Small Oven Controler For Voltage Reference

[blackdog]

Hi,


On this forum, several oven projects are discussed for heating voltage references.
This is one of my controllers equipped with an LM723 and three Power Mosfets.

These are the requirements for a proper controld oven:
Good coupling between the sensor and the heat source.
Good distribution of the heat source over the housing to be heated.
A stable controler.
Good isolation of the heathed unit, best is to use a doubble isolation, two layers of polystyrene with a few millimetres of space around it.
Treat the oven control as if it were your best voltage reference, use good resistors and other components.
If you want to be cheap, you will get a bad oven!

I use "old style" Power MOSfets here the IRFP2410 with a relativ high Rds on resistens, do NOT use modern low Rdson MOSfets or Logic MOSFets.
It wil be troublesome to get is stable with modern MOSfets.
The old style MOSfets are cheap and i chose these also for the large housing, the TO247.
It is posible, if you want to connect the drains to the raw power supply en connect the uA723 to a 78L12 regulator, but i wil not do this.
For maximum stability use a 1,5A regulator like a 7812 or LM317, maximum current will only by drawn for say a few minutes.



At a local flea market I bought three of these boxes for 10€, 50x90x33mm and the bottom is 7.5mm thick, probably used for HF projects.



Here you can see how the IRFP240 MOSfets is mounted isolated on the housing.
There is also an LM35 in TO220 version mounted, which was for an extra test but not easy to use because of the high noise value, it wil be removed.
I don't want to have a galvanic connection between the over circuit and the voltage reference for this kind of project, so the MOSfets are mounted on aluminium oxide plates.
To make the distribution of heat as good as possible, I applied a small selection to a handful of the IRFP240 MOSfets.
I tested at 12V Vds, at 50ma, selected 100mA 200mA drain current and the MOSfets that were within a few percent equal.


I used some stripboard for ease connecting te components around the MOSfets.



Yellow circle is de 5K Thermistor connected to one of my 34461A DMM's i use to measure the oven temperature.
The blue circle is the 10K thermistor for the oven controler and it is directly under the IRFP240 for a tight coupling with the heat source.



The oven controler, to the right of the uA723 you can see the 100uF capacitor with the 100K resistor, which is used to keep the loop stable.
The values of these components depend on how the big housing is, and how well the Thermistor is connected to the heat source, if you try this controler start with these values
and if its stable, make the capacitor smaler until the heather current starts oscilating, make the capacitor now 2x as big, this is a rule of thumb :-)
Red is +12V for the uA723, Blue is de Gate driver, Green is the current sense and 2x white is the 10K Thermistor, and black is ground.



This is the first version, with a different loop controle, on the spot of the trimmer resistor there is now the 100uF capacitor.
Look how i use the space in the IC socket to place the 1uF decoupling capacitor and the connection between pin-3 and pin7.



Room enough for some nice voltage references :-)



This is the temperature drift after the oven has been on for about 15 minutes.



The power used is the oven is stable, arond 1,5-Watt's if the LAB temperature is 23C, the Max. power on startup conditions is at 12V 16-Watt's.
You can play with the source resistors to make it lower or higher is necessary for your oven housing.
My housing needs aboud 0.1-Watts per C.



Things to think about...
The drift you see on the picture is from about 1C of dift in my LAB.
It wil, just like a voltage reference needs minimaal 1000 hour to get stable.
Keep is as air tight as you can make it. (mine is not at the moment of these measurements)
Think about heath leakage of the wiring.
Use a stable powersupply for the oven controler


Shoot at it!

Kind regards,
Bram

[zhtoor]

hello Bram,

a couple of humble suggestions:-

1. how about removing the controller from heated zone, why subject the controller to temp. cycling?
2. mounting sensors (thermisters) in middle of 4-sides by drilling small holes in the thick plate and inserting the sensors.
3. mounting the heat-sources (iRFP240's) on the 4-faces instead of the bottom to lower the chance of temp. gradients.
4. using accurate pt-100/1000 sensors (at least 1) like the TE-connectivity ones. NB-PTCO-002/006 (farnell.co.uk)
5. schematics?

best regards.

-zia

[blackdog]

Hi zhtoor,

This is not a perfect oven 
It is more to let other people see what is posible with a old nice box with a thick bottom plate.
4 Mosfets on the side, yes thats is posible, how mutch it would be better...
This wil take a lot of measurements.

The best oven will be a diamond spere placed in a good dewar 

The reason i used a uA723 is because it has all the electronics you need to make a nice controler (not the best controler!)
I have build many differenty ovens, maybe later today i will show some, also a setup with 4 MOSfets on the side of a aluminium square tube with more sensors.
Sometimes i use good Thermistors and sometimes i use PT1000 sensors, depend on how good i want it to be.

I have also a better controler than the uA723, is uses a instrumentation amplifier, but that one is expensive, it is using a INA125.

The controler wil be more stable in the oven.

The schematic wil be placed later, i forgot it, i'am getting old...

Kind regards,
Bram
 

[blackdog]

Hi, 

I had promised the zhtoor to show some other ovens.
This is a made of rectangular aluminum pipe, 60x40mm and 80mm long, the wall thickness is 4mm.
I drilled two holes in the aluminium under two of the MOSfets in which a 100K thermistor was glued.
This oven uses 4x IRFP140 MOSfet for heating.
The used control unit, is also equipped with a uA723 IC.

But first now the schematic, it is a little different than the first one i showd.
This is using a dual parallel 100K Thermistor.



This picture is a plot of the stability, a small overshoot and than stable, i think i can tune it a little bit better, but I'm not going to do this until both covers have been fitted.
These lids also have a thermal mass and they also change the loop constant.
I also measured  the oven current and there is no sign of instability there.



On the aluminium i glued a one piece of thin double-sided circuit board to make the connection.
And also the uA723 wil be held temperature stable this way.
On the bottom left, is the blue hole that I drilled for the 100K sensor under the IRFP140.
The blue stuf is WLK30 thermic glue.



Sensor hole, sorry for the bad picture.



This is the PT1000 sensor that is connected to one of my TEKTRONIX DMM 4050 multimeters with the 4 wire technology to read out the temperature.
The sensor is fixed with Kapton tape, the connecting wire is also fixed to the aluminium along its entire length.
This reduces measurement errors by dissipating heat through cabling.



Maybe later, some others ovens i played with.

Kind regards,
Bram

[zhtoor]

Hello Bram,

i am most grateful for your insights on the subject.
what would you say the temperature stability of this oven would
be when holding say an un-heated LM399 or an un-heated LTZ1000?
say +- 10mK temperature ripple?

best regards and keep up the good work.

-zia

[blackdog]

Hi Zia,

I think a single oven wil not do the trick.
I have also a nother design, i uses a double oven, the outside oven controled by a Arduino on about 38C, PID controled.
and the inner oven linear controled to about 42C.

I do not have the time to show everithing, too busy with work...

A single oven attenuates the ambient temperature between 50 and 200x, better for a single oven is difficult.
there are so many variables that play a role in keeping the oven at the right temperature.

I like it to play with these techniques, but its also rely time consuming.
It is also difficult to readout 0.001C, the noise of the DMM4050 is high when measurening temperature with a PT1000 and a little drifty.

A good 5K thermistor connected to a KeySight 34461A wil give you a higher noise free resolution, but i'am still not shure how mutch drift this meter has...
I have to do a test with a Vishay 5K resistor "S" series or better to see how the 34461A drift on the temperature range.

Kind regads,
Bram

[blackdog]

Hi,

To make things clearer, if you want to build an oven for an unheated LM399 or a LTZ1000, you will have to make a double oven.
It is difficult to make an oven that is better than the one in an LM399 or LTZ1000.
This is because there is a very good connection between the sensor and the heating element.
But if you make the standard circuit for an LM399 or LTZ1000 and put it in an oven, then you have the best of both worlds.
My opinion is that a single well built oven is a nice addition to a well built LM399 or LTZ circuit.

If you need a low noise reference, you can use an unheated LM399 or LTZ1000 in a double oven.
But it takes quite a bit of "skill" to make up for this.


Here some links to a Dutch Forum where i show some oven developpement, use google translate if you cant read Nederlands.

https://www.circuitsonline.net/forum/view/107384#highlight=oven

https://www.circuitsonline.net/forum/view/114717#highlight=oven

https://www.circuitsonline.net/forum/view/133966#highlight=oven

Happy reading 

Kind regards,
Bram

[blackdog]

Hi,

During my lunchtime I changed the loop constants of the oven with the 4x IRFP140, now the temperature overshoot has disappeared...
And the current through the oven is still stable, this current value is a good indication if the loop is stable.

If the current through the MOSfets goes to almost 0 mA during warm-up, then you know that the loop constant is not yet good enough.

This is the schematic with the different values for the loop controle, C1 is now 47uF and R6 is 221K.



This picture shows that the overshoot is gone...



But again, the loop controle will be different for your oven, play with C1 and R6 and watch the current trough the MOSfets, Bang-Bang is wrong! 

Kind regards,
Bram

[zhtoor]

Hi,

To make things clearer, if you want to build an oven for an unheated LM399 or a LTZ1000, you will have to make a double oven.
It is difficult to make an oven that is better than the one in an LM399 or LTZ1000.
This is because there is a very good connection between the sensor and the heating element.
But if you make the standard circuit for an LM399 or LTZ1000 and put it in an oven, then you have the best of both worlds.
My opinion is that a single well built oven is a nice addition to a well built LM399 or LTZ circuit.

