|
Frequency Electronics FE-5680A Ribidium oscillator
|
inleiding
|
Rubidium atomen bewegen op een zeer constante frequentie van 6,834.682.608 GHz. Dit vormt de basis voor een stabiele oscillator.
 De Ribidium oscillator zoals deze ontvangen is van de eBay verkoper.
|
20160920 - This week I bought on eBay a Frequency Electroncics FE-5680A Rubidium oscillator. This is one of the desired components for my hybrid (GPS/Rubidium) timestandard. It is received in good shape and the original circuit board is still attatched to it. It seems the oscillator was cut out of the rest of the printed circuit board. This is the revision B model. Beware! There are different models with different pin connections! Be sure you use the right pin connections before applying power to the device.
I desoldered the wiring from the obsolete printed circuit board underneath the oscillator enclosure. I hooked up the +15 VDC power supply and the oscillator draws current as it should. (1,9 A at startup and 770 mA in steady state.) This looks fine. With the oscilloscope ground attatched to the common ground I measured singnals of the other seven wires. Only one signal is found. This is the lock indicator signal. After a while, the signal goes "low" indicating the oscillator is locked as it should. So far, so good.
Unfortunately the desired 1 PPS and 10 MHz output signal is absent... I read here there's a test point inside. Therefore I wanted to open up the device so I can see if the test signal is present inside. I removed the rivets from the obsolete pcb underneath the oscillator and removed all the screws. (As shown in the EEVBlog video of Dave Jones.) It looked nice but a quite shocking discovery is a loose screw inside! Since there's no screw missing or an empty threaded hole, its likely the screw was accedentally dropped inside during production. (It's the same screw type as used for the outside enclosure.) This seems a serious quality issue at Frequency Electronics Inc... A seal was placed over one of the screws and the device was rivetted to the pcb underneath, the screw was already there when it left the factory.
 Measuring the test point for the frequency test. The "extra" screw is also shown.
 Power supply delivering +15 VDC @ 770 mA and counter showing the test point signal.
20160921 - To verify the correlation between the loose screw and the missing output signals, I checked if there was a 10 MHz signal at the test pont. Unfortunately there wasn't 10 MHz, there is a 8.388.608 Hz signal present. 2^23 = 8.388.608 Hz. It's likely the DDS is programmed to this "odd" frequency. If it's divided 23 times by two, the result is 1 Hz. Probably the signal is used for a digital system since 8.388.608 Hz doesn't make sense unless it's divided by two for digital purposes.
You can imagine I'm not happy. I spent quite a lot of money for this oscillator and it seems the output board is broken since there isn't any rf signal present. Both 1 PPS and 10 MHZ (nor 8.388.608 Hz) are available on the sub-d connector. It's likely the lose screw made a short circuit somewhere distroying the output board. I'll contact the seller soon, maybe he/she has a solution...
I also made an image of the physics package during operation. There's a realy nice purple glow from the Rubidium discharge lamp visible. It's hard to capture this glow, but I guess you see what I mean. (I flipped the insulation material away to get clear view to the physics package.
 The purple light from the Rubidium discharge lamp is visible.
20160923 - I received a notification from the eBay seller that my mesage is received. In the advertisement was said that the device was tested and has a 10 MHz output signal. As said earlier, this is verified otherwise. I'm awaiting the test report and a message of the seller.
|
J1-1 = +15V DC power input (1,7 A warmp/600 mA steady state)
J1-2 = Common ground (10 MHz, power, RS232)
J1-3 = Loop Lock Indicator (high=unlock/low=lock) Max 1mA!
J1-4 = Not used or +5 VDC (80 mA )activates 10 MHz out on pin J1-7
J1-5 = Common ground (10 MHz, power, RS232)
J1-6 = 1 Pulse Per Second (PPS) output
J1-7 = Not used or 10 MHz output (if activated via +5 VDC on pin F1-4)
J1-8 = RS232 RX 9600 baud N,8,1
J1-9 = RS232 TX 9600 baud N,8,1
|
|
levensduur
|
De levensduur van een Rubidium oscillator is ongeveer 243.000 uur = 10.125 dagen ≈;; 27,7 jaar. Symmetricom, een fabrikant van soortgelijke oscillatoren, geeft aan dat de levensduur kan verdubbelen als de temperatuur van de behuizing lager dan 60 graden Celcius blijft. Dus door de behuizing te koelen kan de levensduur verlengd worden tot 591.000 uur = 24.625 dagen ≈; 67,4 jaar. Een stabiele temperatuur komt de stabiliteit van de oscillator ten goede. Montage op een koelblok met temperatuur geregelde ventilator zou een goede oplossing zijn om de levensduur te verlengen.
