Distribution amplifier for 10MHz lab standard clock

[tkamiya]

I've asked before about a year ago when I was starting to build my lab.  Now that I understand a little more and have a system working, I'd like to ask again for better understanding.

Subject of course is having a 10MHz master clock with GPSDO and distributing it to each equipment to sync their base clocks.  I know many folks here use video amplifier.  I know some use Chinese made one by famous individual or copy there of.  What I really want to know is importance of phase.

Say we have one clock.  Say we have 16 ports going to 16 equipment.  How important is the phase to be in sync? 

It is often discussed phase shift at the distribution port, but I've never seen discussion on cable end.  And, sync shift in the equipment itself.  The signal being 10MHz, it's wavelength is 30meters.  Most coax has a dialectic delay factor of about 70%.  So in 20 meters, it will shift 360 degrees.  In 5 meters, more realistic length, it will shift 90 degrees.

Is this important?

In what situations does it become critical?  How do we know the phase is not shifting internally in equipment?

I  have two GPSDO running in parallel.  They are exactly the same unit.  Antenna is fed via splitter and cable length is exactly the same.  They are NOT in sync as far as phase is concerned.  They are about 45 degrees off with each other.  Either that or 45 + (n * 360) off.  So I already know if we have two master clocks, phase will NOT be in sync.

So, here's the big question.  Is phase difference important within a lab?

[KE5FX]

It's rare to encounter an application where absolute phase on a 10 MHz 'house standard' is relevant, or (for that matter) the phase shift from one distribution output to the next.  Typically you feed the distribution amplifier outputs to different instruments, each of which will do their own thing in terms of synchronizing their internal circuitry to it.  1-pps is more commonly used when time or phase synchronization is important.  You may also see 8 kHz used for that purpose in telecom applications.

Phase stability is still worth paying attention to, though.  For instance, you should avoid the temptation to run your 10 MHz signal through a narrow bandpass filter.  That's basically a recipe for building a thermometer. 

For general distribution of 10 MHz, video amps should be OK.  You may be able to use a passive CATV splitter if you have enough input power (+15 dBm or more).   If/when you reach the point where the shortcomings of these methods become apparent, you can deal with them then. 

[RobBarter]

Slightly off topic....can someone point me to a suitable Distribution Amplifier to split a 10MHz signal on ebay.  Doubt I will need more than 4 outputs at most.  Not sure of my 10MHz power output yet as not measured yet.
Cheers
Rob

[LapTop006]

Slightly off topic....can someone point me to a suitable Distribution Amplifier to split a 10MHz signal on ebay.

Exactly what makes most sense depends on the physical layout of your situation, but I've just finished getting a Spectracom 8140 setup ready.

The upside is it's a wide area distribution amp, with pods coming off four distribution lines, and an amp plus some pods comes in at a few hundred dollars.

The downside is you do need the pods, and if you son't actually need to distribute across at least a few rooms you're probably better off with one of the modified video amp options.

[RobBarter]

Thanks Laptop - The Spectracom 8140 setup does look good although I don't need to go across rooms.  Also all the ebay listings I've found are in the US, pushing postage costs right up.   e.g. One tap is going for £18...postage £118 !!

I'll keep it in mind though, not all are daft postage costs.

[MrW0lf]

So what about BG7TBL 2017-06-03. Some are with OCXO, some not.
I have Leo Bodnars 2ch GPSDO which is seemingly stable enough by itself.
Somehow do not see what random used OCXO they put in units would add to it, or missing something?
Need to sync about 5 devices: counter, signal gens, scope etc.

[glarsson]

Somehow do not see what random used OCXO they put in units would add to it, or missing something?

The OCXO in the BG7TBL amplifier is only used if the 10MHz input is missing. It is not disciplind by the input; it is simply adjusted using a screw driver.

If you intend to use the amplifier to distribute the signal from a GPSDO, and don't need a backup source, you don't need the version with a OCXO.

[ArthurDent]

If you're looking at one of the many Chinese BG7TBL amplifiers or clones on eBay and actually want the version with the internal OCXO, be careful and read the entire listing. Some of the less expensive ones that say they are a distribution amplifier show the inside of the unit with a OCXO mounted on the board but the title and the fine print says the oscillator isn't included.

[tkamiya]

So the EEVBLOG's consensus is, BG7TBL version is perfectly acceptable?  (only speaking of distribution amplifier)

I also noticed there are versions with OCXO and without.  AND there are square wave version and sine wave version.

[ArthurDent]

Internally they have a digital ic that outputs a square wave and that square wave is filtered to produce the sine wave outputs.

[tkamiya]

I wonder if I can go backwards.....

I have a 16 port mini-circuit splitter.  Port-to-port isolation is 45db minimum.  Can I, say, take an output from the standard, amplify it 12db or so, then split it 16 ways?  Seems a cleaner way to do this. 

[N2IXK]

Many analog video distribution amps will work at 10 MHz, and they tend to go really cheap these days.

[Gerhard_dk4xp]

I wonder if I can go backwards.....

I have a 16 port mini-circuit splitter.  Port-to-port isolation is 45db minimum.  Can I, say, take an output from the standard, amplify it 12db or so, then split it 16 ways?  Seems a cleaner way to do this.

