EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: linux-works on March 18, 2014, 07:56:30 am
-
here's the start of an Rb frequency standard for my lab. finally motivated by gerry sweeney's video, I bought one and got it running on the bench tonite. some early pics:
(http://farm4.staticflickr.com/3686/13237639975_8365d2bcaf_o.jpg)
(http://farm8.staticflickr.com/7008/13237679525_d331e53728_o.jpg)
after a short delay (that's a story in itself) the ebay Rb box arrived. it was not at all ready to be used, though; lots of things had to be done. on the pcb photo you can see a short length of coax. that's the 10mhz (or tuned freq) sine out. there's also a 3 wire ribbon cable going to a db9 female. this Rb box did not have the courtesy to supply me with a pin2/3/5 output and so I had to run one myself. as usual, its real rs232 levels (not TTL) and so you go to a db9 and then to an FTDI serial/usb dongle and into a pc. run any term program you want (9600 baud) and type right F=deadbeef stuff (you have to calculate your own deadbeef hex code) and you'll see the new freq on the coax-out line.
no external 5v needed, at least that was done internally.
interestingly, you don't need 15v for this. 12v runs it *just fine* and at lower temp and current draw. also interesting is that I was running a fan on this and while the fan was keeping it cool, the current draw was HIGH! when I turned the fan off the current dropped almost immediately. it seems that you don't want to just blow 'lots of air' at this; you want some air but actually not very much. I'll find what that right amount is via experimentation and observation.
that's it for tonite. it does work and I was able to change it from the default 8.mumble mhz to a proper lab 10 without too much trouble. you do have to open the chassis and do stuff, so thas was not really turn-key like some of them are.
(more to come later)
-
(more to come later)
:clap: and thanks for sharing this.
I've been drolling to have this as well for quite a while, but I know for sure I DO NOT need it. :'( :palm:
-
also interesting is that I was running a fan on this and while the fan was keeping it cool, the current draw was HIGH! when I turned the fan off the current dropped almost immediately. it seems that you don't want to just blow 'lots of air' at this; you want some air but actually not very much. I'll find what that right amount is via experimentation and observation.
Thanks for sharing this. I'm currently working on a GPSDO but might get a Rubidium as well.
I'm not surprised that the fan increased the current draw. My understanding is that the Rubidium lamp needs to be heated to something like 105C so if you cool it with cold air the heater has to work that much harder to reach temperature.
It is similar to an OCXO - I am using one for my GPSDO and I found by experimentation that if I insulated it well the current draw dropped from 300 mA to less than 200 mA, from discussion on this forum I realise that it is not a good idea to over do it else the oven control goes awry but it seems ok so far.
I guess what you want to do is insulate the lamp bit very well and cool the rest of the circuitry.
-
I've been drolling to have this as well for quite a while, but I know for sure I DO NOT need it. :'( :palm:
Different captain, but same boat here!
-
All Rb standards have to be operated regularly with closed case, and with a big cooling plate on the bottom.
The two ovens in the physics package operate at 74°C and 106°C, or so, but the electronics around should be cooled by the plate to < 70°C, otherwise damage will occur. Quite opposing requirements, but it's working. Lifetime of > 10 years is usual.
Btw.: all the GPS satellites rely on Rb clocks nowadays.
Visit time-nuts for detailed discussion of setting up Rb and GPSDO in one case, and for thermal management problems..
I did it that way, anyhow, i.e. both devices in a 2 units 19" rack case. Disadvantage is the heating of the GPSDO by the Rb, and thereby decreasing stability of the OCXO in the GPSDO a little bit.
Anyhow, good for 1e-12. And it's very handy for calibration of the Rb by the GPSDO, afterwards using the Rb only (better short/mid term stabilities).
Frank
-
I've had mine just laying around for quite a while (you may be able to tell from the dust in the photo below). I've recently made some progress though, I'm starting at the output by making a VCXO PLL to clean up the phase noise/spurs on the output of my Rb, and a distribution amplifier.
For later, I have a surplus LEA-4T GPS module and a cheapo FPGA development board laying around that will make up a digital PLL to lock the Rb to GPS for long term stability. Finally I want to add an ARM board to run as NTP and PTP (1588) server for my home network. Then I'll finally be able to tell the time with some confidence :phew:
Overkill? Never heard of the term... :D
-
the local NTP server idea is a good one. maybe instead of an arduino for correction I'll just put a rasp-pi board in there, have it do the gps-do stuff and it can also serve time and even have a web interface, just for grins.
-
I have two GPSDOs running and my house standard is a Rb oscillator. I have more of both in my lab. Yes, I am a time-nut! ;D
I've never considered linking the Rb to GPS. There's really no point. The only real reason to link them is if GPS is so unreliable that you need the Rb to keep everything in sync during a long duration GPS outage. A typical Rb has a drift in the range of 5e-11 per month. That's 0.0005 Hz per month! Rather than use GPS to discipline the Rb, just compare them occasionally.
My house frequency standard is a free-running Rb that I use to lock my signal generators and counters to so that I can trust the readings that they're giving me. I can take a 10 MHz source, multiply it to 6 GHz , and my counter measures it to the nearest Hz - correctly! It's so nice when reality agrees with the math!
Ed
-
the problem with the Rb boxes is that there is no automatic calibration of them. you have to enter an integer for the dds chip to use as a divider. the output will be very stable but I have no way to know how close to 10mhz it really is.
but, if I let the seconds count on this unit and also on a gps unit, I can know how far apart they are and tune the Rb box at sub-hz levels to align it.
once its aligned, it probably won't need the gps help anymore.
that's the issue with the Rb boxes. they don't come magically tuned to 10.000000mhz exactly. there still is some setting you have to do.
-
the problem with the Rb boxes is that there is no automatic calibration of them. you have to enter an integer for the dds chip to use as a divider. the output will be very stable but I have no way to know how close to 10mhz it really is.
but, if I let the seconds count on this unit and also on a gps unit, I can know how far apart they are and tune the Rb box at sub-hz levels to align it.
once its aligned, it probably won't need the gps help anymore.
that's the issue with the Rb boxes. they don't come magically tuned to 10.000000mhz exactly. there still is some setting you have to do.
Yes, it's nice to have a GPSDO available to check your Rb. I think you'll find that the calculation gives you an integer that's REALLY close to 10 MHz. Maybe +- 1 from there. One reference said that the resolution was about 6.8 microHz. Then use the C-field to tweak it if you need to.
There are many ways to compare the frequencies. One of the easiest is to use your scope. Trigger on the GPSDO on one channel and adjust the Rb on the other channel so that it stands still. Use a stop watch to measure the drift and you can measure frequency differences in the e-11 range. Beyond that, there are things like Time Interval Counters and Phase Comparators.
