On 24/12/2016 03:48, "Brent Hilpert" <hilp...@cs.ubc.ca> wrote:
>> On a whim I managed to solder new legs onto the old LM385Z and it works, >> giving 1.2V at IN-, but the output is still only 0.2V. >> >> I don't mind admitting I'm stumped :) > > > The op amp is configured as a schmitt trigger or comparator with hysteresis: > There is no negative feedback so it is operating at full gain and functions > like a comparator. > However there is positive feedback via R412 (*1), this adds hysteresis to the > trip point(s). > (Brief hackneyed, not rigorous, theory of op: As the input differential varies > past the trip point, the output pulls the + input > further above or below the point at which it just tripped, so the inputs now > have to 'overcome' a greater differential > to trip as the input differential varies in the opposite direction.) OK, that explanation helps a bit. My brain still has trouble 'slowing down' how these components work, like when I first dropped from 3GL programming to assembly. Fortunately there are a lot of tutorials out there. > One input to the (now) comparator is the 1.2V from V407 regulated down from > some power rail. > The other input is the ~ 1/4 voltage divider down from +5V formed by R398 & > R406 (netting 1.34V @ 5V). > > It appears the idea is that as the +5 supply ramps up at power-on the > comparator trip (and hence release of reset) is delayed till the +5 reaches > something around +4.5V. That's what I figured yesterday which is why I was surprised that reinstating the LM385Z didn't make it all spring into life. > The ICL7611 is, I expect, a very-low-power (Intersil's niche) op amp. Together > with the CMOS 4081 the circuit appears tailored for low-power operation. > Is it supplied by the battery? > It may require the battery presence for stable supply at time of power-up to > get reliable reset operation from this power-on-reset circuit. That's a very good point so now I'm going to have to scrabble around to find the battery to find out what voltage it was since I removed it ages ago and may have recycled it. I don't recall any markings on it though, it looked like a 'normal' 3 terminal NiCAD wrapped in blue plastic. I did take hi-res pictures of it however. The 5V input at R395 does head off towards the battery location before it hits the resistor so I'll trace that out too. > The diode-resistor pairs at the 4081 AND gate inputs turn the AND gates into > 'asymmetric edge delay gates': the resistor together with some capacitance > delays the switching of the gate for an edge of one direction, while the diode > shorts the delay for an edge of the other direction. Gotcha, I wondered why the diode was there. > If there is no cap at the input, they must be relying on the gate capacitance > of the CMOS inputs, making for a pretty short delay. > > *1: Are you sure that's a 220 resistor? It's awfully small compared to the > impedance of the voltage divider feeding it. My bad, it's a 1.4Mohm and tests ok. Thanks! -- Adrian/Witchy Binary Dinosaurs creator/curator Www.binarydinosaurs.co.uk - the UK's biggest private home computer collection?