Active differential power rail probe...

[Dominic Chan]

I\'ve been doing some power rail noise/ripple measurements lately. It seems like the generally agreed best practice is to solder a coax lead as close to the load as possible, taking care not to probe directly across backside decoupling caps to avoid underestimating HF noise. This coax lead is then connected to a power rail probe / 50R scope input.

Since the boards I\'m working with have 0V tied to Earth, there is a ground loop formed when making these measurements. I\'ve found that if I power the scope and board off the same power strip the environmental noise coupling into this loop is sufficiently small for good measurements.

This has got me wondering though, why isn\'t an active differential probe optimized for power rail measurements a thing? Perhaps it does exist? - If so please let me know!

Cheers, Dominic

 

[olaf]

Dominic Chan <dominic@arista.com> wrote:

This has got me wondering though, why isn\'t an active differential
probe optimized for power rail measurements a thing? Perhaps it does
exist? - If so please let me know!

Tek offers the TPR4000. :-D And I think there is one from Agilent
in the same price range, too.

https://www.tek.com/en/datasheet/active-power-rail-probes

These probes are AC-coupled, so offering a direct connection for
High frequency and add the DC-Level to the output again.

Olaf

 

[Dominic Chan]

On Wednesday, May 4, 2022 at 2:15:09 PM UTC+10, olaf wrote:

Dominic Chan <dom...@arista.com> wrote:


This has got me wondering though, why isn\'t an active differential
probe optimized for power rail measurements a thing? Perhaps it does
exist? - If so please let me know!
Tek offers the TPR4000. :-D And I think there is one from Agilent
in the same price range, too.

https://www.tek.com/en/datasheet/active-power-rail-probes

These probes are AC-coupled, so offering a direct connection for
High frequency and add the DC-Level to the output again.

Olaf

Olaf those are the classic power rail probes - the DC level circuitry is \'active\' but the AC path is purely passive, and of course it is a single ended probe.

 

[whit3rd]

On Tuesday, May 3, 2022 at 8:48:08 PM UTC-7, Dominic Chan wrote:

I\'ve been doing some power rail noise/ripple measurements lately.
....
This has got me wondering though, why isn\'t an active differential probe optimized for power rail measurements a thing? Perhaps it does exist? - If so please let me know!

For a quick check, I just use two channels, with x10 passive probes, and difference \'em.
It\'s not terrific for accuracy, but power isn\'t required to be accurate.
Biggest drawback: you need both hands to hold the probes.

 

[Dominic Chan]

On Wednesday, May 4, 2022 at 2:49:41 PM UTC+10, whit3rd wrote:

On Tuesday, May 3, 2022 at 8:48:08 PM UTC-7, Dominic Chan wrote:
I\'ve been doing some power rail noise/ripple measurements lately.
...
This has got me wondering though, why isn\'t an active differential probe optimized for power rail measurements a thing? Perhaps it does exist? - If so please let me know!
For a quick check, I just use two channels, with x10 passive probes, and difference \'em.
It\'s not terrific for accuracy, but power isn\'t required to be accurate.
Biggest drawback: you need both hands to hold the probes.

Yep that works for a quick check, I\'ve done the same for jank current measurements through a resistor.
The issue with x10 probes for power rail measurement is the measurement noise. E.g. Xilinx transceivers guidelines allow for 10mVpp from 10kHz to 80MHz. Using active differential probes on rails with even more stringent requirements also runs into the same measurement noise issue.

Since input cap / loading of a probe is pretty much irrelevant for power rail measurements, it seems to me that an active diff probe that is designed intentionally for power rail measurements could obtain much better noise performance.

E.g. the RT-ZD40 active diff probe has 3mVrms input referred noise, while 50 ohm oscilloscope frontends easily achieve sub 1mVrms.

