Probing SMPS with oscilloscope

[tony359]

Hi all

Surely a silly question but better safe than sorry.
I am trying to fix a SMPS and on the primary I have 420V p/p which is normal. My scope says "400V p/p max" on inputs but I am using a 10:1 probe, rated 500Vp/p.

Am I right in thinking that with a 10:1 probe the actual voltage reaching the scope is 42V and hence I am perfectly safe to use it that way?

Thanks!

[tony359]

Thanks,

I suspected that - even though my scope is indeed 400V p/p input so while I should be safe with a 10:1 probe, it would be dangerous and borderline.

[PKTKS]

(..) Where do you hook your ground clip is an interesting question, and there is not an answer to that unless you have a differential probe.

interesting comments but nevertheless - NEVER cross sides.
A common mistake - taken sometimes by distraction - is to
hook your clip on the secondary "ground".. and go probing
the primary...  bad things happen

You can get around by using isolated DUT or isolated scope, the former can electrocute you without triggering GFCI, and the latter, while is protected by GFCI, greatly increase your chance of getting zapped, and GFCI may fail from time to time.

The only proper way is to pay for a proper HV differential probe set.

ALL MY BENCH IS ISOLATED.

DUTs by dedicated trafo and scopes by another set of trafos and so forth

USE GLOVES.  - I never check mains powered without them.

It is not about getting zapped...  the panic of touching a 400V
hot spot can do a lot of bad things... even without killing you,
it is a fast reaction hard to control

USE GLOVES... USE GLOVES... USE GLOVES... USE GLOVES...

Paul

[Per Hansson]

Thanks,

I suspected that - even though my scope is indeed 400V p/p input so while I should be safe with a 10:1 probe, it would be dangerous and borderline.

It would not be borderline: the second you put the negative of the probe to B- the probe will vaporize in your hand.

EEVblog #279 - How NOT To Blow Up Your Oscilloscope:

[bdunham7]

You can avoid killing yourself for now by getting a 100X HV probe set, like this:

https://www.ebay.com/itm/P2301C-High-Voltage-Oscilloscope-Passive-Probe-5kV-100X-300MHz-For-Tektronix-HP/332835001398?hash=item4d7e831436:g:2vcAAOSwEIhZg-yw

AND removing the ground lead and probing where you like with just the tip.

If you can get by only measuring points that are ground-referenced, than you're good.  Anything with bridge input, for example, will not typically not have DC- as ground referenced. 

To measure non-ground-referenced voltages--IOW to use a scope like you would a DMM--you need either a HV differential probe, an isolation interface or an isolated scope.  If you already have the scope, the cheapest solution that won't shock you is likely the MicSig line of HV differential probes.

[nctnico]

Get a differential probe. Very useful. Not just for measuring mains but there are cases where being able to do a floating measurement is very handy.

[james_s]

The only safe way to do this is with a HV differential probe. I picked one up last year and now wonder how I ever got by without it. Poking around with a scope on the primary side of a SMPS is a classic way of destroying your scope and the PSU you're trying to fix.

[David Hess]

I am trying to fix a SMPS and on the primary I have 420V p/p which is normal. My scope says "400V p/p max" on inputs but I am using a 10:1 probe, rated 500Vp/p.

Am I right in thinking that with a 10:1 probe the actual voltage reaching the scope is 42V and hence I am perfectly safe to use it that way?

There are two different limits.

The operating range which produces a reasonable display is simply based on the probe attenuation factor and the oscilloscope sensitivity.  So a 10x probe at 5 volts per division, which is typical, yields 50 volts per division.  8 vertical divisions then are 400 volts peak-to-peak.  When the position control is taken into account, that means that voltages up to about + or - 400 volts can be displayed hich is generally sufficient for line side circuits.

The 400 volt peak input rating has nothing to do with the above.  Instead, it is related to the input protection, switching, and coupling circuits.

