Clamp on cable ferrite for scope lead

[treez]

Hello,
Does anyone have a part number for a clamp on ferrite for a scope lead of diameter 3.5mm?
All the ones we find are too big internal diameter, and i believe they have to be a snug fit to work properly.
We are getting noise injected into our LED power supply through the scope lead...soon as we remove the scope lead, the circuit works again.

 

[betwixt]

I doubt noise injection is causing the problem. Its far more likely to be capacitive loading or a voltaic connection back through the power supply. I assume your scope is fed through a mains isolating transformer.

Try finding a short ferrite rod, the kind found in cheap AM radios and wind a few turns of the scope lead around it. Not exactly the same but easier to do and it will prove the point. The other thing if you have a dual beam scope is to use the probes differentially so you can omit the ground clip. Invert one channel and add them so it measures across probe points without the ground connections but if you do that, check there is at least a leakage path from the scope to ground so it's body can't become 'live'.

Brian.

 

[FvM]

Clamp-on ferrite cores are made of low or medium µr RF ferrite, larger high µr toroid cores are much better suited to reduce common mode interferences on cables, something like Epcos/TDK B64290L0082X830.

 

[treez]

larger high µr toroid cores are much better suited to reduce common mode interferences on cables, something like Epcos/TDK B64290L0082X830.

Thanks, and i suppose we just rap the probe lead a few ties round the torroid?
https://www.digikey.co.uk/product-detail/en/epcos-tdk/B64290L0082X830/495-3871-ND/1830201

I assume your scope is fed through a mains isolating transformer.

Thanks, the scope is actually straight into the mains...the led power supply is powered through an isolated AC PSU.

 

[c_mitra]

All scope leads I have seen are coax: you are unlikely to see any significant improvement with a wrap around ferrite.

It is more likely an earth fault: you disconnect the scope earth and the fault may disappear.

 

[FvM]

Thanks, and i suppose we just rap the probe lead a few ties round the toroid?

Sure, five turns or so.

Thanks, the scope is actually straight into the mains...the led power supply is powered through an isolated AC PSU.

Presume the power supply has considerable output capacitance, isolation for high frequency common mode noise is hardly perfect.

All scope leads I have seen are coax: you are unlikely to see any significant improvement with a wrap around ferrite.

The ferrite is acting as common mode choke.

 

[treez]

This is a related post, so please may i post it here....

Hello,
Please can you tell why the attached offline LED driver test setup makes the LED driver not work properly whenever the coaxial probe is connected to the scope. As soon as we disconnect the BNC of the coaxial probe from the scope, the LED driver works properly again.

Do you know what is happening?

The strange thing is that this set-up was working fine a few weeks ago and giving nice scope shots of the current waveform via the sense resistor.

The scope is a Tenma 72-8705A….
**broken link removed**

 

[FvM]

We can just guess how much high frequency isolation is provided by the "AC output PSU". Probably not much.

Why not use a current transformer on the AC side of the rectifier, why not a state-of-the-art differential probe?

 

[treez]

We can just guess how much high frequency isolation is provided by the "AC output PSU". Probably not much.

Thanks yes i see your point.......incidentally, when we raplaced the isolated output AC PSU with a simple Carroll and Meynell 50Hz isolation transformer, the problem still occurred.

- - - Updated - - -

Isolating txformer
https://uk.rs-online.com/web/p/safety-site-transformers/4368805/

 

[FvM]

In this case, I suppose a design fault of the LED driver causing unexpected susceptibility to conducted or capacitively coupled noise.

 

[treez]

In this case, I suppose a design fault of the LED driver causing unexpected susceptibility to conducted or capacitively coupled noise.

Thanks, but it works fine without the coaxial probe attached.
Also, i added a heavy diff mode and common mode filter and still got the problem.

Why not use a current transformer on the AC side of the rectifier

Thanks, i did a search for such a 50Hz current sense transformer, or even just the core so we could wind it ourselves, but there is absoluteley nothing off the shelf that isnt stratospherically expensive.

 

[FvM]

Thanks, i did a search for such a 50Hz current sense transformer, or even just the core so we could wind it ourselves, but there is absoluteley nothing off the shelf that isnt stratospherically expensive.

