I purchased this GPS Disciplined Oscillator a few weeks ago. The reason being, I wanted to make sure that this frequency counter was accurate.
This Hewlett Packard 5315A was last calibrated in 1990. That made me suspicious. While I could send it back to Agilent and have it recalibrated, I thought it might be interesting to check it with a known standard.
When I connected the frequency counter to the 10 MHz GPSDO, it was -2.1 Hz off. At first, I thought perhaps the GPSDO was off; however, the spec for the LBE-1420 is 1 x 10-12 with a resolution of 1 Micro Hz. I let the HP unit warm up for 3 hours thinking maybe it was cold and would come back in tolerance. Nope, the frequency stayed about 2 Hz low.
It took about five seconds to find the full service manual online, which gives the alignment and calibration procedure in detail.
The first step is to use a DVM and check the +3, +5, and -5.2 power supplies. If they are off then adjust each accordingly. The next step is to check the +5 VDC pin on the Option 4 OCXO module and adjust as needed.
The calibration procedure for the HP 5315A is to connect a known 10 MHz reference to one channel of an oscilloscope and the output of the frequency counter OCXO to the other channel and look for slippage of the two signals. If the counter is on frequency, there should be no movement between the two waveforms. This is more accurate than trying to adjust the counter while looking at the frequency display on the counter.
When I first connected it, the HP’s waveform was running backward at a pretty good clip. I adjusted the OCXO until there was no movement relative to the two waveforms. I let it sit like this for about three more hours before buttoning the HP unit back up. I am confident that the frequency counter is accurate +/- 0.3 Hz, which is good enough for my purposes.
What I like about this Leo Bodnar GPSDO is that you can change the output frequency to any value between 1 Hz and 1.4 GHz. The output level is +13 dBm (per data sheet) with low phase noise, making it an excellent portable signal generator. The output is a squarewave, however, installing an LC type bandpass filter such as a Mini-Circuits SBP 10.7+ will round that out into a nice sinewave.
The Leo Bodnar website has a portable Windows executable for download, which can be used to program the output frequency and monitor performance.
I measured the output with my precision power meter; at 10 MHz it was +10.35 dBm. The low power output setting is about +5 dBm.
Another use for the LBE-1420 is as an external 10 MHz reference for test equipment. My Network Analyzer (and many other pieces of test gear) has an external 10 MHz input and if I use the spectrum analyzer to measure carrier or pilot frequency, it is nice to know that the test equipment is exactly on. I confirmed this by measuring the WWV carrier with my Siglent SVA-1032X spectrum analyzer using a long wire antenna.
Continuing with this interesting topic, I purchased a fairly cheap version from Ebay for further research. This particular unit is a clone of a BG7TBL, which is itself a clone. The interesting thing about these units is that they are using recycled OCXOs, which appear to be from decommissioned telecom equipment.
This diagram shows how these units work. The GPS signal is received by the GPS module, in this case, a uBlox M-7. The NMEA sentences and 1PPS are fed into the CPLD (Complex Programmable Logic Device). The NMEA sentences are also available on the RS-232 DB-9 connector.
The CPLD takes the output of the OCXO, in this case, a CTI OC`12SC38A, and compares the 1PPS from the GPS module to the 10 MHz from the OCXO module and adjusts the OCXO module by varying the voltage on the frequency adjust pin to keep it on 10 MHZ. It then sends the corrected 10 MHz and 1PPS signal out to BNC jacks. I found the 10 Mhz output level was +13.58 dBm as measured with my precison power meter. There is a built in bandpass filter, so the output is a good looking sinewave.
Judging by the CTI model number, it was made before 2015. There should be a date code on the bottom of the unit, but I did not feel like unsoldering it.
The one issue with this; OCXOs frequency drifts over time and eventually it will be out of the adjustment range. A closer look at the circuit board shows that it will accept several different OCXO modules. These modules run about $40-60 US new and $10-15 US used.
If an OCXO is suspected of being out of adjustment, they can be measured using the osciliscope method noted above.