providing remote power

[JohnT]

I have a couple questions on powering a remote device:
3.3V (+/- .1V), 100mA nominal, and surges to .75A for 2 seconds every 10
seconds.

I read somewhere to keep wire losses below 10%, I'm guessing to keep heat
down. Running 12V over 22 gauge (I have to use 22AWG) allows only 95 feet
of wire. I need to go more than 200 feet (400 foot of wire) resulting in a
4.1V loss (33%). Considering my ~25% duty cycle what loss is tolerable?

When the device activates it draws .75A and I found (trial and error) I had
to add a big 2200uf cap to the remote device to keep the voltage from
dipping too much. This is when going to a 3.3V LDO regulator. What causes
this dip? Should this cap be placed on the device side of the regulator or
the 'wire' side...or both?

I could really cut the amps by going to a 90% efficient DC-DC
converter...but these are quite expensive. Big caps are cheaper if they are
a viable solution.(?)

jt

 

[Ian Stirling]

I have a couple questions on powering a remote device:
3.3V (+/- .1V), 100mA nominal, and surges to .75A for 2 seconds every 10
seconds.

I read somewhere to keep wire losses below 10%, I'm guessing to keep heat
down. Running 12V over 22 gauge (I have to use 22AWG) allows only 95 feet
of wire. I need to go more than 200 feet (400 foot of wire) resulting in a
4.1V loss (33%). Considering my ~25% duty cycle what loss is tolerable?

Wire losses below 10% is only a guide.
If you don't care about efficiency, then simply stick 30V or so in at
the end.

When the device activates it draws .75A and I found (trial and error) I had
to add a big 2200uf cap to the remote device to keep the voltage from
dipping too much. This is when going to a 3.3V LDO regulator. What causes
this dip? Should this cap be placed on the device side of the regulator or
the 'wire' side...or both?

Wire side.
This way it can supply current as its voltage sags.

 

[JohnT]

Seems I'd be wasting 'lectricity by placing it on the wire side since its
current would still have to go through the inefficient regulator. On the
device side it would only have to power the device. One side is ~12V and
the other is 3.3V and I have no idea how that figures in.
jt

 

[The other John Smith]

I have a couple questions on powering a remote device:
3.3V (+/- .1V), 100mA nominal, and surges to .75A for 2 seconds every 10
seconds.

I read somewhere to keep wire losses below 10%, I'm guessing to keep heat
down. Running 12V over 22 gauge (I have to use 22AWG) allows only 95 feet
of wire. I need to go more than 200 feet (400 foot of wire) resulting in a
4.1V loss (33%). Considering my ~25% duty cycle what loss is tolerable?

When the device activates it draws .75A and I found (trial and error) I had
to add a big 2200uf cap to the remote device to keep the voltage from
dipping too much. This is when going to a 3.3V LDO regulator. What causes
this dip? Should this cap be placed on the device side of the regulator or
the 'wire' side...or both?

I could really cut the amps by going to a 90% efficient DC-DC
converter...but these are quite expensive. Big caps are cheaper if they are
a viable solution.(?)

jt

How fast can the load cause the current to change from 100 mA to 750 mA? And
how much inductance do you have in your power wires?

Very crude assumptions:

- 200 feet straight #22AWG is about 150 uH (using a straight-wire equation
which may not apply to 200 feet)

- Two wires (+ and -) doubles it to 300 uH.

- You have about 6.5 Ohms of DC resistance which will give about 4.9 V drop.

If so, the current rise time must be no greater than L*di/E or about
(300e-6)*(.75-.1)/(12-4.9-3.3). This is 51 uS. If the current rises that
fast or faster, the regulator will not have sufficient headroom.

Or maybe this is all wrong. After all, I'm old and rusty.

The capacitance on the load end needs to be about dV = I*dt/C or 5 uF to
support the voltage during the 51 us. Use more than that, but go through all
this yourself and prove that the circuit is well damped. You don't want the
voltage at the load end to go skyrocketing (due to line inductance charging
the capacitor) when power is applied. Wire resistance may be your friend.

Twisting the wires together may help.

Good luck.

John

 

[Ian Stirling]

Seems I'd be wasting 'lectricity by placing it on the wire side since its
current would still have to go through the inefficient regulator. On the
device side it would only have to power the device. One side is ~12V and
the other is 3.3V and I have no idea how that figures in.

There is very little waste of current in the regulator, the waste is
the voltage that it drops.
Putting the capacitor on the wire side is the right way to do it, as
that way the capacitor can supply current as it drops down from
12V-6V or so.
On the load side, it can only supply current as it drops from 3.3 to
3.1V or so, so you'd need a capacitor of around 20 or more times the
size.

 

[JohnT]

I 'thought' the voltage difference would come into play some how.
thx

 

[N. Thornton]

I have a couple questions on powering a remote device:
3.3V (+/- .1V), 100mA nominal, and surges to .75A for 2 seconds every 10
seconds.

I read somewhere to keep wire losses below 10%, I'm guessing to keep heat
down.

you do whatever suits your app. In this case there is probably no
reason to keep loss down to 10%. But as usual you dont give us nearly
enough info to know and advise accordingly.

Running 12V over 22 gauge (I have to use 22AWG) allows only 95 feet
of wire. I need to go more than 200 feet (400 foot of wire) resulting in a
4.1V loss (33%). Considering my ~25% duty cycle what loss is tolerable?

you tell us! How do we know?

When the device activates it draws .75A and I found (trial and error) I had
to add a big 2200uf cap to the remote device to keep the voltage from
dipping too much. This is when going to a 3.3V LDO regulator. What causes
this dip?

wire R: more current, more R drop.

Should this cap be placed on the device side of the regulator or
the 'wire' side...or both?
wire


I could really cut the amps by going to a 90% efficient DC-DC
converter...but these are quite expensive. Big caps are cheaper if they are
a viable solution.(?)

to what? I dont know what you expect if you dont tell us what youre
doing.