Saw tooth output from LM34CAZ temp sensor

Dave E

Here is one I ran into today.

I have 2 LM34s feeding signals into the two inputs of a LTC1298 ADC. A BS2 reads the ADC and outputs the temps to an LCD.
I noticed every once and a while that the temp reading on one of the inputs would start reading from 122 degF to 50 degF (readings updated every 3 seconds). I reseated the ADC, no luck, replaced the LM34, no luck. Finally I put a scope on the input and found a saw tooth wave for the output instead of a steady (varies 0.01 Vdc per degF) dc voltage.

The LM34 is soldered to one end of a 3-wire, shielded cable that terminates on a·Super Carrier Board. I also noticed that when I power the board up all is well until I turn on a power strip. This triggers the saw tooth wave even if nothing is plugged into it. I also get the saw tooth with only my scope lead attached to the output.

I took another look at the data sheet and it looks like I may need to add a 2K resister in line with the output or even add a RC circuit at the sensor end. The data sheet states this is for decoupling when feeding the signal into a high capacitance / high impedence input. I will assume for the moment that the LTC has high impedence inputs and the sheilded cable may be acting as capacitance.

By the way, the saw tooth is a fairly steady 37kHz, 0.6 V p-p.

I will try the decoupling circuits asap but was wondering if anyone has ever seen this before. I have used a number of LM34s without a hitch.

Any input will be appreciated.

Dave E

Dave E

Looks like a resistor soldered in line with the output at the LM34 may do the trick. The data sheet recommends 2k. I tried 1k and it didn't work, a 4.7k does. Didn't have a 2k. I still get a small wave pulse but it dampens out very quickly.

Just an FYI.



Dave E

Tracy Allen

I too found that the LM34 is sensitive to capacitive loading that comes from the length of cable. It is often more stable on either no cable at all or on a long cable, and a a short cable is worst. With the resistor alone I found that 3k gave more certain results than 2k. It can be done with a smaller resistor if you also add a capacitor, something like this:

390 ohm
LM34 output -----/\/\-----o--------cable to ADC
                                     |
                                ===== 0.68 uF
                                     |
                           ----common----

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Dave E

Thanks for the confirmation, Tracy.

I bought some 2.2k today but didn't get a chance to try them.
The odd thing is, I have 2 sensors wired to the same type of cable, same length (+- an inch or so), same·ADC·and only one of them is giving a problem. The wave is still there but it dampens out within a few cycles.

Thanks again Tracy, I figured that you probably had some experience with these.

Dave E