Measurement Techniques without Earth Grounding an Offline smps

[D from BC]

I'm thinking about methods to test a non-isolated homebrew 130W
offline smps.
I need to detect RF bursts "birdies", ringing, spikes and overshoots.
An earth grounded scope will blow a fuse.

I'm trying out:
Diode detector circuits with an analog meter.
I suspect wrong readings from my digital DMM due to the HF.
Also, it's only rated to 350Hz without true RMS.
Just touching a pulse node with one probe scrambles the DMM display.
(300Vpk square wave at 100khz.)
I also get a nice hiss from my stereo speakers.

How about...a pick up "sniffer coil" on a scope probe.
If that works??
Just a coil at the end of a scope probe.
EMI is not such a bad thing... It's a test signal.

I have extra isolated windings on the power inductors. That's safe to
earth ground for scope use..

I'll be interested in any nifty poorman test tricks...
D from BC

 

[Gibbo]

I'm thinking about methods to test a non-isolated homebrew 130W
offline smps.
I need to detect RF bursts "birdies", ringing, spikes and overshoots.
An earth grounded scope will blow a fuse.

I'm trying out:
Diode detector circuits with an analog meter.
I suspect wrong readings from my digital DMM due to the HF.
Also, it's only rated to 350Hz without true RMS.
Just touching a pulse node with one probe scrambles the DMM display.
(300Vpk square wave at 100khz.)
I also get a nice hiss from my stereo speakers.

How about...a pick up "sniffer coil" on a scope probe.
If that works??
Just a coil at the end of a scope probe.
EMI is not such a bad thing... It's a test signal.

I have extra isolated windings on the power inductors. That's safe to
earth ground for scope use..

I'll be interested in any nifty poorman test tricks...
D from BC

You have a dual trace scope?

Can it display an "A+B" trace?

Can it invert one trace?

If so then you can display the voltage difference between *any* two
points anywhere in your SMPS

 

[Frithiof Andreas Jensen]

I'm thinking about methods to test a non-isolated homebrew 130W
offline smps.
I need to detect RF bursts "birdies", ringing, spikes and overshoots.
An earth grounded scope will blow a fuse.

It will not if you use an isolation transformer!

Otherwise you are stuck with differential measurements, which suck because you
use two channels on the scope for each trace.

 

[Eeyore]

You have a dual trace scope?

Can it display an "A+B" trace?

Can it invert one trace?

If so then you can display the voltage difference between *any* two
points anywhere in your SMPS

A *real* differential input is better though.

Graham

 

[D from BC]

It will not if you use an isolation transformer!

Otherwise you are stuck with differential measurements, which suck because you
use two channels on the scope for each trace.

I only have a single channel scope.

The challenge is making measurements and test techniques without a
isolation transformer nor an isolated scope to detect RF bursts
"birdies", ringing, spikes and overshoots.

I suspect it's a little bit like making radio tuner/demodulator ccts
and measuring a DC out with a DMM.

D from BC

 

[D from BC]

You have a dual trace scope?

Can it display an "A+B" trace?

Can it invert one trace?

If so then you can display the voltage difference between *any* two
points anywhere in your SMPS

Single trace.
It's the stone age over here
D from BC

 

[Frithiof Andreas Jensen]

I only have a single channel scope.

The challenge is

pointless ... like deciding to run the 100 m sprint with your left hand tied to
your right foot behind your back?

 

[Joerg]

I'm thinking about methods to test a non-isolated homebrew 130W
offline smps.
I need to detect RF bursts "birdies", ringing, spikes and overshoots.
An earth grounded scope will blow a fuse.

I'm trying out:
Diode detector circuits with an analog meter.
I suspect wrong readings from my digital DMM due to the HF.
Also, it's only rated to 350Hz without true RMS.
Just touching a pulse node with one probe scrambles the DMM display.
(300Vpk square wave at 100khz.)
I also get a nice hiss from my stereo speakers.

How about...a pick up "sniffer coil" on a scope probe.
If that works??
Just a coil at the end of a scope probe.
EMI is not such a bad thing... It's a test signal.

