LM7171: Inverting Amplifier with High Impedance input

Hello Sahand,

If possible move the sensor as close as possible to the inverting input (any capacitance on this node (i.e. your coax cables), causes unwanted phase shift / phase margin loss. If you cannot move the sensor as described, at least isolate the cables from the inverting input using a small form factor resistor placed close to the inverting input. Shorter cables also help.

Probing the output directly with a 12pF scope probe is also something to avoid, if you can. An isolation resistor (or a 50ohm series output resistor leading to the scope 50ohm termination) is another method that avoids loading this high speed amplifier with capacitance.

The non-inverting input 1k resistance adds noise and could possibly affect stability. If the steps above don't resolve your issue, touch a scope probe to the non-inverting input (while monitoring the output) and see if that is a sensitive node or not? cap de-coupling to ground could be something to try (or alternatively a small resistor in series with a decoupling cap).

Also, the LM7171 is only stable for noise gains of 2V/V or higher (your noise gain is 1V/V = 1+ 1k/5M). You may have to employ lead-lag compensation across your input:

http://www.ti.com/lit/ml/sloa079/sloa079.pdf

Finally, use the LM7171 EVAL board if you suspect your bread board layout:

http://www.ti.com/tool/lmh730227

Regards,

Hooman

Hello Sahand,

Here is your email (put here for reference).

"Hooman,

Thank you very much for your reply and help. Sorry, TI community website was down that's why I am reaching you out from here. I used BNC terminator at the output and the problem of 20V (20MHz) output signal that I was getting got resolved. Now, the problem is a signal with high frequency noise on it. The two coax cables stranded wires are not connected to any thing at this moment. When I touch them with bare hand, the noise goes away. Also, when I move the coax cables my output changes. I am not able to make these cables shorter. Meanwhile, would you please explain more about small form factor resistor that I need to place close to the inverting input. What can it be the value of it and does the other end of the resistor need to be connected to ground?

Thanks,

Sahand Pirouzpanah"

I've responded here instead of email for easier future reference.

My response below:

1. I'm not sure what you mean when you say "The two coax cables stranded wires are not connected to any thing at this moment.". Do you mean you remove the sensor all together and you are getting the "noise"? As I noted yesterday, the LM7171 won't be stable with a gain of +1V/V (which is your effective gain). You must use a unity gain stable device or otherwise add compensation. At least that's what I think is going on.

2. Small form factor resistor: I meant you could use a 0406 size SMT resistor placed very close to the inverting node between the sensor and the amplifiers. This way, the long trace / capacitance to your sensor is somewhat isolated from the sensitive inverting input. If you go to a unity stable configuration (either by using a different device or by using compensation), then you could vary this resistance and see if isolating the sensor side eliminates your problem or not. But first you must address the unity gain issue.

3. Making the cables shorter: If I were you, I'd find a way to bring the amplifier close to the sensor (for bench testing purposes only). This may not be doable in the real application but this gives you insight as to what is making your circuit unstable.

4. Use the EVM layout views as guide (i.e. ground plane with cut out near sensitive areas, power supply decoupling cap placement and return path routing, etc.).

Regards,

Hooman

Thank you Hooman for the quick response. 

1.  I am using the solid-central wire of coax cable and the braided wires around the cable are free and not connected to anything. The solid wire of one coax cable is connected to a function generator and the other solid wire of the other coax cable is connected to the opamp. When I touch the free braided wires with bare hand, the noise goes away. Since, I am using the inverting amplifier so the gain is close to zero (-1k/5M). I would be very thankful if you suggest the proper value for 1/RC in lead-lag compensation. 

Thank you very much,

Sahand

Hi Sahand,

Using the center conductor of a coax (with the return left floating) may be worse than just using a piece of 26 gauge hook up wire for the connection(s). You are probably not getting any "shielding" from the return / braid and you're probably getting a higher capacitance hanging on a sensitive node. Once again, I recommend you debug your setup with a tight sensor / amplifier arrangement, if at all possible.

If you tell me the set of requirements (e.g. supply voltage, bandwidth, slew rate, offset voltage, noise, input voltage range), I'd be possibly be able to recommend a unity gain stable device which may have a better chance. Is this something you can provide?

Let's see if you can supply the information above and if a better-suited device is available before we get into the compensation scheme. Agreed?

Regards,

Hooman

I forgot to mention:

The gain I'm talking about is "noise gain" which in your circuit is close to 1V/V. Noise Gain is what determines phase margin / stability.

Regards,

Hooman

Thank you again Hooman,

Below are the information,

 Supply voltage:+-15V,

Bandwidth: 200MHz,

Slew rate: 3800V/us

Input Offset Voltage: 1mv

Input Voltage Noise: 7nv/(Hz)^.5

Input Voltage Range: +-2V

8pin-Dip Package (since I want to use it on Bread Board)

Closed loop stable unity gain up to 15MHz 

Thanks,

Sahand

Hello Sahand,

The closest  device to your list of requirements is the THS3001, as shown below to the left. However, its offset voltage is 3mV and it is not available in 8 pin DIP! However it is unity gain stable with RF= 1kohm. Since it is a current feedback device you must be careful with the feedback you choose (but Av= 1V/V feedback matches your 1kohm). Its noise is quite a bit lower too:

You can get the Selguide (image above) device selection tool from the link below, to narrow down the list yourself as well:

http://www.ti.com/tool/opamps-selguide

As far as compensating the LM7171 for stability, here is what I think might help a lot based on simulation (it remains to be seen on the bench in your application):

Increase CF (across RF, currently 1pF) to a value close to what the capacitance you are adding due to your setup. For example, with 6" of RG59 cable (100pF/meter) hanging on the inverting input, that would be about 15.2pF total capacitance. Use CF= 15pF to see if that improves stability or not? TINA-TI simulation shows that this should get your phase margin to around 60 deg (which should be adequate for stability). The step response won't look pretty but it might be good enough for your application.

Regards,

Hooman