RG Amplified Zener - heat

[lowell]

Ok yeah please do tell. FYI my chassis has top mounted tubes so heat inside should be negligible. It wouldn't hurt to measure it as you say.

[J M Fahey]

OK, I'll dust my camera and post tomorrow night.

[lowell]

Just found this old thread. JM, you mind sharing the diy temp sensor? The IRF644 amp'd zener for the B+ is arcing in my amp. Amp has been working just fine for about 7 years so I'm not too concerned. I thought maybe this MOSFET needs a flyback or reverse pulse protection diode but looking at the data sheet it seems it has one built into it. Think I'll just replace it and hope it sticks.

[J M Fahey]

Oh, it's so simple that they'll throw me out of the Cheapskate club just for being way too cheap

It's just a 1N400x diode epoxied to a stamp sized piece of aluminum, with a 15k resistor inseries, feeding it from a +15V supply, I measure voltage drop across the diode:



In a nutshell:
silicon diodes drop about 700mV at 25 deg C and that voltages goes down by 2 mV by each 1 deg C change up, very linearly and with great precision.

There are circuits out there with added Op Amps and stuff plus own internal meters , nice cases, etc.which directly display temperature.
They might even have a PIC or some other microprocessor inside.

There are also multimeter add on Temp probes.

Since Murphy's Law RULES the World, I often happen to need such a meter on a Saturday night or in the middle of a blizzard, eartquake or Tsunami, but I can quickly improvise and calibrate this one ..... and trust it more than any cheap probe.

The 15k resistor passes about 1mA through the diode, I measure voltage across it.

Calibration involves putting the diode for a minute or two in a glass with water and ice, stirring the mix with the diode; whatever I measure (say, 740mV) means 0 degree C (sorry Fahrenheit fans, you'd need Ammonium Chloride for this, or moving to Danzig on a cold Winter, guess why the World turned Metric).

Then you put it for a minute inside a boiling water kettle, and whatever you measure, say, 540mV means 100 deg C .

You draw two perpendicular XY scales on the graph paper, say the vertical one goes from 540mV to 740mV (or whatever you measured) , the horizontal one from 0 deg C to 100 deg C.

03:41 AM here now, tomorrow I'll get a camera and post my own graph, which I taped to the wall.

So now you:

* apply the measuring probe to the heatsink or speaker magnet or transformer iron or whatever you want to measure (when designing a new transformer I may embed a diode inside , close to the core, and solder two test wires to it, , it's so cheap as to be disposable),

*measure voltage after a minute (there's some thermal inertia) , some grease improves contact, I do not push it with the finger but with a chopstick or plastic pen and then

* read voltage (say, 590 mV) and read temperature on the graph drawn on the paper (in this case about 75C , already unbearable for Human hands).

Of course, use your own handmade scale.

Although calibrating and drawing the graph may take 30 minutes (or less) , you do it *once* , it avoids Math, you just read values on a scale.

A linear scale to boot, so if you are , say, halfway between 50C and 60C you can trust it's 55C and not some wild value.

Easier done than said.

FWIW poor old Fahrenheit used human body temperature as 100 degrees definition, I don't want to imagine where he'd have to stick that diode to properly calibrate it .

[Dave H]

Oh, it's so simple that they'll throw me out of the Cheapskate club just for being way too cheap

It's just a 1N400x diode epoxied to a stamp sized piece of aluminum, with a 15k resistor inseries, feeding it from a +15V supply, I measure voltage drop across the diode:

To measure chassis temperature I just stick a mercury in glass thermometer through one of the vent holes. It's glass so it doesn't matter if it touches a live part. If there isn't a vent hole it fits nicely in a jack socket.

I'd add one component to your cheapskate circuit. Connect a trimpot across the 15V supply and measure between the pot wiper and diode. Put the diode in melting ice and twiddle the pot for a reading of 0V then it will read -2mV/degC with no offset.

[R.G.]

