Those who throw out gratuitous insults are unlikely to be overwhelmed with helpful responses. --
"Andrey Semyonovitch really was rather stupid; he attached himself to the progressive cause and 'our younger generation' from enthusiasm. He was one of the numerous and varied legion of dullards, of half-animate abortions, conceited, half-educated coxcombs who attach themselves to the idea most in fashion, only to vulgarise it and who caricature every cause they serve, however sincerely."
Cursitor Doom would know about the gratuitous insults. His capacity to to distinguish between gratuitous and well-deserved insults is less well-developed and his capacity to distinguish between a helpful and a less than helpful responses is pretty much nonexistent.
He's pretty good at working out what is the right-wing nonsense of the month and posting his approval of that is predictable - but right-wing nonsense isn't remotely helpful.
After Fyodor Dostoevsky, as quoted by Curistor Doom, who doesn't seem to realise the he is precisely the type Dostoevsky was commenting on.
I needed a simple controller for charging a gel-cell from a low-power PV about 30 years ago and came up with this.
formatting link
Its main virtue is that it has a very low quiescent consumption of <100uA. So it wastes very little of the PV array output power. It can pass up to a few amps depending upon the FET and heat sinking.
It also uses an N-channel MOSFET by exploiting the fact that the solar array is floating so the control can be on the negative side.
I'm using the BE junction of a transistor as a Zener here feeding a diff pair with a current mirror in the collector.
The output voltage should be adjusted to 13.8V with the pot for a typical 12V lead-acid battery.
Transistors are generic silicon - 2n3904/3906. The FET can be pretty much anything that can handle the current.
The FET does need a heat-sink although the power dissipation is actually not very high because the load-line of the array and battery mean that the current is getting lower by the time the battery reaches the 13.8V float voltage.
I did have a non-reverse diode in the feed from the PV array that isn't shown in the schematic.
I subsequently used it to charge from a 60W PV array and for the last few years, it has been fed from a wall wart.
Huh... Well it's most of an opamp... But thanks! Does the voltage reference account for differences in temperature? I guess that's my main (silly) problem. How much current do I need to run an lm317? (have it turn off a pfet above some voltage.) That looks like a cheap (robust) voltage ref. ? (can I add hysteresis to the lm317 feedback?) George H.
What a lot of rot. It's easy to see you don't do much reading, Bill. And your forgery's not up to much, either. --
"Andrey Semyonovitch really was rather stupid; he attached himself to the progressive cause and 'our younger generation' from enthusiasm. He was one of the numerous and varied legion of dullards, of half-animate abortions, conceited, half-educated coxcombs who attach themselves to the idea most in fashion, only to vulgarise it and who caricature every cause they serve, however sincerely."
Cursitor Doom doesn't like getting a taste of his own medicine.
It's easy for Cursitor Doom to think that. It's a ridiculous assertion, but he's so ill-read himself that he hasn't picked up the evidence that I do read widely. I wonder if he knows who Thomas Love Peacock was or how I caught John Woodgate cheating with a pseudo-Celtic song that John had poached from Peacocks 1824 satire "The Misfortunes of Elphin" which I'd happened to have read (and liked).
It's not a forgery. It's a parody - explicitly playing with the text to illustrate how well it fits your obvious deficiencies. If I'd tried pass it off as anything else it might have been plagiarism, but for it to be a forgery I'd have had to try to pass if off as something that it wasn't.
If Cursitor Doom was a well-read as he seems to imagine he is, he'd have used a more appropriate word, but he's not famous for getting things right.
That's actually a pretty good design except for the voltage reference. The gate-source capacitance of the NFET integrates the error current resulting in infinite gain at DC, meaning zero error, which is what you want. But the avalanched NPN is problematic. You say 100uA quiescent, but 100uA through the 470k would be 47V, so something is amiss with that number. Then the avalanched NPN probably has a positive tempco which is in effect doubled by your voltage sample being divided in half. Lead acid cells have a negative temperature coefficient, being about -3mV/oC. The basic circuit idea is good, but you have a few loose ends that need fixing.
Don't forget the current through the zener (ok avalanched BE junction) (26uA) and through the divider chain feeding back the battery voltage to the right hand PNP(60uA). There will only be about 12uA tail current through the 470k. The total adds up to about 100uA.
I agree - it could do with some temperature compensation to be more generally useful - I took advantage of the location it was being used where the temperature of the battery wouldn't change much.
I used the reverse biased BE junction because it worked well at only 10's of microamps - the zeners I had at the time didn't work well at that current.
Initially I was using it with a solar array that only produced a small current so I wanted low parasitic current.
It isn't. The collector base diode of the avalanching transistor at the top left is being wasted. The avalanche voltage typically has a temperature dependence of around +2mV/C and putting the collector base diode in series with that would add in another -2mV/C. Apparently with some gold-doped transistors you can find a current here the cancellation is exact (at least at close to room temperature). There was a short paper on doing exactly that in the UK Journal of Scientific Instruments some fifty-odd years ago.