Een OCVCXO heeft temperatuur nodig om stabiel te blijven, dus daar heeft warmte een functie. In het geval van een Rubidium oscillatr is warmte een bijproduct dat juist niet gewenst is...
|
- Brand: Frequency Electronics Inc. (FEI)
- Model: FE-5680A
- Output frequency: Programmable from 1 Hz...20 MHz with a resolution of <1E-12 Hz
- Signal type: Sinusoidal
- Amplitude (minimum): 0,5 Vrms into 50 Ohm (+7dBm)
- Adjustment Resolution: <1E-12 over range of 3,8E-5
- C-field potentiometer resolution: 1E-11 over range of 3E-9
- Drift: 2E-9/year / 2E-11/day
- Short Term Stability: 1 s < 100 s: 1,4E-11
- Retrace: 5E-11
- Phase Noise (fo = 10 MHz): -100 dBc @ 10 Hz, -125 dBc @ 100 Hz, -145 dBc @ 1000 Hz
- Input Voltage Sensitivity: 2E-11 / (15...16 VDC)
- Frequency vs. Temperature: ±3E-10 (-5...+50 °C)
- Spurious Outputs: -60 dBc
- Harmonics: -30 dBc
- Loop Lock Indication: > 3 VDC = unlocked, < 1 VDC = locked
- Input Power (@ 25°C): 11 watts steady state, 27 watts peak (startup)
- DC Input Voltage/Current: 15...18VDC @ 1.8A peak (startup) and 700 mA steady state
- Ripple: +15 VDC: <0.1 Vrms
- Warm-up Time: < 5 minutes to lock @ 25°C
- Size: 25 mm thick, 88mm wide and 125 mm deep
- Weight: 434 grams
|
|
nieuwe Rubidium oscillator aanschaffen?
|
De doelgroep van deze website is hobbyisten in elektronica en radiotechniek. Helaas is het budget voor hobbyisten klein, dus ligt de focus op nieuwere goedkopere apparatuur of duurder oudere apparatuur dat betaalbaar is geworden voor hobby gebruik. Een Rubidium oscillator heeft een beperkte levensduur (lees: ongeveer 20 jaar), dus na een zekere periode kan de betrouwbaarheid niet meer worden gegarandeerd en is het tijd om dergelijke professionele apparatuur af te danken. Deze afgedankte apparatuur (zoals in dit geval uit basis stations van mobiele telefonie posten) is betaalbaar geworden voor hobbyisten.
Om een beeld te geven van de nieuwprijs van dergelijke apparatuur heb ik in de markt gekeken wat de actuele prijs is voor een Rubidium oscillator. Ik ben zo bij ACAL BFi Nederland in Eindhoven terecht gekomen die Rubidium oscillatoren van Stanford Research Systems leveren in Nederland. De sales dame heeft mij van twee prijzen voorzien dat een indruk geeft van wat een actuele (anno 2016) prijs is voor dergelijke apparatuur.
Een Stanford Research FS725 is een Rubidium oscillator met distributie versterker ingebouwd. Dit is een kant en klaar apparaat dat direct gebruikt kan worden. Een richtprijs is € 2.800,00 inclusief transport, invoerrechten en verzekering. Het bedrag is wel exclusief btw.
Een Stanford Research PRS10 is een Rubidium oscillator module. Omdat het een module is, moet deze nog ingebouwd worden in een apparaat zodat de module van voedingspanning kan worden voorzien. Een eventuele distributie versterker, deler en/of filtering van het signaal kan dan naar wens worden ontworpen en aangesloten. Een richtprijs is € 1.500,00 inclusief transport, invoerrechten en verzekering. Het bedrag is wel exclusief btw.
Voor de volledigheid: Ik krijg geen geld of andere tegenprestatie voor het vermelden van een leverancier of fabrikant. Zie het niet als reclame maar als blijk van waardering voor het leveren van een bijdrage (informatie) voor deze website.
