EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: HSPalm on February 17, 2017, 10:50:02 pm
-
Hi,
For a project I want to step up the voltage from two lithium 3.6v cells to 7v for a device that accepts 7-13v.
- If I wire the batteries in parallell I step it up with an efficiency of somewhere around 85-90%.
- If I wire them in series, the "efficiency" is 100% all that time when the battery voltage Vin is greater than programmed Vout. When the voltage goes below 7v, the boost circuit will "step in" and maintain 7v.
Can I do it like this? Will the boost circuit go crazy while Vin > Vout, or will it just turn off the mosfet? The switcher circuit I'm looking at in particular is the SX1308.
-
From the datasheet:
http://www.datasheetspdf.com/datasheet/download.php?id=921054 (http://www.datasheetspdf.com/datasheet/download.php?id=921054)
On P5 look at the Efficiency VS Vin graph.
For all efficiencies to match the spec your layout and component selection will be critical to achieving good results that those frequencies.
-
A boost converter by definition cannot have higher Vin than Vout, well, not more so than the forward drop across the diode. If the input voltage tries to go higher than the output, the whole converter is effectively bypassed by the diode. How the IC handles that I'm not sure. I would expect most of the modern stuff will tolerate it.
-
If the input voltage tries to go higher than the output, the whole converter is effectively bypassed by the diode.
I believe that's the idea: boost the battery voltage if it's under 7V plus a Schottky diode drop, otherwise bypass the regulator and deliver the full battery voltage to the load, less a Schottky diode drop.
How the IC handles that I'm not sure. I would expect most of the modern stuff will tolerate it.
The only way to know that is to try it out and see what happens.
-
Hi,
For a project I want to step up the voltage from two lithium 3.6v cells to 7v for a device that accepts 7-13v.
- If I wire the batteries in parallell I step it up with an efficiency of somewhere around 85-90%.
- If I wire them in series, the "efficiency" is 100% all that time when the battery voltage Vin is greater than programmed Vout. When the voltage goes below 7v, the boost circuit will "step in" and maintain 7v.
Can I do it like this? Will the boost circuit go crazy while Vin > Vout, or will it just turn off the mosfet? The switcher circuit I'm looking at in particular is the SX1308.
If you have an issue with that setup you may need to implement a Buck/Boost IC that will output 7V.
-
Use a buck boost controller.
https://www.maximintegrated.com/en/products/power/switching-regulators/applications/battery-powered.html?gclid=CPjJg7fSmdICFVtMDQod9gwAgw (https://www.maximintegrated.com/en/products/power/switching-regulators/applications/battery-powered.html?gclid=CPjJg7fSmdICFVtMDQod9gwAgw)
http://www.linear.com/products/buck-boost_regulators (http://www.linear.com/products/buck-boost_regulators)
http://www.ti.com/lit/an/slva059a/slva059a.pdf (http://www.ti.com/lit/an/slva059a/slva059a.pdf)
Regards, Dana.
-
What device do you want to power?
It's quite likely there's a linear 5V regulator inside it, which you could remove and replace with a low drop-out regulator.
-
You must try. There is no way to say definitely whether it will work.
It's likely it will work if input voltage changes slowly and gradually. If input voltage is higher than output in this moment, and in the next ms it goes below and the condition oscillates, you have a trouble.
I can try, but I'm not sure how much I'm able to extensibely test for conditions like the ones you describe as a possible outcome. Clearly from your answers this is not the normal way to handle this, so I guess testing is the only way.
If the input voltage tries to go higher than the output, the whole converter is effectively bypassed by the diode.
I believe that's the idea: boost the battery voltage if it's under 7V plus a Schottky diode drop, otherwise bypass the regulator and deliver the full battery voltage to the load, less a Schottky diode drop.
How the IC handles that I'm not sure. I would expect most of the modern stuff will tolerate it.
The only way to know that is to try it out and see what happens.
Thanks, seems like I have to try it.
Hi,
For a project I want to step up the voltage from two lithium 3.6v cells to 7v for a device that accepts 7-13v.
- If I wire the batteries in parallell I step it up with an efficiency of somewhere around 85-90%.
- If I wire them in series, the "efficiency" is 100% all that time when the battery voltage Vin is greater than programmed Vout. When the voltage goes below 7v, the boost circuit will "step in" and maintain 7v.
Can I do it like this? Will the boost circuit go crazy while Vin > Vout, or will it just turn off the mosfet? The switcher circuit I'm looking at in particular is the SX1308.
If you have an issue with that setup you may need to implement a Buck/Boost IC that will output 7V.
Use a buck boost controller.
Regards, Dana.