This stuff is addictive.
Don't go there.
You've been warned.
It's too late for me, save yourself!
Ed
-
My house frequency standard is a free-running Rb that I use to lock my signal generators and counters to so that I can trust the readings that they're giving me. I can take a 10 MHz source, multiply it to 6 GHz , and my counter measures it to the nearest Hz - correctly! It's so nice when reality agrees with the math!
But you can do the same thing with a cheap reference. Just connect all the test gear to the same common reference. eg a cheap OCXO from ebay. The counter will still agree with the sig gen to 1Hz with GHz input signals from the generator because it's all locked to your own version of 10MHz..
The relevant question is: Does it really matter if the version of 10MHz you have at home is 0.1Hz or 0.01Hz or 0.001Hz different to a genuine 10MHz?
Unless you want to synch things extremely accurately with someone else (far away from your house) then I think there really isn't any point to having an accurate 10MHz or 1pps source at home for hobby use.
Maybe someone can convince me otherwise? :)
-
most of us don't -need- DMM's accurate to 6.5 digits. but we want them.
well, same here. if you are going for an Rb box, might as well do it right. right? yes, its probably very close when you plugin the hex integer but I like the idea of a feedback/control loop that auto calibrates it. that way I KNOW its dead-on accurate. I already have the gps hardware and some software to get data from it, so there isn't a lot of cost left; just some coding and integration.
why climb tall mountains? same thing - because we can! ;)
-
The relevant question is: Does it really matter if the version of 10MHz you have at home is 0.1Hz or 0.01Hz or 0.001Hz different to a genuine 10MHz?
Unless you want to synch things extremely accurately with someone else (far away from your house) then I think there really isn't any point to having an accurate 10MHz or 1pps source at home for hobby use.
Sometimes the gear enables the hobby, sometimes the gear is the hobby.
-
"the power of accuracy compels us!"
;)
-
"the power of accuracy compels us!"
;)
Nothing can be closer to the truth. Do I need Vishay 0.001% resistors? Nope. But I still have them.
-
tonite I modified the db9 connector to support wires it should have had but didn't. I had to remove the top 2 boards and get at the right angle db9 pins, cut them and intercept them with my own wires.
(http://farm4.staticflickr.com/3820/13260855993_55b33005e2_o.jpg)
there it is on the left. in fact, I ended up using 4 on that area; 2 for tx/rx rs232 and 2 for 10mhz out/gnd. 2 of the bottom pins are supposed to be tx/rx so I used those for that purpose, but the top pins that I could get access to easily were already defined as 5v and gnd, so I just cut them and intercepted them for 10mhz out. this box supplies its own 5v so that's not a needed external pin. and there is already a gnd pin there so we didn't need a 2nd.
all this saved me having to cut holes in the box and so, for me, this was worth the hacking.
view after adding the 10mhz sine out coax and the tx/rx/gnd wires.
(http://farm8.staticflickr.com/7393/13260717005_5933c3a82d_o.jpg)
I've seen people solder directly to the pin and gnd of the coax connector on the board but I could not bring myself to ruin that jack ;) some pads nearby gave a signal and that was fine enough.
-
If you are running a portable microwave station, your LO needs to be really accurate as it typically multiplied 100 to 1000 times to get to say 10GHz, for a ssb contact even without rain scatter, a drift/inaccuracy of more than 100Hz makes finding your contact very difficult or the speach very difficult to comprehend. This gets down to about 0.01Hz accuracy or stability of a 10MHz source.
-
tonite I modified the db9 connector to support wires it should have had but didn't. I had to remove the top 2 boards and get at the right angle db9 pins, cut them and intercept them with my own wires.
there it is on the left. in fact, I ended up using 4 on that area; 2 for tx/rx rs232 and 2 for 10mhz out/gnd. 2 of the bottom pins are supposed to be tx/rx so I used those for that purpose, but the top pins that I could get access to easily were already defined as 5v and gnd, so I just cut them and intercepted them for 10mhz out. this box supplies its own 5v so that's not a needed external pin. and there is already a gnd pin there so we didn't need a 2nd.
all this saved me having to cut holes in the box and so, for me, this was worth the hacking.
view after adding the 10mhz sine out coax and the tx/rx/gnd wires.
(http://farm8.staticflickr.com/7393/13260717005_5933c3a82d_o.jpg)
I've seen people solder directly to the pin and gnd of the coax connector on the board but I could not bring myself to ruin that jack ;) some pads nearby gave a signal and that was fine enough.
Nice pictures, I too am working on a rb frequency standard for my lab. One helpful hint, if you can find an old laptop that has built-in wifi, usually the wifi antennas connect to the wifi card with a u.fl connector. which is the same type of connector on the rb. Just simply cut the antenna off and the hard part (u.fl) is already done, solder an SMA connector and you're good to go. It's hard to believe but it's actually coax so the SMA part can be a bit tricky.
Completely off topic, but I couldn't help but notice the excellent lighting in your photos. Is that your standard lab lighting or are you using an off camera flash/strobe setup?
Tim
-
hi tim,
the lighting is just overhead fluorescent (which is not even close to white; I have to dip the yellow about -20 or so on pshop) but the trick is to use the shadow/highlight tool to bring out detail in the dark shadow areas. never use flash; flash is too hard to control and often too bright. I like static lighting and being able to see what is going on thru the viewfinder (or display) and once I angle the light the way I want, then I snap the pic.
equipment: olympus e3 body, 12-60 zoom and a decent tripod. set the self-timer, get things setup the way you want, press and let the camera fire itself. set the depth of field to mid way (f14 or even more) so that you get enough in focus.
for processing, I import the jpg image, upres to 16bit color, do some color balancing, curves, levels, shadow/highlight, then resize to 1200px across for 'printing'. after resize, always do a sharpening pass (smart sharpen) and then finally downres to 8bit color before saving as jpg. that's my general workflow for photos. if I had to resort to higher iso, I'll also add a pass of neatimage to kill the chroma noise (oly has always had a bit of color noise in thier sensors).
-
I am about to embark on this as well, got the same video amplifier (although picked up a slightly different model, so there is a bit less room in the case). I am just waiting for the UFL -> SMA pigtail from fleabay. :D Good work!
-
I found the same tall case for the video amp that gerry used; but I'm not sure I'm going to use it. for me, I need multiple freq outs more than I need a single freq with many buffered outs. if anyone wants that box, let me know as I think I'll find another housing solution for this.
(http://i.ebayimg.com/00/s/MTIwMFgxNjAw/z/jlwAAMXQVERS-T-H/$_12.JPG)
there's no point in wasting this box if I'm not going to use the video amp part ;)
-
I found the same tall case for the video amp that gerry used; but I'm not sure I'm going to use it. for me, I need multiple freq outs more than I need a single freq with many buffered outs. if anyone wants that box, let me know as I think I'll find another housing solution for this.
