-
Found home made rubidium reference that seems to really work well
Posted by notfaded1 on 12 Feb, 2020 21:56 -
I recently acquired this rubidium standard that came from an estate sale. I haven't opened the wood box inside yet to see what's inside. It may have taken a few hours to slowly reach temp and then lock up (it didn't lock in half hour so I went to watch TV a while). It looks like after running for 25 hours it's stabilized around 86.7°F on it's temperature gauge. Attached are some pictures I was curious if anyone had an idea what's inside the wood box? It has 5 MHz square, 10 MHz square, and 10 MHz sine wave outputs on the front. From initial observation it's very stable once it locked. Before it locked the frequency wandered up and down more. I'm going to make some better measurements over time.
I was mainly wondering if seeing this rings any bells for anyone on what it might be. Since it's working really well I'm more hesitant to tear the wood box apart with everything mounted to the wood on the outside. My curiosity is going to get me soon anyhow because I can't not look inside the wood box to see how the rest is built and out of which parts. I only have two good working rubidium references so you can see my dilemma.
It has a fan mounted inside and vents on both sides but it doesn't seem to be too warm to me and require turning the fan on. There's a manual switch inside I think's for the fan. A sticker on top lists a measurement Sept 7, 2004 5,000,000.004 Hz @ 66.5°C / 151.7°F. the gauge on the front only is reading 85.8°F but after 20 hours is still rising slowly. It may slowly reach a higher temp. I would think since it's locked up now could there be a good reason to want it higher? Wouldn't you usually prefer the lowest temp that achieves lock?
Perhaps until I know for sure the package inside it's hard to say for sure?
Bill
-
Locked pic the LED is red until locked
-
Running also my 53131A has the 010 HS oven.
-
More pics
-
Pic I'm also hoping this reference will make my 53131A much better as a good external reference. Comparing to a new GNSS gpsdo I ordered from BG7TBL and also comparing to symmetricom syncserver gpsdo outputs I'm thinking my quest for timenuthood will be one step further along!
Last weekend I loaded up labview 2020 and it's licensed so between that and using the raspberry pi image I got from @branadic and my 82357B or GPIB-USB-HS hooked to a pi3 I'm hopefully on track to collecting and graphing better stats. I'm finding out windows 10 and labview sometimes acts kinda funny. Weird timing errors and stuff with GPIB that's hard to diagnose. The pi seem to just work and seem maybe more stable... Maybe simple is sometimes best!
The ultimate goal is more stable digits and less uncertainty to get a standard that can calibrate everything else.
Bill
-
I almost bought it just to find out what was in the wooden box!
I couldn't decide if the purpose of the box was electrical or thermal isolation, or both. The fact that it takes hours to lock suggests that the box is a fix for something that's broken. Most Rb standards lock in 20 minutes or less. Also, the temperature of 86F makes no sense at all. Nothing in an Rb standard runs near that temperature. I see the label says 60C. That could be the temperature of an OCXO.
In general, it's not a good idea to add thermal insulation to any oscillator. The manufacturer designed it with a specific value for heat dissipation to the ambient. If you upset that by adding insulation, you upset the design. Worst case, the oven temperature just keeps going up until it fries itself due to the oscillator's normal power dissipation.
Avoid drafts and sunbeams, but don't add insulation to an oscillator unless you're the manufacturer.
Ed -
Pic I'm also hoping this reference will make my 53131A much better as a good external reference. Comparing to a new GNSS gpsdo I ordered from BG7TBL and also comparing to symmetricom syncserver gpsdo outputs I'm thinking my quest for timenuthood will be one step further along!
Last weekend I loaded up labview 2020 and it's licensed so between that and using the raspberry pi image I got from @branadic and my 82357B or GPIB-USB-HS hooked to a pi3 I'm hopefully on track to collecting and graphing better stats. I'm finding out windows 10 and labview sometimes acts kinda funny. Weird timing errors and stuff with GPIB that's hard to diagnose. The pi seem to just work and seem maybe more stable... Maybe simple is sometimes best!
