Testing the HP 5315B with a Rubidium frequency standard

[JuiceKing]

Inspired by Dave's blog on the topic, I bought one of those decommissioned FE-5860A Rubidium frequency standards from eBay, and put it in a vented plastic box with a switching power supply. The box cover is aluminum and serves as the necessary heat sink. I found an inexpensive HP 5315B on eBay, too, and planned to use the Rubidium standard as its external time base. I'm pleasantly surprised that the 5315B (vanilla model -- no TXCO or oven timebase options) is so stable that with some patience I was able to calibrate it to the final digit. So, it looks like I won't need to use the standard except for periodic tune-ups.

This frequency standard takes about a minute to warm up and lock on the 10Mhz output. Until it locks, the red LED is lit, and its output is unstable.

[djsb]

Do you know of any instructions on how to program the FE-5680A via it's RS232 interface? It would be nice to document the procedure here if possible?
Thanks.

[JuiceKing]

Do you know of any instructions on how to program the FE-5680A via it's RS232 interface? It would be nice to document the procedure here if possible?
Thanks.

Here's the technical documentation, which includes serial port protocols. If you have a version which supports this interface (not all do!) then you program the output. One command will write the new settings to EEPROM, which will eventually wear out if you write too many times, so for frequent updates you should use the command that writes to volatile memory only.

I considered getting an FE-5680A with the option to use this port, but it was considerably more money on eBay. Also, although I can't find the details right now, I do recall reading some cautions that there were limits to the flexibility of this interface--i.e., you couldn't select arbitrary frequencies or that signal quality suffered at lower frequencies, or something of that sort.

In the end, I decided that using the standard as a one-trick pony to discipline other signal/function generators would be best.

[saturation]

There is a large amount of support for many commercial rubidium clocks online, mostly from time-nuts hobbyists. 

Ham enthusiasts seem particularly interested, so you'll find several pages like this:

http://www.ka7oei.com/10_MHz_Rubidium_FE-5680A.html

[ivo.knutsel]

I'd like to do this as well : use the FE-5680A as a 10 MHz frequency standard.  I have bought one via eBay and it works fine in the test setup.

Some simple tests : My HP 33120A has to be set to 10 MHz + 81 Hz to generate the same frequency as the FE-5680A. My PM6666 frequency meter (without fancy crystal oven option) shows 9999928 Hz.

I have two questions :
* I cannot get the 1 pulse per second signal. The 10 MHz shows up at 1 Vpp, the lock signal goes low after warming up but i cannot find a 1 Hz signal on the other pins. How long and what voltage is the one PPS pulse ?
* The unit get quite warm, this is normal and expected because of the oven. Does the unit need a big heat sink to cool it down, forcing the oven to work harder ? Or does it just need some ventilation ?

It draws about 10 W.

[JuiceKing]

I'd like to do this as well : use the FE-5680A as a 10 MHz frequency standard.  I have bought one via eBay and it works fine in the test setup.

Some simple tests : My HP 33120A has to be set to 10 MHz + 81 Hz to generate the same frequency as the FE-5680A. My PM6666 frequency meter (without fancy crystal oven option) shows 9999928 Hz.

Interesting. My HP 33120A has very similar behavior. I've not been able to fix this. The calibration setup will not allow me to change it, even though calibration is unlocked. (Other settings, such as amplitude, can be changed.) I get an "value out of range" error, which suggests that the discrepancy is too large to correct. But even a 1 Hz change up or down gives me that same error, so I wonder if my 33120A is broken. What is your experience? (I can use the FD-5680A as the timebase for the HP 33120A--it locks on and then it just works.)

I have two questions :
* I cannot get the 1 pulse per second signal. The 10 MHz shows up at 1 Vpp, the lock signal goes low after warming up but i cannot find a 1 Hz signal on the other pins. How long and what voltage is the one PPS pulse ?
* The unit get quite warm, this is normal and expected because of the oven. Does the unit need a big heat sink to cool it down, forcing the oven to work harder ? Or does it just need some ventilation ?

It draws about 10 W.

I've not experimented with other pulse settings, but I can tell you that it's normal for the unit to get pretty hot. he eBay seller did put caution labels on my 5680A not to let the unit overheat, so a good heat sink and ventilation are important. There's a brochure from the manufacturer (easy to find with Google) that says the upper operating temperature limit is 50*C. There's another model which will work all the way up to 80*C. They are designed for harsh environments. Even though my FD-5680A is fastened to the aluminum cover of my box, it gets hot in the box--I measured it getting just to 50*C after being on for more than an hour. The box cover doesn't get that hot, so I think I need to improve thermal coupling.

[T4P]

Use some sliver paste/thermal paste as surfaces will never be flat but you can also slap on a heatsink using thermal tape or just bolting a heatsink on

[JuiceKing]

Use some sliver paste/thermal paste as surfaces will never be flat but you can also slap on a heatsink using thermal tape or just bolting a heatsink on

Thanks, this helped. The aluminum cover is now getting up to around 45* C in its hottest spot, but surface temperature of the standard inside the box is still going over 50*C. I ordered tape-on heatsinks to apply to the top of the case. Hopefully that will be enough.

[bingo600]

* I cannot get the 1 pulse per second signal. The 10 MHz shows up at 1 Vpp, the lock signal goes low after warming up but i cannot find a 1 Hz signal on the other pins. How long and what voltage is the one PPS pulse ?

The 1-PPS is just 1uS wide , and it will NOT show if you load the /Lock line (ie. with a led).
The 1-PPS is enabled (gated) via /Lock , and a LED will not let /Lock go to 0.2v for "Low".
 
To use a LED for /Lock use a "Gate" or a transistor to drive it.

To see the 1-PPS on an analog scope should be a challenge i have read , on a DSO it's no problem.

/Bingo