EEVblog® Electronics Community Forum
Products => Test Equipment => Topic started by: Bryan on August 07, 2014, 08:21:02 pm
-
All:
Occasionally see these rubidium oscillators with a 1pps output ONLY . Has anyone successfully been able to hack or modify it to output a 10mhz output. Not sure if the output of 1pps is derived from firmware or perhaps just divided down.
-
Hi Bryan,
The 5680s are a royal PITA. There are many different versions with many different options. Some have 1 PPS or 10 MHz or both. Some are programmable, some aren't. Some require 15V, some also need 5V. Some have EFC, some don't. :palm: :wtf:
If a particular unit is advertised as programmable, it is possible to change the frequency to 10 MHz, but it requires a little hardware hacking. The ones that output only 1 PPS take a frequency of 8.288608 MHz and put it through a fixed divider to get 1 PPS. You can reprogram that frequency so that it puts out 10 MHz. Of course, that blows the 1 PPS way off frequency. I have one that works like that. You then have to bring the 10 MHz signal out via a coax. Start googling. There's lots of info out there where people have done similar hacks.
Ed
P.S. There's a couple of Trimble Thunderbolt up for sale on ebay that have a red label. Don't buy them! The red label units don't perform nearly as well as the plain units. Apparently the plain units were built with a better quality oscillator to meet the requirements of cellular carriers. The red label units also tend to be much pricier than the plain ones.
-
Well the seller is apparently quite knowledgeable with teh GPS items he sells and he can't figure it out so probably not a easy hack. With regards to the OCXO, this site does a comparison, and I believe the red labelled ocxo is referenced.
http://www.ke5fx.com/tbolt.htm (http://www.ke5fx.com/tbolt.htm)
-
Did you buy one of those 1 PPS units? If so, you'll just have to wait for it to show up to see what you ended up with. I got mine some years back and got lucky. Hopefully, you will too.
Regarding the Thunderbolt, I was actually referring to the red labelled Thunderbolt. I didn't realize that the OCXO also had a red label. But the phase noise graph certainly shows the difference between the 'old' red labelled Thunderbolt and the 'new' plain version.
Ed
-
No, I ended up buying a Trimble GPSDO re branded for Nortel. I hope it works well. Not a standard size, but for the price I couldn't do better. Looking around for a decent priced Rubidium that I can "discipline" with the GPSDO.
-
I have a 5680 that originally only had 50Hz output and no serial port. I have now managed to add the serial port and reconfigure it to 10MHz output.
Would you mind sharing on how you hacked it at least for the 10Mhz output.
-
No, I ended up buying a Trimble GPSDO re branded for Nortel. I hope it works well. Not a standard size, but for the price I couldn't do better. Looking around for a decent priced Rubidium that I can "discipline" with the GPSDO.
I've seen those, but never played with one. What kind of antenna do you have? Have you discovered Lady Heather? You'll like Lady Heather. She's into discipline. ;)
Ed
-
put it through a fixed divider to get 1 PPS.
A 1pps signal is a 1hz signal with very low duty cycle: the output is high typically for 1-2us and low the rest of the time.
When you divide a high frequency signal, you get a 50% duty cycle signal - in this case, high for 0.5s and low for 0.5s. Not exactly a 1pps signal.
You can, however, use other circuitry, sometimes as simple as a capacitor, to derive a 1pps signal from a 50% duty cycle 1Hz signal.
-
I've seen those, but never played with one. What kind of antenna do you have? Have you discovered Lady Heather? You'll like Lady Heather. She's into discipline. ;)
Ed
I have not sourced an antenna yet. Any suggestions. I believe any GPS timing antenna, either passive or amplified, would work.? Yes Lady Heather I have downloaded, but not sure if it will work with the board, haven't really looked into that aspect of it yet.