The ultimate goal is more stable digits and less uncertainty to get a standard that can calibrate everything else.
Bill
the inside maybe is FE5680A, you can use it as reference to test your GPSDO, as my testing, the GPSDO's frequency offset may exceed your expection. -
I'll have to take it apart... I don't see much reason to build a big wood box to insulate an FE5680A the way this was done. It's a big box for that and I thought usually those are just mounted on a metal heatsink often so it doesn't get too hot when mounted in an enclosure.
I was hoping someone might recognize the style of build and have a really good idea what someone put inside. I'm going to see how high the temp gets to see if it gets closer to the temp written on the old label. There are a bunch of BIG screws holding the wood box closed and the wood looks to be 3/4 to an inch thick! This things pretty heavy too. -
Looks like a native 5Mhz unit and a frequency doubler on top of the wooden box make a 10 Mhz output.
-
It looks like it shouldn't be difficult to open up the wood box without damaging anything, the screws are exposed.
Regarding adding insulation, I'm not an expert on oscillators but I do recall seeing a teardown of a double oven oscillator that had an off the shelf OCXO within an additional oven, had the normal labels on it and everything and I think the whole assembly was HP. The oven controllers are closed loop so if you add insulation it shouldn't cause the thing to get any hotter. I suppose it's possible that some may have issues, I don't know. -
It looks like it shouldn't be difficult to open up the wood box without damaging anything, the screws are exposed.
Regarding adding insulation, I'm not an expert on oscillators but I do recall seeing a teardown of a double oven oscillator that had an off the shelf OCXO within an additional oven, had the normal labels on it and everything and I think the whole assembly was HP. The oven controllers are closed loop so if you add insulation it shouldn't cause the thing to get any hotter. I suppose it's possible that some may have issues, I don't know.
That oscillator is somewhat (in)famous. It's a slightly modified version of the HP 10811. There are a few on the auction site right now. There are two stories regarding why the outer oven was added. Whatever. The point is that it was the manufacturer who went through the effort to determine that the extra insulation and oven winding could be safely added. That's totally different from one of us amateurs throwing on some extra insulation. If the internal heat generated by the oscillator can't get out fast enough, the temperature will rise. By adding insulation, you're reducing the rate that heat can escape.
Ed
-
I almost bought it just to find out what was in the wooden box!
I couldn't decide if the purpose of the box was electrical or thermal isolation, or both. The fact that it takes hours to lock suggests that the box is a fix for something that's broken. Most Rb standards lock in 20 minutes or less. Also, the temperature of 86F makes no sense at all. Nothing in an Rb standard runs near that temperature. I see the label says 60C. That could be the temperature of an OCXO.
In general, it's not a good idea to add thermal insulation to any oscillator. The manufacturer designed it with a specific value for heat dissipation to the ambient. If you upset that by adding insulation, you upset the design. Worst case, the oven temperature just keeps going up until it fries itself due to the oscillator's normal power dissipation.
Avoid drafts and sunbeams, but don't add insulation to an oscillator unless you're the manufacturer.
Ed
I do notice that the wood box isn't sealed. I'm going to open it once I figure out how high this temp gauge goes up to... also there is a fan in there for some reason??? It hasn't turned on by itself there's an extra switch inside by the power supply you can see where the power comes in.
I can say this... once it locked it's been very stable on the counter... you're right about my other rubidium standard Ed. The Symmetricom Datum locks in 15 minutes or less. On another note I got the seller to take less than he was asking and it was reasonable so it was worth a shot. Since it did lock I'm pretty happy with it really so far!
