Wish to scope the secondary diode voltage on our offline Flyback SMPS

[ocset]

Hi,
We wish to scope the ringing  voltage waveform on the secondary diode of our  Offline  isolated 240VAC  Flyback  SMPS  LED driver.  (The Flyback  works in DCM, but we also want to try it in CCM with lower voltage LED strings.)
The  LED string voltage is around 75V, and the Transformer NP/NS = 26/16.
I usually use one of those scope probes with a “spring probe” on the end, which allows me to avoid using the “dangling ground lead”. However, we  do not have any here, and I can’t seem to find them on Farnell any more.
Do you know where I can get one of these  “spring probes”?
For lower voltage signals  (eg sense resistor waveforms), we just use a piece of coaxial cable and strip one end and solder it to the sense resistor…and connect the BNC at the other end  into the scope. –That gives great results but our coaxial pieces can’t be used  across the Flyback secondary diode, as the voltage is too high.  As you know, the secondary diode voltage can ring up to { 2 x [v(pri)_reflected + vout]}.

[capt bullshot]

I usually use a piece of wire as the ground spring.
One can build quite well working 1:10 or 1:100 "low impedance" probes out of a 450R or 4450R resistor soldered to one end of a coax cable and connect the other end into a 50Ohm terminator or 50Ohm scope input.

[ocset]

Thanks, ill bear that in mind, ...for scopeing scross the secondary diode, with its high frequency ringing, we wish to use a high Z probe, so as not to interfere with the actual waveform.
I think the best way seems to be to buy a 1:100 scope probe and just wrap wire round the ground barrel as Capt Bullshot says?
We are wondering about actually cutting through the coax of the 1:100 probe and soldering it directly  across the diode, so we dont have to fiddle with  wrapping wire round the ground barrel....since we still have the problem of how to connect it across the diode after we have done the "wire wrapping".
Has anyone tried this?

[Phoenix]

I've used thin tinned copper wire instead of the probe spring clips (D=0.5mm ish). You can form it however you want; use it for the ground connection; use it for the tip connection; and it will hold the probe in place.

[capt bullshot]

We are wondering about actually cutting through the coax of the 1:100 probe and soldering it directly  across the diode, so we dont have to fiddle with  wrapping wire round the ground barrel....since we still have the problem of how to connect it across the diode after we have done the "wire wrapping".
Has anyone tried this?

No, haven't tried that, but it won't work anyway. There are substantial components inside the probe body to make it work as a 100:1 probe. If you cut the coax, you'll end up with some lousy kind of a 1:1 probe.

One can buy adaptors from "probe tip" to BNC (sometimes they are supplied with the probe), we've abused these here (@work) sometimes to make an soldered probe connection to a circuit.

[ocset]

No, haven't tried that, but it won't work anyway. There are substantial components inside the probe body to make it work as a 100:1 probe. If you cut the coax, you'll end up with some lousy kind of a 1:1 probe.

Thanks yes, though i was wondering if i could find a probe with the cap and resistor in the BNC side....then i would  cut it at the other end and be OK?

[David Hess]

I usually wrap copper bus bar wire around the probe ring to make an instant ground spring attachment but copper is pretty soft.  If I had to do this enough, I would get some small diameter piano wire and wrap it around a smaller diameter drill bit shaft to make a perfect springy attachment.

[lordvader88]

What is the reason for doing this? Whats wrong with the standard ground clip, is it that the wire acts like an antenna or something? Or the inductance of it or something ?

I'm trying to fix some SMPSs

"That gives great results but our coaxial pieces can’t be used  across the Flyback secondary diode, as the voltage is too high.  As you know, the secondary diode voltage can ring up to { 2 x [v(pri)_reflected + vout]}"

So standard cable TV coax can't handle what voltage ? whats the reflected voltage, I've heard of it, and whats the actual # Its a big beefy copper wire so whats the problem ? What am I missing

2x (240V +75) ?, or maybe 2x((2x240V) +75V )?