If you need a low noise reference, you can use an unheated LM399 or LTZ1000 in a double oven.
But it takes quite a bit of "skill" to make up for this.


Here some links to a Dutch Forum where i show some oven developpement, use google translate if you cant read Nederlands.

https://www.circuitsonline.net/forum/view/107384#highlight=oven

https://www.circuitsonline.net/forum/view/114717#highlight=oven

https://www.circuitsonline.net/forum/view/133966#highlight=oven

Happy reading 

Kind regards,
Bram

thanks again Bram.

please have a look at using the LM399 heater as a die-temp sensing diode for accurate temperature control of the die here:-

https://www.eevblog.com/forum/metrology/a-low-cost-oshw-voltage-calibration-reference-project/75/

now, if this sensor is used to set the oven temperature which is in thermal contact with the LM399
(possibly nine of them with their temp. sensing diodes being averaged) and also holding the attendant
electronics (opamps / resistors etc.)

ie; using a set of LM399's as temperature sensors mounted strategically in the "oven" to make a
lower temperature double oven @ 25 degC, maybe peltier modules would be required for the outer/inner oven.

best regards.

-zia

[zhtoor]

hello Bram,

how about enlarging the tab hole of :-

https://www.digikey.com/product-detail/en/infineon-technologies/BTS282ZE3230AKSA2/BTS282ZE3230AKSA2-ND/4841056

to ca. 4.75mm to insert an LM399 (without the polysulfone cap) and using this contraption as the "inner" oven.

btw. the part can also be used with other ovens with an in-built temp. sensor diode.

or:

a more complex / cheaper way would be to use the source-drain body diode in iRFP240 as an on-die temp. sensor
but the disadvantage would be a discontinuous control with interspersed temperature sampling periods.

best regards.

-zia

[blackdog]

Hi zhtoor, 

I've tried to use the back diode or a MOSfet, it's too complicated.
The Infinion TEMPFet is not usable, the sensor is a thyristor, not a normal diode.

For the two oven design i shown in this topic, it is posibel to use these transistors from ON Semi: NJL3281D with a internal diode.
Datasheet: https://www.onsemi.com/pub/Collateral/NJL3281D-D.PDF
A nice side effect is the large housing of this transistor, it uses a TO-264.

You can't fit a LM399 in the hole, but you can use thermal glue, it fits 6x LM399 
Here you can see the NJL3281 transistors on some aluminium tubing.


Maybe for some inspiration?

Kind regards,
Bram

[zhtoor]

You can't fit a LM399 in the hole, but you can use thermal glue, it fits 6x LM399 

now that is GOOD 
one can make a heatsink-like adapter out of solid aluminum which houses 9x LM399's (with thermal grease/glue) and fits the TO-264 package neatly.
or a solid square block with 4x these transistors on 4 sides and a cavity milled in it for housing electronics / refs.

best regards.

-zia

[blackdog]

Hi,

Just to show that it fits, four LM399 pieces without insulation is possible to glue on this transistor...
These are LM185 zeners in the same TO-46 housing as the LM399.


No, i will not build it 

Kind regards,
Bram

[David Hess]

There is also an LM35 in TO220 version mounted, which was for an extra test but not easy to use because of the high noise value, it wil be removed.

To me that looks like a TO-220 LM350T 3 amp variable voltage regulator (3 amp version of an LM317) and not an LM35 (LM35DT?) temperature sensor.

The reason i used a uA723 is because it has all the electronics you need to make a nice controler (not the best controler!)

The 723 is a great choice for this application since it includes a differential input error amplifier, reference, and power output stage if needed.  There are some combination operational amplifier and reference ICs but they are more expensive.

[blackdog]

Hi David,

It really is an LM35 in TO-220 housing and he has also been connected working.
The other ICs I used for oven projects are the LM10 and the INA 125.

And of course an Arduino Nano I hacked, in order to get a more stable ADC.
I have used a good 1.23V reference IC and better decoupling of the powerlines on the Arduino(PID controler)

Kind regards,
Bram

[David Hess]

It really is an LM35 in TO-220 housing and he has also been connected working.

So it is.  I am sure I have seen that before but then forgotten.  Providing the LM35 temperature sensor in a TO-220 package is a great idea but sure makes for a confusing part number.

The other ICs I used for oven projects are the LM10 and the INA 125.

The LM10 is great for this sort of thing but sure is expensive compared to a 723 unless you need its low voltage and low power features.

[blackdog]

Hi,

The nice thing about the LM723 is that it has all, also the max. Current/Power limiter, you dont need no extra active components.
The loopgaing en reference stability is more than enough for making a reasonably good component oven.

But keep in mind that the impedances that the inputs see remain below 10K.
My second design I showed here is just above that for the Thermistor input, if you use a 100K Thermistor under each transistor you meet the requirement of less than 10K of impedance.
Also don't go to low in impedance, otherwise the reference section will be overloaded.

A rule of tump is that at 42C the thermistor is about half its resistence value, easy to remember. (100K thermistor is at 42C about 50K)
And if a component oven is 41.45C or 42.67C, i dont care, but maybe for you guys it is important,
trim one of the bridge resistors to tune the temperature of your liking.

Kind regards,
Bram

[blackdog]

Hi,

I had had had enough of the work today.... and thought, let me look for some relaxation in the electronics 

I have made some adjustments to the loop compensation.
This I did, after some research of the internal construction of the uA723.
The loop compensation connection (13) has a high impedance and when the loop compensation is on that pin, the open loop gain goes down.
This in turn results in poorer control properties, so instead of pin 13, I used the output pin 10 which has a low output impedance.

After this adjustment the capacitor could also become smaller, and C1 is now 10uf and the overshoot is minimal with this value.


I don't think I can get any more performance from this circuit, remarks, shoot at it!

Kind regards,
Bram

[blackdog]

Hi,

This afternoon I thought about trying to make one of the ovens even better.
Again this time four MOSfets, but this time directly mounted on the aluminum tube.
I then decided that power supply from the oven electronics is connected with the + to the oven tube.
This is not a problem if two separate power supplies are used.

The oven tube looks smoot, but it is not! 

I had to file with alcohol for a long time to get the sides fairly flat.



One side ready, not perfect, but mutch better than it was.



The aluminium tube looks like this when it comes from the supplier, but i drilled the holes.


Here you can see the four holes for mounting the 100K thermistors.
Make sure that all holes are deburred.



Here I have prepared a piece of 0.5mm of double-sided print material to glue to the oven tube.
The material is fixed with 10 second adhesive.



Ready for the next step, the Flux.



Here the tube is wrapped with capton tape and the copper is sprayed with flux,
the cotton tips ensure that the holes for the thermistors do not fill with flux.
It is almost a peace of art 



Shiny Flux!



Making poison!
When you read the manual of this glue, you barely dare to use it....



The prepeard 100K thermistors, teflon sleves and kynar wire, it was a lot of work.



The thermistors glued in the oven.



If i have time tomorrow, i will select four of these MOSfets for the oven



Thats all for today!

Kind regards,
Bram

[David Hess]

I have made some adjustments to the loop compensation.
This I did, after some research of the internal construction of the uA723.
The loop compensation connection (13) has a high impedance and when the loop compensation is on that pin, the open loop gain goes down.
This in turn results in poorer control properties, so instead of pin 13, I used the output pin 10 which has a low output impedance.

Pin 13 is the transconductance (current) output and is used in the same way that similar external compensation pins on operational amplifiers are used.  Schematics usually show a capacitor back to the inverting input but a capacitor to ground also works.  Unfortunately there is not all that much data for the 723 on different compensation schemes.

For voltage regulators which are relatively fast, I doubt you can get away with *not* using the compensation pin but in your case the external frequency compensation is so heavy that this may not be a problem.

[zhtoor]

Hello Bram,

do keep up the good work, make lots of money and retire to do this full time for our benefit 

have a look at an interesting implementation from the distant past: fluke 335 A/D.

best regards.

-zia

[ramon]

Bram, thanks. Nice thread !

How do you attach the mosfet? Aluminium Nitride? Send me a PM if you need four.

[blackdog]

Hi Ramon,

For this oven the MOSfets wil be mounted whitout isolation.
Only some thermal compound will be used.

There will also be a small revision of the schematic.
This is because I now use four thermistors and the impedance of the node of the thermistors is now half of what is was.
This will probably also change the loop constant a little and need some adjustmend.

Furthermore, this week I also calculated some data from the NJL3281D transistors.
The sensitivity with the four diodes in these transistors is about 8x less than with the thermistors,
this will result in poorer control properties with a uA723, this because the open loop gain of this IC is only 30 to 40dB.
Also it is just outside the input common mode of the uA723 without extra precautions.

It is very well possible to use these transitors, but you may need a different IC for proper functioning, but that is outside the scope of this topic as far as I am concerned.

I used these thermistors for the oven:
https://www.banggood.com/10-Pcs-3D-Printer-Thermistor-NTC-100k-100ohm-Match-MK2a-1-Sensor-p-932646.html?rmmds=search&cur_warehouse=CN

In the Netherlands the used MOSfet is cheap.
https://www.eoo-bv.nl/zoeken?controller=search&orderby=position&orderway=desc&search_query=irfp140&submit_search=

Also in Germany
https://www.reichelt.de/IRFP-IRFRC-Transistoren/IRFP-140N/3/index.html?ACTION=3&LA=446&ARTICLE=41674&GROUPID=2893&artnr=IRFP+140N&SEARCH=irfp140&trstct=pos_0

If you are going to use other MOSfets, that's your problem. 