I'm still working on the Datum Cesium. I think I've got some bad caps in the power block on the back of the unit. The next step is to check every one with the DE-5000 for ESR. I'm not exactly sure how long it took to lock because it didn't in a half hour last night and went and watched TV for a while and when I came back a couple hours later it was locked. I'll have to time it from cold to be sure. I'm guessing that's why someone put the clock on the front with the temp gauge... maybe to watch how long it took to lock since it's starts at 12:00 when you turn it on and counts like a regular clock. Or maybe that's just the temp gauge he had and just used it. It does have a nice built in backlight! 
The temp of the whole metal box is mildly warm... it's not hot by any standard so the now 86.8 degrees F on the front is probably close to correct.
-
Got any data to support this? I'm not saying it's not possible, but I'd like to know if it's backed by anything or just speculation because unless you've got such amazingly high performance insulation that the thermal loss is lower than the minimum power the oven controller is capable of going to (which I'd expect to be approaching zero) then I don't see how the temperature is going to continue to rise. Certainly the handful of milliwatts consumed by a typical oscillator itself is not going to do it. I just looked at a datasheet for a random Crystek OCXO and that lists a max ambient temperature of 80C, I would be absolutely shocked if you could insulate any similar oscillator to the point that the case temperature reaches 80C.
While anecdotal at best, I've never had any issues with oscillators in high ambient temperatures. I haven't specifically tried adding insulation to a standard OCXO but I have had them running inside gear that gets quite hot. Either way given surplus OCXOs can readily be found for <$30 it's not exactly a major risk to try. In fact I have an old Datum 10MHz OCXO here I might tape a thermocouple to and pack it in styrofoam to see what happens. -
More pics from the original listing... also some of labels inside were labeled heater. Also it's got an awful lot of trimmers on there with wires going into the box???
Bill
-
If you look at that last pic I think maybe those few screws on the side where the cables go in might be the place to try and open it up without interfering with all the other stuff mounted on the top and sides... a lot of the circuit boards overlap the box being screwed together with their own screws o.O!
-
I would unscrew the two boards on the top, move them out of the way, and then remove the top of the wooden box. That should give best visibility of whatever's inside.
-
Well let's do a little experiment, I don't really know what the result will be but it's easy enough to try. First I wired it up to power in free air and monitored the output while it warmed up, after about 10 minutes the case temperature seemed to have stabilized at around 97F, drifting around +/- a degree or two with drafts. I wasn't able to find any suitable foam that I could use to insulate it so I wrapped it up in an old towel and stuffed that into an insulated cooler. This is probably much better insulation than would be practical to include inside an instrument.
I'll check the temperature on my way to bed tonight and again in the morning before I head to work and see what it's gone up to. I suspect it will stabilize at about as hot as it's going to get within 30-40 minutes but we'll see.
-
James,
The other thing that would be interesting to measure is the current drain. The oscillator will have a constant drain and the oven will vary with ambient temperature. Unfortunately, few oscillators have seperate power pins for oscillator vs. oven. You'd expect that as you add external insulation, the oven should draw less current. If the oven shuts off entirely, that means that the temperature is uncontrolled i.e. starting to rise. If that happens, you might be able to see a shift in the frequency. -
This one does actually have separate pins for power, although I tied them together for this experiment. I can try separating them and hook up meters to monitor the oven and oscillator current tomorrow.
-
Another thing I noticed since the temp went up a little more while locked... it looks like the last digit on my meter has now settled to varying between two digits instead of 4 or 5. It's almost rock solid and that variation might even be just the drifting of my counters built in 010 option oscillator possibly. I've only had this counter for a couple weeks so I'm still getting used to it. I do like the features it has over my others. I like that it does time interval and other stats vs. just counting like some counters do.
-
James,
Somehow I missed a couple of messages in this thread. Time to catch up.
You asked if I had any data to support the idea of burning up the oscillator by adding extra insulation. No, I don't. I think I read about this on Time-Nuts, but it was probably also listed there as a potential worst case. It could even have been a discussion of putting an oscillator in a Dewar flask or in a vacuum! Anyway, even if it happened, it wouldn't burst into flames. But it might cook itself to death over a longer period of time.
