Analogue circuit question -advice needed please !

You should cross-post to sci.electronics.design I imagine you will get suggestions to use comparators or op-amps. Here's an idea using transistors:

Vcc n-channel mosfet out _____________c e_________________________________________ | |___| | | | / |_/\\/\\/\\_____|_________c \\ 2k2 \\ / about 50K \\ \\ | / npn |----------------| | | probe 100 / / signal-->____/\\/\\/\\_____/e \\ | / 100K | \\ adjustable pot | / | \\ | | | | |__________________| | | ground

After you finish twiddling the resistor divider on the right it will impress a certain fraction (here, you will end up setting it at about half) of the output voltage on the base of the transistor. This gets compared to the probe signal voltage by means of the base-emitter junction of the bjt. If this (half) the output is .6 volt (that's the b/e drop) more than the probe signal, the bjt pulls down the bias on the mosfet gate. So the output voltage is regulated at (input + .6volts) X (divider ratio). Linear, though not quite a direct proportion. You may have to amplify the input. If so: get another npn, connect its collector to Vcc, its base to the probe signal, and you will get amplified current out the emitter at a voltage one diode drop below the probe signal. This would be a good thing because the two diode drops we get in this circuit would thereupon cancel each other! Cute.

I forgot to mention: in my post with the ASCII schematic, use a proportional font to view it. If you are in Google, click options and "view original."

D'oh! I labeled the mosfet like a bjt. Instead of c and e, it should be labeled d for drain (in place of the c) and s for source (in place of the e).

Edit: you need a bipolar as shown above instead of the mosfet. The

2k2 resistor changes to a 1k 1/2 watt resistor or a few hundred ohms, 1 watt.

Thanks to all for the suggestion so far.

Couple of answers to the above :- I am not too concerned with absolute accuracy due to the horribly non linear fuel guage. But yes, when the tank is heading towards empty the probes output tends towards zero volts. (This is for a restored classic car with a new petrol tank and fuel level sensing, and I merely need some indication of the tanks contents, using the original dial ! )

One thing I am wary of is loading the probes output, and the first suggestion above has a 100ohms to ground, where as that would have to be a few K to avoid dragging down the probes output. Maybe an input amplifier will be needed.

Anyone else got ideas/comments ? I'll have a crack at trying it out tommorow. Thanks.

Thanks to all for the suggestion so far.

Couple of answers to the above :- I am not too concerned with absolute accuracy due to the horribly non linear fuel guage. But yes, when the tank is heading towards empty the probes output tends towards zero volts. (This is for a restored classic car with a new petrol tank and fuel level sensing, and I merely need some indication of the tanks contents, using the original dial ! )

One thing I am wary of is loading the probes output, and the first suggestion above has a 100ohms to ground, where as that would have to be a few K to avoid dragging down the probes output. Maybe an input amplifier will be needed.

Anyone else got ideas/comments ? I'll have a crack at trying it out tommorow. Thanks.

Mr. Holme has provided a good, workable circuit for what you want. It might be able to use some input protection for the op amp, but other than that, it should work fine.

Here's a redrawing with a bit of input protection and his in-line fuse (view in fixed font or M$ Notepad): ` VCC ` + ` | ` VCC o ` + 1/2 LM358 ( FU1 ` | ) 1/2A ` D - VCC o ` ^ + | ` Vin ___ | ___ |\\| | ` o----o-|___|-o-|___|--|+\\ ___ |/ ` 0-5V | 1.8K | 3.3K | >--|___|--|TIP31C ` .-. - .--|-/ 100 |>

` R| | D ^ | |/| | Vout 0-10V @ 0-250mA ` | | | | === o---------o-----. ` '-' === | GND | | ` | GND | .-. | ` === | 10K| | | ` GND | | | / \\ ` | '-' (M 1) 45 ohms ` | | \\_/ ` '-----------------o | ` | | ` .-. === ` 10K| | GND ` | | ` '-' ` | ` === ` GND (created by AACircuit v1.28.5 beta 02/06/05

This should work well with a 13.8V supply. With low battery or with the key on ACC, a full tank may look just a little low, but that shouldn't be a problem. Choose whatever value for R that you feel you need for circuit input impedance, or leave it out altogether. The 1.8K and 3.3K input resistors and the diodes (use 1N4002 or better) provide protection for the non-inverting op amp input (use an LM358 or other single supply op amp).