Where you can get into trouble is that while a 10x probe attenuates the input signal to one tenth, it only does this for DC when the input coupling is set to DC.  When the input coupling is set to AC, then the full DC input voltage is passed to the input of the oscilloscope as the AC coupling capacitor charges, so the maximum input voltage remains 400 volts.  The same applies to most 100x probes.

For safety, you really should be using a high voltage differential probe but with care, it is usually possible to use 10x probes on line side circuits, but be very careful about where you attach the probe's ground lead, or just remove it for safety.  I use an isolation transformer on the device under test whether I use a high voltage differential probe or not.

[bdunham7]

Where you can get into trouble is that while a 10x probe attenuates the input signal to one tenth, it only does this for DC when the input coupling is set to DC.  When the input coupling is set to AC, then the full DC input voltage is passed to the input of the oscilloscope as the AC coupling capacitor charges, so the maximum input voltage remains 400 volts.  The same applies to most 100x probes.

Just FWIW, I'm sure my legacy scopes all behave this way, but I did just check the input resistance (by DMM) on my Tek TPS2024 (isolated) and Siglent 1104X-E.  The Tek does read 1M on DC-coupled and open circuit on the AC and GND settings, but the Siglent has a steady 1M input resistance regardless of the coupling setting.  Hmmmmmm.  Cheap scopes save the day?  I wonder if the very robust Tek has additional input protection against the scenario you outlined.

[tggzzz]

(..) Where do you hook your ground clip is an interesting question, and there is not an answer to that unless you have a differential probe.

interesting comments but nevertheless - NEVER cross sides.
A common mistake - taken sometimes by distraction - is to
hook your clip on the secondary "ground".. and go probing
the primary...  bad things happen

You can get around by using isolated DUT or isolated scope, the former can electrocute you without triggering GFCI, and the latter, while is protected by GFCI, greatly increase your chance of getting zapped, and GFCI may fail from time to time.

The only proper way is to pay for a proper HV differential probe set.

ALL MY BENCH IS ISOLATED.

DUTs by dedicated trafo and scopes by another set of trafos and so forth

USE GLOVES.  - I never check mains powered without them.

It is not about getting zapped...  the panic of touching a 400V
hot spot can do a lot of bad things... even without killing you,
it is a fast reaction hard to control

USE GLOVES... USE GLOVES... USE GLOVES... USE GLOVES...

Paul

1) isolation solves some problems and introduces others. It is neither necessary nor sufficient.

2) you had better explains what you mean by trafo, and its relevance; I have no idea.

3) mains isn't hard to control; it is impossible to consciously override muscle contractions. I know, from personal experience.

 Nowadays the only sane way to do this is with an isolated HV probe. End of story.

[PKTKS]

1) isolation solves some problems and introduces others. It is neither necessary nor sufficient.

TRUE. However one of them most hard to overcome problem
is to actually hook your scope chassi to the mains or other
bad reference.

Having instruments and DUTs isolated solves this particular bad
problem. You will be isolated from the mains which ultimately
is the primary cause of scope dead shorts.

2) you had better explains what you mean by trafo, and its relevance; I have no idea.

ops.. that one is a more or less ancient term (acronym) for TRAnsFOrmer
(TRAF was also found on some ancient schemas)

that term may be nostalgia for some HAMs

3) mains isn't hard to control; it is impossible to consciously override muscle contractions. I know, from personal experience.

 Nowadays the only sane way to do this is with an isolated HV probe. End of story.

Despite the fact that those DIFFERENTIAL HV probes are very handy...

They are NOT ISOLATED. Their input is mostly a cascade resistor
ladder with very high impedance... but not isolated.

Add that the required power needed by these probes and
sometimes you have introduced another factor to consider

Having DUTs and INSTRUMENTS "isolated de facto" (like having
battery scopes and DMMs)  is very handy most of the time.

For those critical things where I WOULD NOT RISK EXPENSIVE
GIZMOS..  i got myself one of those handy battery operated
hand scope.

Combined with 100x probe operated on battery only
they are a peace of mind.. in case o faulty devices doing
some harm they are cheaper than expensive bench gizmos.

