Fluke 732A temperature control

[1audio]

I have a 1982 vintage Fluke 732A that seems to all work except the thermistor fluctuates around 7-8K. I first thought the transistor driving the heaters was bad but digging in the transistor is fine and there is a control signal coming from the reference board but its not working right. It swings as one might expect, however prematurely. The drive voltage to the heaters goes from 0 to 6V but all indications are that its around 30-35C inside and the thermistor shows changes moving from 8k to 7K. 

Looking at the sensing/servo circuit I'm not sure where to start troubleshooting and I don't want to open the oven unless I need to. The circuit has 3 thermistors and a selected resistor.  Does anyone have experience troubleshooting this temperature servo? The multiple opamps and other loops make it difficult to follow. I have attached scans of the circuits and included details on the wiring to figure out what the connections are.

I opened up the oven, a real PITA since the foam had all stuck together after 40 years and it was a real battle. Nothing obviously under stress inside and the components seem to be what they should be. The heaters don't heat up very fast. but they do draw current. The thermistors seem OK except the two attached to heaters are lower resistance at room temp than the other two at around 6K Ohms.

[TiN]

Good idea would be to check all carbon resistors, they tend to drift a lot with age and fail. Typical mode of failure is big change of resistance.

[1audio]

I was afraid you would say that. I'll mark up the component placement and check them. You would have thought something like this would have metal film resistors throughout.

I have it open on the bench heating up. Takes a long time. I'll see what the internal temp is when its regulating.

[TiN]

Newer 732A's have better resistors, but old ones are full of carbon ones. I had to replace many on my 732A, but it was not a heater problem with my unit (however it was all of nothing, as issue was for zener SZA263 chip itself).
I've posted pics of my unit here.

[1audio]

This unit has been running fine for years and I had never checked the thermistor. However it was 8K so I could not just ignore it. i'll upgrade resistors if i have to do anything. its a bear to work on the way they have the service loops etc.

[picburner]

I opened up the oven, a real PITA since the foam had all stuck together after 40 years and it was a real battle.

I don't have an answer to your particular problem but this last yours statement makes me raise another question addressed to all the voltage references with oven owners:
How much more efficient is this old (even more than 40 years) thermal insulation, leave it as it is or replace it?
I don't have a F732A/B but a pair of KOEP Trancell VTS6001-1-2-LN with oven too and I assume with insulation in the same conditions...

[1audio]

Very relevant since the insulation did not come apart in one piece. It looks like the stuff in crystal ovens. Some high temp foam. Its certainly worth exploring modern options.

[Kleinstein]

Especially the high value resistors may very well still be carbon type.

For the old insulation it depends on the type. Foam can degrade and is easily be damages when taking it out. The more fiber type insulation is often some kind of glass and does not degrade (at least not over normal time scale).  The oven may not like a much better insulation and the temperature set point is chosen for the original heat losses. So much better insulation may not be a good thing - it may not work well anymore at a high room temperature.

...  the thermistor shows changes moving from 8k to 7K. 

Looks pretty cold 

[1audio]

Update-
With the reference assembly out of the box I started to test various aspects. I stuck a thermocouple inside to measure the internal temperature. Out of the box with the heaters on full it would not get to even 40c. Wrapping and encasing in some styrofoam it would get hot but now it would not stabilize even above 50c. Bridging a variable resistance across the two series thermocouples I managed to get it to stabilize at 45C so the basic circuitry is working and the opamps don't seem to be defective. next step, order a collection of resistors and then check the resistors on the reference board. Stabilizing at 45C gets a resistance of 4.8K on the externally accessible thermistor. looking at the Fenwal table that seems about right. The two critical resistors for the bridge are metal film but there are others that may affect the operation.

The bridge is really not intuitive since there are thermistors on both the top and bottom of the bridge so if they track the voltage at the connection won't change. I can see how that would respond faster to the heaters since those thermistors are attached to the heaters but I don't see an obvious way for this to arrive at an absolute temperature, yet.

[Kleinstein]

The temperature control circuit really looks strange. It looks like the 2 thermistors are simply in series, but the rest looks odd: there seem the be a current mirror and some resistors in parallel / series to the sensors. The resistor in parallel could be some kind of linearization - odd for a regulator working at a fixed temperature.  In addition there are way more caps than needed.  I doubt the regulation uses more than just PID - possibly not even a classical PID but more like PI only.

It does not look like really thought through optimized for low costs, more like hacked together on a breadboard with try and error until it works and than copied to a board.
It's also odd combining the precision LM308 with just LM358 followers.

[1audio]

That's reassuring. I thought I was just not experienced enough in advanced temperature servos. The really high value resistors setting time constants look iffy to me but I just want to get it working. . .

Resistor order in today. Check parts and replace next week? I'm anxious about doing rework on that board. its difficult to work on connected to the box and the temp servo is close to the voltage reference so lots of potential for contamination. Any advise on cleaning after soldering? I saw a reference that isopropyl was not good enough somewhere.

[1audio]

The 732B oven servo is much simpler and more straightforward.

[branadic]

Any advise on cleaning after soldering? I saw a reference that isopropyl was not good enough somewhere.

Use IPA first and clean with pure (99%) alcohol afterwards, this way you can get things as clean as possible.

