lab standard clock buffer module.

[tkamiya]

I am designing a sort of "fan out" board for various timing frequency sources - the kind that keeps lab full of gear in sync.  I am finding out there is no agreed upon values and levels on these things, so I'm trying to accommodate most of situations.  First in order is a unit that plugs directly into the frequency source to bring the output to usable level AND protect the source from reverse input.

For example, PRS-10 puts out 10MHz, 0.5Vrms, 1.41Vpp, and +7dbm (50ohm)
HP8644 expects 0.5Vrms (7dbm) to 2Vrms (19dbm)

Account for loss and mismatches, it seems prudent to provide 15db or so amplification, adjustable to 0db.  That would probably mean using OP amp. 

There are plenty of video amp that seem appropriate for this kind of use but problem is, almost all of them are surface mount.  I would very much like to avoid this.  So I'm looking for 8 pin can or DIP packaging.  In addition to compensating for loss, I plan to use an attenuation and transformer to further isolate/protect the source.  So minimum 20db of gain is necessary.

Given these parameters, can anyone give me a recommendation?  This really isn't a distribution amplifier but sort of a buffer.  So I don't mind spending few dollars on it.  For distribution amps, I already have symmetricon set up.

[David Hess]

Good devices should be intolerant of signal levels but many are not.  The most universal simple solution is to provide TTL levels at the receiver so low is below 0.8 volts and high is above 2.0 volts which can be satisfied by a 5 volt CMOS double terminated output without exceeding 5 volts.  More sophisticated solutions can provide a full 0 to 5 volts from a 5 volt supply with proper double termination without exceeding 5 volts if the load termination is missing by using a linear driver and positive feedback but this is probably unnecessary.

[tkamiya]

Many of my equipment specifies sine wave though....

[awallin]

I tried to collect a few links on this page, for design inspiration:
https://www.ohwr.org/projects/pda-8ch-fda-8ch/wiki/similar-projects

if you find open designs and/or measurements that are not listed on that page - please post here! thanks.

[LapTop006]

For some very similar use cases I've been playing with the Linear Tech LTC6957 as the frontend.

My case is fixed frequency (just 10MHz) so I then just use standard gates to buffer (74S120 if you're driving 50-ohm loads), and LC filters to get it back to a sine.

You might also want to look at the Spectracom 8140 system which may be useful as an existing solution available cheaply used.

[David Hess]

If you really want that much gain then 20dB times 10 MHz requires a 100 MHz gain-bandwidth product which is nothing special for a fast operational amplifier.  Driving a double terminated 50 ohm load is more of a limitation at 4 volts peak even without extra attenuation; my first choice of the inexpensive AD8055 can only just make that output level because of supply voltage limitations.

For a completely integrated solution in a DIP package, I would probably use something like an LT1252 if the supply voltages are trimmed (+/-8 volts?)for minimum power dissipation or perhaps even better, the dual LT1253 version with both sections in parallel driving a 100 ohm series resistor to support the necessary power dissipation.  The dual AD812 or single AD811 could also work.

Be careful about filtering if the application requires accurate phase.  The filter poles need to be removed from the 10 MHz signal to prevent temperature dependent so bandpass LC filtering can cause problems.  Better is an LC filter with a higher cutoff and notches at the 2nd and 3rd harmonics if necessary.

[ArthurDent]

I've used a modified RBGS video amp that was easily changed from 75 ohm to 50 ohm for years. You can use the inputs separately for different standards or daisy chain the inputs and use just one standard. The 'S' outputs gives me square wave logic level outputs and the 'RBG' outputs give me good clean 10Mhz sine waves. The amp I'm using has separate buffer ICs for each output giving good isolation to eliminate any interaction between outputs.

[DaJMasta]

Opamps or dedicated driver ICs are common for this, but using TTL and a square wave is going to get the job done fine.  As said, if you really need a sine wave, you can filter it, but generally equipment doesn't.  I'd aim for the 10dB or so ballpark and most stuff should run off it (especially more modern stuff which seems to be more standardized/tolerant), laying out opamps as buffers or dedicated buffer amps are fine ways to do it, just make sure the package/circuit can put out the power required and has the slew rated needed to give you a good looking sine instead of an underleveled triangle wave.  In my experience, most 10MHz inputs do not require the 50 ohm power levels and are higher impedance, but if you take an amp that can drive the output at 1Vpp or so of a 50 ohm terminated load, then 2Vpp or a little more of a high impedance load, you should cover most input signal requirements and you can still use logic level single supply stuff for your distribution box.

[tkamiya]

Thanks everybody.  I see *some* dip packages still left to use.  I'll be checking out today.

[tkamiya]

I found what seem to be perfect for my purpose.  AD8055 and AD8056.  Only issue is, they are pretty pricey.

I wonder if there is anything wrong with using mini-circuit's gain module amplifier?  MAR-4 and MAR-8 series appears to be perfect.

[David Hess]

I found what seem to be perfect for my purpose.  AD8055 and AD8056.  Only issue is, they are pretty pricey.

The AD8055 was my first choice based on price but maybe I do not understand your requirements.  Its limited +/-3 volt output range (2.12 volts RMS) is less than 2 volts RMS plus 3dB loss plus 3dB of operating margin.