Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > sci.electronics.design > #452617 > unrolled thread

DS1054 LF common mode aberation

Back to article view | Back to sci.electronics.design

Contents

  DS1054 LF common mode aberation legg <legg@nospam.magma.ca> - 2017-03-19 17:04 -0400
    Re: DS1054 LF common mode aberation Frank Miles <fpm@u.washington.edu> - 2017-03-19 23:56 +0000
      Re: DS1054 LF common mode aberation legg <legg@nospam.magma.ca> - 2017-03-19 22:21 -0400
        Re: DS1054 LF common mode aberation Frank Miles <fpm@u.washington.edu> - 2017-03-20 16:30 +0000
          Re: DS1054 LF common mode aberation Jon Elson <jmelson@wustl.edu> - 2017-03-20 14:31 -0500
    Re: DS1054 LF common mode aberation legg <legg@nospam.magma.ca> - 2017-03-19 22:31 -0400
      Re: DS1054 LF common mode aberation Phil Allison <pallison49@gmail.com> - 2017-03-19 20:11 -0700
      Re: DS1054 LF common mode aberation Jon Elson <jmelson@wustl.edu> - 2017-03-20 14:33 -0500
    Re: DS1054 LF common mode aberation legg <legg@nospam.magma.ca> - 2017-03-20 11:58 -0400
      Re: DS1054 LF common mode aberation George Herold <gherold@teachspin.com> - 2017-03-20 09:12 -0700
        Re: DS1054 LF common mode aberation legg <legg@nospam.magma.ca> - 2017-03-20 21:30 -0400
          Re: DS1054 LF common mode aberation Phil Allison <pallison49@gmail.com> - 2017-03-20 18:41 -0700
            Re: DS1054 LF common mode aberation legg <legg@nospam.magma.ca> - 2017-03-21 01:43 -0400
              Re: DS1054 LF common mode aberation Jon Elson <jmelson@wustl.edu> - 2017-03-21 17:04 -0500
            Re: DS1054 LF common mode aberation Jon Elson <jmelson@wustl.edu> - 2017-03-21 17:02 -0500
    Re: DS1054 LF common mode aberation Jon Elson <jmelson@wustl.edu> - 2017-03-20 14:28 -0500
      Re: DS1054 LF common mode aberation legg <legg@nospam.magma.ca> - 2017-03-20 21:36 -0400

#452617 — DS1054 LF common mode aberation

Looking at Low-Frequency noise on a 60Hz power supply, using a Rigol
DS1054 and not getting very far.

One puzzling issue is, when ch3 or ch4 scope probe grounds are
connected to a common 0V reference point, a low frequency common mode
signal shows up in the shorted probe of CH1 or Ch2, when connected to
the same 0V reference. Disconnecting CH3/CH4 ground clips removes the
signal in shorted CH1/CH2. 

Measurement using a current probe show >20mAppk of LF current passing
in the CH1 or CH2 probe lead, generating a >20mvppk voltage on the
display for that particular channel, but only when CH3 or CH4 ground
leads are connected.

The scope probes are those supplied with the retail unit.

It's perfectly possible that this unit has ground leakage issues, but
the safety ground lead on both DUT and scope are secure and share a
common return point. The 0V reference has a parallel RC connection to
the frame, so a small scope probe ground current is anticipated.

The 0V reference effect is only present when the largish 60Hz power
unit is on. The shape of the voltages and currents look ~capacitive,
with displacement coincident with line zero-crossing, where most
low-frequency noise in the DC regulator outputs is anticipated, and so
cannot be isolated or ignored.

Is there some issue with DS1054 ground impedance imbalance between
channels that can be corrected?

Is there some issue with basic LF CMRR of the DS1054 inputs that needs
looking into?

I have other scopes to use for this work, but expected the DS1054 to
be useful in simple low frequency measurement.

