Meter shunt with Zener diode and comparator

[THX-1138]

Probably not the right forum; perhaps 'experiments' or 'projects' would be better, but..

 

I'm noodling up a variable-voltage power supply (~0-35VDC), and I'm using a Radio Shack 1mA FSV 15V panel meter to display the output. (15Kohms feed resistor to get the appropriate voltage drop.) Since the voltage range of the supply goes to about double that, I thought I'd put in a switch to make the appropriate changes to the circuit to allow the meter to display 0-30VDC, and add the notations to the meter face.

 

I want to put a shunt across the meter to keep from damaging it in an over-voltage situation, and also give an indication of the fact with an LED. Currently I'm simulating this and using a 16V Zener (to go a wee bit off the scale) and an LM339 comparator.

 

To half-scale the meter is simplicity itself; just switch in an additional 15K resistor to its input path. And adjusting the shunt appropriately could be done essentially by changing the Zener diode.

 

However, I'd like to stick with just one Zener, since the voltage drop across the meter, with its 15K resistor, will still be 15V. I haven't done a lot with Zeners in the past, and I'm not completely conversant with all their behaviours. In other words, I'm having trouble figuring out a circuit that will use a 16V Zener to indicate overvoltage in both full-scale (15V) and half-scale (30V) situations. Also, one that doesn't go all wonky when the low-impedance meter gets added to the circuit! I've tried some isolation diodes, but no joy. Just not doing it right, I guess.

 

I'm attaching a PDF showing the current circuit with which I'm playing, including a SPICE input deck. In the PDF, the highlighted areas on the schematic and netlist show what breaks the results: if the meter isn't connected, all's well; if it is, the break point drops from 16V to less than 1V.

 

Since this is essentially play and a learning exercise for me, I'm interested in any constructive feedback.

 

Thanks!

MeterShunt.pdf

Edited September 25, 2008 by THX-1138
Forgot some key points!

[big314mp]

A good 15V zener in parallel with the meter may be all of the protection you need.

 

Otherwise you may have to connect a MOSFET in series with the voltage divider/meter, and switch it with your zener connected to the inverting output of a comparator. noninverting input of the comparator would need a reference voltage (anything less than 14V or so would work), and the output of the comparator would switch the MOSFET.

 

When the voltage exceeds 15V, the zener breaks down, applies 14V or so to the inverting input of the comparator. This 14V is higher than some reference voltage, so the comparator output is forced low, switching off the MOSFET, cutting power to the meter.

 

That seems overly complex though. There is probably an easier way than that.

 

You can also use an op-amp as your comparator. I know radioshack sells a small MOSFET, as well as op-amps (don't know about comparators though), so you can get all of the parts there.

 

A NOT gate on the comparator can switch a separate MOSFET for your LED.

[THX-1138]

What's the best way to model a meter in SPICE, anyway?

[THX-1138]

I haven't gone beyond modeling in SPICE (i.e., no actual wiring yet), and I'm probably doing something wrong, but I can't get switching behaviour out of MOSFETs at low voltages (5-12VDC). Just gradual low-gain changes, which certainly won't serve for a low-resistance path to anything.

 

Of course, maybe the models I'm using are bogus, since SPICE seems to think the MOSFET gate is a current sink deep enough to drop the comparator high-level output voltage significantly unless I include a 10M resistance between them.

 

There's an excellent chance I haven't a clew at all, too. But it's all about learning anyway..

[big314mp]

Try modeling using a MOSFET driver instead of just a comparator.

 

MOSFETs are voltage driven, so it looks like you aren't applying sufficient voltage to the gate, and the MOSFET is acting as an amplifier.

 

I think something is wrong with your model though, as the gate switches on once the gate voltage reaches a certain level. Switching speed is (mostly) a function of amps applied and the gate capacitance. As the gate capacitance of MOSFETs is usually pretty low, they switch fairly quick.

 

Someone with more electronics experience should post also, as I am definitely not an expert on this stuff.

[YT2095]

I think you`re really over complicating what is essentially a very simple task.

all you really need is a PD across the output with 2 resistors in a 3:4 ratio, IE a 15 ohm and a 20 ohm resistor in series with your meter across (in parallel with) the 15 ohm.

that`s the very bare bones basics behind the idea, and of course if you wanted to use a zenner as crowbar protection as well then that`s cool too.

[THX-1138]

YT: More complicated than it needs to be? Sure! I have a need, yes, but I'm using it as a learning experience, fiddling with new things. Once I have something that works (and that I understand ), I'll pare it back to the basics I need. As it is, though, I haven't used SPICE in a quarter of a century, and then not in this way. And I've never used FETs in my hobby projects.

 

Okey, trying an ideal circuit: SPICE's generic N-channel MOSFET; supply voltage 20VDC; 1Mohm resistances at drain, gate, and source; input voltage (across gate and ground) from -20 to +20 by 0.5V steps. I'm attaching the circuit, the SPICE layout, and the SPICE results.

 

That surely doesn't look like a fast turn-on to me. What am I doing wrong?

mosfet.pdf

mosfet-spice-input.txt

mosfet-spice-output.txt

[THX-1138]

Try this graphically. I changed the source resistor to 1 ohm and plotted the input voltage against the drain voltage -- so the drain-source path makes the negative half of a voltage divider with R1. Shouldn't the drain voltage have dropped quickly to near zero, indicating R_DS dropping likewise?

[big314mp]

The only thing I can think of is that the MOSFET isn't getting proper gate voltage, as it appears like it isn't opening the channel fully (i.e. the linear region of the transistor).

 

That's a pretty poor explanation, so I think someone else who knows more than me needs to help you out on this one.

 

Sorry

[THX-1138]

I asked in another forum and my mistakes were pointed out to me. Among other things, the substrate needs to be connected to the drain node.

 

Thanks!