-
The traditional timing antenna is one of the bullet or cone shaped ones. I think a timing antenna has more filtering than a navigation antenna. Not totally sure about that. You definitely want an active antenna. Most of the GPSDOs that show up for auction are 5 - 10 years old and are a bit deaf by modern standards so they need lots of signal. Don't buy an antenna unless they state what the gain is - e.g. 26 dB unless you can find a data sheet. Check your GPSDO to see what voltage is running up the antenna cable. Older systems are usually 5V, newer ones are 3V3. Strange ones are 12V. The antenna has to match that voltage. I see some cheap mushroom-shaped timing antennas for sale. I don't know anything about them. If you have or can borrow an active patch antenna give it a try - won't hurt (as long as it's the right voltage).
Lady Heather should work with your board. I think you've got one of those NTGS50AA units.
Ed
-
Any recommendations for the gain. I have seen on eBay most are around 20db to 26db gain. I did manage to find what I think is a manual for the GPSDO and yes it appears to be a NTGS50AA. It says the antenna outputs +5v on the center pin, so pretty sure a 3.3-5v timing antenna will do. I think something like this should do
http://www.ebay.com/itm/NEW-PCTEL-26db-GPS-TMG-26N-GPS-Timing-Reference-Antenna-/291208295462?pt=US_Ham_Radio_Receivers&hash=item43cd5df426 (http://www.ebay.com/itm/NEW-PCTEL-26db-GPS-TMG-26N-GPS-Timing-Reference-Antenna-/291208295462?pt=US_Ham_Radio_Receivers&hash=item43cd5df426)
Did see some posts about on the board using Lady Heather, seems a bit of fiddling is required and the recommendation is too get it working with the Trimble software first.
-
Sorry Bryan, but I'm not sure how much gain you should use. In your case - north facing, right? - you'd think that more gain is better, but you could have a lot of multipath. More gain will just make that worse. 26 dB is not really high by timing standards so that might be okay. I've seen antennas with 38 dB gain. Do you have a little patch antenna available that you can test with before you buy a good one? That would tell you how deaf your receiver is. Maybe it needs all the gain you can give it just to get a good signal. I don't suppose you're on the top floor are you? I am and I can get a decent signal through the roof with a good antenna.
Ed
-
Ed:
Yes, facing north on a bottom floor condo in Vancouver. Holding my cell phone out the window I can pick up 3 satellites (GPS Logger II), so think I should be ok with a decent GPS timing antenna. May as well just order one of the 26db ones, cheap enough and if it works great, if not then going to go with another option.
-
OK, sounds good.
Don't forget to get some cables and/or adapters to connect to the board. The RF connectors look like SMB.
Good luck,
Ed
-
Good that you mentioned SMB, already had some SMA's sorted out and had to have a second look at the listing, yes they are SMB. Did find some crimp SMB connectors on EBay that can use RG58 coax
-
Sorry Bryan, but I'm not sure how much gain you should use. In your case - north facing, right? - you'd think that more gain is better, but you could have a lot of multipath. More gain will just make that worse. 26 dB is not really high by timing standards so that might be okay. I've seen antennas with 38 dB gain. Do you have a little patch antenna available that you can test with before you buy a good one? That would tell you how deaf your receiver is. Maybe it needs all the gain you can give it just to get a good signal. I don't suppose you're on the top floor are you? I am and I can get a decent signal through the roof with a good antenna.
Ed
More gain will not make multipath worse. The multipath problem is an SNR issue.
The gain required depends on the Rx sensitivity and the antenna cable loss. Acq. power required is probably around -145dBm. If GPS signal at ground level is -160dBm, then need at least 15dB + cable loss.
Sent from my One S using Tapatalk
-
Sorry Bryan, but I'm not sure how much gain you should use. In your case - north facing, right? - you'd think that more gain is better, but you could have a lot of multipath. More gain will just make that worse. 26 dB is not really high by timing standards so that might be okay. I've seen antennas with 38 dB gain. Do you have a little patch antenna available that you can test with before you buy a good one? That would tell you how deaf your receiver is. Maybe it needs all the gain you can give it just to get a good signal. I don't suppose you're on the top floor are you? I am and I can get a decent signal through the roof with a good antenna.