(http://i.ebayimg.com/00/s/MTIwMFgxNjAw/z/jlwAAMXQVERS-T-H/$_12.JPG)
there's no point in wasting this box if I'm not going to use the video amp part ;)
Hmm, I will take a photo of the one I have, but mine is a slimmer model with fewer outputs per channel. If that fits your needs better, perhaps we could make a swap.
-
I've seen the slimmer models and did my best to pick the 'tall boy' model, just like gerry's. I removed the rear amp (all those nuts!) and was about to convert the 75ohms to 50 but then decided not to even use it at all.
I could do the resistor swap and then repack it and mail it, or just leave it as-is. either way, I am pretty sure I won't use this box. it costed me $80 (not cheap) and I can get empty metal boxes for a lot less than that and machine what I need on my own (will probably even add an lcd front display for some configuration, so I'll have to use a laser cutter or milling machine/cnc).
-
I have the exact box on the way. I got it for 1 pound, but paid 60 pounds for shipping :(
As for the frequency standard, I find it weird that yours has so many stacked boards. The one I got had only one level of boards.
-
some need external 5v and some supply it locally. some have the rs232 and 10mhz lines wired to the db connector and some do not (mine did not). some don't have 1pps output but mine did.
too many variations! drives diy'ers nuts.
I did buy that $80 video dist box on impulse but I just don't have the heart to use the box and let the built in psu and video amp go to waste.
what I think I'll do is to wire up a bunch of dividers (pic-div, maybe) and feed a few bnc's with different multiples. my heathkit nixie counter needs a 1mhz ref whereas all the rest of my gear would prefer a 10mhz ref. I would also like to have some audio freq's as stable refs, as well; and I want to pulse stretch the 1pps output to show on the front via an led and also a rear panel output.
still not sure which controller I'll use. I like using arduinos but I might use an rasp-pi board since it can then be a proper ip-stack and have a webserver, ntp server and whatever else I might want from it.
-
I think you can get multiple frequencies out of that box as well (I don't have it yet, so I can't look inside), but IIRC from Gery's video, you get one opamp for a few BNCs.
About the various versions, I wonder if there's a schematic at least for the base board. That would help debug some issues. One of the ones I have is locking but the frequency is still oscillating up and down - a schematic could really help here.
-
at this point, some may say I'm done. for now, I might be ;)
I found a $10 surplus chassis at Halted today:
(http://farm3.staticflickr.com/2884/13303148743_308cf17b98_o.jpg)
the top is removed to show the circuit that is/used to be there. my existing ebay Rb box is shown to the left, for size.
next, we try out the wiring. I found a 12v 2amp + 5v enough amp psu strip today for $15. looks like a decent enough unit. it fits exactly in this chassis, too.
(http://farm4.staticflickr.com/3769/13303000155_1eedb3834f_o.jpg)
close the case lid, plug it in, let it warm up and it gets to 10mhz soon enough.
(http://farm8.staticflickr.com/7404/13303152973_02785ce2e7_o.jpg)
I did not (yet) wire in the locked-on led or any other status indicators. that front window is calling to me and it wants me to do something there.
for now, version 1 is complete. it works, its self contained and it routes its output to a rear bnc jack. it even has built in cooling.
-
One thing I have been wondering… I of course saw Gerry's video as well, and I was screaming at the monitor to not heatsink the rb standard (since it needs a certain operating temperature). I was thinking that I might try and insulate my rb standard a bit, to keep the heat in. My thinking behind this was after I saw the .7-1.5A draw on the 15V line for the heater, I thought that if I made it easier for the rb to keep at operating temp, that it should draw less power. So I thought perhaps an insulating pad between the rb and whatever it is mounted to.
Any thoughts on this?
-
@idpromnut, I've been thinking not adding too much heatsinking either, but the problem with this theory is that some of these units come connected to a big ass heatsink (a huge, thick chunk of metal).
I suspect the guys who decided to mount them on those pieces of metal knew more about the units than we do. I guess in the end it can be experimented and the temperature can be monitored over a longer period of time.
For isolating it, you can use air :) just mount it on a bunch of standoffs:
(http://cfnewsads.thomasnet.com/images/large/467/467750.jpg)
-
One thing I have been wondering… I of course saw Gerry's video as well, and I was screaming at the monitor to not heatsink the rb standard (since it needs a certain operating temperature). I was thinking that I might try and insulate my rb standard a bit, to keep the heat in. My thinking behind this was after I saw the .7-1.5A draw on the 15V line for the heater, I thought that if I made it easier for the rb to keep at operating temp, that it should draw less power. So I thought perhaps an insulating pad between the rb and whatever it is mounted to.
Any thoughts on this?
Obviously, I did not make it crystal clear in my foregoing description:
A big cooling plate is strictly required!
The bottom side of the RB unit MUST be kept cool to a certain temperature.
Otherwise the electronics around the hot physics package will fail more earlier than later.
That's also defined in the manual of every RB standard!
Baseplate temperature of the Rb is 65..70°C max., but such high temperatures may degrade MTBF by a factor of 2 or more.
My own Rb (FRS-C) draws about 12W during regular operation. Max. baseplate temp. might be 40°C.
Frank
-
Obviously, I did not make it crystal clear in my foregoing description:
A big cooling plate is strictly required!
The bottom side of the RB unit MUST be kept cool to a certain temperature.
Otherwise the electronics around the hot physics package will fail more earlier than later.
That's also defined in the manual of every RB standard!
..
Frank
Bugger, I missed that. So a heatsink it is >.<
-
The thermal management of that project box looks totally inadequate for a small 12W brick. Even if you blow air over the top of it with that little fan the temperature at the bottom face of the little enclosure will still get very high. Especially if you sit it on a wooden bench or shelf.
You could maybe build it upside down with a heatsink in the lid (like a small RF power amplifier) or just put it in a bigger box. However, there may be an ideal orientation for the module for best performance as the thermal gradients inside the module will be different if it is upside down or on its side etc.
The spurious performance of the FE5680A is pretty woeful already but you might want to see if it gets degraded by local interference. eg from the stray fields from a nearby PSU module etc. This may mean you need a bigger box anyway.
If you are running a portable microwave station, your LO needs to be really accurate as it typically multiplied 100 to 1000 times to get to say 10GHz, for a ssb contact even without rain scatter, a drift/inaccuracy of more than 100Hz makes finding your contact very difficult or the speach very difficult to comprehend. This gets down to about 0.01Hz accuracy or stability of a 10MHz source.
