Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Big thermocouple ADC problem

Status
Not open for further replies.

Oznog

Active Member
I have been trying to interface 8 thermocouples, which generates low imedance signal around 20 mV, to a PIC's ADC stage, with high accuracy. There are two thermocouple types involved, and each requires a different gain. The signal is far to small to measure in itself, it's also so small that the offset of most opamps create too much error. So I settled on an LTC1053 dual chopper opamp, which has an extremely low offset. They signals are switched through a CD4051 analog mux.

The problem arose when I found out that one of the thermocouple styles used doesn't even isolate the junction from ground! This is a problem! The negative may or may not be above ground. At worst, it could be 20mV below ground.

I need a minimum of components here. This is difficult. A proper differential stage requires a lot more amps, for one. I could do a single opamp differential amp, but I'm not sure if the low input impedance will be a problem (the analog switch increases the signal's impedance into the amps), and I'm a bit concerned about the resistor matching. And I need a negative supply generator (MAX1044). I would need a pair of CD4052. The PIC pins require protection (probably just a resistor) for the case where the amp could put out -5V to the ADC input, likely due to a disconnected lead. The small box I had just got a lot larger.

The only good solution I could come up with was to use a LTC2418 ADC, which solves all these problems- with a 24 bit output, resolution is 300nV without gain, has 8 differential inputs, and has a fairly unique capability of digitizing signals up to 300mV below ground with no negative voltage supply.

It seems like a miracle, but problem is, this is a 28 pin SSOP pkg (not just SMT, but the closest SMT pins around). I want to make a bunch of these, but this is a terrible problem to solder. I'm having a protoype board made to see how hard it's going to be anyways.

Any advice? Ideas? Does anyone know of another ADC capable of doing the job within a DIP pkg, or at least a SOP?
 
Because the voltage offset alone is 0-5mV on even the best TL082. That leaves you with 25% accuracy at best on a 20mV signal, and I need more like 1-2%. It is useless for this app.

The input impedance problem on a single opamp in differential mode doesn't depend on the input of the opamp itself. Problem is the same.
 
Soldering the SSOP package is not difficult, using solder paste and vapor phase solder system. The surface tension of the solder will align the IC on the pads as long as the leads touch the pad. If you run a thin line of solder paste across all the pads, the solder will pull to the pads when it melts, leaving very few solder balls, if any. I don't like the infra-red solder system because it heats the black IC too much.
 
How would I go about doing the vapor phase step at home?
 
The vapor phase system is too expensive for home use but you could take the assembled PC board to someone for soldering. If you dry the solder paste, the populated board can be transported. A home soldering method would be to heat the board on a temperature controlled hot plate. I tried using a spatula on my electric stove but could not control the temperature well enough (burned the first board). I hand solder SOIC devices, but don't use SSOP.
 
If I were to use the hot plate method, would smeared solder paste "ball" itself onto the pads and away from the gaps between the pads, or is it essential to use a template to keep the paste application confined to the pads?

BTW, it's not even a .0254 SSOP, it's .0250 pin spacing. Glad I figured that out before I sent the stock outline to the PCB mfg.
 
It is not necessary to confine the solder paste to the pads, but too much will cause shorts and solder balls. A thin line of solder paste will work. If you can see that it is soldered without using a magnifier, you have too much solder.
 
Is there a particular paste you would recommend?
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top