RL

[toc] | [next] | [standalone]

#452641

On Sun, 19 Mar 2017 17:04:48 -0400, legg wrote:

> Looking at Low-Frequency noise on a 60Hz power supply, using a Rigol
> DS1054 and not getting very far.
> 
> One puzzling issue is, when ch3 or ch4 scope probe grounds are connected
> to a common 0V reference point, a low frequency common mode signal shows
> up in the shorted probe of CH1 or Ch2, when connected to the same 0V
> reference. Disconnecting CH3/CH4 ground clips removes the signal in
> shorted CH1/CH2.
> 
> Measurement using a current probe show >20mAppk of LF current passing in
> the CH1 or CH2 probe lead, generating a >20mvppk voltage on the display
> for that particular channel, but only when CH3 or CH4 ground leads are
> connected.
> 
> The scope probes are those supplied with the retail unit.
> 
> It's perfectly possible that this unit has ground leakage issues, but
> the safety ground lead on both DUT and scope are secure and share a
> common return point. The 0V reference has a parallel RC connection to
> the frame, so a small scope probe ground current is anticipated.
> 
> The 0V reference effect is only present when the largish 60Hz power unit
> is on. The shape of the voltages and currents look ~capacitive, with
> displacement coincident with line zero-crossing, where most
> low-frequency noise in the DC regulator outputs is anticipated, and so
> cannot be isolated or ignored.
> 
> Is there some issue with DS1054 ground impedance imbalance between
> channels that can be corrected?
> 
> Is there some issue with basic LF CMRR of the DS1054 inputs that needs
> looking into?
> 
> I have other scopes to use for this work, but expected the DS1054 to be
> useful in simple low frequency measurement.
> 
> RL

If the signal of interest is at low frequencies, then by all means get rid
of the extra ground connections.  They're only making matters worse.

Is the origin of the problem in the 'scope?  You should be able to check 
that out without your power supply, testing for proper gain, frequency 
response, input impedance, and leakage currents.  If you're using say, Ch3
and Ch4 as cheapo differential input (via subtraction) you should test its
CMRR too.  Once you're sure your test equipment are working properly you
can measure your power supply properties knowing its limitations.  Until 
then us outside non-observers can only make guesses.

	-F

[toc] | [prev] | [next] | [standalone]

#452666

On Sun, 19 Mar 2017 23:56:56 -0000 (UTC), Frank Miles
<fpm@u.washington.edu> wrote:

>On Sun, 19 Mar 2017 17:04:48 -0400, legg wrote:
>
>> Looking at Low-Frequency noise on a 60Hz power supply, using a Rigol
>> DS1054 and not getting very far.
>> 
>> One puzzling issue is, when ch3 or ch4 scope probe grounds are connected
>> to a common 0V reference point, a low frequency common mode signal shows
>> up in the shorted probe of CH1 or Ch2, when connected to the same 0V
>> reference. Disconnecting CH3/CH4 ground clips removes the signal in
>> shorted CH1/CH2.
>> 
>> Measurement using a current probe show >20mAppk of LF current passing in
>> the CH1 or CH2 probe lead, generating a >20mvppk voltage on the display
>> for that particular channel, but only when CH3 or CH4 ground leads are
>> connected.
>> 
>> The scope probes are those supplied with the retail unit.
>> 
>> It's perfectly possible that this unit has ground leakage issues, but
>> the safety ground lead on both DUT and scope are secure and share a
>> common return point. The 0V reference has a parallel RC connection to
>> the frame, so a small scope probe ground current is anticipated.
>> 
>> The 0V reference effect is only present when the largish 60Hz power unit
>> is on. The shape of the voltages and currents look ~capacitive, with
>> displacement coincident with line zero-crossing, where most
>> low-frequency noise in the DC regulator outputs is anticipated, and so
>> cannot be isolated or ignored.
>> 
>> Is there some issue with DS1054 ground impedance imbalance between
>> channels that can be corrected?
>> 
>> Is there some issue with basic LF CMRR of the DS1054 inputs that needs
>> looking into?
>> 
>> I have other scopes to use for this work, but expected the DS1054 to be
>> useful in simple low frequency measurement.
>> 
>> RL
>
>If the signal of interest is at low frequencies, then by all means get rid
>of the extra ground connections.  They're only making matters worse.
>
>Is the origin of the problem in the 'scope?

