Moronically Simple LVC Circuit

Knuckles

10 kW
Joined
May 2, 2008
Messages
971
Location
Wrong Island, NY
AFAICS, this just uses the very temperature sensitive, highly inaccurate and uncontrolled voltage drop across two diodes plus one LED, plus the conduction voltage of a transistor in an optocoupler to very approximately try to detect a cell low condition.

The LED in the optocoupler can have Vf of anywhere between an unspecified low voltage (probably around 0.9V to 1V) to a maximum of 1.4V. The diodes either side of the LED in the optocoupler will have a Vf of between around 0.6V and 0.7V each and the conduction voltage of the transistor will be around 0.5 to 0.7V. All these will vary with temperature to a marked degree.

The worst case warm weather low cell voltage detect could be around 0.6 + 0.6 + 0.9 + 0.5 = 2.6V, probably OK, and the worst case cold weather low cell voltage detect could be around 0.7 + 0.7 +1.4V + 0.7 = 3.5V, no use whatsoever, as it will trigger when the cells are still well charged.

It might be a fairly simple circuit, but it isn't worth building, as there are far better, and simpler, cell LVC solutions than this, I'm afraid, ones that don't have such a massive variation in trigger threshold and that will predictably trigger a warning or brake signal no matter what the temperature or component tolerances are.

Jeremy
 
ergo "moronically" simple ...

but a LVC at 2v and room temp is better than no LVC at all.

:)
 
Knuckles said:
ergo "moronically" simple ...

but a LVC at 2v and room temp is better than no LVC at all.

:)

And an LVC of 3.5V straight out of the box (for components at the high end of the specs) is no use whatsoever. There's no way of telling if you *might* get a working circuit, apart from building and testing each stage, and swapping parts around that are at the ends of their permitted spec limits. You may need to buy four or five times the number of components in the circuit, just to find a set that are all at the low end of the spec sheet tolerance so the circuit will work.

I don't know who designed this, but one things for sure, they haven't ever had to design something that works for more than just a lucky one-off. One of the very first things you learn with analogue circuit design is to take the min and max values from every component data sheet and run the numbers, just to check that the thing will work with normal tolerances on stock components, rather than just the prototype on the bench.

Ordinary spec sheet variations will stop this circuit working, so it's effectively useless. You may get one to work by luck or good fortune, but the next one built may well cut off when one or more cells are near fully charged.

Jeremy
 
Jeremy Harris said:
The worst case warm weather low cell voltage detect could be around 0.6 + 0.6 + 0.9 + 0.5 = 2.6V, probably OK, and the worst case cold weather low cell voltage detect could be around 0.7 + 0.7 +1.4V + 0.7 = 3.5V, no use whatsoever, as it will trigger when the cells are still well charged.Jeremy
I calculate ... 0.6 + 0.6 + 0.9 + 0.1 = 2.2V and 0.7 + 0.7 +1.2V + 0.2 = 2.8V

Plus I like the "key off" dark current of only 100 nA.

And I'm also a cheap bastard! :lol: And Lazy too!
 
Well, take a look at the specs sheet variations:

The diodes aren't specified, but I'd assumed fairly common, fairly well-specced, low voltage signal diodes, like the 1N4148 or 1N914 and I'd seriously under-estimated their spec sheet max Vf value. The diodes actually have a datasheet max Vf of 1V (although I think that 0.7V would be fairly typical), so two of them at the top end of the allowable tolerance could give 2V. The LED in the optocoupler has an upper spec sheet Vf of 1.4V, so with just the two diodes and the LED we're now up to 3.4V. The conduction voltage of the transistor in the optocoupler is lower than I'd anticipated, at 0.2V max, but that still takes us up to a total of 3.6V as the trigger point for upper spec sheet tolerance components. Not a lot of use for cells that may well sit at 3.3V when reasonably well charged.

Apart from being useless for anything other than a selectively assembled one-off, that might need component changes to even work as intended, I doubt that it's cheaper than the really simple and accurate LVC that a lot of us are using, either. It certainly has a far larger component count which would add to the overall cost.

