DIY 10 MHz Distribution Amplifier - Requirements Gathering

[Dwaine]

I bought  Extron ADA 4 300M off eBay for $20.00.  I think I’m going to add a rubidium module and have one input for my GPSDO. 

Nice little project.

[TurboTom]

I also opted for the isolated outputs -- if you've got (or just like) to syncronize a function generator that's not ground-referenced, you're in trouble without. Moreover, ground loops can really ruin your day... Figured out a very compact circuit that directly bolts onto the LPRO101 Rb oscillator and uses the LT6553 video distribution amplifiers. It provides five +7dBm sine outputs and three digital ones (2.5V, 3.3V, 5V). Fits in a very compact enclosure, see here if you like: https://www.eevblog.com/forum/projects/made-a-nice-_home_-for-my-efratom-lpro-101/msg1250122/#msg1250122

Cheers,
Thomas


P.S. I used LAN isolation transformers which also include common mode chokes to reduce capacitive coupling - works very well. Edit: Link corrected

[Gandalf_Sr]

OK, I am convinced that isolation transformers are better, I like Coilcraft for quality and they now offer a range of transformers; this PFD3215-222ME part is pretty cheap at about $0.70 each.  Any thoughts as to which inductance value would be better and why?

Also, in my schematic above, I use 3 of the 4 Op Amps in the MAX4383 package as voltage follower buffers (the 3rd is unconnected but the idea is that it will drive the TTL logic output circuitry), so that means that they simply convert from very high to very low impedance but what (if any) interface components are recommended before they drive the input of the LT6551 video Amplifiers?  At present I have them driving through 10n capacitors (C13/C14) into the voltage -biased high impedance inputs of the LT6551s.  Is this OK?  Do voltage followers need a load?  Also I read in The Art of Electronics that it's preferable to have 1k resistors in the 100% -ve feedback loop.

[EDIT] I think this project would be better (more universal) if it had some form of AGC on the input stage.

[Cerebus]

OK, I am convinced that isolation transformers are better, I like Coilcraft for quality and they now offer a range of transformers; this PFD3215-222ME part is pretty cheap at about $0.70 each.  Any thoughts as to which inductance value would be better and why?

Just pick it on the quoted bandwidth it will handle. If there are graphs of insertion loss versus frequency, then pick for lowest insertion loss at your frequency of interest. You're in luck with that datasheet as it quotes both. So for 10MHz the *-472 and *-682 parts would seem to be your best bet, the *-103 would also perform OK. If you ever want to squeeze a 1MHz signal through it, go for the *-103 as a compromise part.

[jpb]

 I've just bought an Extron ADA 6 300MX HV off ebay for £25 inc postage, it arrived in a couple of days and seems to work nicely (just feeding a 10MHz in from my generator and connecting the output to a scope).

Thanks to all who suggested this route. I can't decide whether to build my GPSDO into it or keep it separate and perhaps try to improve it a 10MHz filter and perhaps a low noise linear supply as well as doing the resistor mod.

[klaus11]

Another Extron project, https://www.jackenhack.com/reference-10-mhz-distribution-amplifier/

I also have ADA 6 300MX HV, rather, waiting for the shipment to arrive.

[Shock]

There is also the Charles Steinmetz mods
https://doc.xdevs.com/doc/Extron/Extron_ADA_6_modifications_for_use_as_10MHz_distribution_amp_STEINMETZ.pdf

[CJay]

It seems you have many devices so an Extron Video Distrubition amplifier mod is probably the way to go.  (I bought a couple of these but still haven't got around to modding them... one day).

I created a 1:4 10MHz and added a pulse stretched 1PPS output for my lab equipment, based on LT6551IMS. Seems to work fine (but I'm no test nut).  Another EEVBlog member built one from a PCB I had spare and seemed happy.  I imagine this has too few outputs to be useful to you.

I also have an Extron, same model I think as the one Gerry Sweeney modded.

They work nicely without the mod, well enough that I am not planning to mod it any time soon unless I suddenly find myself with a pile of resistors and a few days where all my other commitments magically vanish.

[Dwaine]

Any recommendations for a good linear power supply to replace the switch mode one?

[CJay]

The Time Nuts guys seem to like the Power One Linear supplies (now Bel) but I'm not sure you'll get one to fit in the Extron case...

[DaJMasta]

Any recommendations for a good linear power supply to replace the switch mode one?

Rather than that, you can just use a linear regulator on the output of a switchmode supply.  If you can keep the voltage drop low, you don't lose a ton of efficiency, and if you use a linear regulator with good PSRR combined with a bit of bulk capacitance and either a ferrite or a smallish power inductor between the switcher and the linear reg, you can get a pretty dang clean power rail without too much fuss.  While a clean signal is desirable, the stuff taking the distributed reference also can deal with a bit of noise from an external source, especially with a power level of several dBm on the primary tone.

[TheSteve]

I have never seen a commercial 10 MHz distribution system that uses isolation. Are they just cheap when they build them or do they figure it isn't required?

