53 Amp current spike in 5W Sepic converter...is it real?

[treez]

Hello,

I have just designed a 5W constant-off-time Coupled-Inductor-Sepic Converter. (LED driver) Schematic is attached.

Coupled Inductor Sepic schematic:
https://i40.tinypic.com/1zvtp8y.jpg

Please can you help in explaining the 53 Amp peak current spike (at the switching frequency) that I am seeing on the scope?
The duration of the spike is 1 microsecond.
The spike is seen in a 13 milliOhm current sense resistor in series with the sepic capacitor.
The ‘sense resistor’ is three 39 milliohm 1812 resistors in parallel.

So, to elaborate, I have put a 13 milliOhm sense resistor in series with the sepic capacitor and am seeing a strange spike across this sense resistor at the switching frequency. The spike is 700mV peak, indicating a current peak of 53Amps. Its duration is approximately 1us. I am thinking this must be a scope lead artefact as surely the current cannot be this high?…..i used a high frequency scope probing technique, and not the normal scope tip and ground lead.

The duration of the 53 Amp current spike is one microsecond.

My scope is a Tenma 72-8225:
https://www.farnell.com/datasheets/1661870.pdf

At maximum load power, the SEPIC inductor is warm to the touch, but not so warm that I can’t keep my finger on it all the time when running at room temperature….so I can’t understand how there can be a 53 Amp current spike in the sepic ?
Also, I’ve left this sepic converter running at 5W power continuously at room temperature for 24hrs, and it runs fine over all that time

I was suspecting that there might be ringing between the sepic inductor’s leakage inductance and the sepic inductor’s interwinding capacitance?…following the 53Amp current spike, there does seem to be a decaying ring at a frequency of approximately 500KHz ……I am also not sure if this 500KHz ring is real or a scope artefact?

V(in) is fixed at any level between 4V and 8V.
V(out) is any of several LED lamps , which could have voltages from 5V to 80V.
Off time is set to 3.88us.
Frequency of operation varies depending on load power and voltage, but is between 33KHz and 137KHz.
The ceramic sepic capacitor is 22uF , 1812 size.
The input and output capacitors are all ceramic.

The Coupled Sepic inductor is MSD1583-47uH :
**broken link removed**

The scope probe ground is one of those little metal protrusions which coils around the scope probe’s ground piece, and looks like a spring (for high frequency measurement)

When I touch the scope probe ground to the switching node of the sepic converter, I can see tiny little sparks as the scope probe ground protrusion rubs along the metal of the switching node pad. Why are these sparks seen? Is it something to do with the power supply that’s powering the sepic? (I assume its an isolated PSU , but don’t know, as the spec doesn’t say…it’s a Maplin GW02C PSU)

Bench Power supply used to supply sepic: (Maplin GW02C)
**broken link removed**

Incidentally, the current spike is 53Amps when measured with the scope probe on 1:1 setting, but when the scope probe is on 10:1 setting, then the spike is seen to be of magnitude 87 Amps.

Adjusting the bandwidth setting of the scope does not affect the size of the current spike seen on the scope.

The scope probe is 60MHz “UT-P03” and came with the scope.
Sorry I cannot take a scope sot as the scope doesn’t recognise the USB stick.

So, do you know if this “current spike” could just be a scope artefact ? (the expected peak sepic inductor current is just 2Amps when the load is 80V and 5W). Also, do you know why I’m getting sparks at the scope’s ground terminal when touching the sepic’s switching node?
Also, what would be the best equipment to use to view the current waveform in the sepic capacitor?..(eg a differential probe?)
Also, if this ringing/spiking is real, how should I best snub it out?

 

[schmitt trigger]

"Sorry I cannot take a scope sot as the scope doesn’t recognise the USB stick."

Use your mobile camera to take a snapshot. Just make sure it is in the macro mode so the picture is properly focused.

 

[FvM]

I don't guess about measuring artefacts without seeing the actual setup.

I presume, you see the current peak at transistor turn-on? If the diode is a schottky as claimed in the schematic, there won't be a large reverse recovery peak, a few A might be observed nevertheless. A slow standard diode can actually show a tens of A snap-off current peak.

