Worlds Smallest Cheapest DIY Temperature Controlled Oven

[paulie]

In another post I mentioned designing a small low cost homegrown OCXO. Somebody offline requested more info and instead of "diluting" that thread let's start a new topic. Originally created for a precision crystal based RTC but recently re-purposed for testing DIY voltage references. Pics of the test setup and schematic attached below.

Parts total less than 10 cents and as usual only takes a few minutes to build (limit of my attention span). Seems to perform much better than other more costly and complicated designs, maintaining about ±0.2C even down near freezing.  In stand alone versions the pot is replaced by an 0805 fixed resistor, The 2 lead device is a calibrated NTC for verification only and not part of the oven circuit. Other than super thin 40awg wire which is crucial for stability and low power nothing too tricky. Glad to answer any questions.

ps. Note that all components are thermally coupled.

[paulie]

Thanks for the complement. Yes, I ran across that quote many years ago and it was a great inspiration for all my projects. Too bad more of the great minds here don't adopt a similar philosophy. Some times less is more.

[SeanB]

I made a similar one years ago, using a TO220 PNP transistor. Works well to heat the crystal it is bound to. All buried in a small box of cardboard and polystyrene. Do not remember what I used, and do not really want to take it apart to see but it is basically the same, just relying on a well regulated 15V supply ( from the psu in the equipment) to keep it stable.

[Rick Law]

I like the board you mounted that on!  Looks like an ice-cream stick.  Creative!

[cosmicray]

What is the white material being used for the thermal wrap ?

[Ian.M]

Looks like ordinary expanded polystyrene. If you zoom in you can see the boundaries between the individual pre-foamed beads.  The lid shows an indentation from the retaining rubber band, local crushing being typical for the material.

[1xrtt]

OK, I'm not that clever. What's is the black magic about using a very thin wire?

[matseng]

So this is same-same as what Roman Black did 8 years ago but with he used a standard BJT...

http://www.romanblack.com/xoven.htm

[mikerj]

OK, I'm not that clever. What's is the black magic about using a very thin wire?

The thinner the wire, the lower the losses from heat conduction.

[1xrtt]

OK, I'm not that clever. What's is the black magic about using a very thin wire?

The thinner the wire, the lower the losses from heat conduction.

   Sounds obvious now. Tks.

[paulie]

You guys are SMART! Yes, that is packing material styrofoam. Yes, the thin wires minimize heat loss.

Yes, that is a popsicle stick. More accurately a Walmart "craft stick" (aka doctors tongue depressor). 500 per box for a little over $3. I use them in most of my projects not only because they are cheap but take superglue very well and easy to drill holes for things like pots and jacks.

And yes, very similar to Roman Blacks OCXO but not quite same-same. I did build one to start but not satisfied with performance decided on some improvements. Some differences:

1. His had twice as many parts
2. Mine gets 0.2C stability vs the 3c for the one of his I built.
3. Styrofoam orders of magnitude better insulator than epoxy which isn't that good.
4. Thick connections vs my 40awg also contributes to more stability and much lower current.
5. Big difference in cost because darlingtons are not cheap. In fact 10x more on Ebay.

He was definitely the pioneer but as time goes by better technology allows improvements. In this case cheap MOSFETs for better control. Kinda like Bells original [hone (Watson... I need you!) compared to modern cell. Well, maybe not that big a difference. LOL

[edpalmer42]

How about a one-piece oven?

Glue a PTC thermistor onto the crystal case.

https://www.eevblog.com/forum/blog/eevblog-236-fe-5680a-rubidium-standard-teardown/

Insulation is optional.  In the 5680a it doesn't matter because the oscillator is inside the PLL.

Ed

[jpb]

You guys are SMART! Yes, that is packing material styrofoam. Yes, the thin wires minimize heat loss.

Yes, that is a popsicle stick. More accurately a Walmart "craft stick" (aka doctors tongue depressor). 500 per box for a little over $3. I use them in most of my projects not only because they are cheap but take superglue very well and easy to drill holes for things like pots and jacks.

And yes, very similar to Roman Blacks TCXO but not quite same-same. I did build one to start but not satisfied with performance decided on some improvements. Some difference:

1. His had twice as many parts
2. Mine gets 0.2C stability vs the 3c for the one of his I built.
3. Styrofoam orders of magnitude better insulator than epoxy which isn't that good.
4. Thick connections vs my 40awg also contributes to more stability and low current.
5. Big difference in cost because darlingtons are not cheap. In fact 10x more on Ebay.

He was definitely the pioneer but as time goes by better technology allows improvements. In this case cheap MOSFETs for better control. Kinda like Bells original [hone (Watson... I need you!) compared to modern cell. Well, maybe not that big a difference. LOL

I'm no expert in this area, but I recently have been looking at HP10811s with the view to buying one on ebay and one of the things that HP put in their manual as an innovation was the simplicity of using the loop control transistors as heaters so I guess historically they got there first - though you couldn't really call an HP10811 cheap!

To quote from their 1991 datasheet/brochure for the 10811D/E "HP quality and attention to every design and manufacturing detail are evident throughout the HP10811D/E. For example, the oscillators' oven power control transistors, themselves, are the source of oven heat."

There is nothing new under the sun.

EDIT: The above is not meant to detract in anyway from the very impressive popsy stick micro-oven described in this thread - I only raise it as an item of historic interest.
It fascinates me the way that electronic circuit ideas re-emerge. The classic one in the field I used to be in (microwave integrated circuits) was the idea of a travelling wave or distributed amplifier. Some one at Thorn EMI research invented the idea for valves in the 1940s some time. Then in the 1980s  it was rediscovered in the context of GaAs MMICs to give super wideband amplifiers going from DC to 18GHz and beyond.

[tggzzz]

I love your design! Very brillant! Remind me of this :
"Engineering's achievement lies in simplicity : When you remove one part, it weights nothing anymore, costs nothing anymore, you don't need to learn how to use it, and its reliability becomes infinite."
[Adapted from a quote of Gabriel Voisin (1880-1973), former engineer and aviation pioneer]

A similar quote, also from an aviation pioneer: "Perfection is achieved, not when there is nothing left to add, but when there is nothing left to remove". -- Antoine de Saint-Exupery

[dannyf]

the simplicity of using the loop control transistors

http://www.romanblack.com/xoven.htm

it is better to use a TO220 or the like.

[Marco]

I'm surprised this works well, you'd think the drain current variation would shift the necessary Vgs ... I guess my intuitive grasp of the magnitudes of the effects is off.

Would Roman Black's circuit really work equally well with all Darlington's? I think the base transistor makes some impact, you'd like the first transistor to be biased to a relatively constant current AFAICS (a Sziklai npn/pnp pair would seem better than a Darlington to me).

[paulie]

Shifting Vgs should not be a factor if all components are thermally coupled. It would be included in the feedback loop.

I tried many variations, both one and two transistor, including pnp common collector followed by common emitter. Some of these did have some performance advantage. Probably because bipolar beta is typically higher effective gain than MOSFET transconductance. The problem I encountered was touchy adjustment requiring at least one additional resistor, sometimes more. Also no need for bypass caps with FETs. So to keep component count (and cost) down this is the configuration I settled on.

For bigger higher powered oven control this circuit is not an option. That's where Darlington comes to the rescue. And since high power FETs cost a lot more than bipolar and much more gain is required we do end up using something very similar to Blacks design there.

[Marco]

Shifting Vgs should not be a factor if all components are thermally coupled. It would be included in the feedback loop.

Vgs has to change to increase the power output at a constant temperature, so the temperature setpoint shifts with it ... but as I said, I guess I don't have a grasp of the magnitudes of the effects.

I would try something like this before I'd try a Darlington. AFAICS the setpoint voltage would change far less for a given increase in power output this way.