Rond ongeveer 2013 zijn dergelijke Rubidium oscillatoren via eBay op de markt gekomen voor ongeveer € 60,00. Helaas worden deze gebruikte oscillatoren schaarser en lopen de prijzen nu in 2016 al snel op richting € 200,00. Een gebruikte kost dus ongeveer één tiende van de nieuwprijs. Dit is inmiddels betaalbaar geworden voor een hobbyist waarbij de factor betrouwbaarheid minder relevant is. Wanneer betrouwbaarheid (zoals voor bedrijven) van belang is, is het aanschaffen van een nieuwe oscillator geadviseerd.
|
|
output frequency programming
|
serial connection
In my case the RX and TX of the 9-pin sub-d connector are not in use. The 5 pins of J2 (in the corner of the DDS print) do work fine for programming. If the pin near the corner is pin 1, the pinout is as follows: 1=+5VDC, 2=+5VDC, 3=data RX (from pc), 4=data TX (to pc), 5=ground. Pins 1 and 2 are not needed for programming. 9600 N,8,1 and flow control "none" works great with Puttytel.
settings readout
Open your communication program (like Puttytel) using settings: 9600 N,8,1 flow control = none.
Press [Shift] + [s] following by [Enter]. This should send (capital) "S" to the RFS. The expected response (for 8,388.608 MHz) is:
R=50255055.059909Hz F=2ABB5050B316D200
OK
"R" shows the intermediate frequency. "F" is the hexidecimal code representing the desired output frequency. For example:
F=32F0AD9D0D00D200 will result in 10,000.000 MHz output frequency.
F=2ABB5050B316D200 will result in 2^23 Hz = 8,399.608 MHz output frequency.
frequency calculation
Some FE-6780A models can be programmed. As read on the internet, there are fixed frequency models and there are models which can be programmed between 20 Hz and 20 MHz. (Although using above 15 MHz is not advised by some people.) All the devices are marked FE-5680A, but it’s likely that there are several different versions.
I’ve got the “Revision B” model, which can be programmed. (Recognizable by the weld marks at the bottom and “Rev. B” marking on the label.) The desired frequency can be set by the serial connector. The programming routine is described further below. The first step is to calculate the hexadecimal division code “R”. The desired frequency in Hz, multiplied by 2^32, divided by the internal intermediate frequency “R” in Hz, will result in a decimal number “F”. If this number “F” is converted to a hexadecimal number, the result is the first eight digits of the “F” code. (The last eight digits should be zero’s.) The calculations below will probably clarify the calculation above.
For example: If the desired frequency is 10 MHz. This frequency in Hz should be multiplied by 2^32 (=4.294.967.296). The result is 42.949.672.960.000.000. This number should be divided by the intermediate frequency of 50.255.055,059.909 Hz. The result is 854.633.885,263.875.217.482.629.351.332.58. This large number should be converted to a hexidecimal numer. The result is 32F0AD9D. It’s not possible to convert decimal numbers to hexidecimal numbers. This will result in a relatively small error. The conversion error is shown next to the calculation. If you want a gps receiver with 10 kHz clock pulse replaced by a FE-5680A, the RFS can be programmed to 10 kHz. This is nod advices since the relative error is approximately three times bigger compared to the 10 MHz signal. An external divider (like the 74HC390 chips) can divide the (more accurate) 10 MHz signal to 10 kHz without losing accuracy.
The “F-code” which is needed for programming the FE-5680A exists of 16 digits. The first eight digits represent the division number from the calculation. For 10 MHz are the digits “32F0AD9D”. The last eight digits are zeroes. The “F-code” for 10 MHz is therefore “32F0AD9D00000000”. The “F-code” for 10 kHz is “000D0A6900000000”. And the “F”-code” for 8,388.608 MHz is “2ABB505000000000”.
settings programming
If you want to reprogram the device because another output frequeny is desired, you first need to calculate the desired hexidecimal "F" code. An output frequeny between 20 Hz and 15 MHz should be fine. Since it's likely you need 10 MHz, the "F" code is given: 32F0AD9D0D00D200.
Send the following line (mind the capital characters): "F=32F0AD9D0D00D200" and press [Enter]. The expected response is:
R=50255055.059909Hz F=32F0AD9D0D00D200
OK
This means the information is received in good condition. To write the new settings into the memory press [Shift] + [e] following by [Enter]. On other words send "E" to the RFS. To abort, don't send "E", but disconnect the power from the RFS. If desired, you can send "R" again to verify the programming.
|
|
