FE-5680A Rubidium standard waveform

[TheDefpom]

Hi all,
I picked up a rubidium standard with a 10MHz output, it outputs ok but noticed that 2 of my frequency counters struggle to read its frequency correctly.

One of them has a level adjustment which will read correctly if set to a level just before it drops off. The other cannot read it properly at all.

I put it onto my nice new SDS2102 (thanks tautech) and can see ripples on the waveform when using runt triggering in single shot mode, in auto triggering you can see them pop up as glitches from time to time on a nice sine wave, these ripples seem to be what is causing the reading difficulty.

I was wondering if anyone else here had much experience with this kind of unit and its output waveform quality? I suspect this unit is slightly faulty, it does actually activate its lock output once it's warmed up.

When I get home later I will have a go at grabbing a waveform screen shot and posting it here.

[Vgkid]

Did you give it time to warm up first.

[TSL]

FE5680 FAQ here...

http://www.ko4bb.com/dokuwiki/doku.php?id=precision_timing:fe5680a_faq

If its not locking, the frequency will wander about, Pin 3 on the DB9 should go low if its locked.

It generally takes 5 - 10 minutes to lock, provided you have +15v as Vcc. Some units will operate lower.

Current draw will drop from ~1.7A to ~0.6A once its warmed up.

See the FAQ for more detail.

[TheDefpom]

HI,
Yes as I stated in my post, it does lock once it has warmed up (current draw did drop also as noted), it is locked when my frequency counters have trouble reading its output frequency, and when I found the ripple on the sine wave output.

I have 4 counters, two could read it (lower speed counters), 2 struggled of 1 could not read it.

The counters that could read it are just cheap ones, with low bandwidth, although one is an AADE 2.5GHz unit with TCXO, using its HF input which drops off at about 40MHz, effectively filtering out high frequencies.

My Philips PM6676 1.5GHz could read the 10MHz signal, but only if I turned the input level down to the brink of losing it, then it would be fine (that one is only 0.3Hz off :-).

My HP 5384A sees 10MHz, but doesn't stabilise, jumping around between 10.000.05xx-10.000.3xxx regardless of input level and trigger settings, using other sources it works fine.

I can see the ripple on my scope within the sine wave output, i will set it up again later and grab a shot of it and post it here to help explain it.

I think what is happening is that there is some higher frequency noise sporadically appearing on the output and adding to the sine wave, causing the faster counters to count the ripple noise as well as the normal sine wave output.

[edpalmer42]

What's the output impedance of the 10 MHz output?  Are you terminating it properly?

Ed

[TheDefpom]

Thats a good point Ed, it was just connected to the DB9 output wire using a scope probe to connect the counters to it, the counter connections have 1MOhm inputs on the A channel inputs, which I used, I am pretty sure I connected to the B channel on the more troublesome unit which is 50 Ohm, but that didn't help, but as it was a wire connection (with ground clip to 0V side of unit) it probably wasn't loaded with true 50 Ohms.

I am in the middle of re-arrangeing my test bench, once that is all done I will grab that screen shot.

[Vgkid]

On my unit(using a 5335a) with a 1Meg input impedance it will.often read 20Mhz.nwith a 10x probe it will read correctly, and at a 50ohm mode it is correct. When using a straight probe, it is possible to get it to trigger correctly, just really tricky.

[edpalmer42]

What output level are you seeing?  Maybe the output is bad and it's almost all noise.

Ed

[TheDefpom]

From memory it was about 3.6V RMS.

[edpalmer42]

From memory it was about 3.6V RMS.

Really?  If that was into 50 ohms, it's +24 dBm (!!).  Even into 1 Mohm it would still imply a 50 ohm output of +18 dBm.  The 5680A spec lists a standard output of +7 dBm, but there are so many versions of that stupid unit that anything's possible!

Ed

[TheDefpom]

Hang on, I might of had the scope probe range set wrong on the scope... vs probe switch... so maybe .36V RMS, that sounds more likely !

[TheDefpom]

Right, I think it was a false alarm... after building on a small 5V regulator and powering the entire unit from a single supply, I found that the small power supply I was using was generating noise, I noticed it when I could see the glitches on the scope before connecting up the power to the rubidium unit... so I connected to a different (linear) power supply, all good!

Output is 190mV p-p into 50 Ohms, or 390mV p-p into 1M Ohm.

The Philips PM6676 counter which I picked up on eBay recently for US$175, is showing 10.000.0002MHz (it has the oven option).

The HP 5384A which I picked up for US$70 is off a bit, but is at lease dead stable now, it also has an oven option so I will just need to adjust it, it shows 10.000003116MHz.

I also scored a Fluke 8842A with the AC and GPIB options for US$150, that is working a treat, I just need to check it out more fully, so far it agrees with my 2.048VDC and 4.096VDC voltage references (within .002V).

[TheDefpom]

Well I just calibrated the HP... reads 10.000,000,000 now... drifting very slightly, +-0.01 Hz, well happy.

[edpalmer42]

Sounds good.  Your output level is a little low, but with the 5680A, anything goes!

If you've got a GPSDO or know someone who does, it would be worthwhile to compare your unit to the GPSDO so you can tweak it to be perfect.

Ed

[TheDefpom]

Although it satisfies my OCD to have it read all those 0000's in reality I only need accuracy to 1Hz so it should be good enough when calibrated from the Rubidium reference.

I was very pleased with the Philips counter, it was last cal in 2008 according to its sticker!, the HP was last done in 2005, still, I now have some very precise counters to use, which I can be confident in, prior to getting the Rubidium I was using a 10MHz Tcxo which had been calibrated to a Rubidium about 7 years ago, when I bought it from AADE, it was about 11Hz out after that long, not too bad for what I need.

[edpalmer42]

Actually, I meant compare the 5680A to the GPSDO to make sure that it's on frequency.  Depending on which of the endless versions you ended up with, some can be digitally programmed for frequencies other than 10 MHz.  It would be frustrating to discover six months from now that everything you've measured is 100 Hz off because the 5680A was programmed for 10.0001 MHz.  Remember that we have no idea where these things came from or what they were used for.

Ed

[TheDefpom]

You make another good point Ed. :-)