Combined voltage regulator + supervisor in TO-220/TO-92 pack

[Erik Walthinsen]

I've got a circuit based on a ATtiny15L and Honeywell's HMR3100 compass
module that appears to have only one remaining problem before I'm done
with it. If the circuit is hooked up to a bench power supply, which is
powered on with its front switch, the compass fails to initialize
properly, and the circuit doesn't do its job. This is obviously due to
a slow ramp-up of supply voltage/current from the bench supply. If the
power is stable and a switch applies power to the target circuit, it
works every time...

Now, the problem is that the ATtiny15L is fully utilized: no free pins
and almost all code space used (think 1010 out of 1024 bytes).
Compounding this, the HMR3100 does not have an external reset pin...
Therefore, I need to find a way to make sure that the compass does not
get power until the supply is stable.

However, the entire design is currently through-hole, and needs to stay
that way for assembly purposes (it's really small and assembled by hand
in really small quantity). Therefore I'd like to find a part to replace
the standard 7805 regulator that will solve the problem.

Basically, I'm after a TO-220 or TO-92 (or equiv 3-terminal) voltage
regulator that will not enable its output until it can guarantee
sufficient supply current to bring the regulated rail up immediately.
Complicating this is the requirement that the regulator be able to
operate with a supply voltage up to 36V...

I could probably build a retrofit module with a regulator, on-board cap
(currently 33uF on either side of the regulator), and supervisor chip
connected to a mosfet/etc, but I'd really rather not have to do that, as
that almost guarantees surface-mount parts.

Any component suggestions would be appreciated, as would any other ideas
on how to solve this. Then again, the problem may be made moot sometime
today by way of product requirements being clarified (will a bench
supply ever be used in the field [think middle of nowhere] with that
particularly unlikely turn-on sequence?), but I'd rather have a solution
available if not.

TIA,
Omega
aka Erik Walthinsen
omega@pdxcolo.net

 

[Jamie]

Erik Walthinsen wrote:

I've got a circuit based on a ATtiny15L and Honeywell's HMR3100 compass
module that appears to have only one remaining problem before I'm done
with it. If the circuit is hooked up to a bench power supply, which is
powered on with its front switch, the compass fails to initialize
properly, and the circuit doesn't do its job. This is obviously due to
a slow ramp-up of supply voltage/current from the bench supply. If the
power is stable and a switch applies power to the target circuit, it
works every time...

Now, the problem is that the ATtiny15L is fully utilized: no free pins
and almost all code space used (think 1010 out of 1024 bytes).
Compounding this, the HMR3100 does not have an external reset pin...
Therefore, I need to find a way to make sure that the compass does not
get power until the supply is stable.

However, the entire design is currently through-hole, and needs to stay
that way for assembly purposes (it's really small and assembled by hand
in really small quantity). Therefore I'd like to find a part to replace
the standard 7805 regulator that will solve the problem.

Basically, I'm after a TO-220 or TO-92 (or equiv 3-terminal) voltage
regulator that will not enable its output until it can guarantee
sufficient supply current to bring the regulated rail up immediately.
Complicating this is the requirement that the regulator be able to
operate with a supply voltage up to 36V...

I could probably build a retrofit module with a regulator, on-board cap
(currently 33uF on either side of the regulator), and supervisor chip
connected to a mosfet/etc, but I'd really rather not have to do that, as
that almost guarantees surface-mount parts.

Any component suggestions would be appreciated, as would any other ideas
on how to solve this. Then again, the problem may be made moot sometime
today by way of product requirements being clarified (will a bench
supply ever be used in the field [think middle of nowhere] with that
particularly unlikely turn-on sequence?), but I'd rather have a solution
available if not.

TIA,
Omega
aka Erik Walthinsen
omega@pdxcolo.net
off the top of my head. use a bipolar transistor that has enough current

handling to take care of the board supply by simply feeding the C and
using E as your output.
mean while, use a 555 timer as a Delay on that will drive the Transistor.
the 555 timer can be connected to the main source after the switch.
you may also want to look into the use of using a HexFex as the switch
instead of the Bipolar transistor, this would give you less loss.