Too true. Of course Fred got the currents wrong, as the OP has already pointed out.
What Fred is saying is the tempco of the avalanche voltage is around +4mV/C when it gets to the battery, when it should be -3mV/C. Exploiting the collector base diode would get it down close to 0mV/C. Adding a second diode could move it to about -4mV/C. A bit of tinkering with a resistive divider could get it closer to -3mV/C.
That's a bit harsh - it met my design objectives and I could build it with available components.
I thought I had read something like that but when I posted the circuit I searched all over the place and couldn't find that info, thanks. ...
Or keep the reference at zero tempo by using the CB junction (or a real zener) and put the compensating diode in the upper part of the feedback divider.
By tapping it down the divider the ~2mV/deg of a single diode could be adjusted to be -3mV/deg at the battery.
John Larkin might have described it as insanely good, if he had put it together.
It's functional, simple and would work. Adequate is as far as I'd go. Fred Bloggs didn't think that the performance would as good as it should have been, and I could see how to improve the performance by changing one connection. That does suggest that you should have done better. "Good" - in this context - was a bit of premature self-congratulation.
The reverse breakdown at around 5V is between Zener breakdown and avalanche breakdown, and has a roughly zero voltage dependence.
formatting link
Zener breakdown is softer than avalanche breakdown, and you get a better reference diode if the junction avalanches at about 6.2V where the temperature dependence is +2mV/C and you back it off with forward diode with a -2mV/C temperature dependence. The temperature dependences are current sensitive, so it is tweakable.
Forget it, it's a wasted effort. The VEBO of the 2n3904 and 2n2222 are 6V, so you probably won't see avalanche until the 12V or more , making them unusable. Then according to at least one source, avalanche voltages in that range come in at +8-10 mv/oC tempco. The battery voltage tempco I cited was "per cell" making the 6-cell stack -18mV/oC. That kind of number would suggest a VBE multiplier might be the better choice of reference.
formatting link
The circuit needs to go with an LM385-ADJ or similar with ultra low bias quiescent, or VBE multiplier.
formatting link
Neither here nor there, error arose from assuming transistor biasing was 10% or so of operating current. That turned out to be wrong. So using 1/3 Iq thru the 470k still gets you BATTERY and half voltage, making adjustment necessary.
where a 2N2369 is used in exactly that way, which means that it has to avalanche at 6.2V to give the +2mV/ temp, which can be exactly compensated by the 2mV/c of the forward biassed collector base diode. The data sheet Vebo is 4.5V for that part
Then according to at least one source, avalanche voltages in that range come in at +8-10 mv/oC tempco.
If you've got the right avalanche voltage. Data sheets only give the minimum base-emitter breakdown voltage, typically at 10uA.
Actually reverse biassing the base emitter junction into breakdown degrades the transistor, so what the data sheet is saying us that reverse voltage will generate less than 10uA of reverse current. Inferring the actual breakdown voltage at 1.5mA from the Vebo is a bit tricky
It doesn't. Multiply -2mV on 0.6V up to 12V and you get -40mV/C not -18mV/C.
The LM385 is a nominally 1.25V reference. Stick a 0.6V diode on top of it and you'd have 1.85V which you would have to multiply up by 6.5 to get to 12V. giving you -13mV/C not -18mV.
The 0.6V is pretty loosely defined, unless you buy a close tolerance diode. Philips certainly used to sell one.
The LM35 might be a better place to start - that gives 10+/0.1mV/C.
You'd have to subtract it from your reference voltage,and multiply your reference by 1.8 to get the right drift, so you'd have to start with about 6.7V, which you can get out of an LM385 and two resistors.
It's also pretty close to what you'd get out a of reversed biassed 2N2369 plus the forward biassed collector base diode
And perfectly useless if you want to compensate -18mV/C drift on the battery as a whole
It's at Figure 37. They recommend a process 21 transistor which is a gold-doped NPN device - 2N2369 and 2N2369A - run at about 1.5mA.
That's pretty much what I remember from the J.Sci.Instrum. paper. No American ever seems to have read the UK Journal of Scientific Instruments (now called Measurement Science and Technology).
Having read much at Battery University (and using it as a reference.) It's not that the battery voltage changes that much with temperature. But that the recommended charging voltage should decrease with increasing temp. (I'm going to guess this is due to some battery chemistry, but IDK). The recommended float voltage (which is less than the charging voltage) has less slope than the charging voltage... (and looks mostly flat around room temperature) So ~ zero TC is not that bad.
I just measured a bunch of similar transistors and they are pretty consistent at about 7.5V +/- 0.5v @ 100uA (there are a couple of outliers that are lower)
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.