 

[Mike Monett]

Dominic Chan <dominic@arista.com> wrote:

On Wednesday, May 4, 2022 at 2:49:41 PM UTC+10, whit3rd wrote:
On Tuesday, May 3, 2022 at 8:48:08 PM UTC-7, Dominic Chan wrote:
I\'ve been doing some power rail noise/ripple measurements lately. ...
This has got me wondering though, why isn\'t an active differential
prob
e optimized for power rail measurements a thing? Perhaps it does exist?
- I f so please let me know!
For a quick check, I just use two channels, with x10 passive probes,
and difference \'em. It\'s not terrific for accuracy, but power isn\'t
required to be accurate.

Biggest drawback: you need both hands to hold the probes.

Yep that works for a quick check, I\'ve done the same for jank current
measurements through a resistor. The issue with x10 probes for power
rail measurement is the measurement noise. E.g. Xilinx transceivers
guidelines allow for 10mVpp from 10kHz to 80MHz. Using active
differential probes on rails with even more stringent requirements also
runs into the same measurement noise issue.

Since input cap / loading of a probe is pretty much irrelevant for power
rail measurements, it seems to me that an active diff probe that is
designed intentionally for power rail measurements could obtain much
better noise performance.

E.g. the RT-ZD40 active diff probe has 3mVrms input referred noise,
while 50 ohm oscilloscope frontends easily achieve sub 1mVrms.

For interest, Marco Reps shows how to get 0.001pf isolation on line power:

eXtReMe iSoLaTiOn ( Low Leakage Power in Precision Electronics )
https://www.youtube.com/watch?v=9JinSfCKuNQ

For even more isolation, a number of battery powered oscilloscopes are
available for under $200. For example, the FNIRSI-1013D is a 100 MHz dual
trace touch tablet:

http://www.fnirsi.cn/productinfo/556152.html

It is available at

https://www.aliexpress.com/i/4000861098295.html

Dave Jones, EEVblog doesn\'t think much of it, but whadda gonna do for CAD
$168.74 ?

https://www.youtube.com/watch?v=5iwtDwJlbWk

The lowest range is 50mV, but you could easily add a differential opamp for
low level work, such as the opa846:

https://www.ti.com/lit/ds/symlink/opa846.pdf

It is under $10 at Octopart:

https://octopart.com/search?q=opa846&currency=USD&specs=0




--
MRM

 

[server]

On Tue, 3 May 2022 20:48:04 -0700 (PDT), Dominic Chan
<dominic@arista.com> wrote:

I\'ve been doing some power rail noise/ripple measurements lately. It seems like the generally agreed best practice is to solder a coax lead as close to the load as possible, taking care not to probe directly across backside decoupling caps to avoid underestimating HF noise. This coax lead is then connected to a power rail probe / 50R scope input.

Since the boards I\'m working with have 0V tied to Earth, there is a ground loop formed when making these measurements. I\'ve found that if I power the scope and board off the same power strip the environmental noise coupling into this loop is sufficiently small for good measurements.

This has got me wondering though, why isn\'t an active differential probe optimized for power rail measurements a thing? Perhaps it does exist? - If so please let me know!

Cheers, Dominic

Tek has a number of fully isolated scopes and outboard isolated
probes. We have a 4-channel isolated scope, TPS2024 I recall, that\'s
wonderful for working on power things. Clip the probe ground things
anywhere.

https://www.tek.com/en/search?keywords=isolated



--

Anybody can count to one.

- Robert Widlar

 

[olaf]

Dominic Chan <dominic@arista.com> wrote:

Olaf those are the classic power rail probes - the DC level
circuitry is \'active\' but the AC path is purely passive, and of
course it is a single ended probe.

But the DC level allow more shifting than the scope alone,
so for most power supply it is enought.

Olaf

 

[legg]

On Tue, 3 May 2022 20:48:04 -0700 (PDT), Dominic Chan
<dominic@arista.com> wrote:

I\'ve been doing some power rail noise/ripple measurements lately. It seems like the generally agreed best practice is to solder a coax lead as close to the load as possible, taking care not to probe directly across backside decoupling caps to avoid underestimating HF noise. This coax lead is then connected to a power rail probe / 50R scope input.

