[time-nuts] Fwd: PRS10 question - revisiting an old one with an accent on small signal detection

Mike Cook michael.cook at sfr.fr
Tue Mar 17 04:09:08 EDT 2015


Hi guys,
 A year ago I posted a question concerning the 1PPS output from my PRS10.  I had thought it an anomaly, but other users indicated that the signal profile was common to other units. I have since added 2 PRS10s (actually TDS12s) to my bench and they also show the same profile.  Jarl Risum was very helpful  in my investigations, and proposed using inverters to get a better shaped signal. I tried this out, but am not satisfied with the induced delay. In the end I settled on an AND gate for the project. 
In the forwarded mail below (hope the pic passes the censors) you can see the object of my irritattion is that although I have a nicely conditioned output, there is too much delay between the initital 1PPS input rise and the gate response for even mildly afflicted timenut . 
I couldn’t figure out what the source of this was until I reviewed it the other day with an object of distributing 1PPS over the bench. From looking at  74HC device data sheets I think the issue is that the start of the 1PPS pulse input, (the bit I am interested in) does not reach the trigger level needed for CMOS gates. 

Does that seem a reasonable assumption?, and if so is there a solution? 

 I am wondering whether I could use a comparator? The trigger level of a LT1719 is better than 6mv but I fear that I might get false triggering  as my scope indicates that the 0V level has <=2mv noise. I don’t have one to test so I would appreciate your input. I also can’t see how I might implement it if it is worth testing. 