-branadic-

[1audio]

I had to redraw the bridge circuit to get any understanding of it. I started assuming the heaters would be at 45C. I pulled the value for R3 from
 a table I found online. There is no way to balance the bridge with all of them at 45C.  So let's assume the heaters are much hotter inside where the thermistors are. With some fiddling and checking curves I can get a balance when the inside of the heaters are around 90C and then its a very sharp balance with the values shown.

I'll measure and confirm actual value of the key resistors next week and I will have replacements for most of them. I have a better understanding now of the bridge. It depends on the heaters getting up to temperature before it starts to balance. All those other networks and opamps are to speed up and slow down the transitions around the setpoint. I suspect this could easily oscillate if the time constants are not carefully optimized. It still feels very much like a breadboard committed to production which it may be. This started as an internal 
project that was pushed into commercial production probably before it was fully reviewed. Never happens anywhere else. . . .

[Dr. Frank]

According to the manual, page 3-2,the oven temperature is 48°C +/- 2°C.
There's an additional thermistor also available to monitor the internal temperature with an Ohm meter.
There also should be a sticker indicating the real resistance / oven temperature.
Please also take notice of various changes of the thermistor designators.

Can't understand how you calculate 90°C.
Impossible to get a few ppm/ year at that high temperature..

Frank

[1audio]

Actually running at 48C makes balancing harder. There are three thermistors in series and somehow two in series need to have less than 1/2 the resistance of the one not in the heaters. I was just pulling numbers from the curve for the thermistors to find ones that would make the system balance. To get the resistacce ratios to match the heater thermistors need to be around 760 Ohms. That matches to 96C on the table I'm looking at.

[1audio]

I just confirmed that the thermistors in the heaters have the same part number as the ones on the PCB.

[1audio]

some updates-
in checking the components all the resistors in the bridge seems to be correct. The Rsel at R21 is 4.53K. All the resistors are on value and are all metal film except the very high value resistors. Those are in series with caps so they probably won't have much impact on anything. The thermistor that connects to the front panel had a cold solder joint. All the others seem fine. No other signs of distress. At this stage I'll to a "test" assembly" with the thermocouple inside and everything back sort of. The foam has some problems. The "lift out" tab was stuck to the foam and tore so I need to do something different. 3 of the nylon standoffs that hold the PCB broke at the end of the screw that goes in from the PCB side. They are odd lengths so I'll need to improvise.

I suspect I'll get it together with no replaced parts, just the fixed solder joint and it will work fine. I still don't see how it can work with the thermistors in series but it does.  When its together i'll bring out several test points to better understand it. Hopefully this will help the next guy who needs to troubleshoot one of these.

[1audio]

I realized that I made a mistake following Fluke's schematic. its not an easy one to follow which is typical of Fluke in that period. Now it makes more sense with the 3 thermistors in series fed from a current source referenced to R28 and R29.  It is a bridge but pretty obscure as drawn.

[martinr33]

We still have a mystery. On the schematic, R21 is labeled as "sel" (it is R7 on your schematic).

So - why would that resistor be selected? There is an adjustment for oven temperature.  Plus, the output pair is clipped together by two diode voltage drops.

[Kleinstein]

The selected resistor would be to set the oven temperature and maybe compensate for differences in the transistors.

The diode clipping of the bridge is likely there to limit the overshoot on turn on and the effect of short spikes due to EMI of similar.

[MegaVolt]

I slightly corrected the previous model.
The semiconductor models do not match the original ones.
Resistor R21 is selected so that the circuit gives out 48 degrees.

The full scheme is on the way )))

[MegaVolt]

I hurried.
1. The thermistor model was wrong. They used 273 instead of 273.15.
2. The thermistor model is now independent of the external library.
3. Thermistor Beta is now closer to the original JA41J1
4. Resistor R21 changed.

Corrected schematics:

[MegaVolt]

I'm not very picky.    The "Kelvin" constant is -273.15 and not 273.15 as I expected.   
Schematics fixed.
Resistor R21 was selected new ))

[MegaVolt]

The full scheme is on the way )))

This happened )))
I finished it ))

Here is a fully working model of the Fluke 732a oven.
You can look at the resistance of the thermistor connected to the front terminals of the device.
Also the temperatures of the furnace itself and the printed circuit board and all other signals )) For example, turning on the power (last picture)

Model limitations:
1. Two elements were added that were not included in the standard LTSpice package:
   - LM358 (model included but you may have to connect it yourself)
   - Current Limiting Diode (model attached and included in the diagram)

2. I didn’t have models for many elements. I installed similar ones. Therefore, subtle nuances of behavior may be incorrect. For example, germanium diodes were replaced with silicon schottky ones.

3. The thermal model of the oven must be close to reality and based on calculations. But there are a lot of simplifications in it. For example, the heating of the board itself was calculated through the air from the bottom of the box. Other heat transfer routes are not taken into account.

4. The result coincides with those measured on a live device

5. The circuit is taken from a later edition without transistor Q14 on board A4.

Conclusions from the model.

As you can see, at startup the iron box heats up more than necessary and subsequently gradually lowers the temperature while the board itself warms up. This ensures very smooth heating of the board itself.

If you find any errors or inaccuracies, please let me know I am often very inattentive.