For use at 10GHz I would expect that a reasonably good OCXO would easily outperform the FE5680 standard when multiplied up. This is because I'd expect the FE5680 to have high spurious content and at best 'average' phase noise. Also, I'd expect the OCXO to have better short term stability across a few seconds. The better medium/long term accuracy of the Rb reference won't really be that important here.
-
someone mentioned 70C as the base plate temp that should not be exceeded. that's actually quite hot sounding.
I plan to put a temp sensor there and monitor it, maybe have variable fan speed to keep it in range.
so, if I'm 70 or under on that hottest part of that plate, I'm ok?
if I have to change cases out, I can do that. I also may move that psu strip out of the box and use an external 15v brick. that gives me more room inside (for controller stuff) and also reduces the heat inside the box. if I move the psu out, I could put some rear fans in there and with 2 of 3 tiny ones, I would think that would be enough to keep this in its temp range (?)
-
someone mentioned 70C as the base plate temp that should not be exceeded. that's actually quite hot sounding.
I plan to put a temp sensor there and monitor it, maybe have variable fan speed to keep it in range.
so, if I'm 70 or under on that hottest part of that plate, I'm ok?
if I have to change cases out, I can do that. I also may move that psu strip out of the box and use an external 15v brick. that gives me more room inside (for controller stuff) and also reduces the heat inside the box. if I move the psu out, I could put some rear fans in there and with 2 of 3 tiny ones, I would think that would be enough to keep this in its temp range (?)
You should cool the base plate passively, the annoying fan noise be avoided.
Pity, the FEI Rbs don't come with a good manual, but the LPRO 101 seems to be similar.
65..70°C is absolute max. rating temperature, so 70°C is nOK.
Better use a big cooling plate, so that base plate is at 35..40°C.
That's how several/most time-nutters recommend to do.
Your unit has already been in use for many years, that means that its life-time is consumed partly.
Therefore, you should take care, that the MTBF rate is not driven to the limit, that implies to cool to a reasonable, practicable level.
I recommend this time-nuts site for many practical hints:
http://ko4bb.com/dokuwiki/doku.php?id=precision_timing:fe5680a_faq (http://ko4bb.com/dokuwiki/doku.php?id=precision_timing:fe5680a_faq)
I assume, you have a similar unit.
Frank
-
A silly question, a 'rubidium clock' is a rubidium disciplined ovenised crystal, a GPSDO is gps (? also rubidium ? x3) disciplined (with possible flight time errors) ovenised crystal, shouldn't they be remarkably close in accuracy, allowing for different 'disciplining' protocols?
-
A GPS is in short time, only 10E-9 on the best, on the long run GPS is always better.
So there is no need to connect a RB to a GPS, because in the short, the RB is mucht better, 10E-11
GPS has also day and night shift.
I also have a BBC4 Quartzlock, for more accuracy, you need more sources, i compared
the BBC4 and the GPS and the RB at the same time, you can see the short time jitter
on these radio signals. Long wave is not usefull in night time.
If you connect a GPS to a RB, you get unwanted corrections,( day and Night shifts)
that makes the RB worse. Even the internall clock of the GPS gives little unwanted corrections on the 1pps.
you need special software to discipline a RB with GPS, for the long run.
I calibrate the RB twice a year by hand, just compare it on a DSO for a few hours.
But these Rubidiums are so good, they are still on track for already a few years.
-
The thermal management of that project box looks totally inadequate for a small 12W brick. Even if you blow air over the top of it with that little fan the temperature at the bottom face of the little enclosure will still get very high. Especially if you sit it on a wooden bench or shelf.
You could maybe build it upside down with a heatsink in the lid (like a small RF power amplifier) or just put it in a bigger box. However, there may be an ideal orientation for the module for best performance as the thermal gradients inside the module will be different if it is upside down or on its side etc.
The spurious performance of the FE5680A is pretty woeful already but you might want to see if it gets degraded by local interference. eg from the stray fields from a nearby PSU module etc. This may mean you need a bigger box anyway.
If you are running a portable microwave station, your LO needs to be really accurate as it typically multiplied 100 to 1000 times to get to say 10GHz, for a ssb contact even without rain scatter, a drift/inaccuracy of more than 100Hz makes finding your contact very difficult or the speach very difficult to comprehend. This gets down to about 0.01Hz accuracy or stability of a 10MHz source.
For use at 10GHz I would expect that a reasonably good OCXO would easily outperform the FE5680 standard when multiplied up. This is because I'd expect the FE5680 to have high spurious content and at best 'average' phase noise. Also, I'd expect the OCXO to have better short term stability across a few seconds. The better medium/long term accuracy of the Rb reference won't really be that important here.
For $210 USD from Mouser, you can get a 10MHz OCXO with stability rated at 2 ppb between 0º C to +50ºC, phase noise @ 10 Hz -120 dbc @ 100 Hz 140 -dbc @ 1000 Hz -145 dbc. http://www.mouser.com/ds/2/3/AOCJY4-39438.pdf (http://www.mouser.com/ds/2/3/AOCJY4-39438.pdf)
For $1470 USD from Mouser, you can get a 10MHz OCXO with stability rated at ±0.1 ppb between -20ºC to +70ºC, phase noise @ 10 Hz -120 dbc @ 100 Hz 135 -dbc @ 1000 Hz -145 dbc. http://www.mouser.com/ds/2/3/AOCJY6-35349.pdf (http://www.mouser.com/ds/2/3/AOCJY6-35349.pdf)
According to the time nuts faq posted earlier in this thread, the FE-5680 stability is ±0.3 ppb from -5ºC to +50ºC, phase noise @10 Hz -100 dbc @ 100 Hz -125 dbc @ 1000 Hz -145 dbc
Seems like a wash? :-// Couldn't you just put a 10 MHz notch filter on the rb and call it a day?
-
keep in mind sometimes if you don't have the tool on hand you won't think of a solution to a problem that does not exist yet because of the tool.
if you spent the 200$ on it because you thought it was cool, then years later you decide to make some kick ass project that does need the 1ps resolution because your purchase is guilting you, then thats probably a good thing! good mental exercise at the least.
-
The thermal management of that project box looks totally inadequate for a small 12W brick. Even if you blow air over the top of it with that little fan the temperature at the bottom face of the little enclosure will still get very high. Especially if you sit it on a wooden bench or shelf.
You could maybe build it upside down with a heatsink in the lid (like a small RF power amplifier) or just put it in a bigger box. However, there may be an ideal orientation for the module for best performance as the thermal gradients inside the module will be different if it is upside down or on its side etc.
The spurious performance of the FE5680A is pretty woeful already but you might want to see if it gets degraded by local interference. eg from the stray fields from a nearby PSU module etc. This may mean you need a bigger box anyway.
