As far as I remember, the thermistors are identical to the 10811 ones.

I've never heard of "thermistor aging".

I guess it wouldn't hurt to do this test to eliminate aging as a
consideration, it you haven't got anything better to do.

I will point out that if there were thermistor aging, it would be a much
bigger deal on the E1938A, since we set each

individual unit to its own xtal turnover temp.  If we had aging, the
tempco of the E1938A would degrade over time.

The 10811 xtals don't have turnovers, just a wide flat spot, and in any
event the 10811 tempco is orders of magnitude

worse than a properly operating E1938A.  Actually, just speaking of
tempco alone, an E1938A will beat a 5061B; its that good.

Rick Karlquist
N6RK

On 2025-06-17 17:58, Morris Odell via time-nuts wrote:

Hi

Three glass encapsulated thermistors all changing value in the same direction is highly unlikely. About the
only way they fail is to crack. That might be due to an assembly issue (= lack of strain relief). When they crack,
they don’t just move a little ….

“A whole lot” of aging on this sort of part might move the temperature a couple of degrees. Based on decades
of testing, (of multiple groups with sample sizes in the dozens) even that level of change not was observed
to happen on this sort of part.

Bob

I have 6 of these beasts, I could measure something for you if  that was
helpful...

On 18/06/2025 9:26 pm, wkb--- via time-nuts wrote:

I would be interested in whatever resistance the thermistors measure at when cold. Just to get another data point.

Thanks you Rick, for your comment on failure or drift being highly unlikely. The thermistors are mounted in the inner ring using some flexible (silicone?) compound, so mechanical stress induced problems I feel are unlikely.

thanks, Wilko

Hi

NTC thermistors have a wonderful parameter on their spec sheets known as beta.

It drives a formula that computes resistance over temperature. Wikipedia “thermistor” and it will give you all the grubby math details.

The problem is that beta is not a tightly controlled parameter. The fabrication process on a thermistor can only control it to a limited degree.

The net result is that you can buy a thermistor that is very accurate at (or near) one temperature. When you get well away from that temperature the values of a group of parts diverge. The values on different groups diverge even further.

Since these are precision matched parts, you can’t just replace one of them. You would have to replace the whole set. They also are not something you can buy on the open market. An OEM contracts with a thermistor outfit to buy a custom part. They then order them thousands at a time. They just might pay many thousands of dollars for each batch.

There are alternative approaches, some manufacturers have a pretty large room full of “raw” thermistor material. They buy it in bulk and sort it. They then chop it up “as needed” and match the design parameters on the OCXO up with the material. This is more common with PTC than NTC materials.

Simple debug answer in this case is to monitor the voltage at the junction of the thermistors and that resistor. See what it does as the unit heats up. For further fun, monitor the voltages across each thermistor as this happens. If you don’t see anything strange, move on to the other parts of the circuit.

Bob

It appears I have swatted the bug/problem: the thermistor flex circuit proved to be intermittent. Some little hack later the oven seems to be doing its think like it should.

For more details, I have updated the webpage at https://people.freebsd.org/~wilko/HP-OCXO-E1938A/

Best,
Wilko

-----Original Message-----
From: Bob Camp via time-nuts time-nuts@lists.febo.com
Sent: 19 June, 2025 14:16
To: Discussion of precise time and frequency measurement time-nuts@lists.febo.com
Cc: Bob Camp kb8tq@n1k.org
Subject: [time-nuts] Re: repair of a HP E1938A OCXO

Hi

NTC thermistors have a wonderful parameter on their spec sheets known as beta.

It drives a formula that computes resistance over temperature. Wikipedia “thermistor” and it will give you all the grubby math details.

The problem is that beta is not a tightly controlled parameter. The fabrication process on a thermistor can only control it to a limited degree.

The net result is that you can buy a thermistor that is very accurate at (or near) one temperature. When you get well away from that temperature the values of a group of parts diverge. The values on different groups diverge even further.

Since these are precision matched parts, you can’t just replace one of them. You would have to replace the whole set. They also are not something you can buy on the open market. An OEM contracts with a thermistor outfit to buy a custom part. They then order them thousands at a time. They just might pay many thousands of dollars for each batch.

There are alternative approaches, some manufacturers have a pretty large room full of “raw” thermistor material. They buy it in bulk and sort it. They then chop it up “as needed” and match the design parameters on the OCXO up with the material. This is more common with PTC than NTC materials.

Simple debug answer in this case is to monitor the voltage at the junction of the thermistors and that resistor. See what it does as the unit heats up. For further fun, monitor the voltages across each thermistor as this happens. If you don’t see anything strange, move on to the other parts of the circuit.

Bob

time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send an email to time-nuts-leave@lists.febo.com