 

[treez]

When in DCM with an 80V lamp, I see the current peak at turn off of the fet.....when in CCM with a 5v lamp, I see the current peak at turn on and turn off.
Incidentally, I placed 4.2uH in series with the sepic primary and still see the artefact.
However, the coupled inductor does feel slightly cooler when the 4.2uH is in series with its primary.....the 4.2uH has 120milliohm of esr.

Any ideas on how to look at the current waveform in the sepic cap much appreciated.
Also, the sparking on the ground of the probe.

 

[treez]

The pics are below.
The HF probe is seen in one pic, but I cant hold the camera and that at the same time, so the normal ground clip probe is used.
The waveform s shown (current in sepic cap), but it comes out blurred, sorry. scope is on 1us/div and 1v/div

 

[ahsan_i_h]

Hi treez
Short circuit the 13 milliohm current sense resistor using a thick copper wire. Then check again.
If the 700 millivolt spike still exists then it is due to the nearby magnetic field.
If the 700 millivolt spike is gone then the 53 amp is real.

 

[treez]

Short circuit the 13 milliohm current sense resistor using a thick copper wire. Then check again.

Ahsan_i_h you are a genius , why didn't I think of this....I just shorted it out and the "waveform" is still there just the same.....

This still leaves me with the problem of how do I view the actual current waveform in the sepic capacitor. Is it just a case of putting a common mode choke in some wires coming off the current sense resistor?
Also, the reason for the ground probe sparks at the switching node.

 

[schmitt trigger]

You really require a differential probe to measure signals that are floating above ground.
And as ashan mentions, it is nearby magnetic fields inducing noise, thus the probe leads must be very, very short.
There are many good articles which show how can this be done.

 

[treez]

You really require a differential probe to measure signals that are floating above ground.

Though its an isolated power supply, so there is no "ground"

 

[FvM]

Though its an isolated power supply, so there is no "ground"

Only true for DC and low frequency.

No power supply is perfectly isolated, there are at least Y capacitors and stray capacitance elements. In addition you have circulating ground currents with multiple connected oscilloscope probes. Before deciding for differential probes, there are some measurement tricks that might be sufficient to measure the shunt voltage with less interferences:
- use ferrite cores as common mode filter for the probe cables. Preferably high permeability (power transformer range) toroid cores large enough to feed the BNC connector and 5 to 10 windings through it
- use a 1:1 probe, or even better a thin coaxial cable (e.g. RG174) directly soldered to the shunt resistor

If the circuit layout allows it, self-made current transformers (small 4 - 10 mm ferrite core, 1:50 - :100 ratio, 50 ohm coaxial cable terminated at the oscilloscope as secondary burden) can give perfect AC current measurements with > 100 MHz bandwidth.

Ultimately, miniature rogowski probes can be used. Not easily made at home and quite expensive, but the perfect tool in power electronic measurements. They can be small enough to be placed e.g. around DPAK package pins. Commercially available from www.pemuk.com

But even these advance tools aren't completely immune against interferences, e.g. capacitive crosstalk. They have to be handled with care, measurements must be checked for plausibility again and again.

 

[treez]

a thin coaxial cable (e.g. RG174) directly soldered to the shunt resistor

...they used to use bits of coaxial cable like this when I was at Alcatel.
They used to measure all rail voltages like that......it was a real pain trying to solder the ground sheath of the coax to the pcb......also, they used to put 47 ohm series resistors in series with the signal before feeding it directly into the scope. I am not sure of the 47 ohm was really needed?...I thought that the scope had a 50 ohm input impedance option.?

 

[FvM]

Some soldering skills are essential for a hardware engineer. If you worry about reusability, you may prefer temperature resistant cables with PTFE/FEP isolation.

You can use source side (50 ohm series) as well as load side (50 ohm parallel) termination, also double sided termination for highest frequency range, also unmatched source side series resistors with load side termination to create a voltage divider (resistive low impedance probe). For a low impedance source like a shunt, simple load side termination seems to be the best. With slow oscilloscope that miss a 50 ohm setting, use through-termination adapters.