High or low side current sensing

[Raveninghorde]

I'm reviewing an existing charger design which I'm expanding to 10
outputs.

At the moment I use high side current sensing with an opamp and fet
level shifting the signal so that it is ground referred.

The disadvantage is that this circuit can not measure the current
during a short and relies on the swticher's own current limit for
protection.

I've used a current mirror circuit before to get around this but it's
too many parts.

If I low side sense then this problem disappears. But then I loose the
simplicity of a common ground.

I can't make my mind up. Time for a whisky.

 

[Joerg]

I'm reviewing an existing charger design which I'm expanding to 10
outputs.

At the moment I use high side current sensing with an opamp and fet
level shifting the signal so that it is ground referred.

The disadvantage is that this circuit can not measure the current
during a short and relies on the swticher's own current limit for
protection.

I've used a current mirror circuit before to get around this but it's
too many parts.

If I low side sense then this problem disappears. But then I loose the
simplicity of a common ground.

I can't make my mind up. Time for a whisky.

You could consider current monitor chips that have a helper voltage and
where common mode includes ground. But it's not as cheap. If each charge
channel has its own switcher, can't you sense current there?

 

[Joerg]

Operate the sensing circuit from a separate supply that does not get
effected from the main supply drop.

In your circuit, I assume you're using the +/- inputs of the opamp to
form a dif shift via a shunt R ? The op-amp and Fet you're using can be
operating from a secondary supply that can also share the same common.

of course, if you throttle back the drive using the current sense, you
wouldn't get a total loss of voltage!.

If the battery terminals short out you do get a total loss of voltage.
But it's no big deal if there's a helper voltage that, for example,
provides a negative supply if the opamp CM range doesn't include VSS.

 

[Jamie]

I'm reviewing an existing charger design which I'm expanding to 10
outputs.

At the moment I use high side current sensing with an opamp and fet
level shifting the signal so that it is ground referred.

The disadvantage is that this circuit can not measure the current
during a short and relies on the swticher's own current limit for
protection.

I've used a current mirror circuit before to get around this but it's
too many parts.

If I low side sense then this problem disappears. But then I loose the
simplicity of a common ground.

I can't make my mind up. Time for a whisky.

Operate the sensing circuit from a separate supply that does not get
effected from the main supply drop.

In your circuit, I assume you're using the +/- inputs of the opamp to
form a dif shift via a shunt R ? The op-amp and Fet you're using can be
operating from a secondary supply that can also share the same common.

of course, if you throttle back the drive using the current sense,
you wouldn't get a total loss of voltage!.

 

[Adam Seychell]

I'm reviewing an existing charger design which I'm expanding to 10
outputs.

At the moment I use high side current sensing with an opamp and fet
level shifting the signal so that it is ground referred.

The disadvantage is that this circuit can not measure the current
during a short and relies on the swticher's own current limit for
protection.

I've used a current mirror circuit before to get around this but it's
too many parts.

If I low side sense then this problem disappears. But then I loose the
simplicity of a common ground.

I can't make my mind up. Time for a whisky.

What are your desired current, voltage and measurement accuracy ?

 

[Raveninghorde]

How about:

---------------------------------------------------------
! !
! ---------/\/\--- !
! ! ! !
+-----------------/\/\---------+--!+\ ! s !
! ! >--+--/\/\--+--!! !
! ------/\/----------+--!-/ ! !!----
! ! ! ! --!! P-
MOSFET
-------+---/\/\---+----Battery(+) --/\/\-- !d
--------
+--/\/\--GND
!
V =
K*I

It is the normal op-amp based constant current source but the op-amp
doesn't have to take
the large common mode and you don't need the 0.1% resistors.

The down side is you needenough voltage for the op-amp.

Thanks.

That looks good and appears to get around the short circuit problem.

It adds 3 resistors to my component count. Under half a cent extra
component cost plus placement time.

 

[Raveninghorde]

What are your desired current, voltage and measurement accuracy ?

I'm happy with 3% accuracy on current. Who cares if a battery charges
at 970mA rather than 1A.

I'm charging lithium ion so voltage is more important. On a 4 cell
pack I expect to be between 16.70V and 16.75V so that's +/- 0.15%
after adjustment.

 

[Raveninghorde]

You could consider current monitor chips that have a helper voltage and
where common mode includes ground. But it's not as cheap. If each charge
channel has its own switcher, can't you sense current there?

In this application I am using National simple switchers so they don't
help with the current measurement. They do provide an internal current
limit which does provide short circuit protection.

I don't like the short circuit current being higher than the charging
current. In the other job I've got to tidy up for production I'm
using a sepic which does need a proper current limit.

I like to use common building blocks and MooseFET's circuit looks good
at the moment.

 

[Raveninghorde]

Protect power on with 'battery good' detect first ?

I have battery good detection in the micro. The charger has to be safe
even if the micro dies.

The customer checks short circuit by putting a switch across the
output and a then a fuse in line to the battery. They then start
charging and throw the switch.

Similarly for reverse battery protection they fit a reversing switch
into the wiring. Again they start charging and throw the switch.

 

[Joerg]

In this application I am using National simple switchers so they don't
help with the current measurement. They do provide an internal current
limit which does provide short circuit protection.

I don't like the short circuit current being higher than the charging
current. In the other job I've got to tidy up for production I'm
using a sepic which does need a proper current limit.

On a SEPIC that's easy. You can sense the current at the base of the
second coil and run that into an inverting opamp.

I like to use common building blocks and MooseFET's circuit looks good
at the moment.

So you owe him a beer then

 

[Raveninghorde]

I guess I'm not visualizing Ravenhorde's setup. I'm assuming Vcc to
the op amp higher than the raw battery charging voltage, just as you
did...

Ah, I see. So use mine ahead of the regulator, if that's feasible.
If not, yours wins.


Nice work.

Your circuit is the same as what I use at the moment except + and -
inputs of the op amp are swapped.

The op amp does come from a higher supply, but as you relaized that
doesn't help.

I'm using a switching regulator so measuring in front doesn't work.

 

[Joerg]

Raveninghorde wrote:

[...]

I'm using a switching regulator so measuring in front doesn't work.

What kind? Any chance to add a C and an L and make it a SEPIC? Then
current sense would become easy.

 

[Raveninghorde]

Raveninghorde wrote:

[...]

I'm using a switching regulator so measuring in front doesn't work.

What kind? Any chance to add a C and an L and make it a SEPIC? Then
current sense would become easy.

I'm using an LM2675 simple switcher in a buck configuration. Changing
it to a sepic doesn't make sense as the extra cost would outweigh the
cost of high side current sensing.