The case has six threaded holes for mounting, two rows of three set in from the edges. They seem to be intended to mount to some sort of plate, consistent with a heat sink. The surface isn't flat, so I've ordered a sheet of material to better couple to the heat sink. It is still quite warm, I'm inclined to agree with CJay that the heat sink protects the rest of the components.
Sent from my iPhone using Tapatalk
Well, I set the counter internal reference to the rubidium standard, but have no way to tell how accurate it is. All those zeros do look impressive though. Reminds one of the difference between precision and accuracy.
I also hooked an led to the lock output, which goes low when locked. The led is on dimly while warming up which is odd. The transition is clear though, and corresponds to the time when the frequency stops hunting.
Since the OCXO is so stabile, I'm questioning the need to actually use the standard continuously. Maybe powering it up from time to time to verify the counter accuracy would prolong its life. I can use the counter's reference output for the signal generator reference input. All of this is primarily interesting rather than useful to me, but it is fun and educational.
Thanks for all the help here.
Brad
Sent from my iPhone using Tapatalk
That grey-haired guy in the back, that is you?
So everything looks alike in my lab, 5335A, 3325B, Rb standard.. and a lot other time-nuts stuff.
Well, the OCXO seems to be adjusted very well, I assume, you did NOT attach the Rb also to the 5335As REF IN?
1E-9 is a reasonable precision for the OCXO.
If you want to get deeper into stability figures, I really recommend to try the Time-Lab, it's really fun to dig into Allans Deviations.
An OCXO needs 48h to stabilize to a stable frequency, i.e. < 1E-9. So it should be powered on all the time, as it's the case of our 5335As. Consumes a few Watts, which I can afford, even here in Germany.
Rb clocks may need up to 5 days, to stabilize to about parts in 1E-11, so it's not recommended to switch it on for a quick measurement only.
You may also let it run, and feed to REF IN of your 5335A.
That would be the most stable oscillator for this counter.
P.S.: The back plane cooling of the Rb standard is definitely necessary.
Please trust an experienced time-nuts, or read the manuals of the LPR-101, or the EFRATOM FRS-K.
These specify the use of a metal plate, to limit the temperature.
In the LPRO manual, the MTBF figures go down drastically with rising temperature, and in the FRS-K manual, 65°C back plane temperature is defined as absolute maximum.
The 2nd effect after the limitation of the life time, is the stability of the internal ovens, which might be affected, or unregulated, if improper cooling lets the internal temperature rise above the 70°C of the first oven in the physics package.
Frank
It is me, I hadn't noticed. My "lab" was a very small yoga room, hence the mirrored wall behind the shelves.
The counter is plugged in continuously. I had to unplug it briefly to pull it out to get at the OCXO adjustment. Probably less than 30 seconds. I let it run until the displayed frequency of the rubidium standard stabilized at the same frequency as it saw when it had run continuously for weeks. That only took a few minutes, probably because the oven had barely cooled.
Here are a couple of pictures of the heat sink
The coupling to the surface of the rubidium standard isn't great, but when I get the thermally conductive pad that I ordered it should be much improved. At that point I'll turn the standard on and leave it for a couple of weeks to stabilize and then repeat the measurement.
The other problem is that the heat sink won't fit in the aluminum box I'd hoped to use. I'll have to mount the standard inside and the heat sink outside. That seems sub optimum since I'd expect the inside of the box to heat up. I hate the thought of a fan running constantly to cool the inside of the box. If I can find a rack panel to hold everything that might be the answer.
Thanks again for the help.
Brad
Sent from my iPhone using Tapatalk
The coupling to the surface of the rubidium standard isn't great, but when I get the thermally conductive pad that I ordered it should be much improved. At that point I'll turn the standard on and leave it for a couple of weeks to stabilize and then repeat the measurement.
The other problem is that the heat sink won't fit in the aluminum box I'd hoped to use. I'll have to mount the standard inside and the heat sink outside. That seems sub optimum since I'd expect the inside of the box to heat up. I hate the thought of a fan running constantly to cool the inside of the box. If I can find a rack panel to hold everything that might be the answer.
Thanks again for the help.
Brad
Such a big heat sink is not needed, see the hints in the manuals I mentioned.