Since the boards I\'m working with have 0V tied to Earth, there is a ground loop formed when making these measurements. I\'ve found that if I power the scope and board off the same power strip the environmental noise coupling into this loop is sufficiently small for good measurements.

This has got me wondering though, why isn\'t an active differential probe optimized for power rail measurements a thing? Perhaps it does exist? - If so please let me know!

Cheers, Dominic

Noise, by definition, is not a DC thing.

For PARD (Peak And Random Deviation), you would include DC.

Differential and common-mode (AC) noise can be measured/contrasted
using passive or active combination, over a specific bandwidth.

Wideband diff probes are available for lower frequencies, but
reduce with BW and accuracy as DC maximum withstand levels increase.

Is somebody asking for this data? Ask for the measurement standards
being referenced. This will usually include a test set-up figure
and equipment requirements.

I built a 2-pin, battery-operated Ppk measurement probe for quick
\'relative\' PSU measurements that gave a digital reading of % in the
ancient past. This used the commodity ICL7106 to do a ratiometric
comparison between the DC (ref) and an ECL-type biased AC peak
detector, though the peak could be obtained as a number, as well.

It\'s not common test gear because it has very low volume demand.
Indications are sufficient, because real world testing usually
requires a >>2:1 performance margin on the test limit, so
tolerances are not a great issue - just repeatability.

Aim is to avoid trouble, not to track it down after the fact;
though it can be used as prediction/symptom of failure elsewhere.
Physical noise : Same thing.

Hence strict test setup and usefulness of field experience, in
tracking down measurement or application errors / physical
configuration pitfalls.

RL

 

[Chris Jones]

On 04/05/2022 13:48, Dominic Chan wrote:

I\'ve been doing some power rail noise/ripple measurements lately. It seems like the generally agreed best practice is to solder a coax lead as close to the load as possible, taking care not to probe directly across backside decoupling caps to avoid underestimating HF noise. This coax lead is then connected to a power rail probe / 50R scope input.

Since the boards I\'m working with have 0V tied to Earth, there is a ground loop formed when making these measurements. I\'ve found that if I power the scope and board off the same power strip the environmental noise coupling into this loop is sufficiently small for good measurements.

This has got me wondering though, why isn\'t an active differential probe optimized for power rail measurements a thing? Perhaps it does exist? - If so please let me know!

Cheers, Dominic

I put some connectors on an AD830 because I have lots of them:

https://twitter.com/chrisgj198/status/330590675806605312

 

[Phil Hobbs]

Mike Monett wrote:

Dominic Chan <dominic@arista.com> wrote:

On Wednesday, May 4, 2022 at 2:49:41 PM UTC+10, whit3rd wrote:
On Tuesday, May 3, 2022 at 8:48:08 PM UTC-7, Dominic Chan wrote:
I\'ve been doing some power rail noise/ripple measurements lately. ...
This has got me wondering though, why isn\'t an active differential
prob
e optimized for power rail measurements a thing? Perhaps it does exist?
- I f so please let me know!
For a quick check, I just use two channels, with x10 passive probes,
and difference \'em. It\'s not terrific for accuracy, but power isn\'t
required to be accurate.

Biggest drawback: you need both hands to hold the probes.

Yep that works for a quick check, I\'ve done the same for jank current
measurements through a resistor. The issue with x10 probes for power
rail measurement is the measurement noise. E.g. Xilinx transceivers
guidelines allow for 10mVpp from 10kHz to 80MHz. Using active
differential probes on rails with even more stringent requirements also
runs into the same measurement noise issue.

Since input cap / loading of a probe is pretty much irrelevant for power
rail measurements, it seems to me that an active diff probe that is
designed intentionally for power rail measurements could obtain much
better noise performance.

