Maybe I have a wrong type of distribution amplifier

[tkamiya]

I'm afraid I may have acquired a wrong equipment.  The equipment in question is a distribution amplifier.

I've purchased a Spectracom 8140 Distribution Amplifier.  I also have Nortel Trimble 10MHz GPSDO.  Obviously, an output of GPSDO goes into the 8140.  My concern is an output.  It puts out sine wave output overlayed with 12volt to power the  distribution tap. 

As target, I have HP8590L spectrum analyzer, HP8384A counter, and HP 8644A and B. 

Can I connect 8140's output directly to those test equipment?  Would 12v overlay be harmful to any of my gear?  What really IS a tap?  Why not just directly wire via coax?  Are all distribution amplifiers like this?

I was under impression that many that has as many as 12 to 16 output are to be connected to one equipment each.  Somehow, what I ended up seem different.

Before I blow this stuff up, can someone guide me to the right direction, or at least away from wrong direction?

[PA0PBZ]

It looks like this distribution amplifier uses what they call "Line Taps", which take the 10MHz and a DC signal from the distribution cable and create a 10MHz output:

"Remote station LineTaps receive both 10 MHz and DC power from the trunk line cable."

https://spectracom.com/sites/default/files/document-files/8140_frequency-distribution-amp.pdf

So yes, if you don't have the Line Taps you have a problem...

[jpb]

I don't see why you can't just use a dc-block i.e. a capacitor between the distribution amplifier and the reference inputs of the kit. Terminate the line in 50 ohms so it is a good match[EDIT: don't do this - see EdPalmer's post below] and then a suitable capacitor to the reference input of the instrument. You may have to take the input impedance into account but at 10MHz it probably isn't too critical.

If the kit expects say TTL (0 to 5V) then there are comparitors that can withstand relatively high voltages and only output at TTL level so and perhaps you could power it via a standard 5V regulator from the 12V + RF.

Am I missing something? I presume the 12V was to power some sort of amplification at the "taps". If the 10MHz level is too low perhaps you could wire your own circuit to use the 12V in the same way to power a local amplifier.

I see from the link posted by PA0PBZ the taps also did frequency conversion (i.e. division down to 5MHz or 1MHz) but I don't think you need that.

[glarsson]

I don't see why you can't just use a dc-block i.e. a capacitor between the distribution amplifier and the reference inputs of the kit. Terminate the line in 50 ohms so it is a good match and then a suitable capacitor to the reference input of the instrument. You may have to take the input impedance into account but at 10MHz it probably isn't too critical.

Doesn't this amplifier just have one output and is intended to drive a number of taps? Each tap having a filter and driver amplifier? Looking at the data sheet indicates that a single amplifier can output several frequencies and it is up to the tap to somehow extract the wanted frequency and drive the cable to the instrument.

Basically; one distribution amplifier with eight outputs equal one 8140 distribution amplifier, a coax bus and eight taps.

Just using a capacitor will only drive one instrument with all frequencies at the same time. Most likely not what the owner wanted.

[edpalmer42]

Since you can add up to 25 taps onto a single output, it appears that the tap presents a high AC impedance to the 8140.  If you just use a DC block directly to a piece of equipment, you might start to load down the 8140 when you add multiple devices.  I've found that the external frequency reference on equipment doesn't really have a standard.  Some equipment has a 50 ohm input and some has a multi-Kohm input.  The tap isolates the 8140 and the other taps from this type of nonsense.  If you don't have any taps, you can experiment with a DC-blocking capacitor on the output of the 8140 and then use T-adapters to each piece of equipment.  I don't have an 8140, but I have successfully distributed 10 MHz to various pieces of equipment like that.  Just remember that you could load it down far enough that one piece of equipment might not see it or worse, the low level might cause noisy readings on the equipment.

If you search ebay for <spectracom tap>, you'll find a few for sale.

BTW, be sure you don't put a standard 50 ohm termination at the end of the chain.  As shown in the datasheet, it MUST be a DC blocked termination.  i.e. a DC-blocking capacitor before you get to the 50 ohm termination.

Ed

[glarsson]

I've found that the external frequency reference on equipment doesn't really have a standard.  Some equipment has a 50 ohm input and some has a multi-Kohm input.  The tap isolates the 8140 and the other taps from this type of nonsense.

Another reason for isolation is stupid implementations of reference inputs like in the Rigol DG1022Z. The Rigol has a shared reference input and reference output. During power on it unconditionally configures the reference as an output and blasts its own 10MHz out on the coax. Later when it checks the saved setup it changes its mind and reconfigures the reference as input. To make it even more funny any other Rigol DS1022Z on the same coax will reconfigure and use its internal reference when they declare the reference unusable (due to the other Rigol mixing in its own 10MHz). No error message, just a small icon change in the GUI.

[tkamiya]

I'm realizing this is really like the old 10Base-5, the old Ethernet standard used to be....  Oh well...  I'll find a use for it someday....