That's how I do it. But not at the end of a scope probe. Take a coax but
make sure it's 50ohms and not 75ohms, they can look deceptively similar.
Then switch you scope to 50ohms input or terminate it via a T-connector.
Now wind a coil of 1/2" diameter or less. About 3-4 turns, nice and
tidy, and use wire that has an insulation that isn't easily worn or
scraped off. Solder to shield and center at the end of the coax. Now you
have an H-field sniffer. If ye olde Hickock scope has enough bandwidth,
that is.

The other trick is the E-field smeller: Take the same kind of coax,
strip the shield about 1/4" but do not strip the center conductor.
Insulate the tip well. I use Kapton tape. Anything that won't scrape
off. Make sure the center doesn't poke through (which it easily does
when bending the coax back and forth). Now you have an E-field probe.
Same here, if the scope doesn't have enough BW you may not see a thing
even when the birdie cometh along.

I have extra isolated windings on the power inductors. That's safe to
earth ground for scope use..

I'll be interested in any nifty poorman test tricks...

Well, those are the tricks, pretty much. Since I do this stuff for a
living I abandoned the poor man's method a long time ago and use an EMCO
probe kit. But that would set you back by about $1000. Believe it or not
but I managed to wear out one of its probes already.

 

[Rich Grise]

I'm thinking about methods to test a non-isolated homebrew 130W
offline smps.
I need to detect RF bursts "birdies", ringing, spikes and overshoots.
An earth grounded scope will blow a fuse.

I'm trying out:
Diode detector circuits with an analog meter.
I suspect wrong readings from my digital DMM due to the HF.
Also, it's only rated to 350Hz without true RMS.
Just touching a pulse node with one probe scrambles the DMM display.
(300Vpk square wave at 100khz.)
I also get a nice hiss from my stereo speakers.

How about...a pick up "sniffer coil" on a scope probe.
If that works??
Just a coil at the end of a scope probe.
EMI is not such a bad thing... It's a test signal.

I have extra isolated windings on the power inductors. That's safe to
earth ground for scope use..

I'll be interested in any nifty poorman test tricks...

If you're crazy enough, you could float the scope, but of course, then
the case could be at line voltage - use one of these or equiv:
http://www.radioshack.com/sm-3-prong-grounding-adapter-2-pack--pi-2104010.html

I do not advise this, it is dangerous and can be lethal - it's up
to you to decide if you want to take the chance. In any case, do NOT
touch any metal part of the scope.

Good Luck!
Rich

 

[martin griffith]

I'm thinking about methods to test a non-isolated homebrew 130W
offline smps.
I need to detect RF bursts "birdies", ringing, spikes and overshoots.
An earth grounded scope will blow a fuse.

I'm trying out:
Diode detector circuits with an analog meter.
I suspect wrong readings from my digital DMM due to the HF.
Also, it's only rated to 350Hz without true RMS.
Just touching a pulse node with one probe scrambles the DMM display.
(300Vpk square wave at 100khz.)
I also get a nice hiss from my stereo speakers.

How about...a pick up "sniffer coil" on a scope probe.
If that works??
Just a coil at the end of a scope probe.
EMI is not such a bad thing... It's a test signal.

I have extra isolated windings on the power inductors. That's safe to
earth ground for scope use..

I'll be interested in any nifty poorman test tricks...
D from BC

Definately read AN70 from Linear.com, nice bit about sniffers


martin

 

[D from BC]

On Wed, 21 Feb 2007 16:42:44 GMT, Joerg

[snip]

That's how I do it. But not at the end of a scope probe. Take a coax but
make sure it's 50ohms and not 75ohms, they can look deceptively similar.
Then switch you scope to 50ohms input or terminate it via a T-connector.
Now wind a coil of 1/2" diameter or less. About 3-4 turns, nice and
tidy, and use wire that has an insulation that isn't easily worn or
scraped off. Solder to shield and center at the end of the coax. Now you
have an H-field sniffer. If ye olde Hickock scope has enough bandwidth,
that is.

Cool...
I'm hoping for low freq. Big Bird size birdies to see on that 40yr old
25Mhz BW Hickock scope.
I'm guessing my scope has a 6dB/oct attenuation.
So....mmm.. down 24dB's@ at 400Mhz....4th harmonic.
Those birdies have a f limit somewhere...
Maybe I can alter my PCB construction and make some "cats"

The other trick is the E-field smeller: Take the same kind of coax,
strip the shield about 1/4" but do not strip the center conductor.
Insulate the tip well. I use Kapton tape. Anything that won't scrape
off. Make sure the center doesn't poke through (which it easily does
when bending the coax back and forth). Now you have an E-field probe.
Same here, if the scope doesn't have enough BW you may not see a thing
even when the birdie cometh along.