E.g. the RT-ZD40 active diff probe has 3mVrms input referred noise,
while 50 ohm oscilloscope frontends easily achieve sub 1mVrms.

For interest, Marco Reps shows how to get 0.001pf isolation on line power:

eXtReMe iSoLaTiOn ( Low Leakage Power in Precision Electronics )
https://www.youtube.com/watch?v=9JinSfCKuNQ

For even more isolation, a number of battery powered oscilloscopes are
available for under $200. For example, the FNIRSI-1013D is a 100 MHz dual
trace touch tablet:

http://www.fnirsi.cn/productinfo/556152.html

It is available at

https://www.aliexpress.com/i/4000861098295.html

Dave Jones, EEVblog doesn\'t think much of it, but whadda gonna do for CAD
$168.74 ?

https://www.youtube.com/watch?v=5iwtDwJlbWk

The lowest range is 50mV, but you could easily add a differential opamp for
low level work, such as the opa846:

https://www.ti.com/lit/ds/symlink/opa846.pdf

It is under $10 at Octopart:

https://octopart.com/search?q=opa846&currency=USD&specs=0

Another approach is to put a Mini-Circuits transformer across the
current sense resistor, and use a normal scope on the secondary. Needs
a ~1 uF coupling cap, of course, but that\'ll get you down to the low
kilohertz. Coupling is very high--their gizmos typically measure over
0.999, often about 0.9997.

Try the T1-6.

<https://www.minicircuits.com/pdfs/T1-6-X65+.pdf>

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[piglet]

On 04/05/2022 17:42, Phil Hobbs wrote:

Another approach is to put a Mini-Circuits transformer across the
current sense resistor, and use a normal scope on the secondary.  Needs
a ~1 uF coupling cap, of course, but that\'ll get you down to the low
kilohertz.  Coupling is very high--their gizmos typically measure over
0.999, often about 0.9997.

Try the T1-6.

https://www.minicircuits.com/pdfs/T1-6-X65+.pdf

What is the pri-sec capacitance of that transformer the datasheet is silent?

piglet

 

[Jeroen Belleman]

piglet wrote:

On 04/05/2022 17:42, Phil Hobbs wrote:
Another approach is to put a Mini-Circuits transformer across the
current sense resistor, and use a normal scope on the secondary.
Needs a ~1 uF coupling cap, of course, but that\'ll get you down to the
low kilohertz. Coupling is very high--their gizmos typically measure
over 0.999, often about 0.9997.

Try the T1-6.

https://www.minicircuits.com/pdfs/T1-6-X65+.pdf


What is the pri-sec capacitance of that transformer the datasheet is
silent?

piglet

About 13pF.

Jeroen Belleman

 

[server]

On Wed, 4 May 2022 12:42:39 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

Mike Monett wrote:
Dominic Chan <dominic@arista.com> wrote:

On Wednesday, May 4, 2022 at 2:49:41 PM UTC+10, whit3rd wrote:
On Tuesday, May 3, 2022 at 8:48:08 PM UTC-7, Dominic Chan wrote:
I\'ve been doing some power rail noise/ripple measurements lately. ...
This has got me wondering though, why isn\'t an active differential
prob
e optimized for power rail measurements a thing? Perhaps it does exist?
- I f so please let me know!
For a quick check, I just use two channels, with x10 passive probes,
and difference \'em. It\'s not terrific for accuracy, but power isn\'t
required to be accurate.

Biggest drawback: you need both hands to hold the probes.

Yep that works for a quick check, I\'ve done the same for jank current
measurements through a resistor. The issue with x10 probes for power
rail measurement is the measurement noise. E.g. Xilinx transceivers
guidelines allow for 10mVpp from 10kHz to 80MHz. Using active
differential probes on rails with even more stringent requirements also
runs into the same measurement noise issue.

