@Sensacell Yeah, and I live in a small country town in Northern NSW Australia, (pop. 40k) so the local shop is tiny. Just for fun I looked up Digikey Australia, and there is one, so I tried ordering a LM4040 from them. It was a couple of dollars, but postage was from the US and was AUD$24. So no Aussie stock at all apparently. There is another big company here that I generally use when I want parts, Element14. (Silicon, if you're wondering). They're usually pretty good, and they do have the LM4040 in stock. aud$15 postage. I guess I can't complain too hard though, it probably costs them that to get someone to pick the part, put it in an envelope, write my address on it and stick a stamp on. c'est la vie, as the French say.

;-)

Try TaydaElectronics.com. They are out of Thailand and shipping is reasonable. Selection is pretty limited but they always seem to have the category of parts I need. $1 for shipping and a TL431 at 10 for $1. If you select the 7 to 30 days for delivery it is $1 but they usually deliver in 5 to 7 days.

https://www.taydaelectronics.com/tl431asf-tl431-precision-shunt-regulator-sot-23.html
.
.
.

 

@MrSalts thanks for the tip! They didn't have the LM4040 or the TL1634 unfortunately, so I've ordered five TL431's. I've got a couple of other old R/C transmitters I'm hoping to use one day, and they also have analog displays in them, so who knows? The extras could come in handy.

;-)

 

I've ordered five TL431's

So you are going to build my op amp circuit?
What op amp are you using?

 

@crutschow I'm attempting the one you posted in post #29, but with a 5.1v zener instead of the TL1634-5 as recommended by @Ian0. I did have it nearly working, the fully charged battery gave full deflection, in fact the needle banged up against the stop at the end, so I'm guessing it was working. There was no magic smoke and the meter is still functional, so, so far so good. But I made a mistake hooking it up to the switch so it was always on (didn't go through the switch) and when I tried to fix that I broke one of the transistors (leg fell off), so it's another trip down to Jaycar for me tomorrow. Maybe I'll get a couple extra this time...

I looked at your circuit in post #27, but it looks mighty complicated to me. I don't think my meagre skills would be up to building it. I had enough trouble with four components...

;-)

 

I'm attempting the one you posted in post #29, but with a 5.1v zener

Then I'm confused.
Why did you buy the TL431, since it won't work with that circuit.

 

Then I'm confused.
Why did you buy the TL431, since it won't work with that circuit.

It looks as though it should work in that circuit, just wired as a basic shunt regulator, replacing the zener; but the minimum 1mA current for it to regulate isn’t met. The TLV431 or LM385 might work.

Now what would be interesting is if we could configure it to eliminate the variations of Vbe from the circuit. Perhaps with a PNP current mirror. I’m at work now, I might give that some thought this evening.

 

@crutschow @Ian0 You're confused! Did I mention my knowledge of electronics is nil? I can follow instructions, but I'm still struggling with v=ir. I thought the TL431 would work in that circuit, or that was my understanding from the previous conversation. Seems I got the wrong end of the stick. Wouldn't be the first time. Oh well, it was only a few dollars, and maybe they'll come in handy somewhere else.

:-(

 

I thought the TL431 would work in that circuit

As Ian0 noted, the TL431 has a minimum current of around a mA, so will not work in the shown current-mirror circuit, which requires the reference carry essentially no current below it's reference voltage, such as a Zener or the other mentioned references.

Incorporating Ian0's suggestion for using PNP transistors I came up with the circuit below:
R2 supplies the TL431 current until its set reference voltage is reached (5v).
The green trace is the voltage across the meter resistance.

Edit: Modified circuit to remove the diode, which was not needed.

 

Apologies to Mr. Harris whilst we go off on a bit of a tangent. I suppose I could have put it in a private message, but other followers of this thread may be interested. . .
This is what I had in mind (I was busy yesterday evening so didn't get chance to simulate it)
Put the Base-emitter junction inside the TL431's feedback loop (I used a TLV431 to get over the 1mA minimum current problem).
That removes the variation due to changes in temperature and random variations in Vbe voltage. However, it does introduce a different error - the current through R1 and R2 doesn't go into the current mirror, so the output current is short of Vreg/(R1+R2). As Vreg is fixed, the error is also fixed, and has exactly the same effect on R3 as having a larger value of Vreg.
So 180uA through R1/R2 has the same effect on Iout as reducing Vreg by 180uA x R3.

 

Here's Ian0's circuit above using a TL431 since Ian stated he had bought five of them.
R2 provides the 1mA of current the TL431 needs to reach regulation at about 4V before the transistors start to turn on.

 

@Ian0 that BCV62 looks like the answer to a maiden's prayer. It seems to contain half the circuit, or at least the hard bit with the transistors. However it's a SMD device. I can barely see SMDs, never mind solder them. Is there a through hole equivalent? I've had a look on Element14, but everything they have is SMD.
Thanks,
Ina

 

It's just two matched transistors. You can use two BC547s or whatever you have. As the device only has three terminals, you can twist the various leads of the BC547s together and solder them. You can glue two TO92 devices together flat-side to flat-side with superglue to keep them in place (and keep them thermally coupled!).

 

@Ian0 yeah that's what I did last time, and one of the legs fell out of one of the transistors. Too much twisting I'm guessing...

;-)

 

The simple way to avoid damaging transistors is to solder them to a small bit of that copper strip board, and then connect to the board strips, Not so good for RF but OK for DC and audio circuits.

 

@MisterBill2 thanks. That was plan A, I had a small PCB board that I was going to solder everything on, but then changed my mind as it seemed to be getting too complicated, trying to work out which leg should poke through which hole. So I opted for plan B, the "just solder everything together" option. Back to Plan A for the next attempt.

;-)

 

it seemed to be getting too complicated, trying to work out which leg should poke through which hole.

If you make a paper copy of the board, it should not be too difficult to draw the connections between the components.

 

@crutschow you assume that I know what I'm doing...

:-\

 

@crutschow you assume that I know what I'm doing...

:-\

I usually try to make that presumption. Or at least think that people consider the results of what they are doing.

 

you assume that I know what I'm doing...

You don't really have to know.
Look at it as sort of a puzzle.
How can the parts be positioned on the board to make the connections from the schematic as direct as reasonable feasible?
Getting the best result likely will result in trial-and-error with some shuffling of parts.
Or do you have some difficulty in visualizing that?

 

You don't really have to know.
Look at it as sort of a puzzle.
How can the parts be positioned on the board to make the connections from the schematic as direct as reasonable feasible?
Getting the best result likely will result in trial-and-error with some shuffling of parts.
Or do you have some difficulty in visualizing that?

Working on the layout on paper is cheap and easy and does not put parts at risk. AND nobody has to see the sketches that did not work. So a failure on paper is by far the best place for it. That has been my experience in engineering for over 40 years! There is a mantra of "don't be afraid to fail", to which always add "on paper, where it is cheaper". Not every first sketch works the first time, which is why we have erasers!