This is the question. None of the other scopes here show the problem,
singly or in varying combinations with the scope in question (all
others are 2ch scopes).

I'm looking for a correctible fault in construction.  

>You should be able to check 
>that out without your power supply, testing for proper gain, frequency 
>response, input impedance, and leakage currents.  If you're using say, Ch3
>and Ch4 as cheapo differential input (via subtraction) you should test its
>CMRR too.

I'm not using them for anything complicated - not even their simple
vertical deflection, while their probes are shorted. 

The interaction seems to be between either of CH1/CH2 and either of
CH3/CH4 in a manner suggesting that the channel groups do not share
the same ground return, or safety earth connection.

The measured current flow in the scope probe cables indicates that the
current flow is out of ch3/ch4 and into ch1/ch2, when the DUT power
supply is on - AND GET THIS - connection to the 0V reference by the
scope's probes or ground leads is not required.

The shorted scope probes do not have to have any galvanic connection
to the DUT, except their input power L and N - use of a cheater cord
to lift the scopes ground wire has no effect.

All units are plugged into the same input line phase, without obvious
L/N substitution.

On second inspection and paying attention to the current probe
amplifier scales, the current flowing between the scope input channel
pairs is actually 140mApkpk, easily modulated by the quality of short
existing between the scope probe ground lead contacts.

RL

[toc] | [prev] | [next] | [standalone]

#452733

On Sun, 19 Mar 2017 22:21:20 -0400, legg wrote:

> On Sun, 19 Mar 2017 23:56:56 -0000 (UTC), Frank Miles
> <fpm@u.washington.edu> wrote:
> 
>>On Sun, 19 Mar 2017 17:04:48 -0400, legg wrote:
>>
>>> Looking at Low-Frequency noise on a 60Hz power supply, using a Rigol
>>> DS1054 and not getting very far.
>>> 
>>> One puzzling issue is, when ch3 or ch4 scope probe grounds are
>>> connected to a common 0V reference point, a low frequency common mode
>>> signal shows up in the shorted probe of CH1 or Ch2, when connected to
>>> the same 0V reference. Disconnecting CH3/CH4 ground clips removes the
>>> signal in shorted CH1/CH2.
>>> 
>>> Measurement using a current probe show >20mAppk of LF current passing
>>> in the CH1 or CH2 probe lead, generating a >20mvppk voltage on the
>>> display for that particular channel, but only when CH3 or CH4 ground
>>> leads are connected.
>>> 
>>> The scope probes are those supplied with the retail unit.
>>> 
>>> It's perfectly possible that this unit has ground leakage issues, but
>>> the safety ground lead on both DUT and scope are secure and share a
>>> common return point. The 0V reference has a parallel RC connection to
>>> the frame, so a small scope probe ground current is anticipated.
>>> 
>>> The 0V reference effect is only present when the largish 60Hz power
>>> unit is on. The shape of the voltages and currents look ~capacitive,
>>> with displacement coincident with line zero-crossing, where most
>>> low-frequency noise in the DC regulator outputs is anticipated, and so
>>> cannot be isolated or ignored.
>>> 
>>> Is there some issue with DS1054 ground impedance imbalance between
>>> channels that can be corrected?
>>> 
>>> Is there some issue with basic LF CMRR of the DS1054 inputs that needs
>>> looking into?
>>> 
>>> I have other scopes to use for this work, but expected the DS1054 to
>>> be useful in simple low frequency measurement.
>>> 
>>> RL
>>
>>If the signal of interest is at low frequencies, then by all means get
>>rid of the extra ground connections.  They're only making matters worse.
>>
>>Is the origin of the problem in the 'scope?
> 
> This is the question. None of the other scopes here show the problem,
> singly or in varying combinations with the scope in question (all others
> are 2ch scopes).
> 
> I'm looking for a correctible fault in construction.
> 
>>You should be able to check that out without your power supply, testing
>>for proper gain, frequency response, input impedance, and leakage
>>currents.  If you're using say, Ch3 and Ch4 as cheapo differential input
>>(via subtraction) you should test its CMRR too.
> 
> I'm not using them for anything complicated - not even their simple
> vertical deflection, while their probes are shorted.
> 
> The interaction seems to be between either of CH1/CH2 and either of
> CH3/CH4 in a manner suggesting that the channel groups do not share the
> same ground return, or safety earth connection.
> 
> The measured current flow in the scope probe cables indicates that the
> current flow is out of ch3/ch4 and into ch1/ch2, when the DUT power
> supply is on - AND GET THIS - connection to the 0V reference by the
> scope's probes or ground leads is not required.
> 
> The shorted scope probes do not have to have any galvanic connection to
> the DUT, except their input power L and N - use of a cheater cord to
> lift the scopes ground wire has no effect.
> 
> All units are plugged into the same input line phase, without obvious
> L/N substitution.
> 
> On second inspection and paying attention to the current probe amplifier
> scales, the current flowing between the scope input channel pairs is
> actually 140mApkpk, easily modulated by the quality of short existing
> between the scope probe ground lead contacts.
> 
> RL

