Zsc measurement issues

[fotis]

Hello
I try to put in operation the LCR meter. Before that, I have to calculate the Zsc using the "Zscdefault.psf" in DUT. As you know my SC is an Asus Xonar Essence STX and in the settings is registered as Sound Card ASIO. I also have a pair of P601 probes and try to obtain the calculation with them. If both probes are used, yet in position 1 i.e. "2 [P1+P1], the corresponding Rr is very big (2KΩ), and in the process window always BAD is shown in all frequencies, as the signal level is small. Next, I tied with only one probe "1 [P1]", but again the signal level was too small i.e. "Peak Percent = 69%" - except in 10K and 20K - and BAD. Next, I changed the input of SC to MIC (it is common for Line IN and Mic IN) from the Asus control panel, and the signal presented hard clipping (over 100% and RED on the bar graph). As a side note, that caused (again) some malfunction in the ADC of SC and the bandwidth fell to 63KHz in the No Loopback test. Next, I tried the SC as "Sound Card MME" to have some control of volume through the Asus control panel, and I got a gain of 90% and above, but in the process column, I got again BAD in all frequencies. Although the gain is sufficient, I observed on the oscilloscope a relatively big offset of 0.5V. Maybe that is also a problem? The signal levels are very different between 16bit and 24bit. I tried to modify "Zscdefault.psf", without success. And is the default for LCR measurements. The default output level of 1V peak is either insufficient for ASIO or very hard for MME. I have calibrated the levels of my SC with an external generator and RMS voltmeter, in 24bit/192KHz and ASIO. Have you any idea of how could resolve this problem?
Thanks

[fotis]

Hello
Following my previous post, I've checked the Line Outputs of Asus Xonar with the oscilloscope. As stimulus I used the same "Zscdefault. psf" in which the MI signal generator output setting is at max (1V as is stated in the Output Amplitude box) and during the No loopback test the MI bar graph is turned red and shows 100% indicating clipping of the signal. Well, the actual level captured by the oscilloscope, is 6.2Vpp for any frequency. Please note that the full-scale voltage of Asus Xonar input is given as 5.65Vpp. That is the reason for the clipping and the following problems with the sound card. It looks like the Asus is some delicate to overdrive. Apparently is not an issue with the Line In operational amplifiers of SC, it is a problem with its ADC converter, it has problems after an overdrive. I injected some square waves from the MI generator to the Line In and the MI oscilloscope shows significant overshoots and bad rise time. That explains the bandwidth of 63KHz instead of the normal of 93KHz (ASIO/24bit/192KHz). BTW it shows the same peak and RMS values as the external oscilloscope.
Thanks

[VirtinsTech]

During the test, the software will check the quality of the measured signal. If it is clipped or distorted, then BAD will be shown and the test cannot be carried on.

Calibration of sound card input impedance requires two steps:

Step1---Set Test Tone Reference Level (See Section 9.6.2.1 of the software manual)

Step2---Test with DUT (Zsc) (See Section 9.6.2.2 of the software manual)

The reference Rr will need to be changed from a lower value to a higher value from Step 1 to Step 2.

[fotis]

Thanks for the reply
I've read and followed carefully these sections you refer to. The problem is that: When the Sound Card is registered as "Sound Card ASIO" in both ADC and DAC device settings, then there is no other level control except the two sliders (Output Amplitude - Vp) of the MI signal generator. That applies to the Asus Xonar Essence because is a PCI sound card and offers only software controls which are disabled in ASIO mode. The "Zscdefault.psf" is a locked plan (44.1KHz/16bit) and the sliders of the signal generator are at max. I had measured in the past the Zin of SC with other software (Speaker Workshop) and found it 4.3KOhms. If both P601 probes are used, the Rr=1K+1K=2KOhms and is very big. If only one probe is used (1 [P1]) again the 1K resistance is big, and the signal is attenuated to 69%. If a common stereo signal cable from the output to the input is used (Zero resistance) the ADC of SC is hardly overloaded (over 100%) and the result is audible (a fast and continuously interrupted sound, something like.... TRRRRRRT). The only solution is to solder an external potentiometer in series with the input of SC to offer some volume control.
SUMMARIZING: 1) The "Zscdefault.psf" is the only way to calculate the Zin of the sound card 2) It is a locked plan with signal generator output level at MAX while the ASIO mode disables any other level control 3) Two probes insert a 2KΩ Rr in series which is very big 4) One probe inserts a 1KΩ Rr in series which is also big 5) With simple signal cable (Rr=0) the SC Line In is hardly overloaded.
The solution to use the probes would be 1) An open "Zscdefault.psf" where I could adjust the output level of signal generator 2) To open and replace the 1KΩ series resistor of probe, with an e.g. 499Ω resistor. OR to build my own probes with a suitable series resistor, but again there is need to open the Zscdefault.psf and insert the value.