Far more likely is the scenario I mentioned a couple of messages back where the temperature rises high enough that the oven either shuts off or becomes unstable. Whether this happens or not will depend on the amount of extra insulation and the design of the oscillator or oven controller. I would also suspect that older oscillators would be more susceptible to problems due to higher power drains and perhaps less sophisticated designs.
It also occurred to me that SC-cut crystals typically have a turnover temperature that are higher than an AT-cut crystal. Does your Datum oscillator use an SC-cut crystal? That might be why it can run in an 80C ambient. That would also mean that oscillators equipped with AT-cut crystals might have more performance-related problems with overheating since the crystal could be running hotter than the turning point.
Ed
-
Some very positive initial observations comparing my we don't know what yet box to my Datum 8040 Rb
It says Datum on the front and has Symmetricom sticker on the top.
Calibrated by Rohde & Schwarz 2015-06-29
2/13/20
Measurement device: Agilent 53131A w/010 option
Unknown Rb standard
shows temp 87 degrees F on front today
05MHz sq. 4,999,999,808 Period .200,000,007,7
10MHz sq. 9,999,999,60 Period .100,000,003,9
10MHz sin.9,999,999,60 Period .100,000,003,8
Datum 8040 Locked in < 5 min.
10MHz ?? 9,999,999,60 Period .100,000,003,9
That can't be all bad Ed? I ordered 3 equal length quality BNC cables from Pamona Electronics for better comparison tests against each other. I need to put the Datum 8040 on the scope to see if it's a square or sine wave... it doesn't say on the back.
The fact that they agree so closely is a good sign I think.
When the new cables come I'll do the port 1-2 comparisons with same exact cables and get some better data. I've got the 53131A working in Labview 2020 but need to figure out how to take the output and with multiple samples and feed it to the analysis and graphing functions. I may fall back on the pi3 if it takes too long. I need good graphs now that I have the pieces all in the right places.
Thanks for all your help Ed... you've been a good asset!
Bill
-
Some very positive initial observations comparing my we don't know what yet box to my Datum 8040 Rb
It says Datum on the front and has Symmetricom sticker on the top.
Calibrated by Rohde & Schwarz 2015-06-29
2/13/20
Measurement device: Agilent 53131A w/010 option
Unknown Rb standard
shows temp 87 degrees F on front today
05MHz sq. 4,999,999,808 Period .200,000,007,7
10MHz sq. 9,999,999,60 Period .100,000,003,9
10MHz sin.9,999,999,60 Period .100,000,003,8
Datum 8040 Locked in < 5 min.
10MHz ?? 9,999,999,60 Period .100,000,003,9
That can't be all bad Ed? I ordered 3 equal length quality BNC cables from Pamona Electronics for better comparison tests against each other. I need to put the Datum 8040 on the scope to see if it's a square or sine wave... it doesn't say on the back.
The fact that they agree so closely is a good sign I think. When the new cables come I'll do the port 1-2 comparisons with same exact cables and get some better data. I've got the 53131A working in Labview 2020 but need to figure out how to take the output and with multiple samples and feed it to the analysis and graphing functions. I may fall back on the pi3 if it takes too long. I need good graphs now that I have the pieces all in the right places. I want to get the good aDEV graphs going. Once I get that down I'll try triple hat comparison method. I got shipping notice on the GNSSDO from BG7TBL and it's coming DHL air mail because I don't want to wait much longer as it's already been a while due to their holiday delay and the virus!
Thanks for all your help Ed... you've been a good asset!
I want to spend some more time on the 4040A Datum too... busy at work and busy at home o.O! Oh my!
Too much cool stuff to figure out.
Bill -
Some very positive initial observations comparing my we don't know what yet box to my Datum 8040 Rb
It says Datum on the front and has Symmetricom sticker on the top.