Tektronix scopes (most of them, specifically excluding those with isolated
channels) have their input BNC outer shells solidly connected to common/
sheet metal.  Is this not true with your Rigol? If you measure the 
resistance between these channel grounds, what do you get, both at 'scope
inputs and scope probe ground clips?

[toc] | [prev] | [next] | [standalone]

#452770

Frank Miles wrote:


> Tektronix scopes (most of them, specifically excluding those with isolated
> channels) have their input BNC outer shells solidly connected to common/
> sheet metal.

Oh, that USED to be true, in the "good old days".  Now, they use PC boards, 
and it leads to lots of ground currents flowing in one BNC shell and out on 
all the others.  You will see this on pretty much all the plastic shell Tek 
scopes, DPO, etc.

Jon

[toc] | [prev] | [next] | [standalone]

#452670

On Sun, 19 Mar 2017 17:04:48 -0400, legg <legg@nospam.magma.ca> wrote:


The scope probe pairs are normally used as bundled pairs to cut down
on desktop clutter.

It appears that the current is being induced in the physical loop,
created by the separation of the bundled pairs, by a strong local
magnetic flux field being generated by the DUT power supply. 

The power supply employes two large low frequency ferroresonant power
transformers inside a skeleton frame.

The current amplitude can be modulated/inverted by manipulation of the
loop's area and orientation.

Methods of reducing these effects (the transformers stay) in
measurement and basic circuit function are being examined.

RL 

[toc] | [prev] | [next] | [standalone]

#452672

legg wrote:
> 
> The scope probe pairs are normally used as bundled pairs to cut down
> on desktop clutter.
> 
> It appears that the current is being induced in the physical loop,
> created by the separation of the bundled pairs, by a strong local
> magnetic flux field being generated by the DUT power supply. 
> 
> The power supply employes two large low frequency ferroresonant power
> transformers inside a skeleton frame.
> 
> The current amplitude can be modulated/inverted by manipulation of the
> loop's area and orientation.
> 
> Methods of reducing these effects (the transformers stay) in
> measurement and basic circuit function are being examined.
> 
> RL


** And you did *not* get that exact explanation from me about two hours ago on SER??

FYI:

legg crossposted the same query to " sci.electronics.repair". 



....  Phil 

[toc] | [prev] | [next] | [standalone]

#452772

legg wrote:

> On Sun, 19 Mar 2017 17:04:48 -0400, legg <legg@nospam.magma.ca> wrote:
> 
> 
> The scope probe pairs are normally used as bundled pairs to cut down
> on desktop clutter.
> 
> It appears that the current is being induced in the physical loop,
> created by the separation of the bundled pairs, by a strong local
> magnetic flux field being generated by the DUT power supply.
> 
> The power supply employes two large low frequency ferroresonant power
> transformers inside a skeleton frame.
OK, this is a totally different problem, and ferroresonant transforers are 
known to have a LOT of fluxc leakage.