[bson]

You need to use taps with it.  Just go to eBay and search for "spectracom 8140 10M tap" - they're dirt cheap.  Just make sure it's a 10MHz model - the amplifier also comes in other frequencies.

Don't ever feed the distribution output to an instrument; the 12V bias voltage could easily cause serious damage.

[ArthurDent]

If you don't plan to use the taps that the 8140 was designed for and just use it to supply 10Mhz to the ext ref input on test equipment, there is a really easy fix. I have a couple of these 8140 units that I use and I've modified mine this way. The output board is mounted on the back panel and at the bottom of the board you will see 5 toroidal inductors marked L9,L10,L11,L12, and L13 and the right 4 (viewed from inside) supply the 12 volts to the output connectors. If you remove L9,L10,L11, and L12 (L13 can be removed as well) the 10Mhz output will be capacitively coupled with no D.C. component. Just snip the leads to disconnect them and remove them completely. Check the attached schematic to see what I mean.

[tkamiya]

Thank you.  I thought about that, too, but decided I didn't want to modify the amplifier.  Instead, I bought a bunch of taps by the manufacturer.  Amazingly, they were pretty inexpensive.  Now, the new problem is, I need a DC insulated 50 ohm terminator at the end of the chain.  But I cannot find such things.  I searched Mouser and Digikey.  No such items.....  Google search brought up too many non-related items. 

Can anyone please point me to one that I can purchase? 

[jpb]

Thank you.  I thought about that, too, but decided I didn't want to modify the amplifier.  Instead, I bought a bunch of taps by the manufacturer.  Amazingly, they were pretty inexpensive.  Now, the new problem is, I need a DC insulated 50 ohm terminator at the end of the chain.  But I cannot find such things.  I searched Mouser and Digikey.  No such items.....  Google search brought up too many non-related items. 

Can anyone please point me to one that I can purchase?

Mini Circuits can sell you a dc-block to which you could attach a 50 ohm termination (which they also can supply) (I think this is what is required) :
https://ww2.minicircuits.com/pdfs/BLK-89-S+.pdf
https://ww2.minicircuits.com/pdfs/ANNE-50+.pdf
Though these may be over the top for something operating only at 10MHz.

[cncjerry]

You should also check your spectrum analyzer reference input specs for phase noise and harmonics requirements prior to running the Trimble into it.  For instance, there is a warning on the 856x spectrum analyzer: Analyzer performance will be degraded unless frequency reference phase-noise and spurious signals are <140 dBc/Hz, referred to 10 MHz at a 100 Hz to 10 kHz offset.

A quick search of the Trimble returned this:

Harmonic level          –40 dBc max
Spurious                   –70 dBc max
Phase noise       10 Hz –120 dBc/Hz
                      100 Hz –135 dBc/Hz
                        1 kHz –135 dBc/Hz
                      10 kHz –145 dBc/Hz
                    100 kHz –145 dBc/Hz

The reference oscillators in the HP analyzers usually exceed these levels.

~Jerry

[tkamiya]

cncjerry, thanks.  I will check that out.

Maybe I can just make one.  Take a BNC connector, put 51 ohm register and 0.01 microF capacitor in series between center pin and ground and call it a day.  At the frequency required and precision (which is none) would anyone think that won't do the job?

[cncjerry]

I've debated the need for a reference into an analyzer.  A lot of HP analyzers have standby circuits for the oscillators so once set they drift very little.  I wish my tracking generator had a reference input as that drifts at startup.  My HP receivers, Selective level meters, counters, frequency generators (with the exception of the lower phase noise type) all benefit.  A saw a few trimbles that had benefited from added HP oscillators and I think some came with DOCXOs so the phase noise was excellent.

Some of the Trimbles do better with improved power supplies.  I did a search and found a couple tested for phase noise and I think a certain low cost switcher performed the best.

Depending on your drive level (my Lucents drive at up to +17dbm), you can use a passive distribution splitter.  I have two made by minicircuits that seem to work well, both have 16 outputs, one SMA and the other BNC.  I worry about phase noise and didn't want to add any switching though it is also argued that a square wave has potentially lower jitter as the zero crossing is faster. Think of | vs / with a sine wave.  I have no way to measure it so don't have an opinion but it makes sense.  I personally love the look of a nice sine wave.

https://www.ebay.com/itm/NEW-MINI-CIRCUITS-16-WAY-POWER-SPLITTER-COMBINER-ZFSC-16-12/222782618448?hash=item33dee0fb50:g:C94AAOSwax5YnzKN

Lastly, BG7TBL makes these nice little 10Mhz band pass filters that clean up all kinds of defects:

https://www.ebay.com/itm/New-10M-10MHz-BPF-Band-pass-filter-Low-insertion-loss/263307031751?hash=item3d4e5274c7:g:RegAAOSwi~5Z2sw-

They are fragile but run a 10Mhz square wave through it and you would be amazed.

~Jerry

[tkamiya]

I have a question....