If you are running a portable microwave station, your LO needs to be really accurate as it typically multiplied 100 to 1000 times to get to say 10GHz, for a ssb contact even without rain scatter, a drift/inaccuracy of more than 100Hz makes finding your contact very difficult or the speach very difficult to comprehend. This gets down to about 0.01Hz accuracy or stability of a 10MHz source.
For use at 10GHz I would expect that a reasonably good OCXO would easily outperform the FE5680 standard when multiplied up. This is because I'd expect the FE5680 to have high spurious content and at best 'average' phase noise. Also, I'd expect the OCXO to have better short term stability across a few seconds. The better medium/long term accuracy of the Rb reference won't really be that important here.
For $210 USD from Mouser, you can get a 10MHz OCXO with stability rated at 2 ppb between 0º C to +50ºC, phase noise @ 10 Hz -120 dbc @ 100 Hz 140 -dbc @ 1000 Hz -145 dbc. http://www.mouser.com/ds/2/3/AOCJY4-39438.pdf (http://www.mouser.com/ds/2/3/AOCJY4-39438.pdf)
For $1470 USD from Mouser, you can get a 10MHz OCXO with stability rated at ±0.1 ppb between -20ºC to +70ºC, phase noise @ 10 Hz -120 dbc @ 100 Hz 135 -dbc @ 1000 Hz -145 dbc. http://www.mouser.com/ds/2/3/AOCJY6-35349.pdf (http://www.mouser.com/ds/2/3/AOCJY6-35349.pdf)
According to the time nuts faq posted earlier in this thread, the FE-5680 stability is ±0.3 ppb from -5ºC to +50ºC, phase noise @10 Hz -100 dbc @ 100 Hz -125 dbc @ 1000 Hz -145 dbc
Seems like a wash? :-// Couldn't you just put a 10 MHz notch filter on the rb and call it a day?
You always have to judge "stability" over the appropriate time scale, i.e. 0.01s, 0.1s, 1s, 10s, 100s,...or in other words, compare the Allan distribution diagrams of the different oscillators.
OCXOs are really superior over anything else on short time scale (0.1s, 1s), but concerning the longterm stability (days, month, year), i.e. ageing, Rb is superior.
In this domain, Rb is about 100 times more stable than OCXO.
Frank
-
OCXOs are really superior over anything else on short time scale (0.1s, 1s), but concerning the longterm stability (days, month, year), i.e. ageing, Rb is superior.
In this domain, Rb is about 100 times more stable than OCXO.
Frank
Does that mean the optimal solution would be an OCXO that is corrected every second or two by a Rb which in turn is corrected every day by a GPS? (Ignoring the power requirements of such a setup.)
-
FYI
The April 2014 Silicon Chip (Aus Electronic mag) has quite a good article about how to get a Rb standard up and going, power supply, how to adjust tuning offset for a FE 5680A.
-
OK I'll try again... :)
The FE5680 will be an extremely poor choice as a 10MHz reference for a typical radio ham who wants to run a portable 10GHz SSB station. The earlier post implies that the FE5680 Rb is 'needed' here. The fact that hams have been using OCXOs (and even TCXOs) as references for 10GHz for decades proves that this isn't the case.
Here's how I think the lessons would be learned...
The first lesson would be that the 'FE5680 ham' would have to take a car battery to the hilltop to power the Rb reference if the portable station was used over a contest weekend. 'OCXO ham' would need a battery with about 1/10th the capacity and would probably use the same battery for the station receiver and for the OCXO.
The second lesson would be learned the moment FE5680 ham received a station on his radio. It would sound noisy even if it was local. The same station would sound clean to OCXO Ham.
FE5680 ham would also get bad reports about his transmitted signal quality too... This is because the FE5680 has very high spurious content close to the carrier. eg lots and lots of spurious terms at about -60dBc to -80dBc within 100Hz to 2kHz of the 10MHz carrier. These will multiply up by 20logN and become very significant at 10GHz!
The close to carrier phase noise will be better with the OCXO
The short term stability will be better for the OCXO so I'd guess a cw tone will sound cleaner on a decent OCXO system.
The OCXO takes less power.
The OCXO is lighter.
The OCXO can be easily bought cheaply as a used item from multiple sources and replaced if it goes wrong (bring a spare?).
FE5680 ham can puff his/her chest out and say the Rb radio is within a few Hz of 10GHz. The OCXO radio might be within 150Hz but this isn't relevant. The error is tiny and is less than 1/10th of a channel bandwidth. The error between stations can be tuned
out manually just like it has been for over 50 years on any ham band from the HF bands upwards.
There's absolutely no need to use a Rb standard at 10GHz for SSB operation. I'm sure there are cleaner Rb standards available than the awful FE5680 but does anyone really need this? Hams have used OCXOs up at >100GHz.
Also, the overall system stability depends on the quality of the references at both stations. Even if you made a complex reference that was trained via GPS etc to be ridiculously accurate and clean it doesn't mean that drift is no longer a problem. The other 10GHz station could be using a multiplied TCXO.
Absolute frequency accuracy isn't important. An OCXO can get you within 200Hz at 10GHz and that is all you need for SSB. The short term drift of the OCXO will give better performance at 10GHz than the drift seen from a typical VFO based ham radio used on the HF bands for SSB.
-
OCXOs are really superior over anything else on short time scale (0.1s, 1s), but concerning the longterm stability (days, month, year), i.e. ageing, Rb is superior.
In this domain, Rb is about 100 times more stable than OCXO.
Frank
Does that mean the optimal solution would be an OCXO that is corrected every second or two by a Rb which in turn is corrected every day by a GPS? (Ignoring the power requirements of such a setup.)
Best are free running oscillators, in general.
And I meant a free running OCXO, in this case.
The disciplining worsens the stability, depending on the time constant.
If the time constant is veeery long, (~hours), the OCXO may keep its short term stability, but will fluctuate due to mid term stability.
Each Rb standard already contains an OCXO or VCO, which is disciplined by the Rb signal.
The pure Rb signal is very noisy (short time scale), so things improve on that scale via the internal OCXO.
But there already are differences, e.g. the Trimble Thunderbolt GPSDO comes with two different versions of the internal oscillators, one good, i.e. the trimble OCXO (grey), and one version bad (red), afaik some sort of cheap piezo oszillator/VCO.
You can see the difference in the Allan statistics, and the phase noise diagram.