Just keep your probe pairs bundled and as far away from the source of the 
magnetic field as you can.

Jon

[toc] | [prev] | [next] | [standalone]

#452725

On Sun, 19 Mar 2017 17:04:48 -0400, legg <legg@nospam.magma.ca> wrote:

Flux banding the transformers, though it looks a bit credulous with
parts this size, reduced current in the unaltered probes by an order
of magnitude.

RL

[toc] | [prev] | [next] | [standalone]

#452729

On Monday, March 20, 2017 at 11:58:15 AM UTC-4, legg wrote:
> On Sun, 19 Mar 2017 17:04:48 -0400, legg <legg@nospam.magma.ca> wrote:
> 
> Flux banding the transformers, though it looks a bit credulous with
> parts this size, reduced current in the unaltered probes by an order
> of magnitude.
> 
> RL

Flux banding?  
http://www.encyclopedia-magnetica.com/doku.php/flux_band

You put a layer of conductor around the transformer?  Does it need to 
be grounded?  At 60 Hz isn't that a thick piece of copper?  
Or are you doing something else?  

I've sometimes had issues with 60 Hz getting into ground/power lines..

George H. 

[toc] | [prev] | [next] | [standalone]

#452828

On Mon, 20 Mar 2017 09:12:13 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

>On Monday, March 20, 2017 at 11:58:15 AM UTC-4, legg wrote:
>> On Sun, 19 Mar 2017 17:04:48 -0400, legg <legg@nospam.magma.ca> wrote:
>> 
>> Flux banding the transformers, though it looks a bit credulous with
>> parts this size, reduced current in the unaltered probes by an order
>> of magnitude.
>> 
>> RL
>
>Flux banding?  
>http://www.encyclopedia-magnetica.com/doku.php/flux_band
>
>You put a layer of conductor around the transformer?  Does it need to 
>be grounded?  At 60 Hz isn't that a thick piece of copper?  
>Or are you doing something else?  
>
>I've sometimes had issues with 60 Hz getting into ground/power lines..
>
>George H. 

I used 20mil thick copper strap, cut to the width of the winding
breadth.

RL

[toc] | [prev] | [next] | [standalone]

#452833

legg wrote:
 
> 
> I used 20mil thick copper strap, cut to the width of the winding
> breadth.
> 

** That will reduce the field by 6dB or so. 

 Do not do what someone I know did and put the copper strap around the winding. 

 The tranny got rather hot after that. 


....  Phil 

[toc] | [prev] | [next] | [standalone]

#452847

On Mon, 20 Mar 2017 18:41:48 -0700 (PDT), Phil Allison
<pallison49@gmail.com> wrote:

>legg wrote:
> 
>> 
>> I used 20mil thick copper strap, cut to the width of the winding
>> breadth.
>> 
>
>** That will reduce the field by 6dB or so. 
>
> Do not do what someone I know did and put the copper strap around the winding. 
>
> The tranny got rather hot after that. 
>
>
>....  Phil 

What do you mean by 'around the winding'. A flux band circles both
core and winding, on the same axis as the windings.

RL

[toc] | [prev] | [next] | [standalone]

#452954

legg wrote:

> On Mon, 20 Mar 2017 18:41:48 -0700 (PDT), Phil Allison
> <pallison49@gmail.com> wrote:
> 
>>legg wrote:
>> 
>>> 
>>> I used 20mil thick copper strap, cut to the width of the winding
>>> breadth.
>>> 
>>
>>** That will reduce the field by 6dB or so.
>>
>> Do not do what someone I know did and put the copper strap around the
>> winding.
>>
>> The tranny got rather hot after that.
>>
>>
>>....  Phil
> 
> What do you mean by 'around the winding'. A flux band circles both
> core and winding, on the same axis as the windings.
> 
> RL
Yes, but it is only linked to the LEAKAGE flux, which you want to get rid of 
(especially in audio and instrument applications), rather than the main flux 
from the primary.