[paulie]

Voltage set point will be much more unstable for two BJT in series but that is not the biggest problem. For one thing in your circuit Q1 can destroy Q2 without additional components to limit base current. I know this from having implemented something similar as described in a previous post and learned the hard way.

And I think the extreme gain (well over couple million for typical general purpose transistors) will create problems. For one thing adjustment will be touchy and additional resistors will be required in the set circuit. Unlikely any off the shelf fixed resistors could be found and even combining multiple resistors to get a proper value would be difficult. And don't forget the need for caps to suppress inevitable oscillations and other instabilities from the positive feedback circuit. Very high gain can increase performance but there are drawbacks.

ps. Regarding set point changes with temperature they are major. But maybe you still don't consider that if all the devices are thermally coupled and inside a negative feedback loop this is not a factor. However the more devices there are in a circuit the more difficult to couple them all so low parts count helps here.

[paulie]

it is better to use a TO220 or the like.

I made a similar one years ago, using a TO220 PNP transistor.

It's not difficult to build Worlds Smallest Cheapest OCXO using TO220. It's impossible. Consider size and cost. Anyway I just got another similar comment offline so for those who don't browse new threads here's a couple pics to prove I have no great fear of TO220:

https://www.eevblog.com/forum/projects/low-cost-voltage-reference-experiment/

[JohnnyBerg]

FWIW I am using a DPAK PNP (MJD127) transistor in my newest voltage reference. In front of that is a P regulator with G=300

The voltage reference is build on top of the collector of the MJD127. Tricky is the layout of the ground and supply. The heater current is not supposed to lift the output of the reference output.

I have 2 versions, the smallest being 20x22 mm and completely shielded

[Alex Eisenhut]

You guys are SMART! Yes, that is packing material styrofoam. Yes, the thin wires minimize heat loss.

Yes, that is a popsicle stick. More accurately a Walmart "craft stick" (aka doctors tongue depressor). 500 per box for a little over $3. I use them in most of my projects not only because they are cheap but take superglue very well and easy to drill holes for things like pots and jacks.

And yes, very similar to Roman Blacks TCXO but not quite same-same. I did build one to start but not satisfied with performance decided on some improvements. Some difference:

1. His had twice as many parts
2. Mine gets 0.2C stability vs the 3c for the one of his I built.
3. Styrofoam orders of magnitude better insulator than epoxy which isn't that good.
4. Thick connections vs my 40awg also contributes to more stability and low current.
5. Big difference in cost because darlingtons are not cheap. In fact 10x more on Ebay.

He was definitely the pioneer but as time goes by better technology allows improvements. In this case cheap MOSFETs for better control. Kinda like Bells original [hone (Watson... I need you!) compared to modern cell. Well, maybe not that big a difference. LOL

I'm no expert in this area, but I recently have been looking at HP10811s with the view to buying one on ebay and one of the things that HP put in their manual as an innovation was the simplicity of using the loop control transistors as heaters so I guess historically they got there first - though you couldn't really call an HP10811 cheap!

To quote from their 1991 datasheet/brochure for the 10811D/E "HP quality and attention to every design and manufacturing detail are evident throughout the HP10811D/E. For example, the oscillators' oven power control transistors, themselves, are the source of oven heat."

There is nothing new under the sun.

EDIT: The above is not meant to detract in anyway from the very impressive popsy stick micro-oven described in this thread - I only raise it as an item of historic interest.
It fascinates me the way that electronic circuit ideas re-emerge. The classic one in the field I used to be in (microwave integrated circuits) was the idea of a travelling wave or distributed amplifier. Some one at Thorn EMI research invented the idea for valves in the 1940s some time. Then in the 1980s  it was rediscovered in the context of GaAs MMICs to give super wideband amplifiers going from DC to 18GHz and beyond.

They were used in Tek scopes as well, to drive the CRT. Ye gads, imagine trying to align one of them...

http://www.ase-museoedelpro.org/Museo_Edelpro/Catalogo/Overview/distiampII.pdf