Ed
More gain will not make multipath worse. The multipath problem is an SNR issue.
The gain required depends on the Rx sensitivity and the antenna cable loss. Acq. power required is probably around -145dBm. If GPS signal at ground level is -160dBm, then need at least 15dB + cable loss.
More gain could raise a multipath signal from a too-low level to an annoying level.
In this case, more gain might be required due to a very poor antenna location. The OP is on the ground floor of a north-facing condo.
Ed
-
in fe-5680b,there is a 20MHz square wave point.
used 74hc393 to 2 divider.can output 10MHz.
-
20MHz point
-
REMOVE SOME COMPENT
-
install 74hc393
-
10MHz output point
-
in fe-5680b,there is a 20MHz square wave point.
used 74hc393 to 2 divider.can output 10MHz.
Do you know if the "20MHz square wave point" output works also in the units that are sold with only 1pps (cripled unit)?
The example chip (in photo), MC74HC393, has status "OBSOLETE" in datasheets. Could you recommend some other Ripple Counter chip that would work as well with the "20MHz square wave point"? (also, how can we tell from the datasheet that the MC74HC393 can handle the 20MHz input on one of the "CLOCK" inputs?)
-
The example chip (in photo), MC74HC393, has status "OBSOLETE" in datasheets. Could you recommend some other Ripple Counter chip that would work as well with the "20MHz square wave point"? (also, how can we tell from the datasheet that the MC74HC393 can handle the 20MHz input on one of the "CLOCK" inputs?)
Other companies make the 74HC393 including NXP, TI, and On. Any of them should work.
The datasheet will either list the minimum clock input toggle frequency or the minimum clock input pulse width.
-
The example chip (in photo), MC74HC393, has status "OBSOLETE" in datasheets. Could you recommend some other Ripple Counter chip that would work as well with the "20MHz square wave point"? (also, how can we tell from the datasheet that the MC74HC393 can handle the 20MHz input on one of the "CLOCK" inputs?)
Other companies make the 74HC393 including NXP, TI, and On. Any of them should work.
The datasheet will either list the minimum clock input toggle frequency or the minimum clock input pulse width.
It seems that user: h572 suggested to power the chip with the 3.3V output. The datasheet for the MC74HC393* guarantees operation with a CLOCK pulse width of 100nS @ 2Vcc and 20nS @ 4.5Vcc (Both at <=85 deg C). Real life interpolation of a 3.3Vcc better be <= 50nS (20 MHz)...
*) http://www.onsemi.com/pub/Collateral/MC74HC393-D.PDF (http://www.onsemi.com/pub/Collateral/MC74HC393-D.PDF)
It would be way better to find a 5V source inside that Rubidium standard.
And then you have the recommended rise time of the CLOCK pulse being nothing less than 400nS, even with a Vcc of 6V.
-
The example chip (in photo), MC74HC393, has status "OBSOLETE" in datasheets. Could you recommend some other Ripple Counter chip that would work as well with the "20MHz square wave point"? (also, how can we tell from the datasheet that the MC74HC393 can handle the 20MHz input on one of the "CLOCK" inputs?)
Other companies make the 74HC393 including NXP, TI, and On. Any of them should work.
The datasheet will either list the minimum clock input toggle frequency or the minimum clock input pulse width.
It seems that user: h572 suggested to power the chip with the 3.3V output. The datasheet for the MC74HC393* guarantees operation with a CLOCK pulse width of 100nS @ 2Vcc and 20nS @ 4.5Vcc (Both at <=85 deg C). Real life interpolation of a 3.3Vcc better be <= 50nS (20 MHz)...
*) http://www.onsemi.com/pub/Collateral/MC74HC393-D.PDF (http://www.onsemi.com/pub/Collateral/MC74HC393-D.PDF)
It would be way better to find a 5V source inside that Rubidium standard.
And then you have the recommended rise time of the CLOCK pulse being nothing less than 400nS, even with a Vcc of 6V.
