cupoftea
Advanced Member level 6
Hi,
We have a burglar alarm circuit which is basically a 2 core power cable which runs
from house to house in a stretch of 300 Houses in a remote Scottish Village.
We often get power cuts , so the 2 core cable is powered from a 24V to 30V battery which is in the end house only.
In each house we have a microcontroller (3V3 , 2mA rail) and a burglar detector (5V, 2mA rail).
This 3v3 and 5v comes off the 2 core cable (one in each house) as in the attached schem.
LTspice and .PNG schem as attached.
Because it often needs to run off battery,
the bias power drawn by the linear regulators needs to be as low as is possible. So the attached one draws only 60uA
of bias current. There is no other way of making such a low-bias-current linear regulator it seems?
Do you know of a better way?
Also the comms between each of the microcontrollers also needs to be over the 2 core power cable, as we cannot have
a separate comms wire. As such, there is a comms module which does FSK comms over the power cable and this gets picked up by each microcontroller
in each house.
Obviously , in each house, we can't have the big electrolytic caps directly connected to the 24-30V bus, since this would
ruin the power_cable comms signalling. So as you can see on the schem, the electrolytic caps are "downstream" of high impedance bias circuitry.
This allows the FSK signalling system to work well over the entire power cable between all houses.
Another point is inrush current. This must be minimal since otherwise the very long power cable would radiate much noise and
destroy the operation of the microcontrollers. So the inrush must be no more than if all the circuits in each house was
drawing its maximum operational current of some 4mA each. This is achieved by way of the attached circuit, where the electrolytic caps
are downstream of high impedance bias circuitry which stops any high inrush current from ever happening.
In every 5th house, the units have a LED flasher circuit, this draws power from the power cable and makes a LED flash brightly for 15ms every 750ms.
-But only if any one of the houses is getting burgled.
As such there is a switch mode LED driver in these circuits which gets enabled after the storeage electrolytic capacitor has been charged up
at a rate off 8mA.
The problem is that the attached circuit has very high resistances in it, and multiple BJTs. We are worried that we
will get oscillations going on in these BJT circuits? How may we avoid this?
Also there are resistance of 10MEGs and 1MEG in the circuit shown. Do you think that the leakage currents in the solder mask surrounding these resistors may actually mean that their "effective resistance" is actually potentially much less than 10MEGs and 1MEG?
Also, even though the simulator shows a no load bias current of only 60uA.....this is with the shown BJTs with their particular hfe values.....in the real circuit, the bias current could be widely varying especially if a batch of BJTs is bought for production that have very low or very high hfe values.....how can we mitigate this? .....these variant hfe batchs could also cause major deviation in the 3v3 and 5v output voltages...how can we mitigate this?
Also, how can we re-design to get even lower bias current?
We have a burglar alarm circuit which is basically a 2 core power cable which runs
from house to house in a stretch of 300 Houses in a remote Scottish Village.
We often get power cuts , so the 2 core cable is powered from a 24V to 30V battery which is in the end house only.
In each house we have a microcontroller (3V3 , 2mA rail) and a burglar detector (5V, 2mA rail).
This 3v3 and 5v comes off the 2 core cable (one in each house) as in the attached schem.
LTspice and .PNG schem as attached.
Because it often needs to run off battery,
the bias power drawn by the linear regulators needs to be as low as is possible. So the attached one draws only 60uA
of bias current. There is no other way of making such a low-bias-current linear regulator it seems?
Do you know of a better way?
Also the comms between each of the microcontrollers also needs to be over the 2 core power cable, as we cannot have
a separate comms wire. As such, there is a comms module which does FSK comms over the power cable and this gets picked up by each microcontroller
in each house.
Obviously , in each house, we can't have the big electrolytic caps directly connected to the 24-30V bus, since this would
ruin the power_cable comms signalling. So as you can see on the schem, the electrolytic caps are "downstream" of high impedance bias circuitry.
This allows the FSK signalling system to work well over the entire power cable between all houses.
Another point is inrush current. This must be minimal since otherwise the very long power cable would radiate much noise and
destroy the operation of the microcontrollers. So the inrush must be no more than if all the circuits in each house was
drawing its maximum operational current of some 4mA each. This is achieved by way of the attached circuit, where the electrolytic caps
are downstream of high impedance bias circuitry which stops any high inrush current from ever happening.
In every 5th house, the units have a LED flasher circuit, this draws power from the power cable and makes a LED flash brightly for 15ms every 750ms.
-But only if any one of the houses is getting burgled.
As such there is a switch mode LED driver in these circuits which gets enabled after the storeage electrolytic capacitor has been charged up
at a rate off 8mA.
The problem is that the attached circuit has very high resistances in it, and multiple BJTs. We are worried that we
will get oscillations going on in these BJT circuits? How may we avoid this?
Also there are resistance of 10MEGs and 1MEG in the circuit shown. Do you think that the leakage currents in the solder mask surrounding these resistors may actually mean that their "effective resistance" is actually potentially much less than 10MEGs and 1MEG?
Also, even though the simulator shows a no load bias current of only 60uA.....this is with the shown BJTs with their particular hfe values.....in the real circuit, the bias current could be widely varying especially if a batch of BJTs is bought for production that have very low or very high hfe values.....how can we mitigate this? .....these variant hfe batchs could also cause major deviation in the 3v3 and 5v output voltages...how can we mitigate this?
Also, how can we re-design to get even lower bias current?
Attachments
Last edited:
