Searching for an Op-Amp

If the backup battery is outside your control then you should make the user provide the correct kind of charger to suit that battery. This is no single charging circuit you could provide that would suit the wide range of batteries that could be connected.

piglet

The tiny Fujitsu FTR-B3GA relays are great.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

I'd be inclined to use a BJT for that job--it'll work well down to a couple of hundred millivolts, and the beta grades are tight enough--a

3:1 variation in a small current is probably fine. If not, spending another couple of hundred mV and using 3 resistors will get the variation down to the 10% level.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
https://hobbs-eo.com

Well, yeah, you're right. I should have done, I suppose, but it isn't exac tly a constraint on the op-amp itself. It is always the case, however, tha t one must arrange for a stable environment whenever the bias voltage for a semiconductor component falls near the shutoff voltage for the component. Designing inside the knee region for a semiconductor is usually not a grea t idea unless one actually wants the circuit to be temperature sensitive.

Oh, absolutely. Indeed, if the design happens to do so, then it is icing on the cake. If it makes the design simpler or easier, then so much the b etter. The bottom line is I am not wedded to a constant current design, as long as the battery remains charged (with or without power applied) and is not damaged by the circuit.

That is pretty puny. It is expected the battery might be as large as 10AH, or perhaps larger. I am using 7.6 AH for the prototype. I would not expe ct the user to deploy anything as small as 500 mAH.

Yeah, OK. I also didn't say it would be temperature controlled, or would s it in an environment with a particular temperature. I have no idea what th e temperature will be where the user puts the device. It is not unlikely i t will be outside. In general any device should operate through a reasonab ly large range of temperatures.

Oh, wait, I misinterpreted what you were saying.

You know, that gives me another idea. I'm not certain it would work. Let me think on it.

Well, yes, and no. It should not vary greatly with temperature, which i s possible if the design is in the knee region of the device's gain vs. bia s curve. More important, though, it is desirable to allow the user to sele ct different battery types with a modest range of terminal voltages and cap acities.

How do you figure? This would be much easier if Vsupply were, say, 24V and the battery 13.6V.

The inflexibility of existing UPS designs for the Raspberry Pi is the main reason for creating this device. If both the charger and the battery have to be supplied by the user, then there is little point in creating this dev ice at all. It also happens to be very undesirable for a user to have to p rovide both a battery charger and a power supply. Would you buy such a dev ice? Given ten users with ten different requirements, will I sell more if the device only suits one of their needs, or suits all ten, provided the co st is reasonable?

Like most designs, fundamentally I am designing this device because it is h as features *I* would like, myself. *I* want to be able to select differen t batteries for different deployments. I am assuming potential users would like the same flexibility. 'A better mousetrap, as it were.

Is such an impossible ideal? I'll never know unless I try. I'm already pr etty close.

could you move everything to the low side of the battery?

a

ood

equirements. Just one, and I am hoping there is a thoroughbred out there t hat can deliver.

igh

n

see if a unicorn is hiding there. There are a lot of parts out there with attributes of which I am unaware. Coming here to ask the question is hardl y a full-scale unicorn hunt. If there isn't, well there is the old circuit which works in a more limited scope than desired. Being flexible for one' s users is a VERY desirable goal, if it can be done at a reasonable cost. So is saving space.

d.

Well the jfet type current sources are my new favorite part, so I want to use 'em everywhere. It certainly keeps part count low if that's a goal.

George H.

With the psu at 0v, the battery pushes i through the tr turning it on in inverted mode.

FWIW Selected FETs can be used that give the right ballpark current at Vgs=0, and, gasp, germanium transistors can run at around Vbe=0.2v. Not convinced those would help. Germanium would limit your T_ambient.

NT

With Phil's input, I thought of the following:

From my original list:

1. Single-ended supply between 8V and 24V
2. Up to 100ma output
3. Rail-to-rail in and out
4. Can withstand +14V on the output when VDD is removed
5. Can withstand +14V on the + input when VDD is removed
6. No more than 50 microamps leakage with VDD removed and +14V on the output and + input
7. No more than 10 microamps input bias
8. Differential gain greater than ~500

It would eliminate #2, #4, #8, and part of #3 and #6.