Regards,Mike

> Début du message réexpédié :
> 
> De: mike cook <cook.michael at orange.fr>
> Objet: Rép : PRS10 question
> Date: 1 mars 2014 16:50:12 UTC+1
> À: Jarl Risum <jarl.risum at gmail.com>
> 
> Hi Jarl,
>   Sorry about the late update but I was waiting for parts.
> 
> I tried using a 74HC04 (I didn't have an HC14 but I only think the they differ in performance) in your suggested config on a separately powered breadboard but that doesn't give me what I want .
> It just gives a slightly attenuated copy of the original, which is to say I still have the slow rise to vcc after the initial 10ns of good rise.  What I was looking for is something that triggers on the first 10ns and then forgets everything else after that, putting vcc on the output, until the input drops to 0, when the output should also go to zero. That is a simple AND operation. How about an AND gate? 
> So I tried a 74HC7001. Now I get 10ns rise to 2V and 20ns to Vcc. There is a bit of overshoot but none of the ringing in the original plus a shorter settling time that I can probably smooth out  if I needed.  There is as you can see quite a bit of delay which I would need to characterize.  However it looks much more usable.
> 
> Regards,
> Mike
>  
> 
> 
> 
> 
> 
> Le 14 févr. 2014 à 16:08, Jarl Risum a écrit :
> 
>> Mike,
>> 
>> Thanks for the feedback. OK - the coax cable explains your result so far. You may try an ordinary 10:1 probe directly on the BNC connector on the breakout board, but be careful to check - and if necessary correct - the probe response before use. The probe needs to have at least 100 MHz bandwidth.
>> 
>> Your idea to use a buffer attached directly to the BNC connector on the breakout board is of course fine, but you should avoid loading pin 8 with anything except what it is intended for: as a supply to a lab. quality 10 kohm 10 turn potentiometer as an external means to correct the C-field in the Rb Physics Package. Pin 8 is a precision 5V reference also used by the internal 10 turn C-field potentiometer and you are almost sure to disturb the calibration if you load pin 8 with anything else.  Please take a look at diagram RB_H1  (10MHz Ovenized Oscillator). The 5V reference circuit is at the top of the page, U100 and U102B.
>> 
>> If you do not wish to add a 7805 5V regulator IC to your circuit yourself you might be able to get a few mA's from the 5V regulator on the breakout board.
>> 
>> As a final remark I would like to draw your attention to the possibilty of using the 10 MHz output instead to feed a divider chain or a PIC divider which could output the 1 PPS where you need it. This solution would be somewhat more flexible with respect to interface problems since the 10 MHz could be distributed in coax cable over 10's of meters without any problem.
>> 
>> Cheers
>> Jarl
>> 
>> 
>> 
>> 
>> 2014-02-14 14:57 GMT+01:00 mike cook <cook.michael at orange.fr <mailto:cook.michael at orange.fr>>:
>> Jarl,
>> Thanks for your tech comments. As far as electronics goes, I don't rise much above the level of a script-kiddie.  I have in fact got a 2 meter 50Ohm coax into my scope.  I will try to probe the output of the 74H140 to see what that looks like.  Rather than use coax to connect in a buffer circuit, I was thinking of a breadboard dongle with BNC male and female connectors soldered to it, maybe slid into a metal tube once tested. I wouldn't be surprised to find it has already been invented. I might be able to steal some juice from the PRBB without braking anything. I should have 5V on pin 8 if I can get at it. I was only thinking of including the single output. If I can get a good square wave, I would use it via a distribution amp. I'll let you know how I get on. 
>> Mike
>> 
>> Le 14 févr. 2014 à 12:17, Jarl Risum a écrit :
>> 
>>> Mike,
>>> 
>>> Thanks for the update. 
>>> 
>>> As you know already from the diagram of the Breakout Board RB_F7, the 1 PPS output is a 74HC14-7 inverter with a series resistor of 1 kohm to protect the gate from external "accidents". In order to speed up the risetime, SRS has put a 100 pF capacitor in parallel to the 1 kohm resistor, but this value cannot be increased while still protecting the 74HC17 output from a short or similar abuse.
>>> 
>>> What I would guess from the scope picture you attached is, that you use a 1 m long piece of coaxial cable between the breakout board and your oscilloscope (or a 1:1 probe with 100 pF input capacitance). That would explain the waveform, since the fast risetime portion is about half of the total height of the total step followed by the slow rise due to the 1 kohm resistor charging the parallel 100 + 100 pF from the initial voltage step and all the way up to Vcc.
>>> 
>>> If you want to examine the waveform without capacitive loading effects you would have to use a (frequency compensated) probe with a very low input capacitance and perhaps even tap the waveform directly on pin 12 of the 74HC14.
>>> 
>>> If you want to make practical use of the 1 PPS output, I suggest that you attach a buffer circuit to the breakout board output with the shortest possible piece of 75 ohm screened cable. The buffer circuit could consist of an inverter followed by 5 inverters in parallel. The output inverters should each have a low value resistor in series (100 ohms ?) in order to protect the outputs from unequal current sharing. With the eqvivalent of 5x100 ohm series resistors in parallel you would have a much lower source resistance available to drive an external load. You will need to decouple the Vcc pin of the buffer circuit with a 0.1 uF ceramic capacitor to GND with very short leads due to the large circulating currents. The buffer circuit would add some delay, but that would be difficult to avoid in any case.
>>> 
>>> Hope you find a solution to your problem, Mike. Please let me know if I can be of any help.
>>> 
>>> Cheers from
>>> Jarl in Denmark
>>> 
>>> 
>>> 2014-02-14 10:54 GMT+01:00 mike cook <cook.michael at orange.fr <mailto:cook.michael at orange.fr>>:
>>> Hi Jarl,
>>>   I received my breakout board from SRS. Thought I'd give you an update.
>>> I had a mechanical issue with one standoff being too long which I reported back to them. They were kind enough to offer a replacement board at no cost. However I am keeping it as the problem doesn't prevent it from working. As you can see from the attached traces my problem with the slow rise is only half solved (upper trace). The inverters on the board certainly give me a quicker rise time for the start of the trace, better than that of the GPS trace below , but it is only useful for about 650mv as I get a little ringing, it then ramps up slowly as per the original. I'd much rather have the shape of the GPS trace. I had an email chat with a guy at SRS and he thought that the trace wasn't good, but couldn't tell me if it was normal or not. He did suggest me terminating into 50Ohms, but that was not the solution. Maybe I can recondition the input somehow, or on output?  The other parts of the interface, 1PPS in and 10MHz out are good.
>>> Regards,
>>> Have a good one as the Yanks say,
>>> Mike
>>> 
>>> 
>>> 
>>>  <PastedGraphic-1.tiff>
>>> 
>>> 
>>> Début du message réexpédié :
>>> 
>>>> De : mike cook <michael.cook at sfr.fr <mailto:michael.cook at sfr.fr>>
>>>> Date : 28 janvier 2014 21:48:29 HNEC
>>>> À : Jarl Risum <jarl.risum at gmail.com <mailto:jarl.risum at gmail.com>>
>>>> Objet : Rép : PRS10 question
>>>> 
>>>> Jarl,
>>>>   I have ordered an interface board from SR. I hope it will lessen the amount of spaghetti I am wrestling with at the moment.
>>>> I had figured that it would be easy enough to put a better shape on the pulse but didn't want to add to my open plan wiring scheme before I receive the breakout board.
>>>> 
>>>>   thanks for the thought .
>>>> 
>>>> Mike
>>>> 
>>>> 
>>>> Le 28 janv. 2014 à 19:40, Jarl Risum a écrit :
>>>> 
>>>>> Mike,
>>>>> 
>>>>> If you are still worried about the rise time of the 1 PPS output signal from the PRS10 I suggest you to look into the description and diagram of the RB_F7 Connector Interface Board for the PRS10, which you can find somewhere on the SR homepage. On this Interface Board the 1 PPS from the PRS 10 passes two 74HC14 Schmitt inverters before being passed on to the BNC connector for external use. 
>>>>> 
>>>>> The 74HC14 has a rise time of around 7 ns, but of course adds some delay to the pulse as well. 
>>>>> 
>>>>> Cheers from
>>>>> Jarl
>>>>> 
>>> 
>>>     < snip the history>
>>> 
>> 
>> 
> 

"Ceux qui sont prêts à abandonner une liberté essentielle pour obtenir une petite et provisoire sécurité, ne méritent ni liberté ni sécurité."
Benjimin Franklin



More information about the time-nuts mailing list