Yes, you can. Drawback is, 45 dB isolation is not world class for a distribution amplifier
and you need a beefy amplifier to drive the splitter. A LMH6702-ish opamp won't do.
But the next 50 dB are then easy if required.
You may need a lot of 50 Ohm terminations until you have enough devices to feed. :-)

[jpb]

There is a review and measurements on the BG7TBL distribution amplifier here:
http://www.reeve.com/Documents/Articles%20Papers/Reeve_10MHzDist.pdf
and previous forum discussion here:
https://www.eevblog.com/forum/projects/what-do-you-all-think-of-this-10-mhz-clock-distribution-system/

[KE5FX]

I wonder if I can go backwards.....

I have a 16 port mini-circuit splitter.  Port-to-port isolation is 45db minimum.  Can I, say, take an output from the standard, amplify it 12db or so, then split it 16 ways?  Seems a cleaner way to do this.

Yes, you can. Drawback is, 45 dB isolation is not world class for a distribution amplifier
and you need a beefy amplifier to drive the splitter. A LMH6702-ish opamp won't do.
But the next 50 dB are then easy if required.
You may need a lot of 50 Ohm terminations until you have enough devices to feed. :-)

One good way to go is a triple amp like a LMH6733 or AD8003 with a shared input termination. 

Port-to-port isolation isn't all that important in a general-purpose distribution amp.  45 dB is more than enough 95% of the time, and nowhere near enough the remaining 5% of the time.  With an LMH6733 you'd see about 70 dB from one port to the next, and close to 100 dB of S12.

[cdev]

The Extron RGB video distribution amps seem to work fine with 10 MHz and they are cheap on the used market. Mine also includes a "TTL" channel which seems to me to work fine for distributing 1 PPS (although I have no way of testing this precisely).

[Housedad]

I bought a Extron  ada 6 300mx hv  on Ebay to modify it.  $21 US plus shipping.  Plenty of information and articles on how to mod the Extrons.  Just more outlets than you will probably ever need.

[tkamiya]

Thanks for confirming phase isn't critical.  I kind of thought so but I wasn't 100% sure.

Also thanks for many options.  I have lots of MMIC.  I'll have to find out its frequency response and use that, possibly. 

One of the reasons I revived this question is that I'm thinking of making a box that combines GPSDO and distribution box.  Currently, I have two gpsdo for redundancy, distribution box, and terminating box for each channel.  Kind of sort of bulky for my small setup. 

[tkamiya]

Talk about too many ports!

https://www.ebay.com/itm/Lot-of-2-Extron-Crosspoint-88HV-Series-12-x-8-BNC-Video-Matrix-Switcher/192609851023?hash=item2cd871168f:g:WOgAAOSwbaxbWfdt:sc:UPSGround!32703!US!-1:rk:49:pf:0

[cdev]

His GPSDOs seem like a good deal and are highly regarded but his 10MHz distribution amp seems overpriced, given the availability of secondhand video distribution amplifiers that likely work just as well.


Of course, some people hate buying used equipment. But the Extrons seem to work well for this.

[PTR_1275]

Gerry Sweeney built a rubidium module into a extron amp



It’s a good video to watch even though you are using gpsdo


Once I get my system setup I have a hp gpsdo with double ovenized crystal, then that’s feeding into a Racal Dana frequency distribution amplifier. It’s a bit of overkill but I got it cheap and will suit any test equipment I’ll be getting in the next 10 years or so.

[jpb]

His GPSDOs seem like a good deal and are highly regarded but his 10MHz distribution amp seems overpriced, given the availability of secondhand video distribution amplifiers that likely work just as well.


Of course, some people hate buying used equipment. But the Extrons seem to work well for this.

Extron ones are good value, even in the UK. And they are not all second hand - there seems quite a few cheap new ones that people have had lying around in boxes. I've now acquired about  four with differing numbers of ports. They are almost cheap enough to be bought as useful project cases.
I suspect though by the time I've added various mods I'll probably end up paying quite a lot in total.

[tkamiya]

Most of you guys know a lot more than I do on this stuff.  My 8644A/B for example, requires 50 ohm 0.5 to 2 volts for reference input.

TO BE SAFE, what kind of spec should I plan for output ports?  Right now, all of my gears are HP, but there is no telling what may show up in future.  I'd like to cover most, if not all cases, without remaking my reference clock system.

[DaJMasta]

I'd aim for 1-3Vpp into a medium impedance (1k or so), with a current limiting resistor in case it's connected to 50 ohm terminated input of some sort.  On a single ended non rail-to-rail amp on a 5V supply (one simple option using an opamp as a driver), that would give you around 3Vpp as a maximum output into a high impedance load (provided you have the required slew rate), and a 1Vpp signal into 50 ohms with a 100 ohm current limiting resistor, which would be plenty.  Generally amps designed for 75 ohm line driving will work fine without much distortion and still adequate signal level.

If you're operating the amps from a single supply, then your output will be DC biased, so also include a DC blocking cap, something around 150nF should give you around the lowest loss for 10MHz.  In picking opamps for this purpose, I found slew rate to generally be the limiter, but especially if using quad packages, you'd also be limited by the current supplying capability of the package.  Sticking with dual amps or looking for line driver ICs should more easily meet relatively high output signal requirements, but use more chips.