Need to heat an IC...will my solution work?

Exo · Aug 12, 2004

I think you're confused by the datasheet. It sais 'typical values (the middle column) are at VCC 3.3V and Ta = 85°C. This doesn't mean maxim considers this the typical and perfect operating conditions. It just means they used this setup to measure the values.
They need some fixed point for comparison, so they say, let's take 85°C and vcc 3.3V...

It doesn't mean running it on other conditions will make it inaccurate...

Graystar · Aug 12, 2004

Exo said:
I think you're confused by the datasheet. It sais 'typical values (the middle column) are at VCC 3.3V and Ta = 85°C. This doesn't mean maxim considers this the typical and perfect operating conditions. It just means they used this setup to measure the values.
They need some fixed point for comparison, so they say, let's take 85°C and vcc 3.3V...

It doesn't mean running it on other conditions will make it inaccurate...

On the data sheet there are no typical values for temperature error rates. The Conditions column clearly specifies different conditions for the temperature error rates. (remember...”unless otherwise specified.”)

When Tambient is between 60C – 100C, remote accuracy drops to +/-1.0C BUT only if the remote temperature is also between 60C – 100C. In my case, the temperatures I’m recording are in the 35C – 50C range. That means accuracy drops again to +/-1.6C

But there’s more. Without some sort of heater my ambient temperature will be around 30C tops. According to the datasheet when Tambient is between 0C – 100C, accuracy remote accuracy drops to +/-3C.

At +/-3C, the range of my testing results is inside the range of my accuracy, rendering them useless. I need the 0.8C accuracy to create meaningful comparative data.

Graystar · Aug 12, 2004

I found an appnote about component temperature management, but it's beyond me. It describes using a thermoelectric cooler, NTC resistor, and their PMW controller to maintain the temperature of a component to 0.1C.

**broken link removed**

I'm stickin' to the transistor!

Nigel Goodwin · Aug 12, 2004

Graystar said:
I found an appnote about component temperature management, but it's beyond me. It describes using a thermoelectric cooler, NTC resistor, and their PMW controller to maintain the temperature of a component to 0.1C.

**broken link removed**

I'm stickin' to the transistor!

As I see it the fatal flaw in all the diagrams you've posted is that the transistor is connected to short out the supply - this is liable to destroy the transistor when it turns on - not to mention possible other damage.

This sort of temperature control has been common place for a great many years - certainly way back in the 60's, and probably a lot farther than that. The trick is to use a resistor as the heating element, switched by the transistor - the resistor limits the current to safe levels, and should be of sufficient wattage for it's dissipation.

Graystar · Aug 12, 2004

Nigel Goodwin said:
As I see it the fatal flaw in all the diagrams you've posted is that the transistor is connected to short out the supply - this is liable to destroy the transistor when it turns on - not to mention possible other damage.

This sort of temperature control has been common place for a great many years - certainly way back in the 60's, and probably a lot farther than that. The trick is to use a resistor as the heating element, switched by the transistor - the resistor limits the current to safe levels, and should be of sufficient wattage for it's dissipation.

First, a comment...
I was completely unaware of how little current is required to heat these components to high temperatures. When I read your message, I happened to have in front of me a 9V battery, 100 Ohm resistor, and a thermometer accurate to 0.1C. I connected the resistor across the battery and measured the temperature change. I also measured the current, which was 88.2 mA (as expected.) I was surprised at how quickly the temperature rose and leaped past 85C. I now realize that I need much smaller currents than I thought I did.

Are you saying that even with a collector current of, say 50 mA, that the transistor will still be destroyed, despite the 85C shut-off mechanism?

The PN2222A that Ron suggested has a dissipation of 625mW. 50mA at 9V is 450mW. Shouldn’t the device be able to handle this, despite shorting the supply?

If not, then I will have to add the resistor. I guess my understanding of what actually damages the transistor is lacking. I thought that damage was cause by excessive heat, and that the shut-off mechanism would save the transistor.