I'm liking John Larkin's idea, though.

Thoughts? Comments?

Hmm. Interesting idea. I'm having a little trouble wrapping my head around the details at the moment, though.

I think that works if you swithc the plus and minus signs (on the opamap) and bias the 1n4148 with ~10 k ohm from the Vin rail. Then the current is a diode drop/ 33 ohms.

GH

If you have a fet-input opamp, you could add a bypassed high-value resistor in series with its input to limit current. Then you only have to deal with the output, which could be disconnected with a small SSR or something.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

No where in this thread have I seen a review that backed up from the title, "Searching for an Op-Amp", and asked, "What are the real system requirements?"

You've picked a nail and are trying to drive it into case-hardened steel.

Maybe a staple will do ?>:-} ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| STV, Queen Creek, AZ 85142    Skype: skypeanalog |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
     It's what you learn, after you know it all, that counts.

Transistor.png

Oh, pfft! I need more sleep. Yeah, of course.

Yes, that is a given. The point here is when Vin = 0, the Op-Amp doesn't fry, and it doesn't need very much in the way of exotic features. In that event, it doesn't matter whether the diode is biased, or not. Obviously, in order to be stable and predictable, there needs to be a bias resistor. It wouldn't hurt to put a large resistor from the negative input to ground, as well.

Of course if I follow John's suggestion, the whole thing is moot.

20Transistor.png

p)

't fry, and it doesn't need very much in the way of exotic features. In th at event, it doesn't matter whether the diode is biased, or not. Obviously , in order to be stable and predictable, there needs to be a bias resistor. It wouldn't hurt to put a large resistor from the negative input to groun d, as well.

That (corrected) circuit misses one point - there will be about zero V drop across the 33R. To get a diode's worth of drop you'd need 2 diodes on the opamp input side not 1.

... unless you used a germanium transistor anyway.

NT

Right. That's where your earlier idea was leading me. I like the whole id ea. Philosophically, it is using the device (an SSR) for its designed and intended purpose, rather than making use of some more-or-less incidental pr operties of a device. An NPN transistor isn't really "supposed" to have it s base-emitter junction used as an isolation diode with no potential on its collector. I mean, it works, as long as one chooses a transistor with a h igh enough maximum Vbe, but it's a little bit... I don't know... more cleve r than wise? It just doesn't seem elegant.

idea. Philosophically, it is using the device (an SSR) for its designed an d intended purpose, rather than making use of some more-or-less incidental properties of a device. An NPN transistor isn't really "supposed" to have its base-emitter junction used as an isolation diode with no potential on i ts collector. I mean, it works, as long as one chooses a transistor with a high enough maximum Vbe, but it's a little bit... I don't know... more cle ver than wise? It just doesn't seem elegant.

With Veb around 5-7v you're likely to run into trouble doing that... unless you use a germanium which have Veb 20-30v

NT

:

-%20Transistor.png

map)

nt

sn't fry, and it doesn't need very much in the way of exotic features. In that event, it doesn't matter whether the diode is biased, or not. Obvious ly, in order to be stable and predictable, there needs to be a bias resisto r. It wouldn't hurt to put a large resistor from the negative input to gro und, as well.

op across the 33R. To get a diode's worth of drop you'd need 2 diodes on th e opamp input side not 1.

I'm afraid you need to look again. (I updated the drawing, BTW.) The (-) op-amp input, the emitter, and the top of the resistor are all at the same potential - they are tied together. Assuming the op-amp has a negligible i nput offset value, the (+) op-amp input is also at the same potential as th e top of the resistor, which means the 1N4148 anode is also at the same pot ential, since it is tied to the (+) input. During normal operation, the an ode potential will be something like .7V above the cathode potential, and t he cathode is tied to to the bottom of the resistor. Thus, there is roughl y a .7V, or 1 silicon diode potential difference across the resistor.

ss you use a germanium which have Veb 20-30v

Yeah, maybe more like 10V. The NiMH batteries I am using for the prototype have a 9.8V terminal potential when fully charged and not supplying curren t. As I said, perhaps more clever than wise. Germanium would work, and th ere are a few silicon transistors that could handle it. The NXP BUJ100LR,1

26 might squeak by.