Author Topic: Heated housing for voltage references  (Read 4381 times)

0 Members and 1 Guest are viewing this topic.

Offline F64098Topic starter

  • Regular Contributor
  • *
  • Posts: 53
  • Country: de
Heated housing for voltage references
« on: February 18, 2018, 07:11:46 pm »
Hello everybody,


after working for some hours i just finished my newest project.

Last year i cracked a double-oven OCXO by reversed power and opened it, to replace the burnt parts.
The inner housing was heated and held by 4 MOSFETs and contains also a MAX6350ESA to produce a stable
reference voltage for the frequency trim.

So the idea was born to build a combination of an oven control and a low noise power supply to take my small voltage
references (AD587) in a heated housing and also supply them by a low noise regulator from varying power sources.

The pcb will fit for a TEKO 371-16 housing which is made of tinned steel. The transistors are mounted with M3 screws 
and a silicone pad on the bottom side provides thermal and mechanical contact to the resistors on the pcb.

The quick and dirty prototype runs at ~43°C (fixed resistors), consuming 86mA @20V when mounted in a double-walled coffee cup
made from stainless steel. Maybe a glass made insulated cup works better. I will find it out...

You can see the schematic and pictures of the pcb and i would be grateful for suggestions and ideas, or any kind of criticism.

Best regards

Frank
 
The following users thanked this post: Echo88, cellularmitosis, Inverted18650

Offline Vtile

  • Super Contributor
  • ***
  • Posts: 1144
  • Country: fi
  • Ingineer
Re: Heated housing for voltage references
« Reply #1 on: February 18, 2018, 07:28:45 pm »
I would say that you have too much volume vs. radiating area allowing drift and radiant heating to form. ....Just thinking out loud, zero experience to back it up.
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 3222
  • Country: de
Re: Heated housing for voltage references
« Reply #2 on: February 18, 2018, 07:57:00 pm »
Hello,

just some thoughts:
- the LT1763 is specced only for maximum 20V. So I would not use 22V.
- the temperature distribution with only some single resistors could be improved.
  I prefer heater foils. But you could also use a bed of SMD resistors.
- I would place the temperature sensor directly at the point of interest (inside the metal housing).
- to get a stable temperature controller I often use 2 temperature sensors.
  one for the I-Part at the point of interest (usually thermally coupled bad to the heater).
  one for the P-part near the heat source to avoid overshoots.
- I would never use sockets for precision references like the AD587
- And shurely I would start with something like the SVR-T cirquit to further reduce the T.C. of the AD587.

with best regards

Andreas


 

Offline Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1928
  • Country: us
    • The Messy Basement
Re: Heated housing for voltage references
« Reply #3 on: February 19, 2018, 06:09:10 pm »
Steel is a dreadful heat conductor. I'd use an aluminum box to even out the heat and give less delay. It makes sense to put the sensor at the point of interest, but you can often do better by closely coupling near the heater because of the reduced delay, as long as the box is highly conductive.
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: Heated housing for voltage references
« Reply #4 on: February 20, 2018, 01:23:54 am »
That the thing i did about week ago:

R6 - 10K thermistor, 5K around +32C, it makes around  -1Kohm / 1C  slope
2x2K - hooked to hp3456 to measure temp deviation, in sequence resistance changing around -75 ohm / 1C  , at 33 C
thermistors from aliexpress, defined as MF58.
heater 2 mosfets, pick 2 from around 6 with close characteristics, mounted on opposite side.
when ON consumed 200ma total, around 1 sec on, 5 sec off, cycle.  ( the hysteresis produce very small level of a temperature jumping, thermistor sensor changing at 0.15 ohm much less then 0.001 C )
the whole board tossed inside the case, so it relatively stable ....

What I'm observing:
assuming 2x2K thermistors are stable - my resistance deviation for 6 hours 2 ohm  == 0.05 C;  over 24h == 8 ohm -> 0.1 C
and drifting per day approx. 0.08C, it drift up and down, around 0.2-0.3 C per week. ( i had just a week of running)
when I use single 2K thermistor, (1K at +43C) due to much less slope i had 0.3C per day stability.

it basically my first attempt, so I wonder:

about overall direction, are thermistors as sensors are stable enought, or I should think about to use diode instead ...
2K and 10K - has a differ material , guess, in my understanding, wonder if 2k are more stable, then schematic need a change to sense much less slope, same for diode.