Since input cap / loading of a probe is pretty much irrelevant for power
rail measurements, it seems to me that an active diff probe that is
designed intentionally for power rail measurements could obtain much
better noise performance.

E.g. the RT-ZD40 active diff probe has 3mVrms input referred noise,
while 50 ohm oscilloscope frontends easily achieve sub 1mVrms.

For interest, Marco Reps shows how to get 0.001pf isolation on line power:

eXtReMe iSoLaTiOn ( Low Leakage Power in Precision Electronics )
https://www.youtube.com/watch?v=9JinSfCKuNQ

For even more isolation, a number of battery powered oscilloscopes are
available for under $200. For example, the FNIRSI-1013D is a 100 MHz dual
trace touch tablet:

http://www.fnirsi.cn/productinfo/556152.html

It is available at

https://www.aliexpress.com/i/4000861098295.html

Dave Jones, EEVblog doesn\'t think much of it, but whadda gonna do for CAD
$168.74 ?

https://www.youtube.com/watch?v=5iwtDwJlbWk

The lowest range is 50mV, but you could easily add a differential opamp for
low level work, such as the opa846:

https://www.ti.com/lit/ds/symlink/opa846.pdf

It is under $10 at Octopart:

https://octopart.com/search?q=opa846&currency=USD&specs=0




Another approach is to put a Mini-Circuits transformer across the
current sense resistor, and use a normal scope on the secondary. Needs
a ~1 uF coupling cap, of course, but that\'ll get you down to the low
kilohertz. Coupling is very high--their gizmos typically measure over
0.999, often about 0.9997.

Try the T1-6.

https://www.minicircuits.com/pdfs/T1-6-X65+.pdf

Cheers

Phil Hobbs

The DRQ series dual-winding inductors have crazy high coupling and are
good for low frequency cases. They come up to 1 mH.





--

Anybody can count to one.

- Robert Widlar

 

[Dominic Chan]

On Wednesday, May 4, 2022 at 11:44:50 PM UTC+10, jla...@highlandsniptechnology.com wrote:

On Tue, 3 May 2022 20:48:04 -0700 (PDT), Dominic Chan
dom...@arista.com> wrote:

I\'ve been doing some power rail noise/ripple measurements lately. It seems like the generally agreed best practice is to solder a coax lead as close to the load as possible, taking care not to probe directly across backside decoupling caps to avoid underestimating HF noise. This coax lead is then connected to a power rail probe / 50R scope input.

Since the boards I\'m working with have 0V tied to Earth, there is a ground loop formed when making these measurements. I\'ve found that if I power the scope and board off the same power strip the environmental noise coupling into this loop is sufficiently small for good measurements.

This has got me wondering though, why isn\'t an active differential probe optimized for power rail measurements a thing? Perhaps it does exist? - If so please let me know!

Cheers, Dominic
Tek has a number of fully isolated scopes and outboard isolated
probes. We have a 4-channel isolated scope, TPS2024 I recall, that\'s
wonderful for working on power things. Clip the probe ground things
anywhere.

https://www.tek.com/en/search?keywords=isolated



--

Anybody can count to one.

- Robert Widlar

Neat! Their isolated probes with 50 ohm probe input impedance is along the lines of what I was thinking of. Though I don\'t think true isolation is necessary, just some CMRR. Why isn\'t this a more common thing?

 

[Dominic Chan]

On Thursday, May 5, 2022 at 2:42:53 AM UTC+10, Phil Hobbs wrote:

Mike Monett wrote:
Dominic Chan <dom...@arista.com> wrote:

On Wednesday, May 4, 2022 at 2:49:41 PM UTC+10, whit3rd wrote:
On Tuesday, May 3, 2022 at 8:48:08 PM UTC-7, Dominic Chan wrote:
I\'ve been doing some power rail noise/ripple measurements lately. ...
This has got me wondering though, why isn\'t an active differential
prob
e optimized for power rail measurements a thing? Perhaps it does exist?
- I f so please let me know!
For a quick check, I just use two channels, with x10 passive probes,
and difference \'em. It\'s not terrific for accuracy, but power isn\'t
required to be accurate.

Biggest drawback: you need both hands to hold the probes.

Yep that works for a quick check, I\'ve done the same for jank current
measurements through a resistor. The issue with x10 probes for power
rail measurement is the measurement noise. E.g. Xilinx transceivers
guidelines allow for 10mVpp from 10kHz to 80MHz. Using active
differential probes on rails with even more stringent requirements also
runs into the same measurement noise issue.

Since input cap / loading of a probe is pretty much irrelevant for power
rail measurements, it seems to me that an active diff probe that is
designed intentionally for power rail measurements could obtain much
better noise performance.

E.g. the RT-ZD40 active diff probe has 3mVrms input referred noise,
while 50 ohm oscilloscope frontends easily achieve sub 1mVrms.

For interest, Marco Reps shows how to get 0.001pf isolation on line power:

eXtReMe iSoLaTiOn ( Low Leakage Power in Precision Electronics )
https://www.youtube.com/watch?v=9JinSfCKuNQ

For even more isolation, a number of battery powered oscilloscopes are
available for under $200. For example, the FNIRSI-1013D is a 100 MHz dual
trace touch tablet:

http://www.fnirsi.cn/productinfo/556152.html

It is available at

https://www.aliexpress.com/i/4000861098295.html

Dave Jones, EEVblog doesn\'t think much of it, but whadda gonna do for CAD
$168.74 ?

https://www.youtube.com/watch?v=5iwtDwJlbWk

The lowest range is 50mV, but you could easily add a differential opamp for
low level work, such as the opa846:

https://www.ti.com/lit/ds/symlink/opa846.pdf

It is under $10 at Octopart:

https://octopart.com/search?q=opa846&currency=USD&specs=0




Another approach is to put a Mini-Circuits transformer across the
current sense resistor, and use a normal scope on the secondary. Needs
a ~1 uF coupling cap, of course, but that\'ll get you down to the low
kilohertz. Coupling is very high--their gizmos typically measure over
0.999, often about 0.9997.

Try the T1-6.

https://www.minicircuits.com/pdfs/T1-6-X65+.pdf

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

I\'m curious if you\'ve ever tried using a transformer in front of a TIA to boost the photocurrent from a PD for effectively lower TIA input noise? (understandably the impedance looking into the transformer + TIA would be greater than the TIA alone).

 

[Phil Hobbs]

Dominic Chan wrote:

On Thursday, May 5, 2022 at 2:42:53 AM UTC+10, Phil Hobbs wrote:
Mike Monett wrote:
Dominic Chan <dom...@arista.com> wrote:

On Wednesday, May 4, 2022 at 2:49:41 PM UTC+10, whit3rd wrote:
On Tuesday, May 3, 2022 at 8:48:08 PM UTC-7, Dominic Chan
wrote:
I\'ve been doing some power rail noise/ripple measurements
lately. ... This has got me wondering though, why isn\'t an
active differential prob
e optimized for power rail measurements a thing? Perhaps it
does exist? - I f so please let me know!
For a quick check, I just use two channels, with x10 passive
probes, and difference \'em. It\'s not terrific for accuracy,
but power isn\'t required to be accurate.

Biggest drawback: you need both hands to hold the probes.

Yep that works for a quick check, I\'ve done the same for jank
current measurements through a resistor. The issue with x10
probes for power rail measurement is the measurement noise.
E.g. Xilinx transceivers guidelines allow for 10mVpp from 10kHz
to 80MHz. Using active differential probes on rails with even
more stringent requirements also runs into the same measurement
noise issue.

Since input cap / loading of a probe is pretty much irrelevant
for power rail measurements, it seems to me that an active diff
probe that is designed intentionally for power rail
measurements could obtain much better noise performance.