Kind regards,
Bram

[ramon]

I want to simplify you schematic as much as possible.

  - instead of four parallel mosfet I will use just only ONE.
  - No big BOX: Just mosfet attached to aluminium nitride sheet and one temperature sensor on the other side.

It cannot be seen clearly on the picture I send before, but the bigest mosfet I have in the drawer is IFRP460.
I think that a single IRFP460 can be used to substitute 4x IRFP240, if the area to heat/regulate is reduced.

  IFRP140: input capacitance = 1700 pF, Power Disipation = 180 W
  IFRP240: input capacitance = 1300 pF, Power Disipation = 150 W
  IFRP460: input capacitance = 4200 pF, Power Disipation = 280 W

Another are that can be modified (improved?) is the resistor divider.
We can add 'filling' resistors on high side, bottom side (or both) to avoid self-healing.

[blackdog]

Hi Ramon, 

It is not clear to me what you are trying to build.
If you do not use a box, the uA723 also can not be co-heated.
This makes the temperature control worse, but maybe that's not so important for your application.

The capacities of the MOSfet are not important at all in this application.
The steepness, however (admittance) will. (need different values for the loop compensation) if you can get is stable 

If you are not going to use a housing, then the leakage of heat will become a problem because of your connecting wires.
Heating the MOSfet only, when the MOSfet is well insulated will not exceed 0.1 a 0.2 Watt.
If you don't pay enough attention to the wiring, the heat loss through the wiring can be as high as 40%...
And there go's your temperature stability...

I do not understand what you mean by: We can add 'filling' resistors on high side, bottom side (or both) to avoid self-healing.

The good results I achieve are due to the fact that I have thought and experimented a lot with this setup.
Applying a well-functioning schematic to an entirely different assembly will never produce the same good results.
But, don't let me stop anyone from doing experiments, because you learn a lot from them. 

With some extra isolation, the power the oven is using dropt from 2.1-Watt to 1.5-Watt at 23C LAB temperature.



And this is 9 hours temperature logging, the wide band is due to the noise level of the TEK DMM4050 meter when you measure with a PT1000 sensor.
The top of the chart is +7 mC and the bottom is -6 mC compared to the start of the measurement yesterday evening.
The LAB temperature had dropped by about 2C during the night.
At the moment, after more than 16 hours of operation, the temperature is approximately +0.004C at approximately 23C LAB temperature.


But keep in mind that also the TEK DMM 4050 used drift with the change of LAB temperature!

zhtoor
The ovencontroler you are showing is almost the same as the internals of the uA723, look at the schematic of the uA723, you wil see a differential amp and a darlington.
The uA 723 uses a current source in the gain stage and Fluke uses a 75K resistor (R3)
Both have a Darlington output stage.

Now it's time to select some MOSfets!

Kind regards,
Bram

[blackdog]

Hi,

Are you all still there? 

This afternoon I measured the four thermistors, the oven was on the LAB table for an hour without having been touched.
That's why I was almost certain that the temperature of the oven would be even.
The value I measured of the four thermistors were these: 2x 115K1 and 2x 115K4 and this is very good, nothing to complain about.

Selecting the MOSfets
In this way I have selected four out of eight MOSfets,
I used the two remaining thick aluminium cabinets which I had, as a heathsink.
There are many holes in it, so it was easy to mount four per box quickly.



And this is the schematic used to measure Ugs at 50 and 200mA drain current.
First i used the Bryman BM869s hand DMM for measuring the drain curren, but it's a drama to read that when you vary the Gate voltage.
This meter is so slowwwwww when measuring current, so i switcht to my Fluke 287, no problemo!



Now the MOSfets are mounted and the Drain in solderd to a soldering lip, why? the MOSfet is in fact mounted without insulation...
Do not fool yourself, I measured all four MOSfets, two measured less than 0.05Ohm to the oven body, but the other two had a resistance of more than four Ohms.
You can be sure that if the resistance is high, you have used too much heathsink compound. 
On some pictures you see dust particles, i will clean this later.



I also pay attention to the way of assembly, like here, the screws are just long enough, no coat hooks in my oven!



The next step will be to assemble the electronic parts, I'll keep you informed!

Kind regards,
Bram

[David Hess]

And this is the schematic used to measure Ugs at 50 and 200mA drain current.
First i used the Bryman BM869s hand DMM for measuring the drain curren, but it's a drama to read that when you vary the Gate voltage.
This meter is so slowwwwww when measuring current, so i switcht to my Fluke 287, no problemo!

This measurement is a lot easier to make with the gate tied to the drain and a power resistor, say 10 ohms, in series with the source.  Adjust the power supply for 50 or 200 milliamps (or measure 0.5 or 2 volts across the 10 ohm source resistor) and then read Vgs.

In a production environment, I would use an operational amplifier to drive the gate and force the drain current to whatever is required.

[blackdog]

Hi David, 

The reason I did it in this way is that I wanted to stay as close as possible to the real-life situation.
The drain voltage also determines the current through the MOSfet, and at around 3.5V (~200mA) it is not de drain voltage de oven will work at.
The MOSfet is used at the point where it is just starting to conduct, so I wanted to be a little sure that I would measure the good value.

Yes, with a opamp it all goes a little bit better, but measuring the 8 Mosfets didn't take more than 15 minutes together with the mounting on the boxes.
The potmeter with the 9V battery, I had already prepared, because I had it already used it for some other projects.

Kind regards,
Bram

[blackdog]

Hi

Today I wanted to show you some of my oven developments in a new topic, maybe you could use that for inspiration. 
But, ^@#%$%^#$%@#%#$

I was soldering the last parts to the oven that I have been working on in the last few days and was doing the first tests.
What I measured and saw did not feel good, there was "mental" behavior of this oven.
I have quite some experience in building this kind of oven, and I started by watching how the current drops as the oven warms up.

If the oven is close to the final temperature, the oven should start using less power quickly and should not become completely powerless.
If the current becomes almost "0" then the loop compensation is not good, that is one problem.
But with this oven I also connected a scoop channel to the output of the uA723. w.t...
A lot of "random" noise, lookt like the infamous popcorn noise, forgot to make a scoop picture of this.

But first a picture when the oven "was" ready.
I didn't have any more 0.1% resistors of the good value, so here are 1% resistors used this week the good resistors come back in here.


Without looking at my test setup, the noisy oven had stopped working.
At the bottom you can see that there are always bursts of current and at the end the maximum current and then nothing anymore....
The oven stopt working, i search myself silly...
Everything at the inputs seemed all right, also the reverence voltage was present with its 7.05V.
The +input of the uA723 also had nicely half the reference voltage as it should have been.
Only the output of the IC it has a voltage of about 1.7V, and then the MOSfets never conduct.


Noise at the output of the uA723 and an output that is far too low in voltage.
It must be a defective uA723, a pity of the neat wiring....
Out you go!



And I've soldered a new one in, "bad language" exactly the same behavior...
Grrr.
I had another hp 34401 on the bench and soldered one side of the thermistors off
and left the other side connected to the circuit and tried to measure the resistance of the thermistors.
Very strange results I got at the resistance measurement.
In the end I cut all four thermistors loose and measured them separately.
When I started measuring the resistance connections to the oven housing, the monkey came out of the sleeve. (This is a Dutch expression)
There was quite a lot of "leak" present to the aluminum of the oven.
This is the first time I have experienced this, when i mount/glue sensors in aluminium.
It's also the first time I've used the glue I've shown for this application.

To make a long story short, I have drilled out all four thermistors and placed four 5K thermistors in series so that I can all use 10K 0.1% resistors at the uA723 inputs.
4x5K makes 20K at 25C and at about 42C the total of the thermistors wil be about 10K.
This time it's not an ebay thermistors but beautiful from Vishay 5K and 0.5% not too expensive and very good.

The current is now nice and stable and the oven works fine, noise is gone.
The wiring is not as neat as it was, but it works fine.
I have to tune the loop just a little more for optimal performance.
The thermistors are put in the holes with termal compound, later i wil blok the hole with some hobby glue.



I can say that this was a learning day... 

Kind regards,
Bram

[blackdog]

Hi all,

Again some steps in the development of this oven.
I would like to show you what you all encounter when developing a piece of measuring equipment.
That's why I'm doing a lot of postings here now, just a few photos and it has these specifications, says little about the problems you encounter when building a project.

This is the modified schematic with now four thermistors and the adjusted value of the bridge resistors.
Because of this configuration with the four 5K thermistors, the other bridge resistors can all be 10K.
The four thermistors in series also result in a value of 10K at an oven temperature of approximately 42C.
So I'm not interested in a precise oven temperature, around 42C is good to about 35C LAB temperature, if you don't put the oven in a warm place.



The loop compensation is now also set for the adjusted impedance at point 4 of the uA723.
It is possible to further increase the control speed, the loop is now a little over compensated.
But I'm not going to change that until the lids are on the oven and a PCB with the voltage references is mounted.


The upper meter shows the current flowing through the oven, the measuring time is approximately 11.5 hours.
The peak current in the beginning is on purpose, I removed the insulation for a moment to be able to see if the current through the oven behaves properly.
The bottom meter shows the deviation of the other oven  i still measure, it is the second oven I showed in this topic.
The suppression of the LAB temperature for this second oven I estimate about 130x with the current insulation, nothing to complain about and more than enough for voltage references.


This is the current trough the oven coming up to temperature, very controlled it runs to the desired value, no over or undershoot.



I deliberately don't show a temperature chart of this oven yet.
This because it is much more important to get the current through the oven stable, by adjusting the loop compensation.
Later this week i hope to have 2 DMM free so i can show the oven temperature and the oven current on the same time.
I use 2x the Keysight 34461A and the horrible BenchVue software for that 

Later more...
Bram

[blackdog]

Hi,

Just a small side step, what could you mount in one of the ovens as a voltage reference...
A selcted 1N829A, LTZ1000A, APEX VRE310AD, AD588JQ, LT6655? no... i am aging 14 pieces of LT1021CH for more than 3 years now.
And i want to use 5 to 8 pieces of them in a oven.

Below is one of the first examples of how the circuit will be constructed.
It uses 5x the LT1021CH in parallel, then a times two amplifier, this will all be in the oven, everything in the right box, will be mounted on the banana sockets (Pomona)
Because its a working reference, you need some protection and i added that on the banana sockets.
I'm not sure how I'm going to do the trimming, the schematic shows a trimpot, but maybe I am going to use an i2c DAC with a very small trimming range, that is optically linked via i2c.
It is still a work in progress...


Shoot @ it!

Kind regards,
Bram

[zhtoor]

Hi,

Just a small side step, what could you mount in one of the ovens as a voltage reference...
A selcted 1N829A, LTZ1000A, APEX VRE310AD, AD588JQ, LT6655? no... i am aging 14 pieces of LT1021CH for more than 3 years now.
And i want to use 5 to 8 pieces of them in a oven.

Below is one of the first examples of how the circuit will be constructed.
It uses 5x the LT1021CH in parallel, then a times two amplifier, this will all be in the oven, everything in the right box, will be mounted on the banana sockets (Pomona)
Because its a working reference, you need some protection and i added that on the banana sockets.
I'm not sure how I'm going to do the trimming, the schematic shows a trimpot, but maybe I am going to use an i2c DAC with a very small trimming range, that is optically linked via i2c.
It is still a work in progress...


Shoot @ it!

Kind regards,
Bram

hello Bram,

how about a low pass filter (maybe bootstrapped) before pin 3 of ltc2057?

best regards.

-zia

[blackdog]

Hi zhtoor,

You mean something like this?



Kind regards,
Bram

[zhtoor]

exactly, only place some suitable low-value (0.01 - 0.1uf) caps in parallel with 330uf ones for taking out fast pulses if any.

best regards.

-zia

[blackdog]

Hi zhtoor,

Where should those fast pulses comming from?
Have you done some measurements on  good 330uF capacitors? these capacitors have a very low impedance over a wide range of frequencies.
And there's nothing in this circuit that generates fast pulses.
There is also a good film capacitor included (C8) at the +input of the LT2057.

I can tell you from a reliable source, that this circuit works very well, it has been my 10V LAB reference for about four years now. (change it a little about 3 years ago)
After my HP 3458A was calibrated again this january, the reference had measured no more than 3PPM in three years.
The LT1021 used in this design, where at least two years aged, before it was included in this reference, it is in a oven of about 43C.
Later this year it wil be a double oven, but is have at the moment no extra time to spare for this project.

Kind regards,
Bram

[zhtoor]

sorry i did not see the 100nf film cap.

choppers are bad-boys when it comes to generation of pretty fast charge-injection pulses at the inputs,
according to MisterDiodes, (who has measured these).

start at page 18 of:-

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0ahUKEwjc8bbJ-uLaAhXEFSwKHaf0DH0QFggkMAA&url=http%3A%2F%2Fe2e.ti.com%2Fcfs-file%2F__key%2Ftelligent-evolution-components-attachments%2F00-14-01-00-00-70-21-03%2FChopper-Noise.pdf&usg=AOvVaw2AKCrO2oJodvZcQSG5mzV5

best regards.

-zia

[blackdog]

Hi zhtoor, 

I hadn't seen that document about chopper opamps yet.
The problems with the charge injection do not occur in my setup because of the low impedances and the extra low pass filter that is set up around the opamp.

Or at least, I can't measure them and never noticed them in my LAB reference.
In any case, I will go through the document carefully.

Thanks

Bram

[chickenHeadKnob]

I don't see the point of starting with 5 volt refs and ending up with 10 volt output. LT1021 is available in 10V versions, you could parallel those and have one less source of drift.

[blackdog]

Hi chickenHeadKnob 

Maybe you can understand if you know, that I have at least 30 LT1021CH 5V and most of them are now well aged :-)

And there is another point that may be also important to others users, when they want to power the circuit out of a 12V battery.
The LT1021CH-5V is always supplied with a sufficient and clean power from the output of the reference circuit.
This is not posible with a LT1021CH-10V
Q1 J310 makes the circuit act as a LDO.

The best setup with a LT1021 is a 7V version, but you will need stange values resistors.

Do not forget that a LT1021-10V needs at least 14  a 15V power supply to be in the sweet spot.

But be completely free to do it the way you want it to.

Kind regards,
Bram

[David Hess]

The problems with the charge injection do not occur in my setup because of the low impedances and the extra low pass filter that is set up around the opamp.

Or at least, I can't measure them and never noticed them in my LAB reference.
In any case, I will go through the document carefully.

Chopper noise is a problem in high impedance applications and where the noise at the input can affect the source.  Usually neither is a problem because chopper wideband noise is high so filtering is used to limit bandwidth anyway.

The big problem I remember from before engineers figured out how to use them effectively was intermodulation producing idle tones and DC offsets.  Preserving the offset voltage drift is not easy either.

[chickenHeadKnob]

Ok, I understand now, thank you. Those are not the engineering choices I would make. I can see 12v battery operation would also then rule out putting two 5v refs in series since you have lots of those. I would have instead gone  to 15 - 18 v BAT. From what I understand of Mr. Diodes posts the laser trimmed onchip resisters in the LT1021 are subject to more aging drift than the best quality wire wounds or Caddoc TaN types and you are stuck with those in either the 5V and 10V chips but not the 7.xx.

[blackdog]

Hi chickenHeadKnob,

My LAB reference with 4x LT1021 dit not drift more than about 3PPM in 3 years, if i believe my calibrathed Agilent 3458A 

Kind regards,
Bram

[Andreas]

the laser trimmed onchip resisters in the LT1021 are subject to more aging drift than the best quality wire wounds or Caddoc TaN types and you are stuck with those in either the 5V and 10V chips but not the 7.xx.

Hello,

that is only part of the truth.
All epoxy packaged resistors will have some kind of humidity sensitivity (seasonal changes)
which you do not have on resistors within a metal can reference.

So the weak point of the cirquit are the resistors R12+R17.
But even those seem to cancel out as Bram has measured.
3 ppm in 3 years is a very good value for a reference.

with best regards

Andreas

[blackdog]

Hi,

I had the precision resistors deleverd today, so it was time to do some soldering 
This are the resistors i mounted today in the bridge circuit of the oven.
These are Vishay PTF56, 10K, 0.1% and 10PPM, i orderd them from Farnell.
Of course you can also use "normal" 10K 1% resistors, i dit this many times, but sorry, i like good/stable performance. 



I regularly test the parts I'm going to use, and I've done that with these resistors as well.
The resistors are measured with one of my two TEK DMM 4050 multimeters on the four-wire mode.


This is a picture that of the first test with the new resistors in the orange circle.
In the two yellow circles you can see that I applied the first layer of plastic glue to close the hole of the sensors.
I also changed the connections of the thermistors a bit, so that the wiring doesn't floats anymore.



The termistor wiring en closing the hole is better visible on this picture.



This is a picture of a KeySight 34461A DMM that measures the current through the oven.
At A I was measuring the voltages of the bridge circuit, this caused interference in the image.
I turned off the oven but let the measurement run,and I thought, let me try not to apply any loop compensation...
I managed to do that at a number of very small ovens, later on more on that in a different topic.
At point "B" is therefore visible that this oven really needs loop compensation, the current went all the way to 0 and back to maximum etc, etc., which of course is not desired.
At point "C" the loop is stable, but also a bit slow, further I would like to use as few electrolytic capacitors as possible in circuits.
So I have replaced this capacitor with a 4.7uF film type and the series resistance I have increased to 1M, TADA!
Beautifully quickly the current drops without any form of aberration, I have already said that I do not like to go to the limit, especially not because the lids still have to be fitted and there also needs to be a PCB in the oven and wiring of the voltage reference.
All of these have a thermal mass and partly determine the loop compensation.



This picture is to show how "quiet" the loop is, the oven currently has two layers of insulation.
The temperature is approximately 41.3C.
This picture shows the variation of the current through the furnace after pressing the meter zero in a measurement that took a little longer than half an hour.
The fact that the current still goes down does not mean that the oven is not at temperature, this is because here in Amsterdam the sun has finally started shining again.
This morning it was really cold for the beginning of the month of May.
But much more important is that in this chart no noise at all is visible, task accomplished. 



This is an update of the schematic, the new components value are now in there.



I will leave the oven on for another day and then glue the four thermistor holes again with plastic and the electronics will be coated with Conformal Coating.
Remarks, I like to hear them!

Kind regards,
Bram

[Wolfgang]

Hi,

just a suggstion: instead of the wobbly small thermistors glued into something you could use a Pt1000 sensor. Its perfectly stable and can be screw-mounted to the surface to be measured.

Regards
   Wolfgang

[blackdog]

Hi Wolfgang, 

The thermistors are there by design...
The uA723 has a low openloop gain and i dit not want to use modern MOSFets for some extra gain.
The whole schematic is now in balance, special the version made with the square aluminium tube and the 4 thermistors.
Almost no loop compenation necessary.

The version made from the old HF cabinet needs much more compensation to become stable due to the large thermal mass and the single Thermistor.

The last few weeks I have been testing several ovens, this takes a lot of time... 

Regards,
Bram

[Wolfgang]

All thermal stuff is never fast when it comes to testing it.
Much success !

[cellularmitosis]

Thanks for this thread, blackdog.

I'm planning on building an oven based on this design over the July 4th holiday.  I started by drawing up a single-fet version of your design.  I modified the R/C values a bit to see if I can get away with an all-ceramic capacitor design.  I'll have to tweak those values for a Hammond 1590B case.

Edit: marking the schematic to indicate non-working status.

[David Hess]

Is frequency compensation between the output and inverting input useful enough to include when the transconductance output is available for frequency compensation to ground?

[eurofox]

I build as well an oven control system based on the idea of kj7e design and reach stability of +/- 0.1°C at 35°C.

It is a simple design easy to tune, use less than 2W to heat a Teco box none isolated, isolation will be done of course when I finalise the LTZ1000 project that will work on battery.

[Kleinstein]

With the rather low frequencies relevant for the thermal setup there is not big difference between having the compensation from the output or the compensation pin (trans-conductance output)  of the LM723.

For the regulation it might be even better to take the slow compensation even further down, from the resistor to ground used to limit the current. this way the nonlinear FET curve and possible windup would have less effect.

[blackdog]

Hi,

First this, all the remarks about the compensation pin of the uA723: "bin there done that" 

Next the schematic of the controler which is built in the thick HF housing with the yellow color.
Because the oven itself has a different thermal mass, the compensation is also different from the oven built of a piece of rectangular aluminum pipe which I also showed here.
Also look at pin-13, it has now a 220pF HF compensation capacitor, it was sometimes oscilating and 100pF was enough, but i doubbled the value to be save..
This oven works wel because of the verry thick walls of the housing, and the small distance between the powersource and the measurements thermistor.


And now we come to what cellularmitosis is trying to build :-)
"You will fail my young panduit" (Yoda) 

The box you ordered is not made of aluminum but of clutter and not of good quality aluminum with a wall thickness of at least 4mm.
And how do you attach the Thermistor to your Hammond box?
There is a error in your schematic, the resistor in de bridge connected to the thermistor is not 10K but about half of this (5K) for about 42C oven temperature.

A oven schematic cannot be seen separately from the oven housing, they belong together.
That is also the reason why I am showing the differences here and also explaining them.

Look at the controler and his compensation voor de square aluminium tube version totaly different.
This schematic and this oven is the best i made with a uA723 controler, and is more than enough to keep voltage references on temperature.


I can only tell you guys to follow my directions to get something usable 
Building a good oven is more complex than building a good Power Supply and everybody think they can build a power supply...

For the uA723 oven controler.
Best is the square tube version with the thermistors drilled in the walls just behind the chip of the MOSFet, you can see it in my pictures of this oven.
Use the components in the schematic, different MOSFets or Thermistors, you are on your own 
Think about the wiring, it is transporting two directions the heat!
Best is to use two layers insulation material, make shure the oven comes only a little in contact with the insulation material.
Second layer also some air arond it and than the second layer insulation material.
Air is a far better isolator than the Polysyreen i normaly use, so that is why less contact between the oven and the insulation material is important.

If you do your best and follow these directives, you can have a oven with a "gain" of 200x that is the temperature in de oven wil change only 1/200C per C.

Can it be better, yes but not much with this controler, than another controler is necessary with direct DC feedback from the power section to get a far higher gain.
But also the isolation, wiring encetera, encetera will need far more attention.

Is this all necessary for a voltage reference or measuring amplifier where i also using is for, that is for you to decide.
My decision is that i do not want to worry about the oven, if there is a good voltage reference in it.
I then choose for a bigger confidence and build it as good as possible with the indicated components.

Designing and building ovens is fun, you can learn a lot from them, so have fun!

Kind regards,
Bram

[cellularmitosis]

Thanks for the feedback, Bram.  Actually, I am building a 25C oven, which is why I used the 10k resistor.

[blackdog]

Hi cellularmitosis,

Do you cool te envirement down to 10C then?
Less then 10C between outside en the set oventemperature wil be extra difficult.

Kind regards,
Bram

[cellularmitosis]

At least for the microcontroller-based PID oven, 3.5C of headroom wasn't a problem.

[blackdog]

Hi cellularmitosis,

I had also good results with a Arduino PID controler.
But... if you have already good results why build this uA723 design?

The same rectangular aluminium oven i want to test with a Arduino PID controler.
The uA723 has not a high openloop gain, but with good design you can work at 5C difference.
But there are so many thing that play a role in your oven stablity, think about the wiring and isolation i talkt about. 

The fact that your oven is set to 25C does not mean that it will continue to do so at varying ambient temperatures, say from 20 to 30 Celsius.
I assume that with a Peltier element you are going to make a colder environment and then put the oven on 25C?
Be aware that you then have two loops that influence each other.

Kind regards,
Bram

[cellularmitosis]

Hmm, perhaps 25C regulation in a 22C environment is not an attainable goal for the LM723.

Currently, I seem to be running into some sort of problem where the current limiting circuit is loading down the inverting input.  I'm not sure what is causing this, but if I disconnect pin 3, the inverting input returns to where it should be.

I just realized I don't have a base resistor on the current limiting transistor.  At one point I managed to have the circuit draw about an amp, while using a 1R current shunt (which should not allow 1 amp to pass).  Perhaps I have damaged my LM723 somehow, and the current limiting circuit is no longer behaving correctly.

Edit: Forgot to respond to this question: The reason I wanted to try the LM723, even though I already had some results with a digital approach, was because I wanted to see if a pure analog approach could work as well, which would be a simpler circuit for hobbyists to construct and possibly also keep any digital noise further away from the circuit.

[kj7e]

Not trying to hi-jack this thread, here is another way to build a very simple and effective analog (low noise) oven controller, adjustable from ~30-40C.

More details in this post;
https://www.eevblog.com/forum/metrology/kx-reference/msg1241349/#msg1241349

[blackdog]

Hi cellularmitosis,

I forgot to mention that most Peltier controler can ad and substract heath, that is a big difference with a lineair controler...

kj7e
Can you show us how you used this controler, some pictures about the mecanics would be nice, please... 
That is then wat kind of box were are the termistors mounted, what kind of wiring dit you use...
Where dit you measure the temperature?

Kind regards,
Bram


 

[kj7e]

Hi Bram, see the link in the post above and here for the oven enclosure details and performance results;
https://www.eevblog.com/forum/metrology/kx-reference/msg1236890/#msg1236890

If I were to build it again, I would add more heating elements for a more uniform thermal distribution and to reduced thermal gradient.

To copy from my post here;
https://www.eevblog.com/forum/metrology/kx-reference/msg1264331/#msg1264331

"Also mounted the heater element pass transistor to outside front of the oven for added efficiently and this also helps heat a third side.  I had some extra 10K NTC sensors laying around so I started experimenting again today and may have made vast improvement.  Currently, the heater resistors are on the ends | ----- | of the oven, one sensor was mounted just offset from the left resistor.  This gave great and stable feedback and the area near the resistor stayed at a very consistent temp.  Where I found some variation is with the internal air temp due to the top, bottom, front and back of the oven not being heated and only relying on the conductive heat from the left and right sides.  Better insulation would help, but there is no more room for more insulation in the box.   So my idea was to place the oven body temp sensor center bottom of the oven,  this surface, while insulated, is in contact with the bottom cover of the box.  I figured this area would be more representative of the internal air temp and the most effected by ambient change.  So far I'm able to hold the internal air temp well within 0.1 dec C from the same 15-30 dec ambient.  The downside is a bit more initial oscillation and time to stabilize, but it seems to be a good trade off.  Ill post of some temp plots and sensor locations after more testing."

Later, I added a second 10K NTC to measure the internal air temp of the oven in series with the oven body temp sensor.   Its the internal air temp I care about.  This creates 50/50 average between the two sensors, the body temp sensor is needed for stability but the air temp sensor helps regulate the internal air temp.

In the photo below, you can see the air temp sensor bent over between the two boards, the oven body temp sensor is mounted to the bottom of the oven below the green board on the left side.

[cellularmitosis]

Thanks Damon.  Was the 3055 also mounted to the aluminum case?

[kj7e]

Thanks Damon.  Was the 3055 also mounted to the aluminum case?

Yes, you can see the TO-220 mounting screw on the left side above the KX board.  In hindsight, I would have placed this more away from the KX board.

[cellularmitosis]

I had been using the LM723 datasheet, but today I happened to take a look at the uA723 datasheet and found its schematic is a bit more approachable.

[cellularmitosis]

Currently, I seem to be running into some sort of problem where the current limiting circuit is loading down the inverting input.  I'm not sure what is causing this, but if I disconnect pin 3, the inverting input returns to where it should be.

Oopsie, I accidentally drew pin 1 as the base of the current limiting transistor.  This should be pin 2.  I effectively had no current limit at all, which explains the strange behavior I've been seeing.   

[blackdog]

Hi 

It is good to see people are building ovens!

I want to share some points that are important for a good oven.

1e
Use copper (difficult to find boxes) or aluminium, but "real" aluminium!
A round aluminium tube with a 4mm wall is good, but maybe for most of you to difficult to make a oven out of it.
Why, you need good resistance wire and expensive 3M glas tape (#69)
If you are carefull you can use resistance wire and Kapton tape.
The easiest way is to use square aluminium pipe with a wall thickness of 4mm with the four MOSFets mounted directly on the pipe as I have shown in this topic.
You need the minimum 4mm wall thickness because you have to drill the four thermistors behind the MOSFets and you need the at least the 4mm for the good transportation of the heat.
The blue glow at the top is the conformal coating I used.



The heat conduction of the Hammond and Bimboxes are not good and also the wall thickness of almost all these boxes is too thin.
The large temperature differences in the walls of the housing, will create thermic in the oven and you do not want this.
The second problem is that it wil be difficult to get the loop stable.
A good oven has no current and/or temperature oscilation!

2e
If you still want to use Hammond or Bimboxes, then take enough heat sources.
My preference is for the older type of MOSFets such as the IRF140, IRF240, then use 4 or 6 pieces and they are cheap.
Use just like i dit 4 thermistors directly under the MOSFets.
Do this as well as possible, as I have done by installing 4 thermistors directly underneath the MOSFets chips.
You have then mounted the MOSFet directly on the housing on the outside without insulation with thermal paste
and then glue the thermistor on the inside directly underneath the place where the chip is in the MOSFet.
You can use 10 seconds glue, but better is to use a blob of thermal glue over the thermistor, like this.

This is another oven setup for my PT1000 measuring amplifier, here I use Kaptop tape and resistance wire for heating the oven.
But it's about the sensor, which is buried under a layer of thermal glue,
Don't forget to also glue a part of the wiring, this here is an Analog Devices TMP37 sensor.


3e
Each part is carefully defined in this uA723 scheme, thinking you know better if you don't feel like following my directions will almost certainly result in a poorly functioning oven.

It is little like this:
You can make an adjustable power supply with an LM317, but is it not a LAB power supply because you can control it from 1.25V to 30V?
No, it has high internal resistance, high noise, no current control, and so on and so forth.
( I realy like the LM317 and is brother and sisters, but it will never be a good LAB power supply)

4e
For a good oven this is necessary:
Good heat conductive material diamond is best 

The worse the heat conductive properties are of the oven housing, the more heat sources you need for an even distribution of heat to prevent thermic in the oven.

Tight coupling between the heat sources and the measuring sensor for a stable running controler.

Good thermal isolation of the oven, there are more ways to do this, some examples.
Here i use "air" as a isolation material for the inner oven, only two nylon screws keep the oven in the middle of the outer oven.



Build in PWM controler, i will never do that again, that is asking for EMC trouble. 



A different oven, and now the oven is hold inside the polystyreen bij small strips of polystyreen,
so there are no large surfaces of the insulating material that makes contact with the oven.



And a different setup of how to isolate a oven.
This time there is a layer of a product called "Blue Deck" around the oven.
http://www.en.bluedec.nl




Make your oven airtight as possible, most Hammond and Bimboxes are not airtight and if they are, the user start to dril holes in it 

Think of the heat leak through the wiring.

Build the controler like you are building a high quality voltage reference, so use good bridge resistors en a high quality thermistor.

If you follow these direction, your oven wil stay probably within 0.05C between 20C and 30C ambient temperature over a lang period of time.
But.. A Chain is As Strong As Its Weakest Link

Do you want a better controler than this uA723, than look for "Jim Williams" and "oven" look for document from its MIT time.
Than you wil see i do not talk BS about ovens 

The high quality controlers from Jim are unnecessary for modern voltage references or measuring amplifiers.
Do you want to go a little step higher in controler performance, than use a TI INA125 IC.
You can have than a gain that is simple to trim, so you can those between high gain and large compensation capacitor or lower gain and a smal capacitor.
You will have to decide for yourself what the "sweet spot" is for your oven.

Happy building, also if you do it your way!

Kind regards
Bram

[cellularmitosis]

Beautiful!  What did you use to cut the polystyrene?  A hot wire setup?

[blackdog]

Hi,

One of my IT clients is an architectural firm, if I need some insulation for an oven, I ask if I can use their professional Hot Wire machine. 

But... Do you want to see more oven photos?   

Lets start then!

This one is designed to calibrate TO92 temperature sensors, the temperature is set between 40 en 45C and is about 5 years old.
Something went bad, the hole u see has a broken dril in it 
This prevents me from inserting the sensor to be tested deep enough into the hole.
In the left hole is one of the bridge resistors and the second with the yellow wires, is the thermistor.
At the top the loop compensation parts , 47K5 and 220uF old style via pin-13 of the uA723.



I connected this afthernoon the oven to a power supply hanging on the wires in a small polystyreen box for some isolation.
The variation is a few mili-C for 1.5 hours.



This is a design from a few weeks ago, the first test version made according to the latest uA723 schematic.
I glue a very thin piece of double-sided pcb material to the BPR10 resistor with thermic glue.
And build carefuly the electronics on this board, the BD137 is a total plastic version.


This is a close-up, just above the "big bridge resistors" (0.1%) you can see the thermistor glued to the BPR-10.
There is a hole in the thin board so that the termistor is touching the resistor layer of the BPR-10.
This was not good enough for me, the yellow and green wire is for the loop compensation and i do not want them in the circuit!
The green wires are a 5K smd thermistor to measure the oven temperature with one of my 34461A DMM's



Another oven, this one is also fitted with Blue-Deck insulation.



And of course I also tested with Hammond or Bimbox boxes.



Sorry for the bad picture, also 4 a 5 years old, to many IC's used, also a uA723 controler would be better fore this box.



This is the latest, about a week old a test setup to find out if a can use this oven for the traveling 10V reference i'am designing.
Also a BPR-10 resistor but 10 Ohm so the current can be high enough for a 3.7V Li-Ion battery.
What is different in this version?
The BD137-16 is closer to the middle of the BPR-10 resistor and the thermistor is almost in the middle under the blue blob of thermal glue.
Again I made a small hole in the copper so that the thermistor makes a good contackt with the BPR-10 resistance.
Is it now perfect, no, but almost, the current while heating has one wave (about 1,5 second) and then its stabel, everithing is now so tight together i do not need loop compensation components.
The temperature has a little overshoot of 0.3C and then its stable.
Are there any other problems? YES! isolated this oven uses with normal temperatures about 140-mWatt.
The uA723 itself uses 45-mWatt, and with these small ovens, the energy that leaks away via the wiring becomes proportionally greater.
Next week I will find out how I will do the mechanical construction and where I can apply very thin wire to limit the heat leakage,
use 0.05 and 0.1 mm wire to connect some parts of the oven.

On the other side i glue 2x a selected LT1021CH-5V on the BPR-10 resistor, this wil use some extra energy, but also at energy of the LT1021 IC's itself.
This is about 20-mWatt for both IC's and I hope that the additional thermal mass added by the LT1021 ICs, will somewhat increase the total consumption,
so the uA723 can do better its work.
Otherwise I will have to raise the oven temperature from 40 to 45C in order to be able to operate around 30C ambient temperature.



And yes, i have even more test ovens 

Kind regards,
Bram

[cellularmitosis]

Thanks Bram, lots of great inspiration in this thread!   

[kj7e]

My next Vref oven I would build in a double enclosure.  First a steel enclosure for EMI reasons to house the oven controller, per-regulator and a smaller aluminum oven for the Vref and buffer.  I would use a smaller oven and work on heating it on as many sides as I could to minimize thermal gradients.  I would even consider a double oven where the outer does most of the heating, then the inner small step and finer control, this would make it easier to lessen thermal gradients.  However, the KX LTZ1000 board is not very temp sensitive and I would not see any benefit to holding the temp any more stable than what I have now.

[blackdog]

Hi cellularmitosis,

Thanks 

What!!! you need more?

Let me show some big falings...
This is a quad 1N827A setup in a oven, and it is not good, to drifty.
Why, I stopped searching for the problem, it took too much time to find out...
but probably too much heat was leaking through the thick wiring.



The schematic of this oven.



The reference zeners 1N827A.



And how the reverence/oven is mounted in the polystyreen box, a lot of air around it.



The next reference oven is the one of which I am most proud and also the one of which I am most ashamed. 
Just soldered open for you to show how this oven is built up.



Here began one of the many mistakes I made at this reference oven.
At the spot of the white wire was a thick piece of copper wire, this gave when heating this oven a lot of mechanical stress on the assembly.
I have cut away the thick wire here and replaced it with the thin wire,
That made it a lot better, but still not good!
The thick wire was connected to a connection of the APEX reference which is glued to the IRF540 heating transistor.



And this is the schematic.


Side view.



And more



And here you can see that even at the bottom, parts are mounted..
I'am proud of how i could fit al the components in this uMetal box, proof that I can build in three dimensions. 
But a BAD reference!



And the last one, a piece of square aluminium tubing with a e-bay temperature controler set to 0.1C difference.
Garbage man, you can also remove this!



Bey!,
Bram

[cellularmitosis]

My next Vref oven I would build in a double enclosure.  First a steel enclosure for EMI reasons to house the oven controller, per-regulator and a smaller aluminum oven for the Vref and buffer.  I would use a smaller oven and work on heating it on as many sides as I could to minimize thermal gradients.  I would even consider a double oven where the outer does most of the heating, then the inner small step and finer control, this would make it easier to lessen thermal gradients.  However, the KX LTZ1000 board is not very temp sensitive and I would not see any benefit to holding the temp any more stable than what I have now.

It is somewhat ironic that it is only the less expensive references which really need the more expensive oven

However, a better oven could be beneficial for a Vref which also has scaled output, and also for resistance references.

(I've also been thinking it might be neat to have a combined LTZ + resistance reference.  One of the 70k resistors could be replaced with a 100k, a 13k R4 could be split into a 10k + 3k, R5 would be 1k, a 120R R1 could be split into 100R + 20R.  With kelvin connections for each, you could power off the LTZ and use the device as a 100k,10k,1k,100R resistance standard.  Or just 10k if you only need artifact calibration).

[cellularmitosis]

Thanks Bram for also posting some failures -- good lessons for the rest of us!

This afternoon I took a second look at the strange behavior I'm seeing from my LM723.

I took a step back from the oven and breadboarded up a test circuit (using a fixed resistor rather than a thermistor, and using a "dummy" 2N3904 heater transistor).

I used a 10k/10k divider for the non-inverting input.

I used a 4.7k in place of the thermistor (above the inverting input), which simulates a thermistor which is heated up a bit.

We would expect 3.5V at the non-inverting input (7V / 2), and about 4.76V at the inverting input (7V * 10k / 14.7k).  With the inverting input higher than the non-inverting input, we would expect the output of the internal "op amp" to go low, which should shut off the output, and the 2N3904 should not conduct.

I breadboarded this up and probed various points of the circuit, and the results are shown in test1.png (attached).

We see that the inverting input is sitting at 3.28V, which is about 1.48V lower than the expected 4.76V.  This means 0.79mA flows through R1, and 0.33mA through R2, indicating the inverting input is "stealing" 0.46mA.  Further, we see 1.46V at the output, when we expect the output to be "off".

I then tried disconnecting the base resistor of the current limiting transistor and again probed the circuit, the results of which are in test2.png (attached).

We find the inverting input is now at 0.93V, indicating the inverting input is now "stealing" about 1.2mA (1.28mA - 0.09mA), and we also find 0.74V sitting on the base of the current limiting transistor.

So, the "op amp" in the LM723 doesn't seem to behave like a pure op amp.  I'm guessing the issue is that the current-limiting functionality is connected to the inverting leg of the circuit, and there is a strange interaction there.  It seems that using the thermistor in the upper half of the inverting divider isn't workable.

Perhaps we can instead use the thermistor in the lower half of the non-inverting divider.

Any feedback or insight is welcome -- this is new territory for me!

[cellularmitosis]

I switched to a discrete opamp-based oven circuit -- it is looking good! 

More details in the morning

[blackdog]

Hi cellularmitosis,

Yes! teasing time.

I shall push you in the wright direction...
If i see your schematic it is different from my uA723 controler, you have to pay for that! 
And then I mean that you will learn what it can cost to customize a schematic, you will pay in time...

If you use a simple setup with a bjt what will the voltage be on the outputpin of the uA723 and if a look at your last drawings is see 0.11 and 1.46V.
And now learn  look at the datasheet think about pin-9 of the uA723, that will be your problem solver, tell us why.

I hope you are not going to dream about this circuit, have a good night's rest.

Kind regards,
Bram

[cellularmitosis]

ahhhh, now this is all making sense   I ran into a similar regulation problem with my OP07, and quickly realized it was because the output couldn't swing below roughly 1.4V, so I added a reverse-biased zener on the output to solve the problem.

And now I finally realize why there is a strange zener hanging off of the LM723 output, exposed as pin 9...

Thanks for the nudge, Bram 

Edit: hmm, your 4.7k resistor is biasing the output voltage up, which is why you don't have this problem when using a mosfet, is that correct?

[TiN]

Sorry about useless post, but I love this thread. Keep it going (esp with photos) 

[blackdog]

Hi cellularmitosis,

You are correct! 

Regards,
Bram

[Andreas]

Sorry about useless post, but I love this thread. Keep it going (esp with photos) 

Ok perhaps I can jump in with my voltage reference ageing box:

Build in PWM controler, i will never do that again, that is asking for EMC trouble.

That is why I use 100 Hz as PWM frequency: this will be canceled out by most instruments as it is a multiple of AC mains frequency at my location.

in the background: lower half of the styrofoam box consisting of 2 parts which are overlapping over the full height.

top and bottom are 2 aluminium plates with a heater foil (car mirror heater) each.
On the top side a NTC for (fast) temperature controlling (P-controller).

Additionally a overtemperature switch which switches off when electronic fails.
(I am missing that in the previous ovens: what can happen if the NTC is shorted or open?)

Then the PCB with the voltage references on the left side.
In the middle of the references a further NTC for the slow temperature controlling (I-controller).

In the middle of the PCB the multiplexers (differential) so that one of the references can be measured.

on the right side from bottom to top:
RS-232 interface (with photocouplers = isolated)
Processor for temperature control (12 Bit ADC = 0.025 deg C resolution)
level shifters for the multiplexers and power stages (PWM)

with best regards

Andreas

[David Hess]

ahhhh, now this is all making sense   I ran into a similar regulation problem with my OP07, and quickly realized it was because the output couldn't swing below roughly 1.4V, so I added a reverse-biased zener on the output to solve the problem.

And now I finally realize why there is a strange zener hanging off of the LM723 output, exposed as pin 9...

The 723 voltage output only has a compliance down to 2 volts from the negative supply so the zener is needed to drive pass transistors which have their emitter at the negative supply.  The metal can package only has 10 pins so the zener output is missing.

[cellularmitosis]

Initial estimate of performance: roughly a 20mK change in a 750mK ambient swing, which makes the attenuation about 37.5x (which also includes the error of the Keithley 2015.

Testing the oven in another oven (rather than relying on ambient swing) would help to isolate the Keithley 2015 tempco error.

[cellularmitosis]

Some notes from my quest for compensation:

• A: no compensation
• B: 0.1uF ceramic + 1M (from inverting input to after the output zener)
• C: 0.1uF ceramic + 10M
• D: 1uF ceramic + 10M
• E: 2uF ceramic + 10M
• F: 2uF ceramic + 10M (attached to top of current shunt).  this was less effective than from inverting input to after output zener.
• G: back-to-back 10uF electrolytics + 10M (from inverting input to after the output zener) (caps are positioned negative to negative)
• H: back-to-back 47uF electrolytics + 10M
• I: back-to-back 47uF electrolytics + 10M (from inverting input to before the output zener)
• J: back-to-back 47uF electrolytics + 10M (caps are now positive to positive)
• K: back-to-back 47uF electrolytics + 20M
• L: swapped around the bridge (thermistor now on bottom half of non-inverting input).  forgot to swap the compensation network, which nulled its effect.
• M: corrected the compensation network
• N: changed out the 10k/10k fixed divider for a 100k/100k divider.  yay the oscillations stopped!
• O: 1uF ceramic + 20M.  this still worked!
• P: 1uF ceramic + 10M.  this still worked as well!
• Q: 0.1uF ceramic + 10M.  this caused the oscillations to return (though smaller than before)

(Ignore the component vallues in the schematics below: only their position is indicated)

[cellularmitosis]

The final circuit ended up being nearly the same as a circuit from Charles Wenzel.

http://www.techlib.com/electronics/ovenckts.htm

(I took the idea of trying a 100k/100k divider from him, which seems to enhance the effect of the compensation RC, and was the change which finally stopped the oscillations)

Edit: made a few corrections to the schematic.

[cellularmitosis]

Here is the template (as pdf and svg) I made to cut out EVA "craft" foam pieces to cover a Hammond 1590B enclosure.

I printed this out on paper, cut the paper pieces out with scissors, then used a straight edge and razor blade the cut the foam.

[blackdog]

Hi cellularmitosis,

All the components connected to the inverting opamp input are part of the loop compensation.
Like i told before, be shure to have a tight coupling between the heath source and the sensor.
This will solve a lot of loop problems.


Andreas
Additionally a overtemperature switch which switches off when electronic fails.
In my other oven controllers is an extra sensor included for too high temperatures.

Kind regarts,
Bram

[cellularmitosis]

The prototype oven has an MJE3055 and two thermistors affixed to the bottom side with Arctic Silver thermal epoxy.

However, the oven did not run at 25C like I had expected.

I lifted up the bottom of the oven and pointed a small fan at the thermistors, ensuring they were both at the same temperature and very closely tracked the ambient room temperature.  I then alternated measuring one and then the other thermistor.

It turns out they have a rather large offset, both from each other and from the Si7021 which I used to log the room temperature.  I believe this explains why I didn't settle on 25C as expected.

