sensing tiny and huge current without dropping too much power

[wingneckedhorse]

I am having a bit of a conundrum trying to come up with a current detection scheme for a battery and power circuit. There is a boost IC between the LiIon battery and a load. We need to sense when a charging voltage/current has been applied at the output of the boost (battery charge is coming from load device) and then switch off the boost to directly route the charging power to the battery. We were looking at a MAX9928 to do current sensing at the output of the boost, via a ~0.02 Ohm sense resistor, but the issue with that is that at the tail end of the charge cycle, current may be very low, like < 5mA, so with that resistor we are talking about only 0.1mV of voltage across the resistor, which I don't think the IC can detect reliably (based on Figure 2 of datasheet). Trickle charging would be interrupted when current dropped to low to be detected. If I increase the value of the sense resistor to 1 Ohm so that we get several millivolts across Rsense even at 5mA, then we have unacceptable voltage drop during normal charge currents and when the boost circuit is in operation.

[LI-ION] =====> [BOOSTER]======> [RSENSE] =====> [LOAD and CHARGER]
^<-------------------(bypass booster when charging detected)------|

Any ideas for different ways to solve this?

Thanks!

 

[2dye4]

Just a though

Put a low offset op-amp across the low ohm resistor in a differential amplifier mode.

Read its output as the value when the output isn't saturated.

Read the resistor directly when the op-amp output is saturated.

 

[VEBill]

In general, trickle charging could be open loop with respect to current and if anything based only on the float voltage (trickle possibly switched off if or when the battery voltage gets to the max). I'm not sure why a charging circuit would have to actually measure the trickle current. It could be a fixed value of current based on design.

Disclaimer - not an expert in this area, so standby for corrections.

 

[wingneckedhorse]

It is true that the trickle charge is just based on float voltage. We don't actually care about the values of the current being detected though, just the polarity (direction). I.e., we just need to know that it is charging so that we can bypass the boost converter. If we have ANY current going towards the battery (negative current detected), we want to disable the boost converter. Otherwise, operate as normal.

 

[ScottyUK]

Can you afford a Schottky diode drop in the paths? If so then use a diode in series with a sense resistor for charging, and a lower value resistor in series with a second diode for discharge. Connect the two diode-resistor strings in antiparallel.

 

[wingneckedhorse]

Hi Scotty, Unfortunately I don't think the diode will work on the discharge path due to needing every tenth of a volt we can get to get all the power out of the battery. On the charge path, the float voltage must be controlled very precisely on Li-Ion cells, so it won't work there either.

 

[LiteYear]

I think I need to know more about your boost circuit to know what the problem is. Typically, boost circuits already have a diode on the output, so you're never going to get negative current. Additionally they typically have some sort of feedback network, so are already monitoring the output voltage/current in some way.

Can you post your boost circuit? Methinks it is critical in understanding the solution.

 

[Skogsgurra]

A flux gate can operate over a wide current range. The voltage drop is negligible and it doesn't cost much if you roll your own.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[benta]

Check at ti.com
They have a large selection of current-sense amps.

Benta.

 

[wingneckedhorse]

Liteyear: Below is the typical application circuit, which is what we would be using. I guess a big question is: What happens when a charging voltage is applied at the output side and can that even be detected? My guess at the moment is that the charge IC (we do not have access to the datasheet yet) will come online, see that the "battery" voltage (actually the boost here) is above the threshold to bypass constant current mode, and then go into constant voltage mode, which would be 4.2V float. At that point, 4.2V would be higher than the boost voltage (3.5V) and current would start to flow *somewhere* (?). If the boost output is still in the voltage where constant current charging is used, perhaps the charge IC will raise its voltage above the boost voltage to try to force constant current in. A lot of assumptions here, but I want to see if there's a way at all to do this and then figure out the details.

http://www.ti.com/ods/images/SLVSAQ2B/TPS61230_3.5Vout_Typ_App.gif

Skogsgurra: I like this idea a lot and took a look at hall effect sensors but they seem quite large. Do you think there is the possibility of being able to detect only milliamps of current with a very limited amount of space? My thought was that it would take many windings to achieve this. It is a 30mm x 5mm board that is designed to go inside a small battery pack.

Benta: The highest gain I see in the TI line is 1000V/V, so if I have a 0.02Ohm sense resistor, that is only 0.0001V of drop at 5mA charge current, which means the charge sense IC output is 0.1V. Did I do that math right for this type of sensor? I could feed that through another 30x gain stage to make it useful for driving other things, but is that too much total gain?