Yeah yeah...I know...I have a hellava lot to learn

Nigel Goodwin · Aug 12, 2004

Graystar said:
Nigel Goodwin said:

Are you saying that even with a collector current of, say 50 mA, that the transistor will still be destroyed, despite the 85C shut-off mechanism?

Click to expand...

50mA wouldn't destroy the transistor, but where have you conjured the 50mA from?.

For the collector current to be 50mA you need some kind of limiting, in your circuits the limiting factor is the internal impedance of the supply, which if you used NICAD batteries could easily be 20A - destroying the transistors in milliseconds (if not faster).

The PN2222A that Ron suggested has a dissipation of 625mW. 50mA at 9V is 450mW. Shouldn’t the device be able to handle this, despite shorting the supply?

Click to expand...

That's a lot of heat for a small transistor, it's good practice to avoid approaching their maximum limits.

But in any case, the voltage won't be 9V, or the current 50mA, so your calculations are incorrect.

If not, then I will have to add the resistor. I guess my understanding of what actually damages the transistor is lacking. I thought that damage was cause by excessive heat, and that the shut-off mechanism would save the transistor.
(

Click to expand...

You seriously need a current limiting resistor, if you want roughly 50mA, work out the value from the HT voltage (9V?) and the current (0.05A), which (off the top of my head) is 180 ohms. The resistor will dissipate most of the heat, the transistor very little.

Roff · Aug 12, 2004

Guys, I'm not gonna spend a lot of effort defending this kluge, but I have a couple of comments.
My first instinct was to use a resistor to heat the chip, but Willy said he wanted to use a TO-92 because it fit the MAX package. I also didn't know at the time that he planned to publish the idea. As I said, it should work for a one-off, and beyond that, ...?
I designed it so that, with 100<Hfe<300, the collector current will not exceed 170ma (well within spec), and the power dissipation would be between 500mw and 1.5w. I am fully aware that the max dissipation is 625mw, but if the transistor is tightly coupled thermally, I don't think it will be damaged. I might be wrong. It may also be the case that much less than 0.5 watts is required to maintain the "glob" at 85C, In which case I would raise the value of the base resistor.
A better solution, as Nigel pointed out, would probably be to use the transistor as a switch and glue a 2010 (200mils by 100 mils) resistor of about 160 ohms to the MAX part. The resistor fits on the MAX part as well as a TO-92 would. Willy, if you do this, change the pullup on *OVERT to 1.3k.

Roff · Aug 12, 2004

Edit - multiple post.

Nigel Goodwin · Aug 12, 2004

I think you already know Ron that it's a pretty horrible idea :lol:

You don't appear to have taken into account the action of thermal runaway - it looks rather optimistic to rely on the control loop (with it's thermal inertia) to prevent this happening.

For a true 'one off', with the base resistor and transistor individually chosen and matched, it 'may' work - but using general components with 5% tolerance resistors (and 20% to 500% tolerance transistors).

For the sake of a single extra resistor, giving good design practice, vastly enhanced reliability and repeatability, I don't think it's a good suggestion!.

But at least it's provided an interesting thread!.

Roff · Aug 12, 2004

Edit -Multiple post.

Roff · Aug 12, 2004

Sorry for the multiple posts. I kept getting some "debug mode" failure message.

Roff · Aug 12, 2004

Nigel Goodwin said:
I think you already know Ron that it's a pretty horrible idea :lol:

You don't appear to have taken into account the action of thermal runaway - it looks rather optimistic to rely on the control loop (with it's thermal inertia) to prevent this happening.

For a true 'one off', with the base resistor and transistor individually chosen and matched, it 'may' work - but using general components with 5% tolerance resistors (and 20% to 500% tolerance transistors).

For the sake of a single extra resistor, giving good design practice, vastly enhanced reliability and repeatability, I don't think it's a good suggestion!.

But at least it's provided an interesting thread!.

You're absolutely right, Nigel. It was a horrible idea. I just noticed that PN2222A's junction-to-case resistance is 83C/W, which means that the transistor chip will be running well above 150C, it's spec limit.
But it is, as you say, an interesting thread, even if it is at my expense.

