[time-nuts] HP10811 current versus temperature

[Didier Juges]

It took longer than I anticipated, but I have the starting point of an
adequate data logger.

For now, I simply recorded the HP10811 heater current versus ambient
temperature in my shack over night.

It was a little more involved than I anticipated because the temperature
variations right now are not much.
So I ended up having to use a processor with a 24 bit ADC (Texas
Instrument/Burr-Brown MSC1210Y5), a stable voltage reference (AD580) and a
precision op-amp (OP-27) to measure the voltage drop across a 1 ohm
precision wire-wound resistor (type has been long forgotten).

The op-amp is wired as a diff-amp to amplify the signal (gain of 5 approx)
and eliminate ground differences between the microprocessor board and the
HP10811 supply board.

I checked the microprocessor data against my HP3456 connected directly
across the 1 ohm resistor and the readings were within 0.1% of each other
over a period of > 1 hour. The weak points are most likely in the 1 ohm
resistor and to a lower extent the 4 resistors used in the diff amp (all
RLR05C metal film type, I will try to get some RNC55). I am not sure what to
do about the 1 ohm resistor, I don't have a lot of choices there at the
moment. 

I only use 16 bits of the ADC, simply because I think it is sufficient for
now, as a proof of concept. After scaling, the current resolution is 10uA
and the temperature resolution is 0.01 degree C.

In the process of developing the code, I learned a lot of about NTCs. I use
a Murata NTSD1XH103FPB30 (10k at 25 degree) with an RLR051002FR as a pull-up
to the 2.5V reference. I ended up writing a linearization routine using the
B factor provided by Murata. This is the first time I use the log() function
with the 8051, previously I used look-up tables for quick and crude
temperature conversion... This gives me the 0.01 degree resolution, with a
little bit of noise that could probably be filtered out analogically and/or
digitally.

Interestingly, Texas Instrument advertises the MSC1210 not as a processor
with a precision ADC, but as a precision ADC that happens to have an 8051
processor, 32 kB of flash and all the amenities on the same die :-)

Plot is at http://www.ko4bb.com/Test_Equipment/HP10811-Current.png
Table is at http://www.ko4bb.com/Test_Equipment/HP10811-Current.dat

The sensitivity looks to be about 2 mA/degree C.

The temperature was captured about 10 inches from the OCXO, in open air but
somewhat shielded though. At the moment, the thermistor is installed on the
processor board (1" leads) and I will move it away.

Please note the time scale on the plot is off, my plotting program got
confused, the actual time scale is in the table, about 8 hours of total run
time. I chopped off the first hour during which the HP10811 stabilized. That
data is available if anyone is interested. At cold, the current is about
0.45 A.

Now, I could record the frequency or TIC to the Thnderbolt GPSDO with the HP
5370 or 5334, but I think I will try to code a TIC within the MSC1210. It
has enough hardwire timer/counters to give 1 PPS at 50% duty cycle from a 10
MHz oscillator clock, all in hardware, leaving the processor itself totally
available, so I think I'm going to be running it from the HP10811. For now,
it runs from an 11.059 MHz crystal oscillator. The MSC1210 can run at up to
33 MHz, so I could conceivably double the 10 MHz and get better resolution.

Didier KO4BB