Update: Adding an output transistor – see below.
So you’d kind of like to know when the battery in your stompbox is getting a bit flat but you’re not sure how to proceed.
The first problem is that you don’t want your low-battery monitoring circuit to make a significant contribution to battery drain – which rules out parts like the MC34064 “volt sense” circuit because it has a quiescent current of about half a milliamp.
The only part that fits the bill and is readily available (in the UK at least) is the TC54 from Microchip. The TC54 is available in a number of packages, output-styles and voltage variations but the one for us is the TC54VN4302EZB. This variant comes in a TO92 package, has a reference voltage of 4.3V and an N-channel open-drain output.
A brand spanky new alkaline PP3/6F22/1604 battery has a voltage of about 9.5V. We shall (quite arbitrarily) assume that it is dead when the voltage drops to about 7.3V. But the TC54 has a reference voltage of 4.3V (and they don’t make a variation with a reference voltage of 7.3V) so we have to cheat.
Here is the TC54:
and here is the circuit we are going to implement:
The TC54 works on a supply voltage (derived from the Vin terminal) from 0.7V to 10V so we don’t need to worry about keeping the TC54 alive.
In a nutshell, the TC54 monitors the voltage at the Vin terminal and when this voltage falls below the internal reference voltage the Vout terminal is pulled low (and the LED switches on to indicate that the battery is failing). If the input voltage rises above Vref + ~200mV (about 4.5V), Vout goes high again (and the LED switches off).
We want to detect a battery voltage of about 7.3V (not 4.3V) so we use R1 and R2 to form a potential divider showing a fixed proportion of the battery voltage to the TC54. We need to find values for R1 and R2 so that when the battery voltage falls to 7.3V, the TC54 input voltage falls to 4.3V.
The Microchip TC54 datasheet suggests a bleed current (Iq) of about 100uA and also advises that the voltage at the Vout terminal when it is pulled low is about 0.5V. Now we have enough data to do some calculations.
When the battery is fresh, Vbat = 9.5V so if Iq is 100uA then (by Ohm’s law):
R1 + R2 = 9.5V/100uA = 95kOhm
We want the TC54 to switch at a battery voltage of 7.3V, so:
4.3V/7.3V = 0.59
R1 and R2 form a potential divider with the voltage across R2 being 0.59 x Vbat, so:
R2/(R1+R2) = 0.59
We can now solve for R2 because we know R1 + R2 = 95kOhm
R2/95 = 0.59
R2 = 0.59 x 95
R2 = 56kOhm and R1 = 39kOhm (i.e. 95kOhm – 56kOhm).
Let’s assume we are going to use a low-current LED such as the Kingbright L7104LID (which has Vf of 1.7V and a nominal current of 2mA). This led will happily shine at 1.5mA forward current (we don’t want to over-tax an already failing battery) so we calculate Rload as follows:
Rload = (Vbat – Vled – Vout)/Iload
Rload = (7.3 – 1.7 – 0.5)V/1.5mA
Rload = 3400Ohm.
So a 3.3kOhm or 3.6kOhm resistor can be used for Rload, giving:
The reason why Microchip suggest 100uA for Iq is that you can then safely ignore Iss (which is only 1uA) making for an easy calculation. As we increase the values of R1 and R2 to reduce the quiescent current (Iq), Iss becomes more significant. For example, if we reduce Iq to about 30uA we can redo the calculations above (ignoring Iss) and find that R1 = 120kOhm and R2 = 180kOhm to give a lo-battery set-point of a nominal 7.16V. However, the effect of Iss raises the set-point back to 7.3V. Neat.
So here’s the absolute, final, Stompville-approved, 9V low-battery monitor which pulls only 32uA in its quiescent state and switches on a low-current LED (1.4mA) when the battery voltage drops to 7.3V:
Microchip doesn’t make a 7.3V version of the TC54 means so we need two more resistors than would otherwise be necessary – but the absolute maximum voltage at Vin is 10V – so adding these extra resistors means we can use a wall wart with an output voltage up to about 16.5V before we fry the TC54. If we used a TC54 with a lower reference voltage, we could increase the maximum wall-wart supply voltage but the 4.3V version is more readily available. Of course the above could all be built into a hybrid LED. Our wish is that Microchip make a version of the TC54 with a 7.3V reference, a Vin(max) of (say) 24V and a constant-current sink output of 1.5mA – all built into a 3mm LED so our low-battery solution is reduced to one component. Dream on.
Update – Maximum drain-source voltage and adding an output transistor
You might be thinking of using a Microchip MCP111 in lieu of the TC54VN (and using similar math to calculate the values of R1 and R2), but the output stage of the MCP111 is less robust than the TC54VN. In particular, the maximum output voltage of the Vout pin should not be held higher than Vdd for any significant time. This effectively prevents you implementing the resistor divider trick to adjust the set-point, unless you use an additional transistor on the output to isolate the MCP111 from the supply voltage.
Also, for the TC54VN we should point out that, according to the Absolute Maximum Ratings on the Microchip datasheet, the output voltage (open-drain) acceptable range is (Vss – 0.3V) to 12V
This suggests that the Vds(max) of the output transistor is limited to 12V, so we shouldn’t implement this design for battery voltages higher than about 10V without adding an extra output transistor with suitable Vceo (or BVdss).
You can use any general purpose small-signal p-channel MOSFET or PNP output transistor with a high-enough breakdown voltage. Note that the bipolar (PNP) option is much cheaper (for through-hole at least) but requires an extra resistor.
Note that the schematics below show a 9V battery supply but the additional transistor should be used when the supply voltage is higher than about 10V nominal.
A wide selection of surface-mount MOSFET devices are available. For through-hole, use a ZVP3306A or similar. The design was prototyped with a ZVP2106A (more expensive than ‘3306).
Use a 2N5401 or BC557 (or whatever) for the transistor version.