But much of my whole point with wiring them in series was to have 100% efficiency above 7v. Since my device handles 4.2V*2 just fine I see no reason to implement a buck/boost unless my idea is bonkers. If this doesn't work Ill just stick to wiring them in parallell and have the usual 80-90% efficiency.
What device do you want to power?
It's quite likely there's a linear 5V regulator inside it, which you could remove and replace with a low drop-out regulator.
This circuit is for those who want to build their own battery packs for Fatshark/Skyzone FPV goggles. Not a chance I get people to open and modify them for this :)
-
I use a 2 cell lithium ion battery directly with my FatShark goggles, no external regulator at all. Never had a problem with that, minimum voltage is about 3.5V/cell for best battery life and fully charged is 4.2V. Not sure why a boost converter is even needed here?
-
It's worth a try.
You could also put a Schottky from Vin straight to Vout to reduce losses in the inductor's parasitic resistance.
Do check that you won't exceed the absolute maximum rating on the FB pin. The feedback voltage divider output will be a little higher than normal, and that can damage some chips.
-
I use a 2 cell lithium ion battery directly with my FatShark goggles, no external regulator at all. Never had a problem with that, minimum voltage is about 3.5V/cell for best battery life and fully charged is 4.2V. Not sure why a boost converter is even needed here?
Yes this is what we normally do. But when the goggles turn off there are still lots of juice left in the batteries, and this is what I'm trying to make a solution for. Many have started making their own batteries with 18650 cells, but can't make use of their full capacity.
It's worth a try.
You could also put a Schottky from Vin straight to Vout to reduce losses in the inductor's parasitic resistance.
Do check that you won't exceed the absolute maximum rating on the FB pin. The feedback voltage divider output will be a little higher than normal, and that can damage some chips.
The voltage divider lands at roughly divide by 10 so this won't be an issue. Max input voltage will be 8.4 volts, and max Vfb is 6 volts.
-
At what voltage do they turn off? You generally don't want to run Li-ion cells down 100%, for best life a cutoff of about 3.5V/cell is ideal and there really isn't that much capacity below that. I use 18650 cells on mine too, I made a pack in the 2S2P configuration and it will run quite a few hours on a charge.
-
At what voltage do they turn off? You generally don't want to run Li-ion cells down 100%, for best life a cutoff of about 3.5V/cell is ideal and there really isn't that much capacity below that. I use 18650 cells on mine too, I made a pack in the 2S2P configuration and it will run quite a few hours on a charge.
For best life, charge and discharge around nominal 3.6V as often as possible...
3.5V is way too high to call ideal! How many cycles do you want from your cells? Cut-off on Li-ion cells are 2.5V, recommended cut-off is 2.75V.
NCR18650PF is rated 500 cycles with capacity dropping from ~2800mAh to 2300mAh, this is with 100mA load and cut-off at 2.5V.
The goggls are mostly rated at 7V. I guess they can go lower but it's best to stay within spec of the goggles I think.
-
http://batteryuniversity.com/learn/article/lithium_based_batteries (http://batteryuniversity.com/learn/article/lithium_based_batteries)
With modern graphite based anode Li-ion batteries 3.5V is about 20% charge remaining and 3V is 0%. According to that document, the shift to graphite occurred back around 1997 so I doubt there are many of the coke anode cells still around.
-
With modern graphite based anode Li-ion batteries 3.5V is about 20% charge remaining
Yes, under no load or very small load.
Below about 15-25% SoC (depending on the cell type), the DC ESR skyrockets, which is why discharge ending voltage is typically set between 2.0 and 3.0V (clearly below the open-circuit voltage of 0% SoC). High-power applications use lower ending voltages. Setting ending voltage to 3.5V only results in 20% SoC with extremely small loads (like C/20), and if there are any spikes to be supplied by the cell, averaging the voltage sense gets very important. Otherwise, 3.5V is easily seen at 50% SoC under modest load, and it's unreliable, sometimes it's 40%, sometimes 60%, depending on load and temperature. It's OK to base an approximate battery gauge on the cell voltage, but cutoffs either require being on the steeper part of the curve (typically between 10% and 0% SoC), or using very small currents (end of the CC-CV charging).
-
Another nice article:
https://www.digikey.com/en/articles/techzone/2014/aug/the-sepic-option-for-battery-power-management (https://www.digikey.com/en/articles/techzone/2014/aug/the-sepic-option-for-battery-power-management)
-
http://batteryuniversity.com/learn/article/lithium_based_batteries (http://batteryuniversity.com/learn/article/lithium_based_batteries)
With modern graphite based anode Li-ion batteries 3.5V is about 20% charge remaining and 3V is 0%. According to that document, the shift to graphite occurred back around 1997 so I doubt there are many of the coke anode cells still around.