[DaJMasta]

I haven't seen any evidence that it's required, and none of the lab distribution amps I've seen use it either.  Basically, while you want a clean signal, it's not really a "signal level" sort of thing (at least, not when it's at a few dBm) and there is no loop to boost noise from a ground loop to a problematic size.

If you've got special power concerns in your lab or want to use your reference into some project that is very susceptible to noise, maybe there's some value in it, but I don't think test equipment with 10MHz inputs will care.

[TheSteve]

Thanks for the reply. I've always figured a clean signal was the most important part. It can be tough to beat the internal oscillators phase noise(on some higher end gear anyway) with an external reference so you don't want your distribution system to add additional noise to the system. I do think it is important for each output from a distribution system to have its own amplifier so there is no change in levels etc when connecting additional devices.

[klaus11]

How much effort to isolate? is it possible in Extron ADA?

[texaspyro]

How much effort to isolate? is it possible in Extron ADA?

It can be very easy with transformer coupling the output...  but using magnetics can open a bit of a can of worms of other issues.

[klaus11]

This is a good reference without transformer coupling:

https://www.allaboutcircuits.com/projects/how-to-eliminate-ground-loops-with-signal-isolation/

[Dwaine]

Got my FE-5680a today.  Starting the project now.  Fed it with 15volts and 5 volts.  The 1pps and 10mhz sine look pretty good. Runs nice and toasty.   Going to see what the outputs look like after I put it into the Extron Video distribution box.

[BravoV]

Subscribed

[Shock]

There are a few Extron ADA 6 300MX HV units on ebay right now for $25.
This gives you about 4-5 inputs each with 6 outputs, enough for the whole neighborhood.

Attached are the specs, users guide and brochure. The Steinmetz mod recommendations can be found here.

[CJay]

Got my FE-5680a today.  Starting the project now.  Fed it with 15volts and 5 volts.  The 1pps and 10mhz sine look pretty good. Runs nice and toasty.   Going to see what the outputs look like after I put it into the Extron Video distribution box.

Be nice to that FE5680A and give it a heatsink, they need one, there are lifespan problems if you run them 'barefoot' and don't manage the temperature..

[mark03]

Has anyone tried powering additional circuitry, e.g. a GPSDO, from the existing power supply in the Extron 6 300MX?  I measure ~ 17.5V open circuit, 16V loaded by the distribution amp.  Would like to piggyback a TruePosition GPSDO board on the same rail if I can get away with it.

[Shock]

Has anyone tried powering additional circuitry, e.g. a GPSDO, from the existing power supply in the Extron 6 300MX?  I measure ~ 17.5V open circuit, 16V loaded by the distribution amp.  Would like to piggyback a TruePosition GPSDO board on the same rail if I can get away with it.

My GPSDO won't fit (understatement) so it's not an option for me, also I have plans to add more functionality. Gerry discuses the power supply a little in his video.

The question that comes to mind is your GPSDOs current requirements. I'd be replacing the caps though for sure, due to the age of the Extron units it's highly recommended (by me hah). The supply design must be ok for noise spec wise, you would think the circuit would be impacted otherwise. A scope is going to let you know anyway and everyone's lab/workshop and definition of a clean supply is different.

I'd have no idea if it would create any noise issues, but if you think about the Extron as another signal and buffering stage on the same supply as the GPSDO I don't see how it could be to adversely affected and could be ruled out with an isolated bench supply and scope performing comparison tests.

Personally though I wouldn't have mounted a FE5680A style RbDO in there to begin with, I like the idea of the Extron as just a distribution box for signals that can be bypassed to troubleshoot. But I can understand if you have a GPSDO PCB and there is empty space to fill.

[mark03]

The question that comes to mind is your GPSDOs current requirements. I'd be replacing the caps though for sure, due to the age of the Extron units it's highly recommended (by me hah). The supply design must be ok for noise spec wise, you would think the circuit would be impacted otherwise. A scope is going to let you know anyway and everyone's lab/workshop and definition of a clean supply is different.

I'd have no idea if it would create any noise issues, but if you think about the Extron as another signal and buffering stage on the same supply as the GPSDO I don't see how it could be to adversely affected and could be ruled out with an isolated bench supply and scope performing comparison tests.

I still need to measure the current requirements.  But the fit will be fine.  I really like the idea of having this all in one box, and I needed a box for the GPSDO anyway.

Re: switchers in or near a GPSDO, I'm not sure how paranoid to be.  FWIW, the TruePosition GPSDO has two switching power supplies right there on the board... perhaps followed by LDOs.

[jpb]

Personally though I wouldn't have mounted a FE5680A style RbDO in there to begin with, I like the idea of the Extron as just a distribution box for signals that can be bypassed to troubleshoot. But I can understand if you have a GPSDO PCB and there is empty space to fill.

I agree with this, though of course everyone's requirements are different. I might want to distribute more than one reference (e.g. square wave and sin wave or an offset frequency for doing dual balanced mixer stability measurements perhaps) and want to switch between them.
Also, I plan to fit a low noise linear supply as well as a mains filter and perhaps splitters and buffers - there is endless fun to be had.
If connecting to an FE5680A which I think has quite a lot of phase noise, another use for the space might be to have an OCXO that is phase locked to the reference to clean things up before distribution.