I might try too..

Well, those are the tricks, pretty much. Since I do this stuff for a
living I abandoned the poor man's method a long time ago and use an EMCO
probe kit. But that would set you back by about $1000. Believe it or not
but I managed to wear out one of its probes already.

I didn't know it was an industrial product...
So I did some googling.. and found some reading material on:
http://www.reed-electronics.com/tmworld/article/CA314013
About near field probes use.

Thanks
D from BC

 

[Joerg]

On Wed, 21 Feb 2007 16:42:44 GMT, Joerg

[snip]

That's how I do it. But not at the end of a scope probe. Take a coax but
make sure it's 50ohms and not 75ohms, they can look deceptively similar.
Then switch you scope to 50ohms input or terminate it via a T-connector.
Now wind a coil of 1/2" diameter or less. About 3-4 turns, nice and
tidy, and use wire that has an insulation that isn't easily worn or
scraped off. Solder to shield and center at the end of the coax. Now you
have an H-field sniffer. If ye olde Hickock scope has enough bandwidth,
that is.

Cool...
I'm hoping for low freq. Big Bird size birdies to see on that 40yr old
25Mhz BW Hickock scope.
I'm guessing my scope has a 6dB/oct attenuation.
So....mmm.. down 24dB's@ at 400Mhz....4th harmonic.
Those birdies have a f limit somewhere...
Maybe I can alter my PCB construction and make some "cats"

With a 25MHz bandwidth I am afraid you may not see them. Unless there is
some serious "slow" ringing going on. Think of humming birds, very
small, very fast and cats usually can't catch them ;-)

I might try too..




I didn't know it was an industrial product...

Yep, even came in a little black suitcase, with calibrations sheets and all.

So I did some googling.. and found some reading material on:
http://www.reed-electronics.com/tmworld/article/CA314013
About near field probes use.

Can't see it. Seems to be one of those members-only sites. You'll be
surprised what you see on the scope when you hold that probe near one of
those laptops power supplies. Makes the toe nails curl.

 

[D from BC]

If you're crazy enough, you could float the scope, but of course, then
the case could be at line voltage - use one of these or equiv:
http://www.radioshack.com/sm-3-prong-grounding-adapter-2-pack--pi-2104010.html

I do not advise this, it is dangerous and can be lethal - it's up
to you to decide if you want to take the chance. In any case, do NOT
touch any metal part of the scope.

Good Luck!
Rich

I've been tempted..
The scope I'm using has an all metal grounded face plate so turning
knobs will be dangerous..
I could do a back away, power up, power off, adj and repeat routine.
Or maybe poke at the knobs with a stick
With no earth ground connection to the scope there's no line power
60Hz loop.. No fuses will blow.
However, should I be concerned with my scopes high frequency isolation
from line and neutral? Just wondering if it will throw off the
measurements when testing the offline smps.
D from BC

 

[D from BC]

Definately read AN70 from Linear.com, nice bit about sniffers


martin

Found it...Page 55 on
http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1040,C1130,P1535,D4159

That's cool... Sleek and compact...
Looks like I could cruise over 30mil traces and not pickup too much
crosstalk from other mag sources..
Not only that...this app note has lots of other goodies too.

In my OP, I called it a "sniffer coil". It's something I probably
heard of some 15years ago. Then forgot all the details but the sniffy
name..

Thanks..
D from BC

 

[Rich Grise]

I've been tempted..
The scope I'm using has an all metal grounded face plate so turning
knobs will be dangerous..
I could do a back away, power up, power off, adj and repeat routine.
Or maybe poke at the knobs with a stick

Sounds like an excellent idea.

With no earth ground connection to the scope there's no line power
60Hz loop.. No fuses will blow.
However, should I be concerned with my scopes high frequency isolation
from line and neutral? Just wondering if it will throw off the
measurements when testing the offline smps.

If your probe has one of those little ground clippies at the end, that
should provide a reference - the freq. response might be affected, and
there could be some 60Hz pickup because the scope case will be capacitive
to whatever's around it, kinda like an antenna, but if it's a decent scope,
it should ignore that and just show what's between the probe tip and its
ground clip.