I have assembled the Rb clock to the bottom metal plate inside a rack size enclosure, using thermal paste, and that's sufficient.
I assume, that the Rb clock should not be used upside down, due to thermal flow.
You won't see any difference on 1E-9 level, between the OCXO and the Rb clock.
The Rb is about 100 times more stable (1E-11) than the OCXO, so that's also the order of magnitude, where you would see any difference over a weeks time.. if you would compare it to a GPSDO, or a Cs clock.
Frank
Frank
This is a slight change of topic.
I connected the 10MHz out from the 5335A counter to the reference in on the 3325A function generator. I set the function generator to 1000 Hz ant connected its output to the A input of the counter. After warming up for several hours I get this
The last three digits vary +- .150 Hz or so.
I recognize that this is a very small change, but wonder what is causing it since the two Instruments are using the same frequency standard. Mostly curiosity on my part.
Brad
Sent from my iPhone using Tapatalk
The 3325A/B uses an M/N counter locked PLL.
So lower frequencies are less stable than those being closer to 30MHz, which is its internal master clock.
Try that measurement at 1 MHz or 10MHz, and you'll see stabilities to 1E-9, or better (with higher resolving counters, like the 5370B).
I have measured the Allan Deviation for different output frequencies, and at 1kHz, red curve, it's on the order of about 1E-7 only, depending on the Gate Time.
So that's the 3325As deviation / fluctuation, which your counter detects also.
This stability parameter / Allan Deviation is NOT specified in the manual, it's a real gap, for some reason. (i.e. principle of operation of the synthesizer).
I always wanted to ask the time-nut community, whether this is all correct, but now I really believe so.
Frank
P.S.: once again, I recommend Time-Lab, which allows to acquire such Allan Deviation measurements.
Frank,
I've started looking at options for GPIB. There are GPIB-USB converters that seem convenient but are expensive. There are also various adapters/cards that seem to be less expensive. NI has drivers, but it isn't clear whether they are free.
What are you using? Any suggestions?
Brad
Sent from my iPhone using Tapatalk
I can't give you an advise, as I'm having standard GPIB cards from N.I.
Time-Lab also works with USB/GPIB adapters, afaik, but please check in the description.
Which cheap adapter to buy, I don't know-there have been many hints somewhere else here in the forum.
Frank
PS: Obviously, Prologix GPIB/LAN adapter is supported, at least.
Mine outputs a sine wave. I am not sure, but a square wave could be better in some applications, but it will be rich in odd harmonics.
Sent from my iPhone using Tapatalk
Also, the prices are at least double what I paid less than a year ago.
Depending on what you are trying to do, my experience with the OCXO in my counter showed it to be accurate to less than one Hertz at 10 MHz, and quite stable at least in the short term. For me, that is probably good enough but I don't regret getting the rubidium standard if only for the experience. Dr. Frank is very knowledgeable and helpful.
Sent from my iPhone using Tapatalk
I am also looking for a Rubidium standard but there are so many on ebay it is difficult to select one over the other.
This one looks nice:
10MHZ Efratom Rubidium FREQUENCY Standard FRS-C , Square-wave,output high level
http://www.ebay.com/itm/10MHZ-Efratom-Rubidium-FREQUENCY-Standard-FRS-C-Square-wave-output-high-level-/281450872886?hash=item4187c78036:m:mr2dt-0VVev04AkzL_2GF8w
But I have no idea how it compares in a real world to others
Any suggestions of which one to get?
You might consider contacting user "yixunhk" on ebay and asking if he has anymore broken LPFRS-01 units. Late 2015 a bunch of us bought them in 5 packs from him for 129 USD + shipping. I was able to get all of mine working great with new electrolytics and the odd tweak to a pot or two. I know you like repairing stuff.
I run one 24/7 and am very impressed every time I compare it to my GPSDO. Most of the time I'd rather use it to feed my dist amp instead of the GPSDO as there is no externally mounted antenna to attract static or worse lightning.
I am also looking for a Rubidium standard but there are so many on ebay it is difficult to select one over the other.