The situation Phil mentions, the person wrapped the copper strip THROUGH the 
window of the core!  That makes a single, low resistance, shorted turn.  
BZZZZT!

Jon

[toc] | [prev] | [next] | [standalone]

#452953

Phil Allison wrote:

> legg wrote:
>  
>> 
>> I used 20mil thick copper strap, cut to the width of the winding
>> breadth.
>> 
> 
> ** That will reduce the field by 6dB or so.
> 
>  Do not do what someone I know did and put the copper strap around the
>  winding.
> 
>  The tranny got rather hot after that.
Oh, yeah, a shorted turn.  if anything but a tiny transformer, there will be 
a blown fuse!  You only want to short out the leakage flux.

Jon

[toc] | [prev] | [next] | [standalone]

#452767

legg wrote:

> Looking at Low-Frequency noise on a 60Hz power supply, using a Rigol
> DS1054 and not getting very far.
> 
> One puzzling issue is, when ch3 or ch4 scope probe grounds are
> connected to a common 0V reference point, a low frequency common mode
> signal shows up in the shorted probe of CH1 or Ch2, when connected to
> the same 0V reference. Disconnecting CH3/CH4 ground clips removes the
> signal in shorted CH1/CH2.
> 
A lot of commodity scopes these days have pretty crummy tolerance to ground 
current.  The way to see theis is to connect the probe's ground clip ONLY to 
some other equipment, such that it might be getting 10's of mA of ground 
loop current.  Turn the attenuator down to some 10's of mV/div and see how 
much signal you get.  Ought to be WAYYY down there around 1 mV, but no, on a 
lot of recent gear, you will see 10's of mV at least.  Instead of the 1/8" 
thick aluminum housings of the old Tek gear, now the BNC shells are grounded 
through PC boards to some other circuitry inside the scope.

Your results above sound EXACTLY like what I'm talking about.

The only hope is to tie big bonding straps from the unit under test to the 
scope, so the ground difference is reduced as much as possible.

Jon

[toc] | [prev] | [next] | [standalone]

#452831

On Mon, 20 Mar 2017 14:28:48 -0500, Jon Elson <jmelson@wustl.edu>
wrote:

>legg wrote:
>
>> Looking at Low-Frequency noise on a 60Hz power supply, using a Rigol
>> DS1054 and not getting very far.
>> 
>> One puzzling issue is, when ch3 or ch4 scope probe grounds are
>> connected to a common 0V reference point, a low frequency common mode
>> signal shows up in the shorted probe of CH1 or Ch2, when connected to
>> the same 0V reference. Disconnecting CH3/CH4 ground clips removes the
>> signal in shorted CH1/CH2.
>> 
>A lot of commodity scopes these days have pretty crummy tolerance to ground 
>current.  The way to see theis is to connect the probe's ground clip ONLY to 
>some other equipment, such that it might be getting 10's of mA of ground 
>loop current.  Turn the attenuator down to some 10's of mV/div and see how 
>much signal you get.  Ought to be WAYYY down there around 1 mV, but no, on a 
>lot of recent gear, you will see 10's of mV at least.  Instead of the 1/8" 
>thick aluminum housings of the old Tek gear, now the BNC shells are grounded 
>through PC boards to some other circuitry inside the scope.
>
>Your results above sound EXACTLY like what I'm talking about.
>
>The only hope is to tie big bonding straps from the unit under test to the 
>scope, so the ground difference is reduced as much as possible.
>
>Jon

Making the external connection between scope ground lug and DUT frame
was an iteration that produced no changes in the readings, at the
original levels of induced probe screen current..

In later measurements, it does result in differences that are still
overshadowed by other influences, in other measurements. Scope probe
lead location is still something that is intentionally misarranged to
separate or identify sources of scope readings of noise levels.

RL

[toc] | [prev] | [standalone]

Back to top | Article view | sci.electronics.design

csiph-web