E.g. the RT-ZD40 active diff probe has 3mVrms input referred
noise, while 50 ohm oscilloscope frontends easily achieve sub
1mVrms.

For interest, Marco Reps shows how to get 0.001pf isolation on
line power:

eXtReMe iSoLaTiOn ( Low Leakage Power in Precision Electronics )
https://www.youtube.com/watch?v=9JinSfCKuNQ

For even more isolation, a number of battery powered
oscilloscopes are available for under $200. For example, the
FNIRSI-1013D is a 100 MHz dual trace touch tablet:

http://www.fnirsi.cn/productinfo/556152.html

It is available at

https://www.aliexpress.com/i/4000861098295.html

Dave Jones, EEVblog doesn\'t think much of it, but whadda gonna do
for CAD $168.74 ?

https://www.youtube.com/watch?v=5iwtDwJlbWk

The lowest range is 50mV, but you could easily add a differential
opamp for low level work, such as the opa846:

https://www.ti.com/lit/ds/symlink/opa846.pdf

It is under $10 at Octopart:

https://octopart.com/search?q=opa846&currency=USD&specs=0




Another approach is to put a Mini-Circuits transformer across the
current sense resistor, and use a normal scope on the secondary.
Needs a ~1 uF coupling cap, of course, but that\'ll get you down to
the low kilohertz. Coupling is very high--their gizmos typically
measure over 0.999, often about 0.9997.

Try the T1-6.

https://www.minicircuits.com/pdfs/T1-6-X65+.pdf



I\'m curious if you\'ve ever tried using a transformer in front of a
TIA to boost the photocurrent from a PD for effectively lower TIA
input noise? (understandably the impedance looking into the
transformer + TIA would be greater than the TIA alone).

I\'ve occasionally used reactive matching networks for narrowband UHFish
things. At ordinary frequencies transformers don\'t help, because in dim
light the limiting noise source is the internal series resistance of the
photodiode.

Getting to that point requires a decent TIA design, of course--barefoot
op amps generally won\'t cut it.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[John Doe]

TIA = Transimpedance Amplifier

 

[server]

On Thu, 5 May 2022 15:13:39 -0700 (PDT), Dominic Chan
<dominic@arista.com> wrote:

On Wednesday, May 4, 2022 at 11:44:50 PM UTC+10, jla...@highlandsniptechnology.com wrote:
On Tue, 3 May 2022 20:48:04 -0700 (PDT), Dominic Chan
dom...@arista.com> wrote:

I\'ve been doing some power rail noise/ripple measurements lately. It seems like the generally agreed best practice is to solder a coax lead as close to the load as possible, taking care not to probe directly across backside decoupling caps to avoid underestimating HF noise. This coax lead is then connected to a power rail probe / 50R scope input.

Since the boards I\'m working with have 0V tied to Earth, there is a ground loop formed when making these measurements. I\'ve found that if I power the scope and board off the same power strip the environmental noise coupling into this loop is sufficiently small for good measurements.

This has got me wondering though, why isn\'t an active differential probe optimized for power rail measurements a thing? Perhaps it does exist? - If so please let me know!

Cheers, Dominic
Tek has a number of fully isolated scopes and outboard isolated
probes. We have a 4-channel isolated scope, TPS2024 I recall, that\'s
wonderful for working on power things. Clip the probe ground things
anywhere.

https://www.tek.com/en/search?keywords=isolated



--

Anybody can count to one.

- Robert Widlar
Neat! Their isolated probes with 50 ohm probe input impedance is along the lines of what I was thinking of. Though I don\'t think true isolation is necessary, just some CMRR. Why isn\'t this a more common thing?

Some of the cheap ebay/amazon battery-powered scopes are actually not
bad. That does single-channel-isolated measurement.

https://www.amazon.com/gp/product/B081Q2KDJT/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1



--

Anybody can count to one.

