I have been more than happy with my MAX 6350 5V ones I cranked out. Vero Board and Socketed 6350 running from a 3S LiPo as the spec is a lot better at the higher voltage.
Pics to follow
MJA version is out in the short term unless anyone knows of a source.
The downside of the JVR is, that it needs a rather good resistor. The reference voltage is not much more stable than the resistors. So the JFET might be cheap, but good resistors (e.g. precision wire wound) can be expensive and it take some time to find the right value(s) to get both a low TC and the correct value.
One problem with the SMD form factor is mechanical stress from the board - so one have to be careful with SMD, especially if mounted the normal way. It might be even worth using a kind of dead bug style mountig instead.
1. Does socketing high precision reference ICs(DIP) affect their performance?
2. It seems that people prefer leaded packages vs SMD for precision reference voltages. Why is this? Or am I seeing something that isn't there?
4. I have a half-baked idea to add a uC, thermistor (and maybe a DHT22) , and LCd display to the project to measure the temperature (and humidity), and print out what the voltage "should be" from a lookup table after a full characterization. However, I'm thinking that adding a uC using a linear regulator and an LCD display using some sort of serial interface will affect the reference ICs voltage by adding noise to the PCB and this probably is just not worth it. Would you agree?
5. Would adding two TL431's in series assist with input voltage stabilization?
6. What about coating a plastic IC in a few coats of clear nail polish or silicone to help seal it?
Hmm,
I would put the heating elements like U2,R5 as far as possible from the reference.
You have introduced them to keep the self-heating (= temperature) of the reference as constant as possible.
But now U2 and R5 instead heat the reference (not constant in case of different battery voltage).
Grounding concept:
is it intentionally that you split the Gnd layer (to give input/output of the reference separate planes ?) or is it accidently.
I fear it is the later because there are several junctions.
I would use different wiring for input and output Grounds to avoid (uV-Level) differences on the output.
For C3 it is not clear: does it belong to the connectors to avoid ESD from entering the cirquit or does it belong to the reference output to avoid EMI influences from the wiring. In both cases C3 is too far away from connector/reference. I would use a additional 100nF capacitor directly at the reference pins (on the bottom side), and put C3 directly between the output connector in a 4 wire connection.
with best regards
Andreas
Do you mean what I should have done is separated the grounds on the right side of the IC (on my schematic), and connected it back directly to the battery negative?