Jeremy
 
Dirt cheap switching diodes ... at 25C ... 0.1 ma - 0.5v drop, 1.0 ma - 0.6v drop, 5.0 ma -0.7v drop

When the key switch is "on" the I(f) is between 10 and 15 ma for EVERY primary emitting diode (the C-E is saturated)

But as a cell V drops the primary C-E voltage decreases to 0.1v and the current to the secondary (feedback) opto drops

This causes the Vce for the low cell 2nd feedback opto to rise dramatically thus reducing the current in the feedback circuit

and thus triggering the +12v e brake circuit.

temp effects are actually not that bad.

:?: WTF. You have the voltages wrong.
by sending 10+ma to the primary emitters, the primary C-E voltage is high and drops down to 0.1 volts as the cell goes low.
As the cell goes low the 2nd emitter current drops dramitally and causes a HUGE increase in the 2nd CE voltage.
This inrease in 2nd CE voltage trips the +12v ebrake

yes no maybe ?
 
texaspyro said:
I've seen optos start conducting at 0.4V in... and that was at room temp.

Yep, the whole circuit's crazy, as designed, as the cut-off voltage could (and probably would) vary to the point where it just couldn't be relied upon to do the job intended without a bit of tweaking. If some of the optos start conducting at 0.4V Vf and other don't start conducting to the spec sheet max of 1.4V Vf, then that's a 1V per cell variation just there.

Anyway, apart from anything else, when you add up all the parts costs, plus the added assembly complexity and the probable need to swap parts around to find ones that work at the right detection point, it isn't a cheap solution, either. A dead simple, ready built and tested, multi-channel LVC that similarly just pulls the ebrake line down with optos, and that has a precise, stable and reliable cell voltage detection point, that handles 8 LiFePO4 cells per board (and can be daisy chained for as many cells as you want) costs just $20 (see this thread for an example: http://endless-sphere.com/forums/viewtopic.php?f=31&t=21695&start=15 ). You could build a simple and reliable LVC with just three or four components per cell, using the same circuit as used on those ready-built boards, at a cost of around $0.80 to $0.90 per cell for the components, plus the cost of the board, connectors, etc, so maybe around $0.10 per cell more expensive than the moronic circuit, assuming that you didn't need to bin any of the components from the the moronic version in order to find ones that were around the middle to lower end of the spec range in order to make it work..

The 100nA current drain when off isn't needed, as the cells have an internal self-discharge current that's massively greater than this. The LVC only needs to be good enough to not significantly increase self discharge, so anything that draws less than maybe 100uA or so will be fine. The circuit mentioned above draws around 1uA, so with no self discharge a single 2650 A123 cell could drive it for more than 250 years.............

Jeremy
 
I would not be all that concerned about variations between the optos in a package (trust but verify). And you should be able to get reasonable matches from a lot purchased at the same time. And you can select combos of silicon/schottky diodes to get the threshold trimmed. Variations with temperature will be a real problem. The circuit could be made to work as a one-off, but would be a production nightmare. There are a whole lot better ways to approach the problem.

My Big Honkin' LED controller use an Atmel ATtiny13A micro. It is continuously powered by the battery and draws less than 20 microamps in standby... half that is from the voltage regulator. I am looking at doing a BMS using one per cell. Standby current draw would be under 10 microamps. It would do LVC, HVC, balancing, and temperature monitoring.
 
I am a moron!
it would be very useful to have an led that comes on if a cell is low volt, or maybe leds on and a low cell led goes off,
could this be incorporated using leds as the diodes?
told you I am a moron
 
whatever said:
I am a moron!
it would be very useful to have an led that comes on if a cell is low volt, or maybe leds on and a low cell led goes off,
could this be incorporated using leds as the diodes?
told you I am a moron

Very easy to do, but don't use this circuit as it will just give you headaches trying to get it to work reliably. Just use the fairly common LVC circuit described in several places on here, or even simpler, just buy the cheap LVC modules from Geoff or Gary, and use the output to drive an LED via a resistor instead of the ebrake line on the controller. The only thing to watch is that you power the LED from a switched supply, not directly from the battery, otherwise the small current drawn by it will add to the battery pack's problem if a cell goes low whilst it's not being used.

Jeremy
 
Back
Top