Calibrated by Rohde & Schwarz 2015-06-29
2/13/20
Measurement device: Agilent 53131A w/010 option
Unknown Rb standard
shows temp 87 degrees F on front today
05MHz sq. 4,999,999,808 Period .200,000,007,7
10MHz sq. 9,999,999,60 Period .100,000,003,9
10MHz sin.9,999,999,60 Period .100,000,003,8
Datum 8040 Locked in < 5 min.
10MHz ?? 9,999,999,60 Period .100,000,003,9
That can't be all bad Ed?
Yes, those numbers seem quite reasonable. It will be interesting to see what's inside the box.QuoteI ordered 3 equal length quality BNC cables from Pamona Electronics for better comparison tests against each other.
I don't see the need for equal length cables. Remember that the sources aren't synchronized, so their phases are random. Also, a difference in cable length has no effect on frequency and just adds a constant offset to phase measurements. There's no effect on typical timing calculations/graphs/analysis.QuoteI need to put the Datum 8040 on the scope to see if it's a square or sine wave... it doesn't say on the back.
When you're making these timing measurements, always use square waves. If your source is a sine wave, you have to square it up before you measure it. Sine waves are guaranteed to give you poor results due to inevitable noise riding on top of the signal. This starts to get into the area of a 'black art'. At these performance levels, it isn't just a matter of throwing in a Schmitt trigger inverter.
You should also consider whether you need to low pass filter the input to remove high frequency noise before the squaring stage. I have a couple of 100 KW FM broadcast stations nearby that are always present - even with double-shielded coax cables. I've sometimes seen my results improve by 2x just by adding filters to the inputs.QuoteThe fact that they agree so closely is a good sign I think.
When the new cables come I'll do the port 1-2 comparisons with same exact cables and get some better data. I've got the 53131A working in Labview 2020 but need to figure out how to take the output and with multiple samples and feed it to the analysis and graphing functions. I may fall back on the pi3 if it takes too long. I need good graphs now that I have the pieces all in the right places.
There isn't a lot of benefit going to Labview when Timelab supports the 53131A natively and includes the most common timing calculations. Labview will become useful when you need to collect data from other devices or multiple devices simultaneously.QuoteThanks for all your help Ed... you've been a good asset!
Glad I can help!
Ed
-
Ok well an update on the Datum quartz OCXO experiment. Last night right before I went to bed I checked and it was 127F. This morning when I got up it was 134.4F and when I got home from work it was 136.2F so to my surprise it is still rising albeit quite slowly. I'm going to let the experiment go a bit longer than I had planned in order to see if it levels off.
I suspect what is happening is that the temperature of the housing is approaching the set temperature inside the oven, unfortunately I don't know what temperature these typically operate although I found mention online of 70-80C which is 158-176F, I suspect mine is not that hot inside though or the outside would be hotter by now than it is.
It's too early for a firm conclusion but at this point I'm going to say I strongly suspect that adding a modest amount of insulation to a quarz OCXO is *probably* fine and will not result in thermal runaway. Putting it inside a Dewar flask is another matter, that is going to be dramatically more effective insulation and losses will be limited to radiated heat and the small amount conducted through the leads. Whether or not it helps with stability I don't know, maybe someone who is set up to measure that can try a similar experiment. I would not suggest doing this on something safety critical, something you're sending up into orbit or your irreplaceable family heirloom oscillator but a cheap surplus OCXO? If you want to give it a shot why not? A few cm of foam may reduce the influence of drafts and room temperature swings without keeping in all the heat, it may work just as well though to simply put it in a closed box to keep it away from drafts although insulation may be of more benefit if the room temperature swings widely. As with anything the usual disclaimer applies, YMMV.
I would also not recommend doing this with a rubidium oscillator, they have parts that run quite a lot hotter and parts that need to be kept cooler, I recall a datasheet for one that said it should be mounted on a metal plate.