Home made coaxial probe for outo measurement?

[grizedale]

Home made coaxial probe for output voltage ripple measurement?

Hello,

Please could you tell me if the following output voltage ripple measurement, which is the standard procedure at a HUGE international telecoms company, is correct?

It concerns the measurement of the voltage ripple on the SMPS's inside the base station.
This method is used for all the different smps's in the base station, which have frequencies from 1MHz to 100KHz, and output voltages from 60V to 0.9V

The method used was:

Get 1 metre length of 50 Ohm coaxial cable.
Solder a 47R axial resistor to the innner conductor at one end of the cable.
Then solder this 47R resistor to the 'plus' trace of the output voltgage.
Then "peel away" some of the outer conductor of the cable, and solder that to the 'minus' trace of the output voltage.
Then put a BNC connector on the other end of the cable.
Then put this into the scope input.

The scope was a Lecroy LC684DM:
**broken link removed**


My query is that page 5-4 of the above scope manual says that the input impedance of the scope channel can be set to 50 Ohms.....and that is what it is set to for this measurement.....


..So my question is that the characteristic impedance of the 1 metre of coaxial cable is 50 Ohms, so why are they soldering the 47R resistor to the end of the coaxial cable?

Also, sometimes it's fiendishly difficult to solder the "peeled away" outer connector of the cable to the PCB under test, so a short, 1 inch piece of twisted wires is then used at the end of this coaxial cable assembly to make it easier to solder it to the output voltage.....is this acceptable procedure?, or would that 1 inch piece of twisted pair wires infect the reading with too much noise?

 

[FvM]

Using a source side cable termination as described, you get several 100 MHz measurement bandwidth when connecting to a high impedance oscillscope input. Twisted pair will achieve less perfect impedance matching but should work for usual SMPS frequency range. I would prefer the coax solution.

 

[grizedale]

sorry i dont understand, why would you want to connect a coaxial cable to a HIGH IMPEDANCE scope input?...........the whole idea of coaxial cable is that you match the scope input impedance to the cable characteristic impedance, which is just 50 Ohms.

And why do i need several 100 MHz of measurement bandwidth for a smps output voltage ripple measurement?

 

[FvM]

the whole idea of coaxial cable is that you match the scope input impedance to the cable characteristic impedance, which is just 50 Ohms.

There are different cable impedance matching options:
- both sides, lowest reflection
- source side only, as in th epresent case
- load side only, can be driven both by voltage or current source

In case of the output voltage ripple measurement, 50 ohm oscilloscope impedance (load side matching) is unwanted due to the high DC current and limited oscilloscope voltage range.

Load side series impedance matching is a good alternative in this case. It's presuming a low AC source impedance (e.g. provided by the filter capacitors).

And why do i need several 100 MHz of measurement bandwidth for a smps output voltage ripple measurement?

Not exactly for the ripple measurement, but e.g. to determine fast switching transients and possible e.m.c. issues. If you can have larger measurement bandwidth with low effort, why not using it?

SMPS output voltage will be at least checked over the "conducted" ( - 30 MHz) frequency range in some e.m.c. measurement, but you may want to use it for a first check in a wider frequency range as well, e.g. as a tool to locate the origin of radiated interferences or to decide about additional wideband filtering of the output voltage. A twisted pair will pick up interferences from everywhere.

All in all, I didn't say but that the measurement setup looks reasonable for me.