To decrease hysteresis more ,  i'm thinking a rectangular aluminium plate , drill a hole for  thermistors or diode , and use it between heaters and the frame,

also thinking 2 transistors, but only 1 are monitoring it may produce more uncertainties ...  so to change to one transistor only,

And I didn't count a self-heating effect of thermistors at all - even don't know if it matter ( when I manually switch resistance range on  HP, a diff voltage apply, and self heating was very-very visible)

So here we go again ... assumed it was a finger snap but ....  like ... drifting a 2C per year - at the best , it will spoil everything ...


schematic:


the box:


10K thermistor as sensor:


2k thermistors:

« Last Edit: February 20, 2018, 01:47:15 am by GigaJoe »
 

Offline Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1928
  • Country: us
    • The Messy Basement
Re: Heated housing for voltage references
« Reply #5 on: February 20, 2018, 03:20:51 pm »
Thermistors can be very stable. How much is very? Not a clue, but I used to use a digital thermometer, a Thinc I think, that read out to 0.001 degrees and had no obvious drift problems over a week or so. Even over years it didn't move much.
 

Offline Vtile

  • Super Contributor
  • ***
  • Posts: 1144
  • Country: fi
  • Ingineer
Re: Heated housing for voltage references
« Reply #6 on: February 20, 2018, 04:18:34 pm »
Anyhow we have been bastards for the OP. He created something nice and all we talk is how it might be improved.  :palm: ;D
 
The following users thanked this post: cellularmitosis

Offline F64098Topic starter

  • Regular Contributor
  • *
  • Posts: 53
  • Country: de
Re: Heated housing for voltage references
« Reply #7 on: February 20, 2018, 05:39:42 pm »
The heat conduction may be not good as with aluminum or copper housings but it's compensated by the use of 4 transistors.
And the material is tinned, so it's very easy to solder it.
 
The heat distribution to the inner pcb will be done by 4 brass standoffs (6mm) which are soldered to the housing.
I also still plan to fill the housing with molten paraffin to get a better heat distribution and reduce the humidity
problem to nearly zero.

The maximum input of the LT1763 is a problem, i haven't looked at. Thanks for the note!
In the first version i had the same tracking pre-regulator as for the 5V-supply on the input of the LT1763,
but there is not enough headroom, when using 3 6V Pb-accumulators, so it was rejected.
Maybe it's possible to reduce the voltage drop of the LM317, when using an LDO as pre-regulator.

Best regards

Frank
 

Offline Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1928
  • Country: us
    • The Messy Basement
Re: Heated housing for voltage references
« Reply #8 on: February 20, 2018, 07:45:07 pm »
Anyhow we have been bastards for the OP. He created something nice and all we talk is how it might be improved.  :palm: ;D

Hey, it's in our DNA, it's what we do!  :box:
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: Heated housing for voltage references
« Reply #9 on: February 21, 2018, 12:54:35 am »
F64098,

I'm really like to see your thing as a Temp. stability in overall, seems 10k glass thermistors would be an extremely sensitive to a minor ( like 0.01 C) degree of a temp change.

I'm still thinking: a sensor should be with direct connection to the heater, eliminating hysteresis (overshooting) .
 

Offline montemcguire

  • Regular Contributor
  • *
  • Posts: 88
Re: Heated housing for voltage references
« Reply #10 on: February 21, 2018, 04:16:47 am »
Having the thermal sensor in close contact with the heater will allow the heater's control loop to operate faster, making it easier to stabilize, but what's not clear is whether you're reading the right temperature. Moving the sensor closer to the thermal target will require a slower control loop, and while that's less desirable, the fundamental goal is to have a control system that attains its set point at the place where the temperature needs to be controlled. Controlling the output side of a heater is useful, but it thwarts the overall control loop, whose goal is to stabilize the 'load', not just the heater. Stabilizing the heater alone is an open loop system.

So, if you're using a control loop, the thermal mass of the controlled stuff needs to be minimized in order to minimize the loop response time. The thermal conductivity between the heater, the thermal target, and the sensor needs to be as high as possible, while minimizing overall thermal mass. And, in the end, regardless of the thermal mass and any thermal sensor lag, the control loop needs to be stable, without overshoot and other easily avoidable problems. The details of a mechanical / materials design will determine the speed of a stable, optimal control loop, so the goal should be more focused on minimizing thermal mass while providing sufficient thermal conductivity from the heater, the thermal target, and the thermal sensor. Given all of those constraints, an optimal control loop can be created, and if that's not sufficient, the materials and mechanics of the thermal system need to be re-designed to increase thermal coupling and decrease thermal mass.
 