Frank
-
G0HZU and Dr Frank :-+ :-+
Thank you guys for the explanation, it's finally starting to sink it. I think I'm going to skip the rb and go straight for a cesium clock :-DD
-
I've been drolling to have this as well for quite a while, but I know for sure I DO NOT need it. :'( :palm:
Ok, the more I read and understand explanations from you guys, the drolling part is fading. :P
-
I've been drolling to have this as well for quite a while, but I know for sure I DO NOT need it. :'( :palm:
Ok, the more I read and understand explanations from you guys, the drolling part is fading. :P
I don't really think many of us need the precision, but hey, precision is fun!
-
G0HZU and Dr Frank :-+ :-+
Thank you guys for the explanation, it's finally starting to sink it. I think I'm going to skip the rb and go straight for a cesium clock :-DD
Fine. There are "small" time-nuts like me (HP5370B / Thunderbolt / Rb level) and really BIG time-nuts:
http://leapsecond.com/hp5071a/ (http://leapsecond.com/hp5071a/)
http://leapsecond.com/pages/atomic-bill/ (http://leapsecond.com/pages/atomic-bill/)
TvB has all kind of clocks, beneath Cs also an active H-Maser.
Here's a complete performance comparison of all clocks, in form of Allan deviation:
http://www.leapsecond.com/museum/manyadev.gif (http://www.leapsecond.com/museum/manyadev.gif)
The best OCXO ("BVA 8607") beats even the H-maser for short time scales.
This diagram nicely shows all characteristics of the different clocks / oscillators and also demonstrates the effect of disciplining.
That manifests itself in those humps at higher time scales, e.g. for a cesium clock(4065B) or a GPSDO (Z3801A).
Frank
-
The best OCXO ("BVA 8607") beats even the H-maser for short time scales.
Out of curiosity I got a quote from Oscilloquartz for a BVA 8607 (for the best option), 34,040 CHF for a one-off! Even the standard unit is 9k.
So buying a Caesium clock is probably a bit cheaper.
-
The best OCXO ("BVA 8607") beats even the H-maser for short time scales.
Out of curiosity I got a quote from Oscilloquartz for a BVA 8607 (for the best option), 34,040 CHF for a one-off! Even the standard unit is 9k.
So buying a Caesium clock is probably a bit cheaper.
I was afraid to ask, but wow! I was totally just joking about the cesium , I couldn't afford one of those. I think the best price I've seen for one on ebay was around $9,000 USD. I think if I had an extra $9,000 laying in a sock drawer somewhere, I would rather buy a network analyzer.
-
A couple of things to remember about Cesium standards:
1. They have a finite life. When the Cs tube is used up, it costs maybe $10K - $20K to replace.
2. If you put a working Cs standard on a shelf and leave it for a few years, the Cs tube may never work again. It needs to be maintained. This maintenance doesn't use up the tube's life.
Because of these two issues, you have to be really careful about buying used Cs standards. I use a GPSDO as a stand-in for Cs.
Ed
-
I was tempted to go the Rb way, but ended up with GPS diciplined OCXO, a.k.a. Thunderbolt. Cheaper and seems to last a bit longer than an ageing Rb-bulb :)
Now in the lab not much of a problem but, what about those going portable needing exact frequency. The Rb should take it's time to get heated up, and our GPS controlled Rb's at work take some 45 minutes to lock fully.
A good quality crystal oscillator should be perfect, and could easily be calibrated against a "master" GPS-locked XO BTW remember HP, they used the 10811 OCXO as a master reference in GHz-equipment, and this was good enough for all of us :D
But of course those true time-nuts really needs their calibrated clocks...
" It was the best extra 22 nanoseconds I've ever spent with the kids. " :-DD
Cited from http://www.leapsecond.com/great2005/ (http://www.leapsecond.com/great2005/)
-
BTW remember HP, they used the 10811 OCXO as a master reference in GHz-equipment, and this was good enough for all of us :D
Yes, I've got a couple of old HP spectrum analysers that use the 10811 OCXO. I've trimmed the one in the HP8566 against an offair standard (maybe 2 years ago) but I can't remember if I've ever bothered with trimming the OCXO in my HP8568 despite owning this analyser for many years. You have to take the analyser apart to get at and set the OCXO. The low close to carrier phase noise is the important thing here.
For some people it appears the pursuit of accuracy is everything and that is the only (obsessive, compulsive?) motive. But for me there has to be a rational reason or need for a given accuracy level.
-
I weigh the cost of the thing against the usefulness and experience I can get out of it.
with second-hand (used) items, us non-corp non-funded folks can get our hands on stuff we, otherwise, would not be able to.
cost of Rb box is $150 or so, these days (all those websites that talk about $50 boxes need to update themselves, or have their human janitors update them for them).
cost of power supply and box could be $50, say. for $200 you get something that is extremely accurate, stable, and generally unique enough to be interesting. I think this device has 'more zeroes' than I realistically need, but its not expensive and its not a hard DIY project, either. therefore, it seemed like a reasonable thing to have at home, for the cost.
if this was $1000 or more, it would not even be marginally interesting to me. but at the $200 level, sure, why not. life is short, right? why not have some exotic equipment fun ;)
-
if this was $1000 or more, it would not even be marginally interesting to me. but at the $200 level, sure, why not. life is short, right? why not have some exotic equipment fun ;)
I couldn't agree more! :-+ :-+
Ed
-
:-+ Fun is the key!
As I said the Thunderbolt fun is not to sneeze at... and there is some really fun software to play with, have a look at:
http://www.ke5fx.com/heather/readme.htm (http://www.ke5fx.com/heather/readme.htm)
The Trimble Thunderbold do cost similar to a Rb, slightly over 100 bucks... But now that you have the rubidium thingy, I guess you better get the ting in order :D
-
cost of Rb box is $150 or so, these days (all those websites that talk about $50 boxes need to update themselves, or have their human janitors update them for them).
They were sold for $50 before Dave decided to post a video about it. ::)
-
I bought a gps disciplined osc. some time ago as my cal standard and a rubidium osc as my work standard, reason being, locking to the satellites just takes too long to wait for, so one would have to keep it running 24/7, not someting I want to do. so on an anual basis I am calibrating the rubidium osc against the gps signals. I admit, the rb osc is not on too often, certainly not every day, but within a year, the deviation is way below 10^-8, certainly better than I need, and all this came at less than 150 € each. so extremely stable, it shows you when its warm and locked, why would i buy a normal ovenized crystal osc? and when i calibrate something against it, I am always on the safe side, even if I only need 10^-7 per cal manual.
-
seems to me that if your GPS takes that long to lock on, that perhaps you need better antennas or something. If your GPS receiver is even seeing a single satellite, then it is getting time signals from that satellite.
I am not a font of knowledge on this stuff, though.
-
Others will know better than I, I think some gpsdo do a position fix (often to very accurate criteria) in the early start up phase, the Trimble you can shorten this by entering an approximate location using the interface program, how the disciplining program fully works in the early start up phase, I think can be varied depending on what you want. Time nuts will explain more clearly.
-
I'm going passive, now ;)
(https://farm3.staticflickr.com/2826/13595889143_09493d1f1c_b.jpg)
(https://farm3.staticflickr.com/2866/13595891533_bfff5c9b03_b.jpg)
this is not a formal build, but I may just leave it all like this for a while. no fans needed anymore ;)
I have an lm317 style (higher current chip; forget the number; lm385?) connected to the BAHS (big-ass heat sink), insulated, while the Rb box is directly connected to it. the regulator board (a 'TREAD' from tangentsoft) can accept a few volts higher than my target, which is 12v, more or less. you don't have to run such a high voltage into the Rb box and 12v works out to be fine. I have a 15v laptop brick that terms in a dc barrel connector. that connects to this barrel jack and that's my power story. real simple; and if the laptop psu goes nuts, my reg board is there to keep the Rb box safe.
the lock-on led is transistor driven. the transistor is air-wired (lol) since, well, I felt like it (and didn't feel like using a board just for this). led turns off when the Rb box is locked and stable.
it may not be fancy but it sure is functional.
-
no actually it is normal that a gps time receiver takes time to sync up (do not mix up with other gps devices such as nav. systems, they are faster). I would say a few minutes. comare this to about 30s fro a rb osc. not a big difference, but if you just want to do a quick measurement, its annoying.
-
no actually it is normal that a gps time receiver takes time to sync up (do not mix up with other gps devices such as nav. systems, they are faster). I would say a few minutes. comare this to about 30s fro a rb osc. not a big difference, but if you just want to do a quick measurement, its annoying.
I don't have a Rubidium, or yet a GPSDO (I'm working on my own as an educational project), but the times I've seen quoted for Rubidium are around 3 minutes when they are new (good lamp voltage) and more like 5 minutes when a bit older. (One I was bidding on e-bay the seller stated 5 minutes.)
The GPSDO has to both lock onto satellites and then bring the OCXO into lock and the time constants for doing so must be long enough for the GPS jitter (order of 10-15nsecs rms even for timing devices) not to matter.
James Miller with his rather nice design quotes:
Warm-up Accuracy
After 30 minutes < 10-10 typical
After 60 minutes < 10-11
though it is probably accurate to 10-8 much more quickly.
See:
http://www.jrmiller.demon.co.uk/projects/ministd/frqstd.htm (http://www.jrmiller.demon.co.uk/projects/ministd/frqstd.htm)
On the other hand the data sheet for the ublox NEO 6T
http://www.u-blox.com/images/downloads/Product_Docs/NEO-6_DataSheet_%28GPS.G6-HW-09005%29.pdf (http://www.u-blox.com/images/downloads/Product_Docs/NEO-6_DataSheet_%28GPS.G6-HW-09005%29.pdf)
states a cold start of less than 30 seconds but a timing initialisation time of 500 secs (recommended) which is around 8 minutes.
But in a Lab context, most instruments state that they need 30 minutes or more warm up time (often an hour) to reach spec, so the times needed by either sort of reference are no worse.
-
At my place of work we have a Quartzlock E8-Y GPS reference
http://quartzlock.com/product/frequency-reference/gps/E8-Y (http://quartzlock.com/product/frequency-reference/gps/E8-Y)
I borrowed this a while back to play with at home and soon realised that it wasn't really a practical thing to have at home. To get it to work properly I ended up downloading some generic Motorola software that allowed me to monitor how it was coping with the various GPS signals.
Without the SW you just get to see a flashing LED that lets you know how many satellites it is picking up. But I found that the performance was poor even when it said it was locked unless I put the GPS antenna outside the house on a flat roof.
Even then it took about 30 minutes to work reliably. So a realistic requirement would be to have a permanent outdoor antenna and set the thing running about 30-40minutes before you need it and also run up the software to see the signal status of each satellite it is picking up.
I did also try to see if I could get it to misbehave and go out of spec even with the lock light on and this was pretty easy to do. Just moving/rotating the main box could show it go way out of spec for quite a number of seconds even though the status LED said it was locked. Maybe these tests are harsh on it but I thought there were worth trying anyway.
-
@linux-works: I think the default behaviour of one of the pins is to transition from H->L on lock (pin 3 on my FE5680A). I used some transistor-fu as well but to get the LED lit on lock.
I have been running some tests on my rb unit, and I have seen some "ghosting" in the output signal (as if there is a slightly out of phase second signal). I will have to get a 'scope screen shot and post it.
-
At my place of work we have a Quartzlock E8-Y GPS reference
http://quartzlock.com/product/frequency-reference/gps/E8-Y (http://quartzlock.com/product/frequency-reference/gps/E8-Y)
I borrowed this a while back to play with at home and soon realised that it wasn't really a practical thing to have at home. To get it to work properly I ended up downloading some generic Motorola software that allowed me to monitor how it was coping with the various GPS signals.
Without the SW you just get to see a flashing LED that lets you know how many satellites it is picking up. But I found that the performance was poor even when it said it was locked unless I put the GPS antenna outside the house on a flat roof.
Even then it took about 30 minutes to work reliably. So a realistic requirement would be to have a permanent outdoor antenna and set the thing running about 30-40minutes before you need it and also run up the software to see the signal status of each satellite it is picking up.
I did also try to see if I could get it to misbehave and go out of spec even with the lock light on and this was pretty easy to do. Just moving/rotating the main box could show it go way out of spec for quite a number of seconds even though the status LED said it was locked. Maybe these tests are harsh on it but I thought there were worth trying anyway.
This is perhaps where James Miller's approach works well as it is very simple.
Certainly having an outdoor aerial works much better. I've just (almost - I've yet to fix the cable properly) installed an outdoor antenna and the jitter I measure on my little gps module is much better with the variation generally being within a +/- 10nsec band and without the wild 50+nsec excursions that I was getting with a puck antenna hung just inside a glass door.
Those Quatzlock E8-Y GPS references are quite pricey (£2,200 + shipping +VAT) so I'd expect them to be a little more robust.
The ublox app note suggests that you want the antenna to pick only satellites high in the sky as those lower lead to multi-path errors. Given that, in timing mode it can use the signal from just one satellite.
-
I didn't expect the E8-Y to perform well in the rotation tests because the test is so harsh but I did (initially) hope it would wink the out of lock LED when I did this.
I didn't really have the right test gear to fully evaluate it at home but I tested it against my old Anritsu counter. I took a couple of videos of it when running and here's a screen grab of one video.
