EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: nouserid on July 24, 2014, 11:18:27 pm
-
I've got a TDS2014 fixed and as a post-repair check, I had the frequency response measured. The measurement setup is the following: Agilent E4432B signal generator --> RG58 coaxial cable --> BNC 50-Ohm feed-through termination --> TDS2014. The measured frequency response is attached. However, the response does not look Gaussian to me and in-band roll-off starts too early at around 10 to 20 MHz. The measured 3-dB bandwidth is 100 MHz and rise-time is 2 ns. I was wondering if the 109-step calibration procedures in the service manual (with DC levels and 5 kHz, 50 kHz, 1 MHz, 20 MHz and 100 MHz sine waves) would correct for the frequency response. Thank you.
UPDATE:
The culprit was the poor RG58 cable. After replacing that with a Pomona 2249 BNC cable, the frequency response looks Gaussian.
-
Is your 50 Ohm feed through up to the task?
BTW what was the repair?
-
The feed-through termination I used is: http://www.caltestelectronics.com/ctitem/72-coaxial-terminators/CT2944. (http://www.caltestelectronics.com/ctitem/72-coaxial-terminators/CT2944.) The advertised VSWR specs are 1.10 DC to 500 MHz; 1.20 500 MHz to 1 GHz. I also tried with Pasternack PE9001 BNC tee with Mini-Circuits BTRM-50 termination and yielded similar response with additional approx. 0.3 dB loss at 100 MHz.
The repair was to replace a shorted bypass capacitor (C703). Before the repair all four channels had random data displayed on the screen.
Is your 50 Ohm feed through up to the task?
BTW what was the repair?
-
Hi,
What's the output power setting on the signal generator, and the input setting on the oscilloscope ?
-
The input power level was 10 dBm, and the sine waves were fitted to approx. 80% of full vertical scale when measured. Averaging (over 64 data sets) was also employed. I also did a measurement with 0 dBm input power and attached the results for comparison. As can be seen from the plot, the response deviates by about 0.4 dB. However, at 0 dBm input, the roll-off at low frequencies is still somewhat faster than expected.
I have no experience with digital oscilloscope calibration. I would like to know if the calibration procedures described in service manual are the ones used by the factory.
Hi,
What's the output power setting on the signal generator, and the input setting on the oscilloscope ?
-
Hi,
The frequency response 0..50 MHz should be almost flat.
Maybe the output of the signal generator varies with frequency.
Did you use a measuring function of the scope ?
-
There are some plots in this discussion thread showing the frequency response of various oscilloscopes including a 100 MHz 2232 that I tested with a leveled sine wave generator at the very end:
https://www.eevblog.com/forum/testgear/frequency-response-of-your-dso/ (https://www.eevblog.com/forum/testgear/frequency-response-of-your-dso/)
If you have a fast clean pulse generator, then you could test the transient response as a sanity check.
-
I found the culprit today. Apparently the RG58 coaxial cable I used previously is terribly poor in frequency response. I did the measurements with Pomona 2249 BNC cable and the frequency response looks Gaussian now.
The rise-time I measured with a fast-edge pulse is 2 ns.
However, I am still curious about if the calibration procedures will actually correct for frequency response?
There are some plots in this discussion thread showing the frequency response of various oscilloscopes including a 100 MHz 2232 that I tested with a leveled sine wave generator at the very end:
https://www.eevblog.com/forum/testgear/frequency-response-of-your-dso/ (https://www.eevblog.com/forum/testgear/frequency-response-of-your-dso/)
If you have a fast clean pulse generator, then you could test the transient response as a sanity check.
-
A 2 nanosecond transition time implies a bandwidth of 175 MHz if the response is Gaussian but some DSOs have a different response and I would not expect that of a TDS2014.
What kind of repair did the TDS2014 need? Was it something that would affect the calibration? Was the calibration data erased?
Is there some kind of sanity check you can do on your signal generator and test setup?
Does the TDS2014 meet its DC specifications?
-
I replaced a shorted decoupling cap array for one of the ASICs. The scope has been serviced by Tek previously (but they failed to fix the problem) and those ASICs seem to have been re-flowed. I am not sure if the chips have been replaced. And based on that I suspected the calibration might have gone.
Now the frequency response looks reasonably good for my "general purpose" usage and I am happy with that. In terms with DC, I haven't looked into that, but I will check it with 34401A. The signal generator itself is OK. The reason for the weird response is the cable. Apparently poor RG58 BNC cable starts to roll-off around tens of MHz.
A 2 nanosecond transition time implies a bandwidth of 175 MHz if the response is Gaussian but some DSOs have a different response and I would not expect that of a TDS2014.
What kind of repair did the TDS2014 need? Was it something that would affect the calibration? Was the calibration data erased?
Is there some kind of sanity check you can do on your signal generator and test setup?
Does the TDS2014 meet its DC specifications?