Offline Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1928
  • Country: us
    • The Messy Basement
Re: Heated housing for voltage references
« Reply #11 on: February 21, 2018, 06:21:43 pm »
Air is a great thermal insulator, but the thing I rely on is that anything in a highly conductive closed box of small size, say of aluminum, will be at the temperature of the box... eventually. And if it's not dissipating significant heat. Building a long time constant control circuit that's well behaved on start-up and doesn't drift, is way harder than a short time constant one. The really serious systems will sometimes use two boxes and two control circuits. Because the outer box does the bulk of the work, the inner box can be just a degree or three above the outer box, making it very stable.
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: Heated housing for voltage references
« Reply #12 on: February 22, 2018, 03:09:02 pm »
2 2K thermistors, NTC, connected to the multimeter
a difference on air or covered by styrofoam brick is  -190 ohm,  I measure slope around -70 ohm / C ; self heating by multimeter
 

Offline thermistor-guy

  • Frequent Contributor
  • **
  • Posts: 365
  • Country: au
Re: Heated housing for voltage references
« Reply #13 on: February 23, 2018, 12:33:09 am »
Thermistors can be very stable. How much is very? Not a clue, but I used to use a digital thermometer, a Thinc I think, that read out to 0.001 degrees and had no obvious drift problems over a week or so. Even over years it didn't move much.

BIPM states that "the most stable thermistors are bead thermistors encapsulated in glass". See section 4.6 in:

https://www.bipm.org/utils/common/pdf/ITS-90/Guide-SecTh-Thermistor-Thermometry.pdf
 
The following users thanked this post: Conrad Hoffman, DiligentMinds.com, cellularmitosis, Magnificent Bastard

Offline F64098Topic starter

  • Regular Contributor
  • *
  • Posts: 53
  • Country: de
Re: Heated housing for voltage references
« Reply #14 on: February 25, 2018, 07:05:57 pm »
I happened to find today the circuit diagram of the OCXO (Morion MV-180) that I damaged. I had to reengineer it, because there were no schematics available.
Nevertheless, I got from Morion the information which types of transistors are used, because they can't be identified (SC-70 housing).

This is only the heater of the inner housing. In this inner housing there is a second heater circuit fo the crystal housing.
There they used a thermistor in 0402, directly soldered to the housing and connected to the pcb with a ultrafine wire,
so i thought it's just a solder blob and tried to remove it, cracking the thermistor  :palm:

Maybe this schematic could help someone

Best regards

Frank
 

Offline CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5173
  • Country: us
Re: Heated housing for voltage references
« Reply #15 on: February 25, 2018, 07:17:57 pm »
While aluminum conducts heat much better than steel, they both conduct much better than a good vacuum bottle.  The thermal time constant of the stuff inside the container will be short wrt the time constant to the outside world.

I have seen wide variations in the stainless steel coffee cups that are available on the market.  Some are near vacuum bottle performance, others seem little better than a solid ceramic cup.  Performance of the ones I have evaluated correlates only weakly with price.  Unfortunately none of the good ones I have encountered has stayed on the market so I can make no recommendations on brand or model number.  It may even be serial number variations (poor control on sealing, the ones that don't leak are good).   If the one you are using doesn't work so well I would try another brand or sample of stainless steel before I went to glass.
 

Offline cellularmitosis

  • Supporter
  • ****
  • Posts: 1111
  • Country: us
Re: Heated housing for voltage references
« Reply #16 on: February 25, 2018, 08:03:18 pm »
Thermistors can be very stable. How much is very? Not a clue, but I used to use a digital thermometer, a Thinc I think, that read out to 0.001 degrees and had no obvious drift problems over a week or so. Even over years it didn't move much.