[VirtinsTech]

The DTP file is a CSV text file with some special characters as separators. It is still possible to edit a locked plan using Windows Notepad, a plain TXT editor. You should be able to figure out where the output voltage is after opening the file.

Your estimated sound card input impedance is 4.3k Ohms which is comparable to 1_[P1]=1k and 2_[P1+P1]=2k. Therefore there should be no problem to achieve good measurement accuracy.

BTW, MME should be good enough for this type of testing.

[fotis]

Calibration.png (24.47 KiB) Viewed 26808 times

Hello
Thanks for the advice.
I spent 2 hours calibrating the sound card as MME. The Zsc has been calculated fine at 4.54KΩ. The output amplitude of the signal generator is also calibrated and its max level is 2.8Vp. Later, I tried some measurements of components. Measurement of linear loads like resistors is very accurate but doesn't apply to complex loads like capacitors and inductors. I tried with all readily offered psf and with different capacitors and inductors but had no luck. E.g. a 470nF cap is calculated as 13uF while on the green window the allowable range shown, is from 240pF to 2uF. I understand that the signal level should be set within the percentage shown on the green window, to obtain a corresponding component measurement. The problem arises from the fact that the level of the signal generator is locked at 1Vp and that level is insufficient for the measurement. Or the minimum resistance of 1KΩ of the probes is so big.
My settings for LCR measurement are 1) Sound Card MME for both ADC and DAC to have control of volume 2) The switches on both P601 probes at position 1, so "2 [P1+P1]. 3) On the Sound Card control panel Playback and Record sliders are all at 100%. The master volume can be set at 66% as much. With these settings, the maximum level obtained is 47.23%. If the master volume is placed above 66%, the indication on the green window is turned BAD because the signal is clipped.
I think the calibration that I did, using the MI signal generator and an external true RMS voltmeter is correct. Please take a look at the picture and tell me what you think.
Thanks

[VirtinsTech]

1. Actually only relative values are needed in LCR measurement here. Calibration is not necessary but there is no harm to do that.

2. The inaccuracy in your measurement of 470 nF seems to be too big. If you want to check the cause of inaccuracy, the input impedance of the sound card at 1kHz (4.54k?), input peak levels (47.23%?) without and with (?%) the 470 nF connected need to be known here. With these values, the capacitance under test can be derived through formula. Moreover, as indicated in Section 9.8, the accuracy will be the best in the middle of the measurement range. The impedance of 470 nF is 0.3386k at 1kHz and 3.386k at 100Hz. The value 3.386k is comparable to the sound card input impedance 4.54k and thus the measurement will have a higher accuracy at 100Hz than 1000Hz. You can load LCR_100Hz.dtp to see if a better result can be obtained.

[fotis]

Thanks for the advice.
I tried the 100Hz psf without success. The green window shows BAD for any level setting. I also tried the 50-10000Hz psf and found that only from 200Hz and above the level is acceptable and shown as a percentage. For 50 and 100Hz test tones, the indication is always BAD for any level setting and the second step of measurement with DUT doesn't open.

[VirtinsTech]

The quality of the test tone is considered as bad if the peak level measured in Channel A exceeds 99.9% or the THD value measured in Channel A exceeds 3%, in which case, adjusting the output level, the reference resistor value, or the input gain may help.