Fred wrote:Ahh, what? I don't even understand what it is that you're *trying* to say. How many of these parts have you handled and measured? I've never seen one over 5.1V, and I've measured plenty. Nor under 4.9, for that matter. Those values (both hot and cold) represent the absolute corners of a bell curve, whereby the vast majority of parts come no where near those values. As with any design decision, you play into statistics with it. If you're happy for 5% to require rework to work properly, then you can pick a voltage based on that and a known statistical distribution.
How may parts did you measure at 0C or at 60C?
The specifications are the speciications, if you don't like them, I guess you can ignore them if you want.
Thanks,
Huff
we could change the part as i understand your point, but Fred is also correct in saying that all the ones we have experienced have been well under 5.1V.
The four I have tested have been very very close to 5.02V (one of them was at approx 40C too, which is my usual summer ambient temp).
Fred wrote: As with any design decision, you play into statistics with it. If you're happy for 5% to require rework to work properly, then you can pick a voltage based on that and a known statistical distribution.
Nope, not for automotive, we are required to prove with a WCA (Worst Case Analysis) that our design will meet the the requirements, ALL the time, under ALL specified conditions including extreme temperatures.
Thanks,
Huff
I'm not ignoring the specifications. I'm paying close attention to them, and specifying other related parts in a way that provides acceptable performance and low risk of manufacturing defects due to tolerance. The nominal value is exactly 5 Volts. The tolerance @ 25 C is maximum 5.149V. If we set the clamp at 5.2V then it would almost certainly never be hit by the reg. If I had a board that was sitting at more than 5.1V I would replace the regulator as that's not good enough for me. Reality differs. Once again, bell curve, tail end of, mean, median, upper, lower quartiles, etc. Reality is much better than the worst case, of course.
Put away the bottle, or go for a walk to cool off, you're clearly in a shitty mood for some reason or other.
Fred.
PS: Spend more time on each post, three in a row replying to one of mine isn't good use of the forum.
Max rating for the CPU is 5.5 under operating conditions, however that would skew battery voltage readings by about 1.44 Volts, which is pretty severe. Not to mention the differential that would form between the CPU supply and the analogue supply. If you recall, the reason this thread started was because the zener allowed a rail voltage of 5.3, which in turn meant the pin could be pulled to 5.6, which was out of range. The ABSOLUTE MAXIMUM that we can tolerate is 5.2 on the rail. 5.1 would be preferable, and will result in mostly working units, however it might be best to set the design at 5.2 or 5.15 and build to taste.
In a mass production OEM environment there is absolutely no room for reworking some resistors and the device goes into the car and is expected to last a million miles without even being so much as looked at. THAT IS NOT OUR USE CASE. These will be installed in an enthusiast vehicle and tinkered with and monitored constantly. At the least they'll be checked over before being put into service. Not to mention the much lower volume they'll be made in and the opportunity for tweaking that this user base affords. We're not OEMs and not constrained by the same forces that they are. We're constrained by one thing: common sense.
I politely disagree with this decision because I feel that the odds of an enthusiast (or noob) wiring even one input accidentally to battery is much lower than than the potential of this circuit to cause issues.
I understand that it is not my call, I just wanted to express my concern.
Though I also am not in favor of the two regulator topology (one always on and one switched), if you do this, I suggest that you first "tune" the switched supply voltage to a value below the constant one before "tuning" the over voltage shunt. This is to ensure that the VRH voltage is < = VDDA for A/D conversion accuracy.
As long as you've calmed down you can disagree all you like :-) And yes, that topology needs some thought, you previously raised some good points about it. Can't comment more now, got other things to do.