I agree that at 3.0 or 3.3 volts, the 74HC393 and 74HCT393 would probably not be fast enough. There are faster direct replacements for the them like the 74VHC393 and 74LV393 which would work and are specified for 3.3 volt operation.
For dividing 20 MHz down to 10 MHz lots of flip-flops could be used like the common '74 dual D flip-flop. Pick the right logic family like AC and it will provide plenty of drive as well.
-
I suggested power the chip with the 3.3V .used 74XX393 is reduce wiring?
-
about the frequency.
in philips semiconductors datasheet ,power in 2.0V,25 degree the typ fmax is 30MHz.if in 3.3V power the fmax more then 30MHz.
what is "typ fmax" ,it means lots of this chip can reach parameter.
But strictly? power in 3.3V? the 74hc393 can not good work.
the best choose is 74vhc393 or 74lv393.
why did i choose 74hc393,because it is easy to buy and unexpensive.
-
I suggested power the chip with the 3.3V .used 74XX393 is reduce wiring?
May I ask why you suggest connecting the 3.3V operating voltage also to the unused CLOCK input pin ("1CP") and the RESET ("1MR") pin?
-
Connect pin-1CP and pin-1MR to 3.3V is easy to connect.
-
Connect pin-1CP and pin-1MR to 3.3V is easy to connect.
Yes, it is easy to connect. But what are the benefits of connecting the unused half of the chip like suggested?
One more thing: How sure are you that the 20MHz signal is present in all (1pps-only) FE-5680B?
-
what is "typ fmax" ,it means lots of this chip can reach parameter.
Typ fmax is the frequency that the parts will typical operate at but it is neither tested nor guaranteed. Min fmax is the guaranteed minimum specification. If you want higher than that, then you should grade them yourself for operation at the frequency of interest over temperature and voltage.
-
Today I've poked different test points & pads with oscilloscope (DS1054Z\hacked), and look what I've found ! :-+
- Point 1 that was mentioned in
20MHz point
Look how strange signal looks. Dunno why is that, I've poked other test points in that area - they are all digital or power.
- Point 2 is on other side of board, looks like DDS generator should be there, look how clean waveform is and how big V peak-to-peak. Poking around with this test point doesn't mess with PLL, on other hand if I poke around VCO that whole unit goes out from lock really fast (1-2 sec)
Any good ideas how to use this 60 MHz point ? I have AD9850 laying around.
PS. If I trigger on this 60 MHz and then scroll for couple of microseconds then the waveform start to look really jittery (on point 1 or 2), is it a issue with time domain oscillator in the scope ? I assume frequency standard is a standard for a reason :) (Reposted this question here (https://www.eevblog.com/forum/beginners/jitter-question/))
PS2. Is it safe to operate this unit without a case ? Rubidium 87 decays by beta emission, so maybe the case is there for a reason ? Also it's getting pretty hot without the case.
-
Picked one up and it is testing out fine. I do have question, does anyone know if the frequency can be tuned with a serial connection. Appears some of the 5680A models with option 02 can be tuned with a serial connection. Once I have it outputting 10MHZ I would like to tweak it to match my GPSDO. The datasheet from the seller seems to suggest their is a serial connection on pins 14 and 15, but I can't seem to find a Max232 onboard so wondering if perhaps this model does not support this option.
-
This is not a topic I'm strong on so excuse a dumb question
is there any value in trying GPS Stabilize a Rubidium standard or is it a case diminishing returns?
-
REMOVE SOME COMPENT
Does anyone know why the components need to be removed. Not sure what the purpose is if the 20Mhz point is there.
-
I've modified my FE-5680B to output 10MHz square wave. I've posted a couple of pics below but there's a full writeup here.
http://andrelubbock.blogspot.co.nz/2015/09/rubidium-frequency-standard.html (http://andrelubbock.blogspot.co.nz/2015/09/rubidium-frequency-standard.html)
Essentially though I found a 20MHz which I brought out on the DB15 and divided it by 2 with a D-type flip flop. My standard also has 5V that I used to power the external circuitry.
-
excellent write up - tnx for sharing :)