I've been known to do this, and I've seen others do it, but never while
alone in the room.

And put your other hand in your pocket, seriously. In the USAF, they
_ordered_ us to keep our other hand in our pocket, at least if we were
probing anything over, say, 28V.

Good Luck!
Rich

 

[D from BC]

On Wed, 21 Feb 2007 16:42:44 GMT, Joerg

[snip]

That's how I do it. But not at the end of a scope probe. Take a coax but
make sure it's 50ohms and not 75ohms, they can look deceptively similar.
Then switch you scope to 50ohms input or terminate it via a T-connector.
Now wind a coil of 1/2" diameter or less. About 3-4 turns, nice and
tidy, and use wire that has an insulation that isn't easily worn or
scraped off. Solder to shield and center at the end of the coax. Now you
have an H-field sniffer. If ye olde Hickock scope has enough bandwidth,
that is.

Cool...
I'm hoping for low freq. Big Bird size birdies to see on that 40yr old
25Mhz BW Hickock scope.
I'm guessing my scope has a 6dB/oct attenuation.
So....mmm.. down 24dB's@ at 400Mhz....4th harmonic.
Those birdies have a f limit somewhere...
Maybe I can alter my PCB construction and make some "cats"

With a 25MHz bandwidth I am afraid you may not see them. Unless there is
some serious "slow" ringing going on. Think of humming birds, very
small, very fast and cats usually can't catch them ;-)

What do you think about catching birdies with a tuner (bandpass) and
diode detector cct.?
In other words, this is the poor mans spectrum analyzer.
Is there a module or inexpensive unit that can tune 100khz to say
1Ghz?
I'm doing a google on that now..
[snip]
D from BC

 

[Joerg]

D from BC wrote:

On Wed, 21 Feb 2007 16:42:44 GMT, Joerg

[snip]


That's how I do it. But not at the end of a scope probe. Take a coax but
make sure it's 50ohms and not 75ohms, they can look deceptively similar.
Then switch you scope to 50ohms input or terminate it via a T-connector.
Now wind a coil of 1/2" diameter or less. About 3-4 turns, nice and
tidy, and use wire that has an insulation that isn't easily worn or
scraped off. Solder to shield and center at the end of the coax. Now you
have an H-field sniffer. If ye olde Hickock scope has enough bandwidth,
that is.


Cool...
I'm hoping for low freq. Big Bird size birdies to see on that 40yr old
25Mhz BW Hickock scope.
I'm guessing my scope has a 6dB/oct attenuation.
So....mmm.. down 24dB's@ at 400Mhz....4th harmonic.
Those birdies have a f limit somewhere...
Maybe I can alter my PCB construction and make some "cats"

With a 25MHz bandwidth I am afraid you may not see them. Unless there is
some serious "slow" ringing going on. Think of humming birds, very
small, very fast and cats usually can't catch them ;-)

What do you think about catching birdies with a tuner (bandpass) and
diode detector cct.?
In other words, this is the poor mans spectrum analyzer.

Won't work very well. Birdies are short bursts of oscillations that
happen during the phase where the FET goes through its linear region.
IOW the duty cycle is extremely low. Even expensive spectrum analyzers
have a hard time diagnosing that. I had to do that a lot in pulsed
ultrasound and had to build myself some fancy limiters to get any kind
of information out out the analyzer. So yes, it can work but it's a
whole lot of effort.

Is there a module or inexpensive unit that can tune 100khz to say
1Ghz?

Probably. People have used TV tuners for that. But I never bothered
after finding out the dynamic range of the average TV set. Sometimes I
wonder whether their designers actually knew what a dynamic range is.

 

[martin griffith]

On Wed, 21 Feb 2007 19:59:02 GMT, in sci.electronics.design D from BC

pick up "sniffer coil" on a scope probe.

Found it...Page 55 on
http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1040,C1130,P1535,D4159

That's cool... Sleek and compact...
Looks like I could cruise over 30mil traces and not pickup too much
crosstalk from other mag sources..
Not only that...this app note has lots of other goodies too.

In my OP, I called it a "sniffer coil". It's something I probably
heard of some 15years ago. Then forgot all the details but the sniffy
name..

Thanks..
D from BC

Think you will need some serious insulation on yours. Remember, always
keep one hand your pocket playing with this stuff


martin