 

[Skogsgurra]

WNH: Can't help you there. It is a very limited space. And it is a very tiny current.
One thing that you remeber for future applications: Hall sensors and GMRs are temperature sensitive. Flux gates are not. Virtually no zero drift at all.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[LiteYear]

Not sure what the charge circuit will do, but that boost circuit is typical, except that the output diode has been replaced by back to back MOSFETs. This allows the device to electrically disconnect Vin from Vout in both directions if the EN signal goes low. Additionally, the device "stops switching" if the output voltage exceeds some threshold, but it is not clear whether that means the output MOSFETs are left on or not. In any case, there are various other behaviours of that chip which haven't been designed with bi-directional flow in mind that may trip you up.

So you're back to your original idea about bypassing the booster. I think you might need to look a bit closer at the behaviour of the various components. I still don't see what current has to be the trigger. Don't you just want to know when the voltage of your load is higher than the output of your booster? If it's not, then the battery needs to supply the load. If it is, then the battery should be charging. So why not just detect voltage?

 

[wingneckedhorse]

Hi LiteYear,

Yes I do just need to detect which voltage is greater, but the problem as I understand it is that the point at which I can measure the charger voltage is the same as the output of the booster. So they are one and the same unless I create a little voltage drop between the two using the sense resistor, which is what I am doing now. Maybe I have misunderstood the question?

 

[ScottyUK]

Can you sense input current to the regulator and infer the output with any confidence?

 

[zappedagain]

Skogs, you got me curious so I had to Google 'flux gate'. The first link was informative. Here is the second one for your amusement:

https://www.youtube.com/watch?v=VkBGfcZjXvM

Z

 

[LiteYear]

Yes I do just need to detect which voltage is greater, but the problem as I understand it is that the point at which I can measure the charger voltage is the same as the output of the booster.

No, not which is greater (which would require the sense resistor), but whether the charger voltage has exceeded a threshold.

I'm still a bit confused by your circuit, but this is my assumption:

Battery is nominal 3.2V?
Booster outputs 3.5V regulated.
Load runs off 3.5V.
Then the load turns into an energy source and the charger switches on. Voltage rises and you want the booster bypassed.

So just set your bypass circuit to be triggered when the booster output voltage rises above, say, 3.6V. The charger will charge the battery, which will eventually float to 4.2V and enter trickle charge. When the energy source disappears, the load reappears, pulling the battery voltage back down below 3.5V. Set your bypass circuit to switch out when the voltage falls below 3.5V, and you'll be back to the battery feeding the load via the booster.

 

[ScottyUK]

Ignore my last post. My brain slipped out of gear for a moment. :-(

 

[wingneckedhorse]

Hi LiteYear,

Yes we considered just monitoring the output voltage, but here was the issue with that: There is overlap because battery voltage could be anywhere between 3.2V and 4.2V at any point in time. Charging voltage could also be anywhere in the same range as well, because most charging algorithms these days start in constant current mode (charge voltage could be anywhere between 3.2V and 4.2V to drive a fixed amount of current into battery) and then switch over to a constant voltage float charge (4.2V) after that cycle is complete.

 

[LiteYear]

Okay, that's pretty complicated then. I'm surprised that booster works at all. My reading of the datasheet suggests that it would disconnect the battery if it rose to 4.2V. I'm also a bit surprised that the charging process can get started at all. If the battery is only at 3.2V, the booster outputs 3.5V and your charger only wants to supply 3.2V and a bit, I don't see how the booster is going to reverse.

So I'm afraid I'll have to back out. I can't think of a simple solution given the parameters you've provided. I can only suggest you: 1) look at something that more directly detects the presence of the charger, 2) try to connect the charger to the battery instead of the load, or 3) swap the boost converter for a bi-directional DC-DC converter.

 

[jimkirk]

Unless you're doing something tricky with the TI boost IC, its output current is limited to 2 amps or so.

You want to detect as little as 5 mA, so that's a 400:1 range.

Here's a TI circuit (eval board available) that can sense from 10 uA to 100 mA, a 10,000:1 range, far more than you're probably looking for. Yeah, it's only 100 mA and they use a 6.8 ohm sense resistor, so you'll want to scale the sense resistor for the range you need, but it looks to me like it would do what you want.

http://www.ti.com/tool/tipd135?keyMatch=high side current sense amplifier&tisearch=Search-EN