Can't take this seriously...

 

[treez]

I appreciate you can get transducers which will output a DC voltage which corresponds to the irms input, but we want to actually see the current waveform....and transformers for that, (our current is 1A peak) are very expensive, due to as you know the high inductance required to keep the magnetising current low so as not to distort the current waveform.

 

[c_mitra]

You can certainly use two channels of the scope as a differential probe; use 1:10 input so that loading is minimal.

Or, did you already tried that and I missed something in between?

 

[treez]

use 1:10 input so that loading is minimal.

Thanks, thats interesting, when we used the DIY coaxial probe it is a x1 probe and we had the scope set to x1.

Strangely, when i measured the internal 14v rail with a x10 "normal" scope probe (with a dangling ground clip), then the led driver did not malfunction then ..even though this again involved me connecting the ground of the scope (=earth) to the led driver circuit.......so i am thinking the x1 diy coaxial probe is more of a conduit for dreaded noise.?

 

[FvM]

Sounds unlikely if the ground connection goes to the same circuit node in both cases.

A possible reason for LED driver noise susceptibility can be an unreliable zero crossing detection algorithm.

 

[c_mitra]

Sounds unlikely if the ground connection goes to the same circuit node in both cases...

I doubt that the ground of the circuit under test is common to the scope ground.

Simple test: disconnect the crocodile clip ground connection and test with the probe the circuit ground.

Connecting two different grounds via the probe lead ground is likely to create a mess.

 

[treez]

Sounds unlikely if the ground connection goes to the same circuit node in both cases.

Thanks, good point the ground connection of the probe was not to the same place in both cases....there is a ET current limiter, and now i show this in the attached...as you can see, the probe ground is more kind of downstream for the OK measurment of the 14v internal rail.

I say "OK", but even the 14V measurment needed me to disconnect the probe to get the circuit to work properly, then when i reconnected it kindly continued working properly. -This is not possible with the sense resistor measurmenet with the coaxial probe....with the sense resistor measurement , it always never works properly.

- - - Updated - - -

So even the 14V measurment with "normal" scope probe is noisy now and again.
This is very bad for us, because when testing the circuit, its nice to be able to clip the scope ground to the circuit ground, then quickly "bounce" about the circuit with the scope probe, taking mulitple waveform shots as you go...however, we cannot do this now, due to the noise, and have to switch the circuit off then on again so that we can reattach the diff probe every time we need to look at any node...this takes ages....because we have to solder in flying wires to every single circuit node we need to look at..................are there any diff probes which look like normal scope probes and have a pointed tip that you can "Bounce" around the circuit with?....our TA041 diff probe just has two croc clips which are very slow to use when you need to quickly examine multiple circuit nodes.

 

[c_mitra]

This is very bad for us, because when testing the circuit, its nice to be able to clip the scope ground to the circuit ground...

But what stops one to connect one end of the differential probe to the circuit-under-test ground and poke other points of interest with the other probe end?

You use channel A and channel B and see on the scope signal for (A-B) and the probe tip of B connected to circuit ground and use probe tip A to test various points of interest.

As long as your scope input amplifier has some decent CMRR, this will work fine; just leave the ground connections of both the probes hanging.

(according to your diagram, the circuit under test has a floating ground and when that is connected to the scope ground anything can happen...

 

[FvM]

You use channel A and channel B and see on the scope signal for (A-B) and the probe tip of B connected to circuit ground and use probe tip A to test various points of interest.

As long as your scope input amplifier has some decent CMRR, this will work fine; just leave the ground connections of both the probes hanging.

You are using two single ended inputs in A-B mode. There's no CMRR available in this case, just gain matching. The worst point is that the input attenuator must be set according to the expectable "common mode" voltage to avoid overload. I think the method isn't really applicable to measure volts with several 100 V common mode voltage.

our TA041 diff probe just has two croc clips which are very slow to use when you need to quickly examine multiple circuit nodes.

The design is a bit unusual. Most HV probes have 4 mm safety jack plugs, so you can connect anything, e.g. test tips. I would replace it.