EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: Sensei on October 25, 2019, 08:02:31 am
-
Hi ladies and gentelmen
I want to perform calibration of internal VCTCXO in Siglent SDG2000x to improve time / frequency measure accuracy from +/-1ppm to +/-0.1ppm (this is resolution of counter in SDG2000x). To do this i want to use 1pps signal from FGPMMOPA6H GPS module as reference signal, put it on COUNTER input of SDG2042x, measure period of this signal and tune VCTCXO.
The hard part of this job is how to tune internal VCTCXO. After investigation andmeasurment i notice that:
- after power up on pin 1 is 0.08V, RF_out is 9,999850HMz
- after boot sequence is done, on pin 1 i measure 1,49V and RF_out is 10,000015MHz
- i decreased voltage on pin 1 by connecting this pin to GND via resistor 180R and 90R, and frequency on RF_OUT decreased.
| Vefc | dVefc | RF_out | dRF_out | dF/dU | ppm/V |
| [V] | [V] | [Hz] | [Hz] | [Hz/V] | [ppm/V] |
| 0,08000 | - | 9999850 | - | - | - |
| 1,48466 | 1,40466 | 10000016,2 | 166,2 | 118 | 11,8 |
| 1,48740 | 0,00274 | 10000016,5 | 0,3 | 109 | 10,9 |
| 1,49020 | 0,00280 | 10000016,9 | 0,4 | 142 | 14,2 |
VCTCXO is marked as G318-10.00 / DAPU1318, propably this is version of T75B series of DAPU Telecom TCXO.
https://www.dptel.com/products-detail.php?ProId=2911 (https://www.dptel.com/products-detail.php?ProId=2911)
Vecf propably is generated by DAC, and propably it can be changed by software.
Maybe this voltage can be changed by modifying configuration files via TELNET connection?
If not, I will make an intermediate PCB between TCXO and SDG2000x board and drive the Vecf pin with my own voltage source.
Any hints are welcome.
Sensei
-
After connecting the 1pps signal from GPS module, I measured on the SDG2042x frequency of 0,99999971Hz, error -0.27ppm. After half an hour of work, the error increased to -0.32ppm, but it is still acceptable. To correct this error, I should increase the V voltage by 25mV because the TCXO is too slow.
-
No need to solder anything extra.
Software calibration (page 58)
http://www.siglentamerica.com/USA_website_2014/Documents/service_manual/SDG2000X_ServiceManual_SM0202X-E01A.pdf (http://www.siglentamerica.com/USA_website_2014/Documents/service_manual/SDG2000X_ServiceManual_SM0202X-E01A.pdf)
-
This is exactly what i needed!!! Thanks very much Tjuurko :):):)
-
Since the internal reference on these generators is just a TCXO (which apparently is internally digitally controlled and corrects its frequency in - even though tiny - increments), you may as well utilize the REF IN BNC and get yourself a GPSDO or a Rubidium oscillator to supply the 10MHz reference clock. This way, you should reach (long-term) accuracy in the low ppb range without any hassle. But be careful, obtaining equipment like this makes you an almost certain candidate for becoming a "time nut" ... ;)
-
Since the internal reference on these generators is just a TCXO (which apparently is internally digitally controlled and corrects its frequency in - even though tiny - increments), you may as well utilize the REF IN BNC and get yourself a GPSDO or a Rubidium oscillator to supply the 10MHz reference clock. This way, you should reach (long-term) accuracy in the low ppb range without any hassle. But be careful, obtaining equipment like this makes you an almost certain candidate for becoming a "time nut" ... ;)
There is nothing wrong with
1. Get a good gpsdo
2. Measure the output of that gpsdo with the counter function in sdg2000x
3. Adjust the 2000x so it displays the correct frequency.
4. Then use that gpsdo as reference in the lab, but know that the 2000x is in itself accurate enough for field work.
-
Since the internal reference on these generators is just a TCXO (which apparently is internally digitally controlled and corrects its frequency in - even though tiny - increments), you may as well utilize the REF IN BNC and get yourself a GPSDO or a Rubidium oscillator to supply the 10MHz reference clock. This way, you should reach (long-term) accuracy in the low ppb range without any hassle. But be careful, obtaining equipment like this makes you an almost certain candidate for becoming a "time nut" ... ;)
There is nothing wrong with
1. Get a good gpsdo
2. Measure the output of that gpsdo with the counter function in sdg2000x
3. Adjust the 2000x so it displays the correct frequency.
4. Then use that gpsdo as reference in the lab, but know that the 2000x is in itself accurate enough for field work.
I know this is a year later but I thought my calibration experience would provide a way to get about an order of magnitude greater accuracy in short order (that frequency counter is just too lacking in digits of resolution to properly do this task full justice).
As well as the above, I connected both channel outputs at 10MHz sine wave 2Vpp 50 ohm (or 4Vpp Hi-Z if you prefer :)) to my 4 CH 'scope with a copy of the 10MHz GPSDO reference triggering the 'scope. When I first started twiddling the knob on the AWG, nothing happened until I jumped a few digits leftward and the first instant sign of change was given not by the digitally challenged frequency counter display but by the traces suddenly getting close to synchronised. followed by the counter readout catching up to display 10.000 000MHz which it did unblinkingly for the rest of my adjustments which were clearly showing their effect on the 'scope display.
I think the DAC had simply been preset to the middle of its range at 16384 but with my twiddling the LSD not showing any perceptible change even in the scope trace, I had visions of it being close to the end of its range until I started adjusting it in larger chunks before settling at the 15621 setting (now 15629 after trying the extremes of the DAC range - shades of retrace?) as being as close to the right frequency (on average - it still wanders up and down in frequency by a few mHz or so, but hey! That's VCTCXOs for you ::) ).
===========================================================================================
[EDIT 2020-10-28]
The datasheet +/-2ppm 0 to 40 deg C specification suggests just a VCXO rather than a VCTCXO and there's no mention of a VCTCXO being installed which confirms my impression that it's just a VCXO rather than the VCTCXO that everyone else seems to have assumed.
===========================================================================================
Curious as to how much adjustment I had left, I entered the value 32768 from the keyboard - it stopped at 32767, fine if zero is included but it's not, you can only go to a minimum of 1. Anyway, it seems I have ample adjustment either way to compensate for ageing (the vcxo's not mine :( ) and a relatively swift way to get it a magnitude closer to frequency than just relying on that low resolution frequency counter function alone.
One final note, it's best to do this adjustment with the AWG in its normal orientation, flat on a shelf or tilted up on its bail stand. VCXOs are no more immune to the 2G tipover effect than OXCOs are. And, one last thing to consider is that having either just one or both channels enabled will also make a small difference to the calibration value.
John