If you have some time and a little money, the LM34 or LM35 provided a pre-calibrated, linear output of voltage with temps. An improved variant similar to this is the
TMP-36, about $1.50

I'm about to instrument my AC units with DS18B20 temperature sensors. They're US$2.00 each, precalibrated and etc., and digitally send out the converted temperature as a digital work when addressed properly. These do need a programmed digital device to work, like a $1.50 uC, but I can run up to two-to-the-64th-power of them on the same wires, all ganged up, and I only need to run a buss of three wires to them, so I can run all 12 or so that I need on one length of cat-3 plain-old-telephone-service wire.

One way to get a bigger output voltage from a diode-forward-voltage temperature sensor is to use a transistor to self-amplify its base-emitter voltage. The temperature compensation in solid state amps is usually a single NPN transistor set up to amplify its own Vbe. A string of two resistors runs from its collector to its base, then to its emitter. If the top resistor is R1, the base-emitter is R2, the voltage from collector to emitter is (R1+R2)/R2 times the Vbe. This works pretty well for a high gain transistor up to 10X or so the Vbe, so you expand the scale of a diode sensor by 10X. The resistor tolerance matters, so, just like the diode-only version, you need to calibrate each instance of the sensor.

[lowell]

Thanks! So how do I use the thing. Lets say I'm using the TMP-36. I'm thinking I could use some 4-lead shielded cable I got layin around and mount it to a chopstick. Build a little box with a 9v power supply, 5v zener etc.. to plug into and some banana plugs for my voltmeter.

Better yet theres probably a voltage display I could buy for the box so I could still have my meter available.

Do i need bird poop on it to get a reading?

[R.G.]

Thanks! So how do I use the thing. Lets say I'm using the TMP-36. I'm thinking I could use some 4-lead shielded cable I got layin around and mount it to a chopstick. Build a little box with a 9v power supply, 5v zener etc.. to plug into and some banana plugs for my voltmeter.

That would do it if you want to build a general-purpose temp measuring device.

Better yet theres probably a voltage display I could buy for the box so I could still have my meter available.

Good idea if you want to build a measuring device and still have your meter.

Do i need bird poop on it to get a reading?

That will help. The goop is to make the thermal resistance from the heat source to the measurement junction be lower. Just like with electricity, thermal resistance slows down the temperature rise of a thermal capacitance - the mass of the measurement device. The measurer comes up to to final temp faster with a low-resistance path - the goop.

One alternate that I have to mention. I've measured temps with thermocouples, thermistors, and now temp-reading ICs for years. But easiest and fastest by far is one of the IR reading "guns". I got one from Harbor Freight for $30. Point and press the trigger, the temperature appears on the readout. It's probably a degree or two off, but most of the time you want temperature differences from the surrounding air anyway. So point at the wall, and measure - that's your ambient temperature. Point at the hot thing and measure - the difference is the temperature rise.

The ICs are useful when you're going to leave them in place - like I'm going to do with my AC units.

[lowell]

Very cool thanks for the info. I may just buy the gun. Add up the enclosure and parts and I'm half way there.

So regarding the devices you mentioned. Do you use those in power amp protection circuitry or have you implemented them for checking temps during repairs?

[R.G.]

I have done both. I like thermal cutoffs for power amp protection, and I've used both the thermal switch buttons and electronic circuits that sense temperatures and then do something - one of which could be turning off the AC or just the DC power to the power amp. For repairs I started off with thermistors where you have to read the resistance then go do a table lookup for temperature. The old Yellow Springs calibrated 10K thermistors were the standard back in the late 60s and 70s. The LM34/LM35/TM36 make all this much easier, with a simple linear voltage per degree output.

If I were building something to replace the temp gun, I'd probably go with something like a temp-to-digital chip like the DS18B20 and a uC to display temps directly.

But the IR gun has a lot going for it. I can measure air temps in AC vents from across the room by pointing the gun and homing in on the hottest or coolest (depending on the HVAC mode) temperature I find.

[lowell]

Very good info and thanks again for the help!

[lowell]

Just got my IR Gun. So cool. And can't wait to use it in the kitchen too....or while smokin ma next brisket.