BIPM states that "the most stable thermistors are bead thermistors encapsulated in glass". See section 4.6 in:

https://www.bipm.org/utils/common/pdf/ITS-90/Guide-SecTh-Thermistor-Thermometry.pdf

It looks like the study he's referring to is "Stability of NTC thermistors", published in "Temperature: Its Measurement and
Control in Science and Industry, Vol. 6"

Code: [Select]
Siwek W R, Sapoff M, Goldberg A, Johnson H C, Botting M, Lonsdorf R and Weber
S (1992a) Stability of NTC thermistors, in Temperature: Its Measurement and
Control in Science and Industry, Vol. 6 (Edited by J. F. Schooley), New York,
American Institute of Physics, pp. 497-502

Google knows about that book: https://books.google.com/books/about/Temperature.html?id=rfFAAQAAIAAJ

Unfortunately, it looks like UT only has volumes 4 and 8:

http://catalog.lib.utexas.edu/search~S29?/XTemperature%3A+Its+measurement+and+control+in+science+and+industry&searchscope=29&SORT=D/XTemperature%3A+Its+measurement+and+control+in+science+and+industry&searchscope=29&SORT=D&SUBKEY=Temperature%3A+Its+measurement+and+control+in+science+and+industry/1%2C8%2C8%2CB/frameset&FF=XTemperature%3A+Its+measurement+and+control+in+science+and+industry&searchscope=29&SORT=D&8%2C8%2C

http://catalog.lib.utexas.edu/search~S29?/XTemperature%3A+Its+measurement+and+control+in+science+and+industry&searchscope=29&SORT=D/XTemperature%3A+Its+measurement+and+control+in+science+and+industry&searchscope=29&SORT=D&SUBKEY=Temperature%3A+Its+measurement+and+control+in+science+and+industry/1%2C8%2C8%2CB/frameset&FF=XTemperature%3A+Its+measurement+and+control+in+science+and+industry&searchscope=29&SORT=D&2%2C2%2C

Volume 3 is up on Library Genesis: http://gen.lib.rus.ec/book/index.php?md5=A371D84737B38CC372E4A4317B91FE1B

But I haven't found a source for volume 6 anywhere.

Could you gents search your local university online catalogs for it?
LTZs: KX FX MX CX PX Frank A9 QX
 

Offline cellularmitosis

  • Supporter
  • ****
  • Posts: 1111
  • Country: us
Re: Heated housing for voltage references
« Reply #17 on: February 25, 2018, 08:07:28 pm »
Here's section 4.6, in case that PDF goes offline:

4.6. Stability

There are several intrinsic effects causing instability in thermistors [Zurbuchen and
Case 1982], including: mechanical cracking of the thermistor body with temperature
cycling, drift at high temperatures due to ingress of atmospheric gases, changes in the
crystallographic structure, and changing contact resistance between the leads and the
thermistor body. The most stable thermistors are bead thermistors encapsulated in
glass. Within the range –20 °C to 60 °C, selected and pre-aged thermistors may be
stable to better than a few tenths of a millikelvin per year. Thermistors with
resistances in the range 2 k? to 10 k? also appear to be the most stable. Glassencapsulated
disc thermistors and epoxy-encapsulated bead thermistors are also very
good but not quite as stable as the glass beads. (See stability studies by Siwek et al.
[1992a], Wise [1992], Edwards [1983], La Mers et al. [1982], Wood et al. [1978],
and Strouse et al. [2012].)

The extrinsic sources of instability relate to the instrumentation: insulation
resistances, stability of the measuring current and stability of the electronic
components used in the measurement.
LTZs: KX FX MX CX PX Frank A9 QX
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: Heated housing for voltage references
« Reply #18 on: February 26, 2018, 03:47:24 am »
Gentlemen,

MV-180 schematic ; Sorry but I dont get it ....   Can, please, someone explain OP2 - meaning, current limiter for a heater?  And NTC2 -  it it really NTC, means negative  coeff, then NTC2 - R7 drive to increase voltage output on OP1 , when temp rising ... ( or I'm wrong ? )

Thanks ...
 

 
 

Offline F64098Topic starter

  • Regular Contributor
  • *
  • Posts: 53
  • Country: de
Re: Heated housing for voltage references
« Reply #19 on: February 26, 2018, 08:52:45 pm »
Can, please, someone explain OP2 - meaning, current limiter for a heater?

Yes, inrush current has to be limited. This means a little extension of the heating time, but you do not need an oversized power supply.

Quote
And NTC2 -  it it really NTC, means negative  coeff, then NTC2 - R7 drive to increase voltage output on OP1 , when temp rising ... ( or I'm wrong ? )

In the next days i will look again at the real circuit.
Maybe i made an error...

Best regards

Frank
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf