[time-nuts] Nifty "MINI TIC" for DMTD work detail Info please read

Bob Camp lists at rtty.us
Wed Nov 21 14:36:13 UTC 2012


Hi

I found that I was past the network analyzers 120 db floor with two stages at 10 MHz.

Actually the reactive termination thing is a bit of a "maybe yes / maybe no" sort of thing. It seems to help with some setups and not with others. What does work just about always is to terminate the mixer in > 10X it's impedance at the IF port. You still terminate properly at the 1X and 2X RF frequencies. Of the two the double is usually the more important. 

Bob


On Nov 21, 2012, at 5:02 AM, Bruce Griffiths <bruce.griffiths at xtra.co.nz> wrote:

> Have achieved around 120db with 3 cascaded 2N3904 CB amps.
> Something like a BFQ18 broadband transistor with its built in emitter ballasting may be a better choice.
> 
> They work well in a pushpull CB Norton transformer feedback amp to boost the +7dBm output of an OCXO to +19dBm.
> Unfortunately Norton CB transformer feedback amplifiers have very little reverse isolation.
> 
> The next most effective step in improving the DMTD would be to capacitively terminate the mixer IF port and eliminate the cascaded RC filter chain at the mixer output.
> 
> Bruce
> 
> Bob Camp wrote:
>> Hi
>> 
>> A pair of common base amps will get you to what ever your layout will permit. A single common base at 10 MHz should get you to at least 60 db with a little care. If you run a reasonable transistor you can run levels that will fry a 10 or a 13 dbm mixer. Again, all the discrete circuits work pretty well. It's the modular stuff (especially when driven hard) you have to watch out for.
>> 
>> Bob
>> 
>> On Nov 21, 2012, at 1:43 AM, Bruce Griffiths<bruce.griffiths at xtra.co.nz>  wrote:
>> 
>>   
>>> Reverse isolation at 10MHz is around 43dB.
>>> This can be improved significantly by using a Sziklai pair instead of a single transistor in each amplifier.
>>> However one could also do this in a CB amp.
>>> 
>>> Since phase detectors require LO and RF input levels of around 10dBm either deign should suffice.
>>> NIST have shown that high level mixers appear to be somewhat noisier than either the ZRPD1 and 10534A.
>>> Whilst their custom mixer utilising diode connected 2N222A's has the lowest flicker noise the increased PCB board complexity may not be worthwhile for this application.
>>> Reducing the isolation amplifier noise contribution should be more effective than any other changes to the DMTD design.
>>> 
>>> Bruce
>>> 
>>> Bob Camp wrote:
>>>     
>>>> Hi
>>>> 
>>>> A few more parts and likely not quite as much isolation. None of that means it won't do a good job though.
>>>> 
>>>> Bob
>>>> 
>>>> On Nov 20, 2012, at 11:20 PM, Bruce Griffiths<bruce.griffiths at xtra.co.nz>   wrote:
>>>> 
>>>> 
>>>>       
>>>>> The attached circuit is somewhat more efficient and a little quieter.
>>>>> With a little elaboration lower distortion is possible.
>>>>> 
>>>>> Bruce
>>>>> 
>>>>> Bob Camp wrote:
>>>>> 
>>>>>         
>>>>>> Hi
>>>>>> 
>>>>>> Given that common practice is to mis-match the IF port on the mixers, it's probably not realistic to depend on exact match for isolation. Simple / cheap common base buffers likely are a better approach. Lots of isolation and not much flicker noise.
>>>>>> 
>>>>>> Bob
>>>>>> 
>>>>>> On Nov 20, 2012, at 9:01 PM, Bruce Griffiths<bruce.griffiths at xtra.co.nz>    wrote:
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>>           
>>>>>>> The details about matching (if any) used in the Czech DMTD would be informative.
>>>>>>> 
>>>>>>> To avoid degrading the performance of the DMTD system below that imposed by the mixers any isolation amps used will need a flicker phase noise floor below that of the mixers.
>>>>>>> Even an opamp based isolation amplifier can be at least 10dB quieter (for offsets of 10Hz and below) than a typical minicircuits RF amp.
>>>>>>> This is still about 10dB or so worse than a good mixer.
>>>>>>> A well designed low gain isolation amp built with discrete transistors can have significantly lower additive phase noise than an opamp.
>>>>>>> 
>>>>>>> To reduce the DMTD system noise one can either:
>>>>>>> 
>>>>>>> 1) Carefully match all ports using series resistors, pads etc as necessary to achieve the required isolation together with a high output low flicker phase noise amplifier to drive the splitter
>>>>>>> 
>>>>>>> 2) Use isolation amplifiers with very low flicker phase noise.
>>>>>>> 
>>>>>>> Some isolation between the 2 RF inputs of a DMTD is usually necessary to avoid injection locking of the 2 sources being compared.
>>>>>>> 
>>>>>>> 
>>>>>>> Bruce
>>>>>>> 
>>>>>>> Bob Camp wrote:
>>>>>>> 
>>>>>>> 
>>>>>>>             
>>>>>>>> Hi
>>>>>>>> 
>>>>>>>> Those isolation numbers are *highly* dependent on very good matching at all ports. That's rarely the case unless you have a bunch of pads running around the system.
>>>>>>>> 
>>>>>>>> Bob
>>>>>>>> 
>>>>>>>> On Nov 20, 2012, at 5:49 PM, Bruce Griffiths<bruce.griffiths at xtra.co.nz>     wrote:
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>>               
>>>>>>>>> Typical Minicircuits SMT RF amps have a phase noise at best 20dB worse (@10Hz offset) than the mixer/phase detector.
>>>>>>>>> Their reverse isolation is quite low (<<40dB)
>>>>>>>>> 
>>>>>>>>> The principle reason that the Czech DMTD has such low internal noise is due to the absence of any isolation amplifiers.
>>>>>>>>> They use the outputs of a 2 way splitter to drive the LO inputs of the mixers.
>>>>>>>>> 
>>>>>>>>> A output to output isolation of 40dB or more at 10MHz is possible with some minicicuits splitters (e.g. SYPS-2-1).
>>>>>>>>> The ZRPD1 has an RF1 - RF2 isolation of around 70dB at 10MHz.
>>>>>>>>> 
>>>>>>>>> With a channel to channel isolation of around 110dB for a 2x ZRPD1 + Splitter combination isolation amplifiers may not be necessary.
>>>>>>>>> 
>>>>>>>>> Bruce
>>>>>>>>> 
>>>>>>>>> Bruce Griffiths wrote:
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>>                 
>>>>>>>>>> Since mixer noise is one of the limiting factors using a mixer with low flicker noise will help.
>>>>>>>>>> NIST found that a custom mixer using diode connected (collector base short) 2N222As had a significantly lower flicker phase noise than either the ZRPD1 or the 10534A.
>>>>>>>>>> They used off the shelf 1:5 impedance ratio transformers (probably from Minicircuits).
>>>>>>>>>> Another issue is the flicker phase noise of any isolation amplifiers used.
>>>>>>>>>> This is particularly critical if each mixer uses its own isolation amplifiers.
>>>>>>>>>> 
>>>>>>>>>> My current amplifier phase noise measurement setup (for measuring the additive PN of a pair of well matched amplifiers) has a self noise of around -170dBc/Hz @ 1Hz offset for a 10MHz input.
>>>>>>>>>> Ideally the additive phase noise of any isolation amplifiers should be well below that of the mixers.
>>>>>>>>>> 
>>>>>>>>>> Bruce
>>>>>>>>>> 
>>>>>>>>>> EWKehren at aol.com wrote:
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>>                   
>>>>>>>>>>> Yes Bruce I have the paper. I am not suggesting to copy it verbatim but if
>>>>>>>>>>> there is a way to reach reasonable priced 1 E-14 members of the list should
>>>>>>>>>>> pipe  in. I am willing to do an other board. the rest of the systems well
>>>>>>>>>>> on its way.  Einally after three years.
>>>>>>>>>>> Bert Kehren
>>>>>>>>>>> 
>>>>>>>>>>> 
>>>>>>>>>>> In a message dated 11/20/2012 3:28:57 P.M. Eastern Standard Time,
>>>>>>>>>>> bruce.griffiths at xtra.co.nz writes:
>>>>>>>>>>> 
>>>>>>>>>>> EWKehren at aol.com wrote:
>>>>>>>>>>> 
>>>>>>>>>>> 
>>>>>>>>>>> 
>>>>>>>>>>>                     
>>>>>>>>>>>> The D/M is being revisited  because  of the counter performance. 1 E-13 is
>>>>>>>>>>>> easily attainable  but the Czech IREE  published a paper and claim 2 E-15.
>>>>>>>>>>>> 
>>>>>>>>>>>> 
>>>>>>>>>>>> 
>>>>>>>>>>>> 
>>>>>>>>>>>>                       
>>>>>>>>>>> Do you mean the paper ""optimization of dual-mixer time-difference
>>>>>>>>>>> multiplier" ?
>>>>>>>>>>> The ZCD developed in this is a bit of a kludge and is far  from optimum.
>>>>>>>>>>> Reverse engineering the circuit from the description given in  the paper
>>>>>>>>>>> isn't too difficult.
>>>>>>>>>>> They claim an instrument limited ADEV of  ~7E-15 @ 1s.
>>>>>>>>>>> Do you have a copy of this paper?
>>>>>>>>>>> 
>>>>>>>>>>> 
>>>>>>>>>>> 
>>>>>>>>>>>                     
>>>>>>>>>>>> Bert  Kehren   Miami
>>>>>>>>>>>>  _______________________________________________
>>>>>>>>>>>> 
>>>>>>>>>>>> 
>>>>>>>>>>>> 
>>>>>>>>>>>> 
>>>>>>>>>>>>                       
>>>>>>>>>>> Bruce
>>>>>>>>>>> 
>>>>>>>>>>> 
>>>>>>>>>>> 
>>>>>>>>>>> 
>>>>>>>>>>>                     
>>>>>>>>> _______________________________________________
>>>>>>>>> time-nuts mailing list -- time-nuts at febo.com
>>>>>>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>>>>>>> and follow the instructions there.
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>>                 
>>>>>>>> _______________________________________________
>>>>>>>> time-nuts mailing list -- time-nuts at febo.com
>>>>>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>>>>>> and follow the instructions there.
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>>               
>>>>>>> _______________________________________________
>>>>>>> time-nuts mailing list -- time-nuts at febo.com
>>>>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>>>>> and follow the instructions there.
>>>>>>> 
>>>>>>> 
>>>>>>>             
>>>>>> _______________________________________________
>>>>>> time-nuts mailing list -- time-nuts at febo.com
>>>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>>>> and follow the instructions there.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>>           
>>>>> <DualCEBuffer3.gif>_______________________________________________
>>>>> time-nuts mailing list -- time-nuts at febo.com
>>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>>> and follow the instructions there.
>>>>> 
>>>>>         
>>>> _______________________________________________
>>>> time-nuts mailing list -- time-nuts at febo.com
>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>> and follow the instructions there.
>>>> 
>>>> 
>>>>       
>>> 
>>> _______________________________________________
>>> time-nuts mailing list -- time-nuts at febo.com
>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>> and follow the instructions there.
>>>     
>> 
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at febo.com
>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
>> 
>>   
> 
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.




More information about the time-nuts mailing list