These are +/-5% parts, which roughly means +/-1C (at 25C).  https://www.digikey.com/product-detail/en/us-sensor-littelfuse-inc/103JG1J/615-1016-ND/1014544

[cellularmitosis]

I hooked up the LM723-based circuit to the same Hammond 1590B + MJE3055, and fiddled around until I came up with this, which seems to be pretty well damped, and is a pleasantly minimalist circuit.

I noticed that there is a tradeoff with the value of the compensation resistor in relation to the set-point divider resistors: as the value of this resistor becomes closer to the value of the divider resistors, the compensation network seems to have a stronger effect.  However, it then also starts to alter the set-point.  Making the compensation resistor much larger than the divider resistors means that it doesn't affect the set-point, but the compensation has less effect.

I tried attaching the compensation network to the top of the current shunt and this seemed to work well (thanks for the suggestion Kleinstein).  Previously, when I had the compensation attached to the output, the voltage could wander around lower and lower during an undershoot, but the current was already zero, because the voltage was below the turn-on threshold of the MJE3055.  This seemed to act like a "wind up" problem, but in the downward direction.  By fixing the compensation network to the shunt, compensation voltage is now directly tied to heater current, which seems to work better.

I also realized that I could split the shunt into two halves, so that I wasn't limited by the 0.65V threshold of the current limiting transistor.  This seemed to amplify the compensation network's effect a bit.

I also realized that with this arrangement, since the set-point (3.5V) is always more positive than the top of the current shunt (or rather, the split current shunt is now chosen to ensure it is always below 3.5V), I can now get away with using a single (polarized) electrolytic in the compensation network, allowing for larger capacitance values than a ceramic (and I can avoid using a back-to-back pair of electrolytics).

I'll test the stability overnight.

[blackdog]

Hi,

It begins to look good 

Now make the capacitor on pin-13 220pF, 10nf is way to much, the openloop gain will be far below 1Hz and you dont want that.
The 220pF is more than enough for HF stability and you wil have a bigger oven gain, that is more temperature stability.
Perhaps a small correction of your loop components is necessary.

Your setup in the bridge  for the negative input is a little on the high side, and yes this helps to keep the loop components smaler.
But keep in mind the bias currents, these are high for a uA723, keep the uA723 in the oven, then you can probably get away with it. 
I would not exceed two resistors of 20K for this bridge section.

I see the connection of the loop components to the emitter of the MJE3055 as no problem at all, it is just a very large emitter follower of the output signal.
Keep op the good work!

Kind regards,
Bram

[blackdog]

Hi,

Andreas,
Some comments about the EMC problem with a PWM controller.

I have tested extensively with a aluminium tube oven with resistance wire wound around it.
I then connected the resistance wire to one of my function generators.
Dit a frequency sweep and measured the field inside the oven, HORROR!

Next step, winding the resistence wire Bifilair on the tube, this was a big improvement.
But even with low frequencies there were big spikes.
This is the oven with the bifilair and also double windings fore parallel (power on quick heating and serial for normal use)
The yellow wire bundle are the three sensors, one for the ovencontroler and the other for measurement and savety (high temp)



So the next step was to remove the scharp edges when using a square wave.
Below is a part of the schematic I use to limit the EMC of a P.ID controler build around a Arduino or Teensy micro contoler.
The 10uF capacitor remove the sharpedges, the P.I.D. frequency is about 1Hz a 3Hz.



And yes.. making this oven is difficult 
The wires are not allowed to touch each other.



Testing the EMC in the oven.



Testing with a square wave.



And this is how it looks after removing the fast edges.


For more info look at this link and use google translate...
https://www.circuitsonline.net/forum/view/133966

Kind regards,
Bram

[cellularmitosis]

Looks like the LM723 oven's attenuation / performance is about 50x (0.5C ambient change results in roughly 10mK change).

Bram, I'll try a smaller cap on pin 13, thanks.

[Andreas]

Some comments about the EMC problem with a PWM controller.

Hello Bram,

of course you are right.
Bifilar winding is essential. (Is also done on the heater foils).

And I also forgot to mention that I usually do not use "normal" FETs when it comes to "power".
I usually use "automotive fully protected" FETs.  (like VNP5N07 or BTS432).
They are optimized for low emissions.
And further they limit the current in case of short cirquits.

with best regards

Andreas

[IconicPCB]

Now that the temperature is monitored and controlled by the LM723...

Has anyone checked performance of the REF voltage of temperature stabilised buried zener inside the LM723?

[Wolfgang]

Hi,

I built a self-heated LM723 reference. It can be seen here:

https://electronicprojectsforfun.wordpress.com/silly-circuits/silly-circuits-a-heated-lm723-reference/

Drift is a millivolt per a few days (out of ca. 7.15V), but this could be a burn-in effect.

[IconicPCB]

Wolfgang,

I saw Your article prior to posting above comment.
I was hoping to see other confirmation of Your experience as a sanity check.

[Wolfgang]

A word of warning regarding high-capacity (X7R, Z5U, ...) ceramic capacitors:

- Their capacitance decreases drastically with applied voltage
- Their tempco is evil
- they are microphonic - See Dave Jones teardown of a Rohde scope where tapping at the touchscreen created voltage jumps - caused by ceramic Cs !
- They age
- In my opinion, the should never be designed into a control loop - Use high quality foil when possible.

Just ranting - much luck experimenting !

[blackdog]

Hi,

+1 for Wolfgang 

Kind regards,
Bram

[Wolfgang]

Hi,

I am continueing the long term tests, but unfortunately I have not seen others doing the same or something similar yet.
Maybe you try yourself ?

[IconicPCB]

My highest resolution instrument is an advantech 16 bit data logger which is not good enough to get me to the resolution and stability  of measurement.

[Wolfgang]

Yeah, you are right. When it comes to checking references, the checking instrument must be at least an order of magnitude more stable than the expected deviations.
Just to be on the safe side, I teamed a RIGOL 6 1/2 digit M300 datalogger with a Keysight 34401A multimeter. At a measured value of around 10V, they disagree for ca. 300uV maximum. When you have a look at my webpage, there is a multimeter comparison at the bottom. 6 1/2 digits at this range means about 300uV accurracy. A 16bit *resolution* (thats not enough, we need accuracy, too) is about 5 3/4 Digits, but again absolute values will be a lot less, so no option.

If you really want more, then you have to go for an 8 1/2 digit multimeter unit, bringing uncertainty down to +/-26uV (Keysight 3485A, 4ppm). But then the whole setup must be improved: Temp control for all instruments, low EMF cables, .... and you need some extra 10k€.

If you can afford all that, then you can also invest in a LTZ1000A ... My LM723 reference creeps around in the mV range, so I would say that 300uV of tolerance is legit.
We are far beyond what the LM723 datasheet was promising, so its fine for me. My reference was meant for radio amateurs with budget constraints.

[iMo]

Or, you may try with a "differential method". Using your 16bit datalogger as the "zero indicator" only. You would need a voltage source more stable as the 723''s ref. On that "stable" source wire a 10turn 10k Bourns (or something like that), from the wiper to your datalogger input, the datalogger's gnd lead to the 723 ref output. Set zero with the Bourns at the beginning.

In past I was using a REF01 as the voltage source for the Bourns.

This method may eliminate the 7.10V offset problem. Your logger will show the actual difference between the wiper and the 723. In case the REF01 and Bourns will be put in a temperature stable box, and you will cool/heat the 723 ref only, you can easily get its tempco.
Just an idea..

[IconicPCB]

Yes, the data logger can be configured as a differential instrument.

[Andreas]

Has anyone checked performance of the REF voltage of temperature stabilised buried zener inside the LM723?

Hmm,

I cannot find the term "buried zener" in the current data sheet.
I am reading only "temperature compensated zener".

So from where do you have the term "buried zener" for the LM723?

On the other side: the specified ageing of 0.05%/1000 hours (500 ppm or 3.5 mV per 1000 hours)
would correspond to the measured data with drift in the mV range.

Buried zener references are usually specced with 30 ppm/1000 hours or better.

with best regards

Andreas

[IconicPCB]

Andreas,

You are quite right. I had seen a number of "documents" in the past week or so and for some reason the temperature compensated zener became subsurface buried zener in my mind.

[Wolfgang]

Or, you may try with a "differential method". Using your 16bit datalogger as the "zero indicator" only. You would need a voltage source more stable as the 723''s ref. On that "stable" source wire a 10turn 10k Bourns (or something like that), from the wiper to your datalogger input, the datalogger's gnd lead to the 723 ref output. Set zero with the Bourns at the beginning.

In past I was using a REF01 as the voltage source for the Bourns.

This method may eliminate the 7.10V offset problem. Your logger will show the actual difference between the wiper and the 723. In case the REF01 and Bourns will be put in a temperature stable box, and you will cool/heat the 723 ref only, you can easily get its tempco.
Just an idea..

When you got a data logger with good resolution, the logger itself would be a better reference. If you take a RIGOL M300, your 1 year DC spec is about 28ppm over +/-5°C temp range. There are references as good as that, but they are not cheap if you want it calibrated and with a long time spec.

This site has one: http://shop.voltagestandard.com/main.sc but spec is only up to 30ppm, with 1.8ppm/°C tempco and only 6months. For some 10€.
If you look for an LTZ1000A based one, its probably a few 100€ if its worth anything.