Prev: Question about AT91SAM7A3 and JTAG
Next: Bluetooth Antenna Design FAIL
From: George Herold on 27 Apr 2010 20:06
Do opamps have a blind spot when they are used to measure the noise on
the power rails feeding them?

I was using the following circuit to measure the supply noise which I
know to be about 1nV/rtHz.


|\
Vsupply----C1C1--+-----+ \
| | >-+--->out..more gain..
R1 +-- / |
R1 | |/ |
| | |
| +-R3R3-+
| R2
| R2
+---+
|
GND

C1 was a 1uF metal film
R1 was 10k
R2 was 100 ohms
R3 was 1k

The opamp was an opa134, though I latter tried an opa228.

When I tied C1 to ground or shorted the (+) input to ground through a
1 ohm resistor I measured about 8nV/rtHz of noise... as expected. When
I hooked C1 to either supply rail the noise not only didn’t go up, it
went down about 10% (in power). (7.7nV/rtHz.)

When I replaced the FET opa134 with the bipolar opa228, I measured
only 3nV/rtHz as expected with the input grounded and basically the
same, (it was a wee bit less.), when hooked onto either supply rail.
The 1nV of supply noise should have caused a 10% increase, which I
didn’t observe.

Is this the same for all opamps.. or is it a Burr-Brown (TI) thing?

George H.
From: Tim Wescott on 27 Apr 2010 20:43
George Herold wrote:
> Do opamps have a blind spot when they are used to measure the noise on
> the power rails feeding them?
>
> I was using the following circuit to measure the supply noise which I
> know to be about 1nV/rtHz.
>
>
> |\
> Vsupply----C1C1--+-----+ \
> | | >-+--->out..more gain..
> R1 +-- / |
> R1 | |/ |
> | | |
> | +-R3R3-+
> | R2
> | R2
> +---+
> |
> GND
>
> C1 was a 1uF metal film
> R1 was 10k
> R2 was 100 ohms
> R3 was 1k
>
> The opamp was an opa134, though I latter tried an opa228.
>
> When I tied C1 to ground or shorted the (+) input to ground through a
> 1 ohm resistor I measured about 8nV/rtHz of noise... as expected. When
> I hooked C1 to either supply rail the noise not only didn�t go up, it
> went down about 10% (in power). (7.7nV/rtHz.)
>
> When I replaced the FET opa134 with the bipolar opa228, I measured
> only 3nV/rtHz as expected with the input grounded and basically the
> same, (it was a wee bit less.), when hooked onto either supply rail.
> The 1nV of supply noise should have caused a 10% increase, which I
> didn�t observe.
>
> Is this the same for all opamps.. or is it a Burr-Brown (TI) thing?
>
> George H.

Try it with a real signal on the supply rail, instead of just what ever
noise you count on being there.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com
From: Jamie on 27 Apr 2010 21:04
George Herold wrote:

> Do opamps have a blind spot when they are used to measure the noise on
> the power rails feeding them?
>
> I was using the following circuit to measure the supply noise which I
> know to be about 1nV/rtHz.
>
>
> |\
> Vsupply----C1C1--+-----+ \
> | | >-+--->out..more gain..
> R1 +-- / |
> R1 | |/ |
> | | |
> | +-R3R3-+
> | R2
> | R2
> +---+
> |
> GND
>
> C1 was a 1uF metal film
> R1 was 10k
> R2 was 100 ohms
> R3 was 1k
>
> The opamp was an opa134, though I latter tried an opa228.
>
> When I tied C1 to ground or shorted the (+) input to ground through a
> 1 ohm resistor I measured about 8nV/rtHz of noise... as expected. When
> I hooked C1 to either supply rail the noise not only didn�t go up, it
> went down about 10% (in power). (7.7nV/rtHz.)
>
> When I replaced the FET opa134 with the bipolar opa228, I measured
> only 3nV/rtHz as expected with the input grounded and basically the
> same, (it was a wee bit less.), when hooked onto either supply rail.
> The 1nV of supply noise should have caused a 10% increase, which I
> didn�t observe.
>
> Is this the same for all opamps.. or is it a Burr-Brown (TI) thing?
>
> George H.
Yes they would, you need R and C as a by pass to the rails of the opamp
so they remain stable.


jamie

From: George Herold on 27 Apr 2010 21:14
On Apr 27, 8:43 pm, Tim Wescott <t...(a)seemywebsite.now> wrote:
> George Herold wrote:
> > Do opamps have a blind spot when they are used to measure the noise on
> > the power rails feeding them?
>
> > I was using the following circuit to measure the supply noise which I
> > know to be about 1nV/rtHz.
>
> >                        |\
> > Vsupply----C1C1--+-----+ \
> >                  |     |  >-+--->out..more gain...
> >                  R1  +-- /  |
> >                  R1  | |/   |
> >                  |   |      |
> >                  |   +-R3R3-+
> >                  |   R2
> >                  |   R2
> >                  +---+
> >                  |
> >                 GND
>
> > C1 was a 1uF metal film
> > R1 was 10k
> > R2 was 100 ohms
> > R3 was 1k
>
> > The opamp was an opa134, though I latter tried an opa228.
>
> > When I tied C1 to ground or shorted the (+) input to ground through a
> > 1 ohm resistor I measured about 8nV/rtHz of noise... as expected. When
> > I hooked C1 to either supply rail the noise not only didn’t go up, it
> > went down about 10% (in power).  (7.7nV/rtHz.)
>
> > When I replaced the FET opa134 with the bipolar opa228, I measured
> > only 3nV/rtHz as expected with the input grounded and basically the
> > same, (it was a wee bit less.), when hooked onto either supply rail.
> > The 1nV of supply noise should have caused a 10% increase, which I
> > didn’t observe.
>
> > Is this the same for all opamps.. or is it a Burr-Brown (TI) thing?
>
> > George H.
>
> Try it with a real signal on the supply rail, instead of just what ever
> noise you count on being there.
>
> --
> Tim Wescott
> Control system and signal processing consultingwww.wescottdesign.com- Hide quoted text -
>
> - Show quoted text -

Yeah, tomorrow I thought I should short out the filters on the power
supply and see how much 'stuff' I can couple through.

George H.
From: George Herold on 27 Apr 2010 21:16
On Apr 27, 9:04 pm, Jamie
<jamie_ka1lpa_not_valid_after_ka1l...(a)charter.net> wrote:
> George Herold wrote:
> > Do opamps have a blind spot when they are used to measure the noise on
> > the power rails feeding them?
>
> > I was using the following circuit to measure the supply noise which I
> > know to be about 1nV/rtHz.
>
> >                        |\
> > Vsupply----C1C1--+-----+ \
> >                  |     |  >-+--->out..more gain...
> >                  R1  +-- /  |
> >                  R1  | |/   |
> >                  |   |      |
> >                  |   +-R3R3-+
> >                  |   R2
> >                  |   R2
> >                  +---+
> >                  |
> >                 GND
>
> > C1 was a 1uF metal film
> > R1 was 10k
> > R2 was 100 ohms
> > R3 was 1k
>
> > The opamp was an opa134, though I latter tried an opa228.
>
> > When I tied C1 to ground or shorted the (+) input to ground through a
> > 1 ohm resistor I measured about 8nV/rtHz of noise... as expected. When
> > I hooked C1 to either supply rail the noise not only didn’t go up, it
> > went down about 10% (in power).  (7.7nV/rtHz.)
>
> > When I replaced the FET opa134 with the bipolar opa228, I measured
> > only 3nV/rtHz as expected with the input grounded and basically the
> > same, (it was a wee bit less.), when hooked onto either supply rail.
> > The 1nV of supply noise should have caused a 10% increase, which I
> > didn’t observe.
>
> > Is this the same for all opamps.. or is it a Burr-Brown (TI) thing?
>
> > George H.
>
> Yes they would, you need  R and C as a by pass to the rails of the opamp
> so they remain stable.
>
> jamie- Hide quoted text -
>
> - Show quoted text -

The power supply rails are heavily filtered. The noise is at the 1nV/
rtHz level.

George H.
 |  Next  |  Last
Pages: 1 2
Prev: Question about AT91SAM7A3 and JTAG
Next: Bluetooth Antenna Design FAIL