate at 3.5MHz with about 20mVpp-40mVpp in a T IA configuration.

e layout has all copper removed from all signal pins on all layers. I have disconnected the TIA from the following stages, so I'm only looking at the output of the TIA.

istor as the TIA gain and I also put a 20Meg ohm resistor from the invertin g input to ground. Changed different cap values and nothing. Dropped the fe edback resistor down to 5Meg, 1Meg, 100k ohm while trying various values an d nothing. Same oscillation.

ed at the output. Same thing, but with more of an offset.

; s y

s.

had both the inverting and non-inverting inputs grounded and no feedback c omponents.

grounded and inverting input is where the PD connects to with the feedback resistor and cap on that pin. 0.1uF caps on the supply pins and nothing els e

The spec sheet shows lots of bypass caps. (not only to ground but also between the supply rails.) Do your bypass caps have good ground connections? What do the supply rails look like? Can you change the 3.5 MHz freq. by changing the bypass cap value?

George H.

Yeah you guys may be onto something, that it isn't the op amp. Gonna head back into the lab and probe around and see if I see that signal else where. I'll see how the supplies look. Try different scope probe and scope while I'm at it.

I'll report back soon

Supposedly a photodiode.

--

-TV

amp as it continues to oscillate at 3.5MHz with about 20mVpp-40mVpp in a T IA configuration.

e layout has all copper removed from all signal pins on all layers. I have disconnected the TIA from the following stages, so I'm only looking at the output of the TIA.

istor as the TIA gain and I also put a 20Meg ohm resistor from the invertin g input to ground. Changed different cap values and nothing. Dropped the fe edback resistor down to 5Meg, 1Meg, 100k ohm while trying various values an d nothing. Same oscillation.

ed at the output. Same thing, but with more of an offset.

; s

is wrong.

y

ting charge pump to generate the negative supply voltage. These voltages su pply the op amp. The inverting charge pump has a 10uF cap at the output and the TIA opamp has 0.1uF on the top layer near its pins.

t I will have to cut a trace so maybe a inverting gain of 1 will be okay to try instead?

Hmm that's not necessary. Phil H. will accuse me of shooting spitballs at the problem, but I think your first job is to find out where the 3.5 MHz is coming from. You seem to have eliminated the feedback circuitry, I'm think ing next is to check the power supplies. Hey , maybe it's capacitive loading on the output? How are you monitoring it? x10 'scope probe?

George H.

One fun thing to try is to look at the oscillation on a scope or spectrum analyzer, and see what happens when fingers mash circuits, or when nodes are touched with a pencil or a small screwdriver.

3.5 MHz is sort of low for this opamp.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

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

C-multipliers are emitter followers, and emitter followers love to oscillate. But 3.5 MHz is pretty low for that, too.

Charge pumps are noisy. Trigger the scope from the cp and see if the oscillation is synchronous to that.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

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

te:

e:

op amp as it continues to oscillate at 3.5MHz with about 20mVpp-40mVpp in a TIA configuration.

The layout has all copper removed from all signal pins on all layers. I ha ve disconnected the TIA from the following stages, so I'm only looking at t he output of the TIA.

resistor as the TIA gain and I also put a 20Meg ohm resistor from the inver ting input to ground. Changed different cap values and nothing. Dropped the feedback resistor down to 5Meg, 1Meg, 100k ohm while trying various values and nothing. Same oscillation.

ooked at the output. Same thing, but with more of an offset.

ils;

in

n is

ot

ally

tors.

, I had both the inverting and non-inverting inputs grounded and no feedbac k components.

ng grounded and inverting input is where the PD connects to with the feedba ck resistor and cap on that pin. 0.1uF caps on the supply pins and nothing else

Trouble shooting is an art. The hard part is often figuring out 'what' is wrong. If I can't make it better I try and make it worse.

(I had this intermittent DC fluctuation in a photodiode signal today. I finally traced it down to a beam splitter with the optic a bit loose.)

Yeah... though if it was only in the TIA....

George H.

p amp as it continues to oscillate at 3.5MHz with about 20mVpp-40mVpp in a TIA configuration.

he layout has all copper removed from all signal pins on all layers. I have disconnected the TIA from the following stages, so I'm only looking at the output of the TIA.

sistor as the TIA gain and I also put a 20Meg ohm resistor from the inverti ng input to ground. Changed different cap values and nothing. Dropped the f eedback resistor down to 5Meg, 1Meg, 100k ohm while trying various values a nd nothing. Same oscillation.

ked at the output. Same thing, but with more of an offset.

s;

is

e is wrong.

ly

rting charge pump to generate the negative supply voltage. These voltages s upply the op amp. The inverting charge pump has a 10uF cap at the output an d the TIA opamp has 0.1uF on the top layer near its pins.

So after probing around, I was able to reduce it quite a bit by adding addi tional capacitors at the connector where the incoming supply voltage enters the board.

I think I am pretty content with this and will move back to setting up the TIA and taking it from there.

Thanks everyone for your help.

By the way, I was looking at the output of the TIA opamp via a 30AWG wire t o an SMA connector on my metal box to a SMA cable to the scope. I was looki ng at the output of this amp the entire time while trying various things. A gain, additional caps on the incoming supply had the most improvement.

Wait, what? The charge pump is _after_ the cap multiplier?

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

Hey, I do some of my best work with spitballs. ;)

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

I did think that was interesting.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

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

I would think he has the capacitive multiplier before the charge pump because the charge pump is actually an inverter. If he inverted first he would need two capacitive multipliers to get +-V.

--

Rick C