- Robert Widlar

 

[John Larkin]

On Thu, 5 May 2022 21:14:45 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

Dominic Chan wrote:
On Thursday, May 5, 2022 at 2:42:53 AM UTC+10, Phil Hobbs wrote:
Mike Monett wrote:
Dominic Chan <dom...@arista.com> wrote:

On Wednesday, May 4, 2022 at 2:49:41 PM UTC+10, whit3rd wrote:
On Tuesday, May 3, 2022 at 8:48:08 PM UTC-7, Dominic Chan
wrote:
I\'ve been doing some power rail noise/ripple measurements
lately. ... This has got me wondering though, why isn\'t an
active differential prob
e optimized for power rail measurements a thing? Perhaps it
does exist? - I f so please let me know!
For a quick check, I just use two channels, with x10 passive
probes, and difference \'em. It\'s not terrific for accuracy,
but power isn\'t required to be accurate.

Biggest drawback: you need both hands to hold the probes.

Yep that works for a quick check, I\'ve done the same for jank
current measurements through a resistor. The issue with x10
probes for power rail measurement is the measurement noise.
E.g. Xilinx transceivers guidelines allow for 10mVpp from 10kHz
to 80MHz. Using active differential probes on rails with even
more stringent requirements also runs into the same measurement
noise issue.

Since input cap / loading of a probe is pretty much irrelevant
for power rail measurements, it seems to me that an active diff
probe that is designed intentionally for power rail
measurements could obtain much better noise performance.

E.g. the RT-ZD40 active diff probe has 3mVrms input referred
noise, while 50 ohm oscilloscope frontends easily achieve sub
1mVrms.

For interest, Marco Reps shows how to get 0.001pf isolation on
line power:

eXtReMe iSoLaTiOn ( Low Leakage Power in Precision Electronics )
https://www.youtube.com/watch?v=9JinSfCKuNQ

For even more isolation, a number of battery powered
oscilloscopes are available for under $200. For example, the
FNIRSI-1013D is a 100 MHz dual trace touch tablet:

http://www.fnirsi.cn/productinfo/556152.html

It is available at

https://www.aliexpress.com/i/4000861098295.html

Dave Jones, EEVblog doesn\'t think much of it, but whadda gonna do
for CAD $168.74 ?

https://www.youtube.com/watch?v=5iwtDwJlbWk

The lowest range is 50mV, but you could easily add a differential
opamp for low level work, such as the opa846:

https://www.ti.com/lit/ds/symlink/opa846.pdf

It is under $10 at Octopart:

https://octopart.com/search?q=opa846&currency=USD&specs=0




Another approach is to put a Mini-Circuits transformer across the
current sense resistor, and use a normal scope on the secondary.
Needs a ~1 uF coupling cap, of course, but that\'ll get you down to
the low kilohertz. Coupling is very high--their gizmos typically
measure over 0.999, often about 0.9997.

Try the T1-6.

https://www.minicircuits.com/pdfs/T1-6-X65+.pdf



I\'m curious if you\'ve ever tried using a transformer in front of a
TIA to boost the photocurrent from a PD for effectively lower TIA
input noise? (understandably the impedance looking into the
transformer + TIA would be greater than the TIA alone).

I\'ve occasionally used reactive matching networks for narrowband UHFish
things. At ordinary frequencies transformers don\'t help, because in dim
light the limiting noise source is the internal series resistance of the
photodiode.

Getting to that point requires a decent TIA design, of course--barefoot
op amps generally won\'t cut it.

Cheers

Phil Hobbs

Somebody, SRS I think, makes a lab amplifier with a transformer in the
front end. The noise level is way sub 1 nv/rthz. A good (well
shielded!) transformer is noiseless voltage gain.

I recall that ribbon microphones like to have a transformer. They are
very low voltage, very low impedance sources.

Given a low impedance source, a DC-coupled transformer-enhanced amp
wouldn\'t be difficult.

--

If a man will begin with certainties, he shall end with doubts,
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon