LEDs at Elevated Temperature

if you are trying to measure temperature by measuring the forward voltage drop of the LED, there's a better way to do this... a 2N3904 max operating temp is 150C, and the forward drop of the B-E junction has a temperature coefficient of -2.4mV/degC. at room temperature (nominally 25C) the forward Vbe will be about 700mV. at 100C Vbe will be 520mV.

unclejed613 said:

if you are trying to measure temperature by measuring the forward voltage drop of the LED, there's a better way to do this... a 2N3904 max operating temp is 150C, and the forward drop of the B-E junction has a temperature coefficient of -2.4mV/degC. at room temperature (nominally 25C) the forward Vbe will be about 700mV. at 100C Vbe will be 520mV.

Click to expand...

I'm using a Microchip MCP9808 temperature sensor. I need good temperature accuracy – this chip is within a degree C or 2, but with sub-degree resolution. I can calibrate in a boiling water bath at 100°C and have very accurate temperature resolution.

The LEDs will be used as indicators only.

If the LEDs are for indication only then why does it matter if they fail? You'll be able to get all the information from the logged data. I like Nigels suggestion of LEDs only and wonder if the current will have an influence on longevity.

Mike.

Pommie said:

If the LEDs are for indication only then why does it matter if they fail? You'll be able to get all the information from the logged data.

Click to expand...

Again, my question was specifically about the performance of LEDs at high temperature. Why I want to have LEDs isn't a matter for discussion. Nor did I ask anybody for design assistance.

So far, only a couple of posts have even touched on the question asked, and nobody has mentioned any real-worth experience. JimB did offer an alternative of using grain-of-wheat bulbs.

Thank you all for the comments. I don't mean to sound ungrateful, but I didn't ask for the "engineering team" to take over my project.

I'm guessing no one knows the answer hence the suggestion to try just the LEDs on their own.

Mike.

And please don't get me wrong. If I have overlooked some fundamental problem, I am happy and glad to hear about that. But I've used all of these components before, so I know my design is fundamentally sound.

As my EE friend says, there's more than one way to skin the electronics cat, and many different ways to solve problems.

There's a simple way to skin your electronic cat, cobble together a battery, LED and resistor any try it.

Mike.

Pommie said:

There's a simple way to skin your electronic cat, cobble together a battery, LED and resistor any try it.

Mike.

Click to expand...

Perhaps a range of different LEDs on a piece of perfboard, one identical row powered, and one not.
See if there is any difference between the rows after the heating/cooling cycle.
My 0.02

I'll probably go ahead and get my boards built and give it a try. Having LEDs is a convenience; it's not essential.

I'm away from home for another week or so, so it's not a matter of soldering a few LEDs together and making a test.

Found this on Stack Exchange, by ETO member Tony Stewart :

From here:

As you know, I'm not an EE, just a hobbyist, so if what follows is a steaming pile........you know what to do.

Soldering and reflow processes subject the LEDs to far higher temperatures than what you will experience during the canning, but they are un-powered.
Your test requires a ramp up, 4 hr soak, and a ramp down, which is way outside the duration of a reflow, but the max temps are below half.

After doing a powered/un-powered perfboard test, should you find that the powered LEDs fail, but the un-powered ones survive, perhaps you could write your code so that the status LEDs only become active after the temp ramp down period?

Another 0.02

The advice that I received from those who know a little more: pulse the LED with a low duty cycle.
If your LED is only status indicator that a particular condition has been reached, it does not require to be on continuously.

Here's a picture of the finished board, which is 1.5" × 3". The size is such that it will position the temperature sensor in the center of a pint canning jar. The connector at upper right will actually be a female right-angle header. A CH340 UART-USB module will plug in directly to unload stored data.

The MCP9808 is located at the center of the board. LEDs 3 and 4 are for through-hole replacements if the SMT LEDs don't survive. They are positioned so a 4 position right-angle female header can be installed to make the LEDs easily replaceable.

The battery holder for the temperature-rated CR2450 is on the back of the board.

In-use, the board will be vacuum-sealed in a plastic bag that can withstand the temperature.

I haven't yet defined the purpose of the two switches and LEDs. I will need some kind of easy-to-use user interface, since these boards may be used by others.

Thanks for the comments. I should have some knowledge about LED operation at moderately elevated temperatures in a few weeks.

I have misunderstood how Instant Pots work.

A traditional pressure cooker uses a weight on an outlet in the cover to maintain steam at the desired pressure. When the steam pressure is too high, that pressure lifts the weight to the point where stream escapes. The reduced pressure allows the weight to drop back down, blocking the outlet. The pressure builds and the cycle repeats.

Other types of pressure cookers use a pressure relief valve. This type may offer an adjustable pressure relief point, allowing pressure and thus temperature to be adjusted.

Instant Pots also feature a weight like a traditional pressurevcooker, but its use is different. It releases steam at 15 PSI and acts primarily as a safety pressure release. The Instant Pot monitors the temperature at the outlet of this valve to determine when pressure has built up as a result of boiling.