[capt bullshot]

No, haven't tried that, but it won't work anyway. There are substantial components inside the probe body to make it work as a 100:1 probe. If you cut the coax, you'll end up with some lousy kind of a 1:1 probe.

Thanks yes, though i was wondering if i could find a probe with the cap and resistor in the BNC side....then i would  cut it at the other end and be OK?

If you find one, maybe.
Afaik, there's no such kind of probe.

[ocset]

@LordVader, Thanks.....the reflected voltage to the sec  is the Vin multiplied by the turns ratio...in this case NS/NP.
The sec diode off state voltage is VOUT +V(REFLECTED).

Yes the gnd clip has too much inductance for fast signal scoping.

Good idea about getting some high voltage coax.....i will try and find some, does anybody know of some thats flexible and has a BNC at at least one end?

[David Hess]

What is the reason for doing this? Whats wrong with the standard ground clip, is it that the wire acts like an antenna or something? Or the inductance of it or something ?

The standard ground clip has two problems:

1. The length adds inductance which causes ringing distorting the measurement.  This by itself may not be a problem up to 100MHz if care is used.

2. Magnetic flux within the area between the probe and probe ground lead will completely corrupt the signal.  A twisted pair or coaxial connection minimizes this problem by reducing the loop area.  If you want to test for this, short the ground lead to the probe tip and wave it around inside the probing area.

The 300MHz and 20MHz example shown below of some low frequency switching power supply waveforms at startup were measured using short RG-179 coaxial connections with coaxial to probe tip adapters and standard x10 probes.  Even with a bandwidth of 300MHz, only very low levels of switching noise is present and that might be only ground noise.  These measurements were impossible when using the ground leads on the probes.

[ocset]

Thanks, we’re actually thinking of using a piece of coax, and putting a 9 Meg resistor at the end of it, in order to make a high bandwidth  scope probe.  Has anybody tried this? We could add a 15pF  cap in parallel to the 9Meg resistor too.
We wish to do this because we wish to have something thats solderable to the test point.
The "spring barrels" we find, are too fiddley and dont always give a good connection to the ground barrel.....(maybe its just us but thats how we are finding it)

[capt bullshot]

Many people tried that, and there's an article with all the results:
http://w140.com/tekwiki/images/6/62/062-1146-00.pdf
Short form: What you propose is good for a 5MHz oscilloscope

[dmills]

For this was 50 ohm mode invented, take a length of 50 ohm coax, stick a BNC on one end and connect the screen to your reference at the other end, a 4950 ohm resistor (or as close as you can find) in a small SMT package goes in series with the centre conductor right at the point you are probing, **STICK THE SCOPE IN 50 OHM MODE**, job done, one 100:1 probe with bandwidth for days. For a 10:1 equivalent use a 450 ohm resistor.

Build that right and you have more then enough bandwidth for any flyback secondary on the planet, and the parts are trivial enough that mounting them and a U.FL or such at likely points on prototypes is well worth the trouble at layout time, because it means you can just plug the scope in.

I do wish people would get with the fact that with modern devices EVERYTHING is pretty much RF, once you get this it all becomes much easier to deal with.

Regards, Dan.

[fourtytwo42]

in a small SMT package goes in series with the centre conductor right at the point you are probing

Please be very careful with the breakdown voltage rating of said small resistor!

[David Hess]

Thanks, we’re actually thinking of using a piece of coax, and putting a 9 Meg resistor at the end of it, in order to make a high bandwidth  scope probe.  Has anybody tried this? We could add a 15pF  cap in parallel to the 9Meg resistor too.
We wish to do this because we wish to have something thats solderable to the test point.
The "spring barrels" we find, are too fiddley and dont always give a good connection to the ground barrel.....(maybe its just us but thats how we are finding it)

What you are suggesting is very difficult to do properly.  Much better is to use a BNC to probe tip adapter on the oscilloscope probe to connect to short length of thin cable like RG-174 or RG-178 which is then soldered to the test point.  That is how I generated the results shown above.