The op amp, TIP31C power transistor and resistors provide a gain block with a precise, linear voltage gain of 2, and the ability to source up to 1/4A into your 45 ohm load.

Try Mr. Holme's circuit, you should like the results. An op amp circuit is much easier, more accurate and less expensive than cobbling together a circuit with discrete transistors.

Good luck Chris

snipped-for-privacy@hotmail.com wrote: (snip)

Just to be sure: Does 0V indicate empty or full?

(snip)

Again, does 0 volts indicate empty or full?

No sense suggesting something that has a 50 50 chance of reading backwards.

You want a DC amplifier with a voltage gain of 2, high input impedance, and high output current drive. You could use an op-amp non-inverting amplifier with an emitter-follower buffering the output e.g. (view in fixed font)

' +12V ' Vin |\\ | ' ---------------|+\\ |/ ' | >---------| Q1 ' .-|-/ |>

' | |/ | ' | | ' | | ' R1 | R2 | ' ___ | ___ | Vout ' .---|___|----o------|___|-----o-------------- ' ' ' === ' GND '

You need an op-amp that runs off a single supply, with inputs that work down to the negative supply rail.

Q1 will dissipate up to 1 Watt.

The two resistors (R1, R2) set the gain. R1=R2 for a gain of 2. I suggest R1=R2=10k

I would include a 10uF power supply decoupling capacitor, and a 0.5A fuse in the power lead.

I would also connect a large resistor (100k - 1M) from the input to ground, in case it's left floating.

You mean a non-proportional font.

If you use a proportional font, you have to know which one was used to create the drawing so you can get things to line back up.

--
Former professional electron wrangler.

Michael A. Terrell
Central Florida

Chris got it right. The op-amp circuit is much better. Piece o' cake, prolly work first time out.

Kell

Simple circuit idea ..

+12v +---+-------+ | | | | R=2k @=meter opt=R | | | | |C | +-----| T2=any med power (to220) npn just simply bolted to something metal | T1| |E 22k| |E | o--R-+--| R=6.8 ohm | |C | 10k=R | | t1= any smal signal (high gain) pnp | | |

-----+---+-------+

The 0-5v sensor voltage is divided by the 22k/10k to ~ 0-1.6v This voltage then apears accros the 6.8 ohm resistor where it sets the curent through T1 and hence the meter to ~ 0-230ma T1 counteracts the vbe drop acros T2 and provides curent gain.

This should be moderatly acurate. bear in mind this is untested and at full scale T2 is just about saturating, wich may mean 'full' scale is not quite acheived depending on type of transistor. ov input might not be quite zero current, this may corespond better to the the guage anyway, the optional R is if you want to increase the curent on empty. (try with 1meg) you can always tweak it ie lower the 10k or the 6.8 ohm etc.

Colin =^.^=

oops line wrap

+12v +---+-------+ | | | | R=2k @=meter opt=R | | | | |C | +-----| T2 | T1| |E 22k| |E | o--R-+--| R=6.8 ohm | |C | 10k=R | | t1= any smal signal (high gain) pnp | | |

-----+---+-------+

T2=any med power (to220) npn just simply bolted to something metal (but insulated!)

Colin =^.^=

Follow up :-

Many thanks for all the ideas submitted. I now have a fully working solution based on the op-amp idea and it works a treat.

I only wish my knowledge of analogue electronics was better so I could contribute something back to this particular forum.

Cheers !

Chris wrote:

One other small thing -- you _will_ need a heat sink for your pass transistor, if you don't have one. With a 14V supply, about 2 watts of power will be dissipated by that transistor when the gas tank is about half full. That's well into the "groan zone" for a TO-220 package by itself.

Just about any heat sink made for a TO-220 package will do, but the bigger, the better -- especially in an automotive environment, where you can get elevated temperatures. A heat sink rated for 4 watts or more is probably a good idea. A thin coat of silicone heat sink compound (so you can just barely see through it when applied) is optional.

Thanks for an interesting question, and taking the time to give feedback. That's contribution enough.

Good luck Chris

Cool!

Ah, contribute whatever you've got - we run the gamut. :-)

Hopefully, it'll be under the dash, and not out in the crud.

No, it's NOT! That's almost as bad as not using a heat sink at all. Personally, I prefer that pasty white stuff, but probably only because I "inherited" about a 16-oz jug of it from the Air Force. ;-)

Cheers! Rich