I guess I should have filmed it during the rotation tests as well but I was really just interested to see how stable it was once I'd set it up correctly. When I had the antenna indoors I could see it sometimes wander on the counter. Once it was using the outdoor antenna with big signals shown on the Motorola GUI it seemed very stable indeed :)
I'm quite happy with the performance of the Anritsu counter although the (option 3) oven in it is huge and it takes about 45 minutes to warm up. For this reason I usually use it with an external OCXO that takes about 5 minutes to warm up. I normally only use it in 9 digit mode where it updates the display really quickly.
Note: I also have an old HP vector voltmeter (circa 1965!) that would probably have been a better thing to test it against because it has an A and B input and can measure phase drift very accurately. But I only really want/use the VVM to measure complex impedances here at home.
Also, I paid considerably less for that old Anritsu counter than people are paying for their Rb standards on ebay. To me it is a far shrewder purchase because I don't need the accuracy it offers from its OCXO but it is a very fast counter that works from something like 0.0001Hz up to 3GHz.
-
I just posted a blog entry about my Rb standard that I finished tonight: http://clog.unrecoverable.org:8060/2014/04/lab-frequency-standard/ (http://clog.unrecoverable.org:8060/2014/04/lab-frequency-standard/)
-
I just posted a blog entry about my Rb standard that I finished tonight: http://clog.unrecoverable.org:8060/2014/04/lab-frequency-standard/ (http://clog.unrecoverable.org:8060/2014/04/lab-frequency-standard/)
Here's a question, do you really need additional heatsinking? I'm about to finish mine too (in a similar box) but I'd like to avoid heatsinking it. I also plan to mount the Rb standard on the bottom.
As for fan holes... sadly mine is too crowded since I added some extra stuff.
-
Here's a question, do you really need additional heatsinking? I'm about to finish mine too (in a similar box) but I'd like to avoid heatsinking it. I also plan to mount the Rb standard on the bottom.
As for fan holes... sadly mine is too crowded since I added some extra stuff.
I was told that the electronics in the FE-5680 really don't like being toasted. I'm still not sure that the way I mounted mine will be adequate as the part facing in the inside heats up quite a bit as well. For sure with an additional heat sink the heater for the Rb will be working more/harder to keep the Rb at the correct temperature.
I was intending on putting passive ventilation, so no fan, just strategically placed holes to allow air to circulate. I also don't think this type of design is enough to allow this to run 24/7, so I will probably only power it up when I need it (or for the 2-3 hours a night I'm at the bench).
-
I also don't think this type of design is enough to allow this to run 24/7, so I will probably only power it up when I need it (or for the 2-3 hours a night I'm at the bench).
I thought these units were originally designed to run 24/7 in telecoms units? Though given that they have a finite life and warm up quickly and use a lot of power it probably makes sense to just turn them on when needed.
-
SPAM removed.
Please keep your advertising to the Buy/Sell forum.
-
I also don't think this type of design is enough to allow this to run 24/7, so I will probably only power it up when I need it (or for the 2-3 hours a night I'm at the bench).
I thought these units were originally designed to run 24/7 in telecoms units? Though given that they have a finite life and warm up quickly and use a lot of power it probably makes sense to just turn them on when needed.
Absolutely! However, they were designed to run IN telecom units, and the unit I've built doesn't have the same design considerations I would imagine. The actual telco gear these Rb standards are part of most likely have the correct ventilation, heat management, etc etc. The aging is obviously effected by excess heat but my understanding is that hours of service is a large factor as well (I think the Rb itself changes in some way).
-
The best OCXO ("BVA 8607") beats even the H-maser for short time scales.
Out of curiosity I got a quote from Oscilloquartz for a BVA 8607 (for the best option), 34,040 CHF for a one-off! Even the standard unit is 9k.
So buying a Caesium clock is probably a bit cheaper.
I was afraid to ask, but wow! I was totally just joking about the cesium , I couldn't afford one of those. I think the best price I've seen for one on ebay was around $9,000 USD. I think if I had an extra $9,000 laying in a sock drawer somewhere, I would rather buy a network analyzer.
Newbie here, been lurk'n this thread awhile.
...pardon me for poking in, and I know its an old post, but I can't resist sharing a story re; one of our two HP cesium standards in the L2C GPS dev lab. These things are delicate. The rack-mount unit had been placed on a large, soft-wheel rollable lab-cart, on which the cesium-clock had been sitting without it having be rolled about for several years, until it was needed to be relocated a dozen or so meters across the lab closer to the environment chamber, where we were doing precision L2C measurement tests while environmentally cycling the RXs. The clock has a built-in battery backup system, so that was no issue. We unjacked the cables, and....very carefully....rolled the clock to its new location. That move required a $15,000 re-cal. We got lots of RBs sprinkled around the lab, mostly Stanford Research, they're pretty tough by comparison. For my home lab, I've got several different types of FEIs and one really nice older unit, a Ball Efratum B100-102 RBO. I've also have a Trimble Resolution T, which I've had running since around 2007? Its ok, However, I thought it was a bit cheesy that Trimble stuck an 'F' connector on the LNA/Bullet Antenna, instead of something more appropriate, such as maybe a TNC or even an 'N'.
Anyway, just dropping in to say 'hello'.
bench knob
-
I thought it was a bit cheesy that Trimble stuck an 'F' connector on the LNA/Bullet Antenna, instead of something more appropriate, such as maybe a TNC or even an 'N'.
Trimble designed the Tbolt for 75ohm impedance (also F-conn), the Bullet was their answer for the antenna.
I personally think they were brilliant in allowing for use of "cheap 75ohm Cable-tv cable" for the antenna run , and it's super easy to mount the F-Connectors. I use quad-shielded 120dB high-quality Sattelite coax cable ($60 for 100m), imagine what the price would be for RG213 , and the holes to drill ;).
No complaints from me on that decision.
/Bingo
-
Quote from: bingo600 on April 14, 2014, 12:22:49 AM (https://www.eevblog.com/forum/index.php?topic=28382.msg424963#msg424963)>Quote from: bench_knob on April 12, 2014, 09:02:15 PM (https://www.eevblog.com/forum/index.php?topic=28382.msg424362#msg424362)
I thought it was a bit cheesy that Trimble stuck an 'F' connector on the LNA/Bullet Antenna, instead of something more appropriate, such as maybe a TNC or even an 'N'.
Trimble designed the Tbolt for 75ohm impedance (also F-conn), the Bullet was their answer for the antenna.
I personally think they were brilliant in allowing for use of "cheap 75ohm Cable-tv cable" for the antenna run , and it's super easy to mount the F-Connectors. I use quad-shielded 120dB high-quality Satelite coax cable ($60 for 100m), imagine what the price would be for RG213 , and the holes to drill ;).
No complaints from me on that decision.
/Bingo
Thaks, actually, your point is a very good one.
bench knob