Graystar · Aug 12, 2004

Nigel Goodwin said:
For the collector current to be 50mA you need some kind of limiting...

You seriously need a current limiting resistor

So now I’m confused again.

I thought that a transistor was a current controlling device. A small base-emitter current will control a larger collector-emitter current, regardless of the collector-emitter voltage (within limits.) I mean, if you short the collector-emitter across a power supply, and there’s no base current, then there’s no current flow across the collector-emitter...right?

With Ron’s latest update, the base-emitter current is controlled by a 1.3k resistor and the 3.3V power supply. That gives an Ibe of 2mA (I think.) With an hFE somewhere around 50-75 (according to the sheet,) that should give an Ice somewhere around 125mA.

Going back to what you were saying about needing limiting, I thought that the 2mA current was what provided that limiting. Are you saying that I need something else to limit current? That there’s a problem if the voltage from the battery is dropped occurs across the transistor?

I guess I’ll start looking into adding a resistor.

What is it they say about a little knowledge?

Graystar · Aug 12, 2004

Ron H said:
I just noticed that PN2222A's junction-to-case resistance is 83C/W, which means that the transistor chip will be running well above 150C, it's spec limit.

So I definitely must add the resistor. I guess one additional part isn't too bad if it greatly improves reliability and operation.

I'm glad I provided an entertaining thread...even though I feel like the court jester!

Roff · Aug 12, 2004

Graystar said:
Ron H said:

I just noticed that PN2222A's junction-to-case resistance is 83C/W, which means that the transistor chip will be running well above 150C, it's spec limit.

Click to expand...

So I definitely must add the resistor. I guess one additional part isn't too bad if it greatly improves reliability and operation.

I'm glad I provided an entertaining thread...even though I feel like the court jester!

Yeah, sorry for the bum steer. As I said, I think a 160 ohm, 2010 surface mount resistor will work. They seem to be rated for 0.5w to 1w, depending on the vendor and the product line. 160 ohms will dissipate about 0.5w when the transistor is on. It will limit the current to about 55ma. This gives you a"forced beta" (Ic/Ib) of about 25, which should be adequate to saturate the transistor. In fact, you can use almost any NPN, such as 2N3904, if you have something else available.

Graystar · Aug 12, 2004

Well now that I know what kind of current draw I'm looking at, I think I should be able to power the heating circuit with the voltage regulator...so I can drop one battery from the design.

Graystar · Aug 12, 2004

I think this is it.

**broken link removed**

Thanks Ron and others for your help. I appreciate it.

Roff · Aug 12, 2004

Looks OK, except the heater resistor needs to be a 20 ohm resistor in a 2010 surface mount package. Where did you get 62k?

Graystar · Aug 12, 2004

Ron H said:
Looks OK, except the heater resistor need to be a 160 ohm resistor in a 2010 surface mount package. Where did you get 62k?

Sorry! That was supposed to be 62 Ohms. I recalculated using 3.3V instead of 9V. But if you say it's supposed to be 20 I'll change it to that.

Any idea where I could get surface mount resistors in small quantities?

EDIT: I fixed the schematic.

Roff · Aug 12, 2004

I picked 20 ohms for 0.5 watt dissipation. The resistor will have 3.3v-Vce(sat), or about 3.1v across it.
R=V^2/P
R=3.1^2/0.5
R~20
There is nothing magic about 0.5w, it's just the target I've been shooting at. 62 ohms might work. You'll get about 160mw, which might be enough to heat your chip
.
How weird. Did you notice in your post, where you quoted me, it said 160 ohms? That was the value I had given when I thought you were using 9 volts. Maybe I edited it, but there is no indication on my post that I did. Oh well.

Need to heat an IC...will my solution work?

Active Member

New Member

New Member

Super Moderator

New Member

Super Moderator

Well-Known Member

Well-Known Member

Super Moderator

Well-Known Member

Well-Known Member

Well-Known Member

New Member

New Member

Well-Known Member

New Member

New Member

Well-Known Member

New Member

Well-Known Member

Similar threads

Privacy & Transparency

Privacy & Transparency