You see..  repair and R&D are quite 2 different things.

Expensive stuff to repair garbage bricked stuff just do not match

Paul

[CDaniel]

Using an isolation transformer for the supply under test is necesary and there are no problems . However measuring the primary side of the SMPS with the bench scope should be done only when absolutly necesary , not jumping to measure if the mains voltage is coming in , the filter cap has voltage or too much ripple or if the power switches are working  For troubleshooting in 99% cases the multimeter is enough , I'm not referring at designing a SMPS and so on

[tony359]

Thank you all for the many replies - I really appreciate the time you spent to warn me!

Well, before I posted here I did actually probe the primary with the ground connected to chassis ground. Nothing exploded and the scope (Siglent 1104X-E) has survived but I won't do that again!

The reason I was poking around is to try and identify where some ripple (more of a spike) was coming from - it's present on all 5 rails and I could see it before and after the switching transformer too - see pics attached.

[AVGresponding]

You could use HV differential probes, or... you could just use a scopemeter.

[nctnico]

Despite the fact that those DIFFERENTIAL HV probes are very handy...

They are NOT ISOLATED. Their input is mostly a cascade resistor
ladder with very high impedance... but not isolated.

For all intents and purposes these probes are isolated. Their input resistance is extremely high so no dangerous current can flow and, more importantly, they are safety rated.

[bdunham7]

For all intents and purposes these probes are isolated. Their input resistance is extremely high so no dangerous current can flow and, more importantly, they are safety rated.

You should specify 'extremely high'. It is actually a lot less than most single-ended HV probes.  One common type of these is 4M each side to ground.  So measuring a small signal at 1400V common mode voltage, you would have a net resistance of 2M to ground, with a resultant leakage current of 700uA.  That's not even getting into what the current might be at the highest frequencies it is rated at.   In many cases this is not problematic, but it is still there and has to be considered in some situations. 

[David Hess]

Where you can get into trouble is that while a 10x probe attenuates the input signal to one tenth, it only does this for DC when the input coupling is set to DC.  When the input coupling is set to AC, then the full DC input voltage is passed to the input of the oscilloscope as the AC coupling capacitor charges, so the maximum input voltage remains 400 volts.  The same applies to most 100x probes.

Just FWIW, I'm sure my legacy scopes all behave this way, but I did just check the input resistance (by DMM) on my Tek TPS2024 (isolated) and Siglent 1104X-E.  The Tek does read 1M on DC-coupled and open circuit on the AC and GND settings, but the Siglent has a steady 1M input resistance regardless of the coupling setting.  Hmmmmmm.  Cheap scopes save the day?  I wonder if the very robust Tek has additional input protection against the scenario you outlined.

There is a clever alternative "two path" input buffer design which newer oscilloscopes may use which does not require the input coupling capacitor, but it is not something to count on for safety.

[MathWizard]

You can avoid killing yourself for now by getting a 100X HV probe set, like this:

https://www.ebay.com/itm/P2301C-High-Voltage-Oscilloscope-Passive-Probe-5kV-100X-300MHz-For-Tektronix-HP/332835001398?hash=item4d7e831436:g:2vcAAOSwEIhZg-yw

AND removing the ground lead and probing where you like with just the tip.

If you can get by only measuring points that are ground-referenced, than you're good.  Anything with bridge input, for example, will not typically not have DC- as ground referenced. 

To measure non-ground-referenced voltages--IOW to use a scope like you would a DMM--you need either a HV differential probe, an isolation interface or an isolated scope.  If you already have the scope, the cheapest solution that won't shock you is likely the MicSig line of HV differential probes.

Isn't the Neg probe of Diff-probes also a few Megs from MainsEG, like the pos. probe?

Isn't the whole point of diff-probes is so that you could (if transients aren't a concern) put them anywhere (both on hot/clod side tho)

[dustooff]

After looking at a few HV diff probes available to me, some state the resistance to ground (4M), and they would be balanced to maintain the differential input.

