500 ppm LCR-meter: open trim fails with Kelvin-clips

[siamcat]

I received the Elektor tested mainboard of the LCR-meter, put it in the alu-case, did the first-use procedure, all is well. Trim without Kelvin-clips goes well: closed with jumper (0,00 Ohm) and open (>1Gohm and 0,00pF).
But when I use the Kelvin-clips the open trim fails. Also the PC-display hangs, or changes far too often (tx/rx-leds do not flash or flash continuously), and values showed are strange: a few k-ohms to a few megs, L-values varying from 100H down to few micro-H's etc. It seems very unstable with Kelvin-clips. The closed trim with Klevin-clips goes well.
Is this issue known by other users? And what could be the cause of this problem?

[snoopy_94]

I received the Elektor tested mainboard of the LCR-meter, put it in the alu-case, did the first-use procedure, all is well. Trim without Kelvin-clips goes well: closed with jumper (0,00 Ohm) and open (>1Gohm and 0,00pF).
But when I use the Kelvin-clips the open trim fails. Also the PC-display hangs, or changes far too often (tx/rx-leds do not flash or flash continuously), and values showed are strange: a few k-ohms to a few megs, L-values varying from 100H down to few micro-H's etc. It seems very unstable with Kelvin-clips. The closed trim with Klevin-clips goes well.
Is this issue known by other users? And what could be the cause of this problem?

The problem is with the power line.
Kelvin clips cables are, on the frequency of the electrical line, like antennas.
To reduce this phenomenon, I recommend:
- shorten the cables up (they are very long original).
- make sure that the housing is connected to Earth (it is not automatic with a laptop!)
- If you have the possibility, place a metallic sheet steel connected with the earth below the device and the cables.

The new versions of Firmware (version 3.0.0) and of the program AU2011 (version 3.0.0), allow to limit the maximum gain in range 8 (and also in range 1 to solve the problem of the measure of resistances < 1 ohm).
They should be made available on the site of Elektor very quickly.

Jean-Jacques

[siamcat]

Hello Jean-Jaques, thanks so far.

Here are my results till now:

Without clips, the closed (J1, R=0,00 mOhm/L=1-2nH@10kHz) and open trims (>1GOhm, 0,01pF) work as can be expected. Though sometimes with J1 closed I see C>100mF! Which is weird of course.

With open clips on the table on 10kHz R is about 80kOhm, and L is a few Henry! The green led next to the coax is flickering about 8 times a second (which is far too fast, it should be once per second).

1) I closed the case, and connected the case to Earth. This made no difference in behavior.

2) As the cables are already shielded, only the clips themselves can catch 50Hz line. So I put the (open) clips in a closed metal box, connected the box to Earth (via the case of the meter). Again, this made no difference in behavior.

3) As soon as I connect about 6cm of unshielded wire to J12 or J13 via the coax, it goes wrong: same behavior as with open clips. This shows (I think) that making the wires of the clips shorter will probably not help.

I will wait for the new firm- and software, and try again. When can we expect it?

[snoopy_94]

Have you connected the circuit board to the housing with a metal screw as specified in article 3? I have also noted an improvement when I touch the box with my hand!

2) As the cables are already shielded, only the clips themselves can catch 50Hz line. So I put the (open) clips in a closed metal box, connected the box to Earth (via the case of the meter). Again, this made no difference in behavior.

The shield of the cables is not very efficient at the power line frequency. Try wrapping the cables together, as shown in figure 7 (a) of article (without the attachment link!).

I will wait for the new firm- and software, and try again. When can we expect it?

The new firmware and programs are in the hands of Elektor...

[siamcat]

Hello Jean-Jaques,

Yes, I attached the screw connecting mass to the case (and checked the connection, because the Alu surface may not conduct well due to the oxide layer!)

When using the Kelvin-clips the meter goes mad with resistors greater than about 100 Ohms.
Without clips (directly to J11 and J12 via coax-entry) even R´s greater than 8M are measured accurately.

NB one problem might be the phase angle calculation: when it results from division of two very small values (which fluctuate due to chance), the result can vary widely.
Example: open without clips:
Ux=403,9mV (100Hz) - 408,2mV (10kHz) (is stable)
Ix=690-1036pA (100Hz) - 85nA (10kHz)
Especially the 100Hz values for Ix vary considerably.
This phenomenon may explain why I sometimes see angle values vary widely, especially in gain range 8 @100Hz, causing wrong values of 100mF en H to appear. What are your thoughts about this?

Maybe it is better to mount a Hameg probe instead of using the Kelvin-clips. However, I will shorten the clips to see if that makes a difference.

Greetz & Thanks from SiamCat!

[siamcat]

I have shortened my Kelvin clips to about 10 cm. It made no improvement. As soon as I connect the black plugs to J12/J13 the meter goes mad: green led flashing continously at high rate but not the light on au2011.exe!), values > 100H are shown occasionally, or 100nF and 2Mohm. R's up to a few kOhm are measured accurately, greater values let the meter derail (SEE ABOVE, same as open clips).
The alternative HZ181 probe from Hameg costs more than 180 euro's... Bit too expensive for me, and it is sometimes not even handy.
I wonder if (and how!) this problem can be addressed?

[snoopy_94]

Hello SiamCat,

Yes the HZ181 probe from Hameg is too expensive, but the same from TONGHUI (TH26001A) is about 50 euro's...
You can find it on eBay. Works fine!

I hope new firmware and application soon on Elektor web page.

[hadihf]

Dear Sir Jean-Jaques,

It will be very helpfull if we have a log of oscilloscope waveforms at various important points of the instrument.

Best Regards,
F

[snoopy_94]

Dear Sir Jean-Jaques,

It will be very helpfull if we have a log of oscilloscope waveforms at various important points of the instrument.

Best Regards,
F

I'll do this, but you should wait until the beginning of September on my return from holiday!

Jean-Jacques

[hadihf]

This is interesting reading

http://www.ni.com/white-paper/3981/en


Regards
F