[time-nuts] Build my own dist. amp ??
Bruce Griffiths
bruce.griffiths at xtra.co.nz
Wed Dec 17 05:54:58 UTC 2008
Brian
BriMDavis at aol.com wrote:
> Bruce Griffiths wrote:
>
>
>> Very few OCXOs and other frequency standards have outputs
>> lower than +4dBm. Most have outputs in the +7dBm to +15dBm range.
>>
>>
> And what happens to that +7 dBm source if you split it 8 ways and
> add some padding for additional matching and isolation, plus some
> input transformer and/or filter losses?
>
That is a very effective way of elevating the phase noise floor.
Its usually far better to amplify the input and then split the output
maintaining a gain to the splitter outputs of at least 0dB.
>
> Or what if the reference being distributed comes off a multiplier
> chain or VCO at a lower output level?
>
>
Again, its better where possible to ensure that the input to a
multiplier is as high as is reasonably possible, most multipliers are
more efficient with higher inputs.
If one has a low level signal it is best to use a low noise buffer driving a
> Is there room in your world view for a simple, low cost design
> that doesn't reach the ultimate noise floor, but covers a wide
> range of input frequencies with decent noise floor performance?
>
>
None, where it severely compromises the phase noise floor of the
reference frequency source.
Even a Thunderbolt OCXO phase noise floor will be degraded by a system
using this device (unless of course it is used to drive N of them in
parallel and their outputs are recombined - however thats more expensive
than implementing a good design in the first place).
>> Using an unnecessarily wide band device in a frequency distribution
>> system isn't usually a particularly good idea.
>>
>>
> It isn't "unecessary" when the circuit needs to operate over
> that wide of an input frequency range in its' intended range
> of applications.
>
>
Its almost always better to design the circuit to suit the operating
frequency, often this just means using suitable output and input filters.
>> The only way to find out what the close in phase noise characteristics
>> are is to construct an amplifier using such a device and measure it.
>>
>>
> Some data sheets contain this information at spot frequencies;
> this one does not.
>
Very few datasheets from Maxim specify much about the noise
characteristics of such devices.
>
> Unfortunately, I do not have access to a phase noise test system
> at this point in time, hence my question to the list.
>
>
>
You can easily cobble one together using whatever sound card your PC has
together with a mixer and a few inexpensive opamps, filters etc.
> Brian
>
> p.s.
>
> I deleted most of my original reply before posting this.
>
> Your posts contain many condescending remarks insinuating
> misuse of the part, and how easy it is to do it properly,
> yet you refuse to answer a simple question, to wit:
>
> Can you point to something, IC or discrete, that will
> cover sources in the 10-200 MHz range with similar or
> better reverse isolation at comparable size and cost ?
>
>
>
The criticism is aimed at the concept not the person.
The question is largely irrelevant.
The use of such a device in the way which you intend is far from optimum.
But of course you are free to do that.
However, I think that its important that the limitations of such an
approach be clearly stated.
Bruce
More information about the time-nuts
mailing list