[time-nuts] Bad batch of HP10811's
Didier Juges
didier at cox.net
Fri Jan 19 23:53:28 EST 2007
Warmup data on the HP 10811:
Here is some data collected on the internal HP 10811 timebase in my HP
5370A counter, which had been turned off and plugged off for about 3 days.
21:39:25, 9999798.6
21:39:35, 9999806.0
21:39:46, 9999822.6
21:39:55, 9999836.8
21:40:06, 9999853.2
21:40:15, 9999866.8
21:40:26, 9999881.7
21:40:35, 9999892.4
21:40:46, 9999904.1
21:40:56, 9999913.1
21:40:58, 9999915.2
21:41:07, 9999923.0
21:41:16, 9999930.1
21:41:25, 9999936.5
21:41:36, 9999944.0
21:41:46, 9999949.3
21:41:55, 9999954.3
21:42:06, 9999959.9
21:42:15, 9999964.0
21:42:26, 9999968.6
21:42:35, 9999971.8
21:42:45, 9999975.2
21:42:54, 9999977.9
21:43:05, 9999981.0
21:43:17, 9999983.9
21:43:26, 9999985.9
21:43:35, 9999987.8
21:43:46, 9999990.0
21:43:55, 9999991.3
21:44:07, 9999992.9
21:44:16, 9999994.0
21:44:25, 9999994.9
21:44:36, 9999996.1
21:44:45, 9999996.9
21:44:56, 9999997.7
21:45:06, 9999998.3
21:45:15, 9999998.9
21:45:17, 9999999.0
21:45:19, 9999999.1
21:45:21, 9999999.2
21:45:24, 9999999.3
21:45:26, 9999999.4
21:45:36, 9999999.7
21:45:45, 9999999.8
21:45:56, 9999999.9
21:46:05, 9999999.9
21:46:08, 9999999.9
21:46:10, 10000000.0
After that, the display was a stable 10,000,000.0 for the next 15
minutes (one reading was 10,000,000.1), then I stopped the data collection.
The counter used was the HP 5334B with the Thunderbolt as external
reference.
Ambient temperature: about 22 degrees C (cool)
Bottom line: about 200 Hz from a cold start and about 7 minutes to
within 0.1 Hz of final frequency.
Didier KO4BB
Didier Juges wrote:
> That is interesting. I have not checked the 10811 for that particular behavior, but I have checked a number of cheap, non TC crystal oscillators, and the initial drift has always been much smaller, in the order of 100's of Hz, not 1,000's, scaled for 10 MHz.
>
> I refer to the crystal oscillators inside my various ham radio gear and non-TCXO controlled instruments, such as the 5334B counter I have.
>
> It may be due to the fact that the crystal in the HP 10811 is optimized for flat temperature sensitivity around the operating point of the oven, when the cheap crystals I have checked are optimized around ambient (if they are optimized at all).
>
> I have several 10811's here (one by itself, the others inside instruments) and now that you picked my attention, I have to do the test :-)
>
> It's easy to plot it against the Thunderbolt with a counter hooked to the GPIB controller and with my logging software running. I shall report here soon :-)
>
> Didier KO4BB
>
>
> ---- Jeroen Bastemeijer <J.Bastemeijer at TUDelft.nl> wrote:
>
>> Dear Time-nuts,
>>
>> Just a quick experiment. I started with the cold 10811 OCXO (switched
>> off for >24 hours). I connected it to a counter and noted the frequency
>> every 30 seconds. (last two measurement each 5 minutes).
>>
>> You can see the curve in the attached PDF. The offset at t=0 was 1706 Hz
>> (Compared to 10.000000MHz).
>>
>> After 10 minutes warm-up the total offset is 0.75Hz (compared to
>> 10.000000MHz) and offset is 0.12Hz from the "final" value. According to
>> the spec the offset should be within 5*10^-9.
>>
>> From this I conclude the oscillator is not faulty. Sorry, Didier ;-)
>>
>> The possible reason for this offset is probably not the crystal, but the
>> tempco of the rest of the circuit. The oven heats up till about 82
>> degrees centigrade. A PN-juntcion at room temperature (22 degrees) will
>> increase 60 degrees in temperrature, it's forward voltage will change by
>> approx. 120 mV!!!!! This change will lead to a significant change in
>> circuit properties (e.g. of the oscillator and the attached AGC circuit).
>>
>> Probably this was the reason HP specfied it's oscillator first after 10
>> mins of warm-up.
>>
>> This makes me wonder: If an oscillator gives a big offset, the
>> ovencontroller may be faulty... not working at all, or not getting warm
>> enough.
>>
>> Jeroen PE1RGE
>>
>>
>>
>> The raw data:
>> Time Frequncy [MHz] Offset [Hz]
>> 0 9,998294 1706
>> 30 9,998405 1595
>> 60 9,998645 1355
>> 90 9,998854 1146
>> 120 9,999038 962
>> 150 9,999197 803
>> 180 9,999349 651
>> 210 9,999466 534
>> 240 9,999569 431
>> 270 9,999666 334
>> 300 9,99974 260
>> 330 9,999804 196
>> 360 9,999858 142
>> 390 9,999906 94
>> 420 9,99994027 59,73
>> 450 9,99996696 33,04
>> 480 9,99998633 13,67
>> 510 9,99999979 0,21
>> 540 10,00000062 -0,62
>> 570 9,99999958 0,419999999
>> 600 9,99999925 0,75
>> 900 9,99999933 0,67
>> 1200 9,99999937 0,630000001
>>
>>
>> Didier Juges wrote:
>>
>>
>>> That sounds like an awful lot of drift, absolutely inconsistent with a
>>> crystal oscillator of any kind.
>>>
>>> Either this OCXO is really sick, and just happen to end up at the right
>>> frequency by accident, or it is an OCO (Oven Controlled Oscillator :-)
>>>
>>> Didier KO4BB
>>>
>>> Jeroen Bastemeijer wrote:
>>>
>>>
>>>
>>>> Dear Rick,
>>>>
>>>> Are you sure the 500Hz is too much for a cold oven? I checked my 5345A
>>>> (recently acquired) counter, which was switched off for some time.
>>>> Measuring the frequency directly after applying power and switching on,
>>>> resulted in about 1640 Hz frequency offset. After warm up, the error is
>>>> below 0.4 Hz. (The unit used here is a 10811a). The error drops pretty
>>>> fast, allthough I didn't measure an accurate curve. Is the frequency
>>>> difference caused by the tempco of the oscillator circuit?
>>>>
>>>> Cheers, 73s Jeroen PE1RGE
>>>>
>>>>
>>>>
>>>>
>>> _______________________________________________
>>> time-nuts mailing list
>>> time-nuts at febo.com
>>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>
>>>
>>>
>>>
>>>
>>>
>> --
>> Ing. Jeroen Bastemeijer
>>
>> Delft University of Technology
>> Department of Electrical Engineering
>> Electronic Instrumentation Laboratory
>> Mekelweg 4, Room 13.090
>> 2628 CD Delft
>> The Netherlands
>>
>> Phone: +31.15.27.86542
>> Fax: +31.15.27.85755
>> E-mail: J.Bastemeijer at TUDelft.nl
>>
>>
>
>
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>
>
More information about the time-nuts
mailing list