This one looks nice:
10MHZ Efratom Rubidium FREQUENCY Standard FRS-C , Square-wave,output high level
http://www.ebay.com/itm/10MHZ-Efratom-Rubidium-FREQUENCY-Standard-FRS-C-Square-wave-output-high-level-/281450872886?hash=item4187c78036:m:mr2dt-0VVev04AkzL_2GF8w
But I have no idea how it compares in a real world to others
Any suggestions of which one to get?
There are several comparisons by well-known time-nuts:
http://www.ke5fx.com/rb.htmhttp://febo.com/pages/oscillators/rubes/The FE5680A is not so good in terms of phase noise, then comes the FRS, then the LPRO-101, but they all perform similar in respect to stability.
No wonder, as the underlying physical principle and practical setup is identical.
The Stanford Research PRS10 is the best performer, but expensive.
The most important questions, anyhow, is the life expectancy, as all of the surplus devices made many years already in telecommunication cells.
I think, the FRS might be older than the LPO, so maybe you find a seller who publishes the lamp voltage, to estimate the time-span left.
There are also ways to rejuvenate the Rb cell, instructions for that can be found.
I have an old EFRATOM FRS-C (I think), which works ok.
Manual for these units is the best, very detailed.
output can be changed from sine to square and vv.
Frank
Hey,
does anyone happen to have any information as to what can be done with the Serial Port on the XHTF1003H by Chengdu Tianao Electronics Co., Ltd., and how the protocol works?
I've emailed the company several times, but with no success at all.
I'm attaching the data sheet I've found for the box.
Thanks
- pit
I remember, I have read somewhere in the web, you can set the frequence output (divider).
But unfortunately I can't remember, where I have read it.
/PeLuLe
I've started looking at options for GPIB. There are GPIB-USB converters that seem convenient but are expensive. There are also various adapters/cards that seem to be less expensive. NI has drivers, but it isn't clear whether they are free.
What are you using? Any suggestions?
I use an E5810A that I bought from EBay - It's a LAN to GPIB adapter and is supported by the current Keysight IO Libraries Suite. If you wanted to save some money you can get the previous one, E2050A for around US$100.
Personally I like these because I can use any network connected PC in my house, including my tablets, to run control software. I also have a Prologix GPIB-LAN, an Agilent 82357B and an old HP 82341C - I prefer the 5810 over all of them.
There are also many many fakes of the 82357B on EBay that seem to work fine - I've also seen an increase in the NI USB adapters so I suspect those are fakes as well.
Lastly, if you have a proper desktop then there are many PCI GPIB adapters (though the PCI-E adapters seem wildly overpriced).
As an aside, I wouldn't trust anything from China that says "New sealed box" as there are many videos showing that they've cloned the packaging as well.
TonyG
TonyG
I don't have the software skill to do this, but it seems like the new ESP-32 could be the basis for a GPIB – Wi-Fi adapter. I think it has enough I/O pins on board that all you would need would be a power supply, a few pull up/pull down resistors, a connector, and a housing. The connector and the housing would probably be the most expensive parts. I think the total cost would be less than a good quality cable.
The software would take some effort, but I think there is arduino/8266 software out there that would be at least a good start.
Am I missing something.
Sent from my iPad using Tapatalk
Am I missing something.
IEEE 488.1-compliant bus drivers? See the thread linked by Tony_G for some discussion.
I had read that. My thought was that the 8266 didn't have enough I/O pins so the Arduino was required. The ESP32 has more pins and runs Arduino code. It can do BLE and wifi.
Could you do a wifi - GPIB adapter with just that one chip?
Doesn't some of the control software work with the existing wifi adapters that cost about $200?
Sent from my iPhone using Tapatalk
Here is some C code that I wrote for doing GPIB on an AVR chip (2561). It emulates the Prologix GPIB-serial adapter (with some extensions). Released under the MIT license. Tested with gpibkit.
Note that the attached code is not complete (it uses LCD/touchscreen/serial I/O routines and some header files that are not included), but should be fairly easy to modify for other systems. The serial I/O routines are interrupt driven and use buffered I/O.
You can find the missing code here:
https://github.com/mega-donkey/Mega-DonkeyThe file also has code for emulating another GPIB serial interface... it should be nuked from orbit.