When you inject a reference 10Mhz into a test instrument, does it discipline the internal oscillator or does it replace the internal oscillator?  My counter HP counter has a switch internal/external but none others do, so it seems external reference "takes over" somehow.

What I'm really needing to do is to assure counters are accurate and RF signal generators are accurate.  I guess counters are fairly easy as it is only used as a gate control.  RF generator is a synthesized type, so I'd think there is a PLL circuit inside and some master oscillators are compared to a reference.  So I'm guessing phase noise may be a concern but a wave form (sine or square) doesn't really matter?

I'm quite new to this level of equipment.

[jpb]

I have a question....

When you inject a reference 10Mhz into a test instrument, does it discipline the internal oscillator or does it replace the internal oscillator?  My counter HP counter has a switch internal/external but none others do, so it seems external reference "takes over" somehow.

What I'm really needing to do is to assure counters are accurate and RF signal generators are accurate.  I guess counters are fairly easy as it is only used as a gate control.  RF generator is a synthesized type, so I'd think there is a PLL circuit inside and some master oscillators are compared to a reference.  So I'm guessing phase noise may be a concern but a wave form (sine or square) doesn't really matter?

I'm quite new to this level of equipment.

The internal oscillator is phase locked to the external reference, at least on the instruments I have. The spec required on the reference is a reflection of the internal oscillator - where there is an optional OCXO the reference must be closer to 10MHz than for a standard TCXO. For counters with relatively long gate times the external reference will determine timing and phase noise is less of an issue than it would be for a spectrum analyser perhaps.

Having said that, some other instruments switch over to the external reference clock completely - Tabor AWGs give you the option of switching to the external clock and if there is no clock then the instrument won't run until you manually switch back to the internal clock (I think this is the case from reading the manual, I don't actually have one).

[tkamiya]

Thank you.

[cdev]

You definitely do not want a 12v DC signal going into any of your test equipment.

You can make a DC block for RF with a bit of impedance appropriate microstrip and a capacitor and two RF coaxial connectors. For 10 MHz you can use through hole parts. I would make it on a scrap of PCB the square SMAs . (sandwiched between two of them) You could perhaps consider using a 1:1 isolation transformer (balun) also on the input of your test equipment after the RF block to prevent ground loops if its a critical use. (the ground loop could introduce error)

You can also find BNCs that have baluns attached to them, I think.

NIST suggests the use of baluns for distributing 10MHz.

I think I saw this in here, but even if not this is a good reference

"The Use of GPS Disciplined Oscillators as Primary Frequency Standards for Calibration and Metrology Laboratories"
Michael A. Lombardi
https://tf.nist.gov/general/pdf/2297.pdf


Another is this one on "Distributing UTC":

https://tf.nist.gov/general/pdf/2428.pdf
I have a Extron video distribution amp I got on ebay for under $20 which has three video channels and one TTL channel so its ideal for both 10MHz and 1PPS. But the input needs to be attenuated a bit, coming straight out of the GPSDO its too strong.

[tkamiya]

Balun and isolation transformers are two different things though, as most baluns do have DC continuity.  I think though, to prevent ground loop, a capacitor in each signal line and ground line will do.  I am not going to be that critical about my installation though.

[cdev]

I meant an isolation transformer then.

Anyway, I am gathering they do this because of junk on the power lines and so on. Just noise that might throw off the precision in a laboratory setting.

I agree with you, its overkill for people who just need it for a frequency reference.

[cncjerry]

cncjerry, thanks.  I will check that out.

Maybe I can just make one.  Take a BNC connector, put 51 ohm register and 0.01 microF capacitor in series between center pin and ground and call it a day.  At the frequency required and precision (which is none) would anyone think that won't do the job?

Not to drag this out, but you are just trying to block the DC, not drop the level, correct?  In that case, just put the .01uf (probably .1uf would be better) in series with the center pin, no connection to the ground at all.  This will block the DC and have about 1/10 ohm (if I did my math correctly) impedance at 10Mhz.  A .01uf would be 1.6 ohms.  Some devices block the DC on the external reference already but I wouldn't risk it.

I think the put DC on the lines as they expect to drive unpowered extender amps.  The simpler method as someone pointed out is to just clip out those inductors if you never plan to use extension amps.  Also making up cables with caps in series with the center pin will be difficult to maintain connector strength and if you are like me, some day you will grab the wrong cable.

~Jerry

[tkamiya]

I'm sorry this thread is confusing.  I changed my mind few times as I learn more.  I talked about both using JUST the distribution amplifier and block the DC, and using taps and terminating the line at the end.  On the latter part, I need to provide a 50 ohm impedance and block the DC.

At this point, I'm thinking I'll just make this as complicated as possible and call it my Rube Goldberg setup.

[bson]

I got a couple of terminators with my 8140, but if I didn't have one I'd just create one from a BNC connector, a ceramic 10-20nF cap, and a 50ohm resistor.