Now I’ve finally got all the parts and the new circuit boards as well. The remaining 4 boards and 5 full DIY kits are available for fellow electronics addicts. You’ll find all the stuff on the projects page.
Packaging was pretty unpleasant. Too many tiny SMD parts to sort through. I think I will stick to PCB-only offers in the future – maybe PCB + awfully hard to get parts.
These are the issues I’ve discovered so far:
* One header conflicts with the 9V battery. Easy fix: make sure it is 100% FLUSH on the bottom and add a bit of sticky tape as insulation.
* C5 conflicts with cable tie used to pin down the battery. Don’t populate it, it works nevertheless.
* The 5V1 Zener diodes aren’t a perfect fit. Need to be replaced with 5V6 ones.
* The 5V regulator was unstable _again_. Fix: 1k bodge resistor to draw more standby current. You can also lower the LED’s resistor a bit, say to 300Ohms.
* The heatsink is about 0.5mm off it’s optimal position (too far away from TO-220 packages). BUT I had taken that possibility into account and made the mounting hole a bit wider, so it can be shifted into position. Not a real issue.
The layout issues were fixed in the V1.01 files, but as the current version works as desired I declare this project as finished – no new boards will be made.
And as a little sweet at the end let me give you a few very obvious tips when designing stuff:
* In case of doubt, read the datasheets!
* Avoid using a voltage regulator as a precision voltage reference
* If you do use a V-Reg. as a V-Ref. make sure it is STABLE – and TEST if it is STABLE
* If it is not stable and you just can’t find out why –> read the datasheet _again_
* Did I mention you should read the datasheet?
* Don’t assume (*) anything
And finally the last one: READ THE DATASHEETS!
Of course I should have labelled this little intermission as “Note to self” ;-)
(*) Just recently I saw a video on YouTube. It was an excerpt of “The Odd Couple” with Jack Klugman as the sports columnist ‘Oscar Madison’ and Tony Randall as ‘Felix Unger’. I don’t know any of the background, but they were in court and Felix said to a woman “Don’t assume! Because if you assume, you make an ASS out of U and ME!”
The 5V6 Zener diodes clamp at about 5.7V – I know that is far from perfect. The 5V1 ones started to clamp at 4.5V, which is no good as well. They would have required a higher current going through the voltage divider, no good either. So best make sure the attached 5V micro-controller can handle that voltage on the input pins. As the output resistance for the V_DUT pin is 30k, you should expect a current of ≤ 1mA that must be handled by input protection diodes (DUT voltage of 25V, which is already 5V above the maximum).
In case of 25V reverse voltage applied to the DUT input, the V_DUT pin will show about -0.7V with no load attached. Short circuit current is ≤ 1mA as well.
A crude LTSpice simulation indicates it is likely to work as expected, but of course I will have to do a real-life test.
An attached Arduino powered by USB (Vcc was about 5.15V) survived a positive DUT voltage of 30V. The internal protection diodes seem to be pretty high impedance – at least compared to the 1N5817 used in the LTSpice simulation, as those clamped much earlier. The voltage on the analog pin was finally clamped at 5.66V, which was 0.51V above Vcc and below the absolute maximum of 6V.
I’ll do the reverse voltage test later. According to the ATmega168 datasheet the absolute maximum value for a negative input voltage is -0.5V. That may be a bit harder to accomplish.
Zener-diode / microcontroller input protection conclusion (new data):
a) overvoltage, correct polarization: DUT voltage: 30V, V_DUT scaled and clamped to 5.68V, 80µA pin current going through the on-chip protection diodes, Vcc: 5.15V, heat dissipated in diode: 42.4µW
b) overvoltage, wrong polarization: DUT voltage: -30V, scaled and clamped to -0.5V, 200µA diode current, Vcc: 5.15V, dissipated heat in diode: 100µW
All of that is still on the safe side. Good.