After boiling and pressurization has been achieved, the Instant Pot is thermostatically controlled to the point where the pressure should be about 12 PSI (I may be wrong about the exact number). No steam should be released during operation as in a traditional pressure cooker, which guarantees a given temperature. Instead, the temperaturevis dependant on the control system and how much that temperature varies is unknown.

Progress is being made. The board is actually boiling away right now.

My first experiment was a flop. I developed the firmware while powered by my PICkit 2. Lots of LED flashing as the user interface, but only a short flash each time a measurement was taken. Reasonably easy to control, with strong features to prevent losing stored data.

So of course the next step is to insert the special high temperature battery, vacuum-seal the board and get to boiling.

A couple hours later, after an hour of boiling and cooling down, I pulled out the board. Nothing happening. Pressed the reset switch and there was the tiniest glow of one LED, nothing from the other which should have been alternately flashing.

My first thought was the LEDs had been fried, but that's not the case. The battery was dead. A CR2032 is rated at 220mA-H, and the micro drew 3mA. The brief LED flash added 5mA* to that for a mS every 15 seconds. 3mA × 2 hours is about 6mA-H. What gives?

The other important parameter for batteries is the rated current draw. CR2032s have a recommended continuous current rating of less than 1mA. Oops. Back to the drawing board!

There aren't many battery types rated for 125°C. Changing the type of battery wasn't a practical option. So the option was to reduce current drain.

With the PIC18F parts, the watchdog timer(WDT) can run from an independent clock, and wake the micro periodically. Doing this I can cut the steady power drain way down, and have a short pulse of a 3mA current draw when data is being acquired. It wasn't too difficult to make changes to my code to do this.

This time, I tested with a standard CR2032 battery at room temperature to see if a test at temperature was even worthwhile. Instead of working for less than 15 minutes like my first test, the battery lasted for 4 hours and 44 minutes, taking a measurement every 16 seconds!

*I had brainfade on the LED resistors, thinking 5mA would be ok based on the mA-H rating of the battery.

The wonder of microcontrollers - being able to fix your mistakes without reaching for the soldering iron (or throwing those brand-new boards straight on the bin!).
Interesting project, and your polite, thorough and concise posting is an example to us all.

This has been an interesting project and a learning experience. There are some obvious questions when you want to measure temperature inside a pressure cooker:

> What's the maximum temperature expected?

> Will micro, temperature sensor and EEPROM work at the maximum temperature?

> What power source is available?

> What limitations are there with other components?

Industrial-grade chips will survive the 115°C temperature expected. Fortunately, that temperature is limited by the physics of a pressure cooker – tbere's no way to exceed that temperature except by boiling it dry (and I think Instant Pots have a safety cutout if that happens), so chips rated for 125°C aren't a worry.

The battery was an interesting bit of research. Coin cells rated for 125°C are available and aren't too expensive. I designed for a CR2450, for plenty of capacity. My research showed these cells to be available. I didn't look closely at the options.

Turns out these high temperature coin cells are designed for in-tire tire pressure sensors, and are expected to last "the lifetime" of the sensor. Batteries of this size are virtually unavailable without tabs for soldering to a circuit board. I figured this out after the boards were ordered of course. A 2032 surface mount holder could be soldered to the through-hole pads for the 2450 holder, and the specs for the 2032 seemed up to the task. Incidentally, the 2450 has the same recommendation current draw limitation.

Will the LEDs survive? The Instant Pot is cooling down even now, so I'll soon know.

I

Visitor said:

Will the LEDs survive? The Instant Pot is cooling down even now, so I'll soon know.

Click to expand...

my guess is, yes if it was an old school 5mm bullet or any 2 to 2.5Vf SMD.

The test results are in. The test was a success!

1156 measurements were made at 16 second intervals – a period of over 5 hours. The data logger was still functioning (i.e., battery not dead) when I removed it from the jar after it had cooled down.

The LEDs do not appear to be diminished at all.

The first graph shows the entire sampling period, starting with hot water in the sealed jar. Section A is the 85 minute period the pot was at pressure and temperature. Section B shows the jar cooling inside the sealed Instant Pot. Section C is the cooling period with the jar removed from the Instant Pot.

This second plot shows an expanded plot of the temperature while at pressure. Note that the span of the measurements of the vertical scale is only about 1.1°C. It looks pretty horrendous but it's actually fairly tight temperature control.

Temperature was measured with a Microchip MCP9808, which has a temperature accuracy of +/– 1°C between 100°C – 125°C.

The temperature monitor has been proven a success.

The results the first test provided, based on one off-brand Instant Pot, aren't what we had hoped for. The maximum temperature of ~ 105°C corresponds to about 1.2 atmosphere.

We had expected about 1.5 atmosphere to reach a temperature of 111°C, which is the minimum safe temperature to can fish.

But this is just the start of the project to document whether any Instant Pot is up to the challenge.

Some good background information can be found at this link.