Having a new scope that can also do multichannel power measurements, I'm looking at HV diff probe adaptor/kit.  In the past any measurement that was more than about 60v was either with the Fluke 87 or PM97, both have served well and still provide relevant results.
Looking at the various available HV diff kit, in no particular order,
https://emona.com.au/products/electronic-test-measure/oscilloscopes/gdp-050.html
https://www.eevblog.com/product/hvp70/
http://www.micsig.com/html/41.html
http://hantek.com.cn/products/detail/6163

I think I will go with the micsig DP10013.
They all seem to have similar CAT ratings, though the micsig prominently displays the MAX rating not +/-700 or  +/-600   DC+AC pk.
The HVP70 does have a CAT III 1000v rating where the others are CAT II 700v or 1000v. I've been thinking about drawing a compatibility chart of which accessories can be connected with devices to maintain CAT rating. ( for all CAT rating and examples)

I do find it interesting that Hantek have added the word Isolated to their description of HV diff probe kit.  This seems a bit misleading / unjustified.

[PKTKS]

(..)
I do find it interesting that Hantek have added the word Isolated to their description of HV diff probe kit.  This seems a bit misleading / unjustified.

Like many others.. there is an assumption that a
HIGH IMPEDANCE is equivalent to ISOLATION.

Obviously there is no GALVANIC ISOLATION.

In the unfortunate event of a cascade failure
the whole surge may enter the diff amp
frying  whatever is in the way

GALVANIC ISOLATION is somewhat in limbo
once the heavy bulky transformers are hard
to deal with and hard to build.

Nevertheless I have never seen a primary fault
entering secondary or vice-versa

Nothing compares the isolation core itself.
There is real ISOLATION in the strict sense

Paul

[nctnico]

(..)
I do find it interesting that Hantek have added the word Isolated to their description of HV diff probe kit.  This seems a bit misleading / unjustified.

Like many others.. there is an assumption that a
HIGH IMPEDANCE is equivalent to ISOLATION.

Obviously there is no GALVANIC ISOLATION.

Just check how galvanic isolation is defined... Several hundreds of micro-amperes is allowed. In they end there is always a possibility for creepage. The fact is that CAT rated differential probes are designed to withstand high voltages. Those probes have a much wider barrier between dangerous voltages and the operator compared to -for example- an opto coupler or windings in a transformer.

[PKTKS]

(..) or windings in a transformer.

When the ENAMEL or the isolation layer on the core fails...

The magnetic flux stops (or drastically tend to zero)
leading the surge to void - no harm in the chain.

Opposite results in the resistor ladder failure.
The dead short surge propagates until the energy dissipates..

Paul

[nctnico]

(..) or windings in a transformer.

When the ENAMEL or the isolation layer on the core fails...

The magnetic flux stops (or drastically tend to zero)
leading the surge to void - no harm in the chain.

The problem is not the enamel isolation from a winding but leakage between the windings! Just look at a teardown of a differential probe. There is not 1 resistor but a whole bunch of them in series. IOW you need a catastrophic failure in multiple parts before there is a dangerous situation. Also when a differential probe starts to fail it won't work right so you are aware of problems immediately.

[PKTKS]

(..) you need a catastrophic failure in multiple parts before there is a dangerous situation.

CORRECT.

Assuming the sub-millimeter dimensions of today SMT
it is easy to high energy surges propagate once the
layout  is cracked - arcs won't even need to go via
smd components they can spread or arc ad hoc.

Old school HV probes are LONG and sharp to prevent
those catastrophic surges to arc ad hoc..

those small things are not. tight and too close
to prevent - in the event of a surge failure - random arcs

Paul

[nctnico]

(..) you need a catastrophic failure in multiple parts before there is a dangerous situation.

CORRECT.

Assuming the sub-millimeter dimensions of today SMT
it is easy to high energy surges propagate once the

Please take a look at how these probes are constructed before making any other comments; the parts they use are huge! There is nothing sub-millimeter about the resistors you find in CAT rated differential probes. CAT rating implies that these designs can handle surges.