I don't know about coke anodes. Search for a recent 18650 Li-Ion cell from a renowned manufacturer. It's all the same, cut-off 2.5V. All other data (cycles, capacity etc.) is specified in relation to this cut-off voltage. The first random popular cell I found is the LG INR18650 HG2 3000mAh which also states:
4.1.2 Standard Discharge
“Standard Discharge” shall consist of discharging at a constant current of 600mA to 2.5V. Discharging is
to be performed at 23 ºC ± 2 ºC unless otherwise noted (such as capacity versus temperature).
4.1.3 Fast Charge / Discharge condition
Cells shall be charged at constant current of 4000mA to 4.2V with end current of 100mA. Cells shall be
discharged at constant current of 10000mA and 20000mA to 2.5V. Cells are to rest 10 minutes after
charge and 30 minutes after discharge.
I attached the datasheet.
-
Search for a recent 18650 Li-Ion cell from a renowned manufacturer. It's all the same, cut-off 2.5V.
Cutoff is under load - you can see this in most datasheets, typically it's defined under 0.2C or 0.5C load. As the cell's ESR is very high near 0% SoC, the cutoff must be a lot lower than the corresponding OCV. You can try fully discharging a cell per manufacturer instructions to 2.5V @ 0.2C, then wait a minute and measure the voltage. It's around 3.4V typically.
This distinction can cause confusion. The difference between OCV and under load voltage becomes very significant below 20% SoC.
-
Yeah I mostly deal with my flight packs which I measure the resting voltage when I pull them from the plane after landing. Rule of thumb has always been discharge them no lower than 20%, both to increase battery life and to ensure enough power remains to do a go-around if I miss the landing approach.
-
Search for a recent 18650 Li-Ion cell from a renowned manufacturer. It's all the same, cut-off 2.5V.
Cutoff is under load - you can see this in most datasheets, typically it's defined under 0.2C or 0.5C load. As the cell's ESR is very high near 0% SoC, the cutoff must be a lot lower than the corresponding OCV. You can try fully discharging a cell per manufacturer instructions to 2.5V @ 0.2C, then wait a minute and measure the voltage. It's around 3.4V typically.
This distinction can cause confusion. The difference between OCV and under load voltage becomes very significant below 20% SoC.
Yes, it can seem confusing to those who have not experienced this. Just to be clear, as from the quote from the datasheet and in my application, I am talking about cut-off voltage during load. In my case the load is 440mA (not stated at which voltage, though) which is about 0.2C on a standard 2-3 Ah cell. If I wire them in parallell as has been discussed, the load will be 0.1 C per cell.
The cells are under a load when current is being drawn, of course. This may sound stupid, but did anyone believe I meant the video goggles will turn off when the cells had a certain resting voltage? How would the goggles know?
Yeah I mostly deal with my flight packs which I measure the resting voltage when I pull them from the plane after landing. Rule of thumb has always been discharge them no lower than 20%, both to increase battery life and to ensure enough power remains to do a go-around if I miss the landing approach.
Do you use any kind of telemetry, OSD or beeper to warn you about flight pack voltage? I usually try to land when my 4S pack is at or below 14V, resting voltage is usually around 3.7 or 3.8V.
-
I have telemetry on one of my planes and OSD in my multirotor but in practice I've found them to be nearly useless for that purpose and rely instead on a flight timer built into my transmitter. The discharge curve is just too flat to take a meaningful capacity reading from the voltage under load. Resting voltage is the only accurate measurement.
-
Add a few more components and you can make a boost-buck with a 34063. See attached. Ignore the stuff on the bottom. We used this design for years.
(http://)
Paul
-
Add a few more components and you can make a boost-buck with a 34063. See attached. Ignore the stuff on the bottom. We used this design for years.
(http://)
Paul
That is an interesting solution, thanks for sharing. Especially since that chip is so damn cheap :) But as I've mentioned earlier, I don't need the buck part if I can feed the battery voltage straight through when higher than 7V (no loss). For this design, the schematic you show has to many and too large components, but I will keep it.
-
Actually the components can be quite small. You need to do the calcs to see how much inductor you need. The MOSFET can be in a small package. There is an additional flyback diode shown which is not needed. We used it to generate a small negative voltage to power LM324 op amp negative rails so we could swing to a zero volt output.
paul
-
I have telemetry on one of my planes and OSD in my multirotor but in practice I've found them to be nearly useless for that purpose and rely instead on a flight timer built into my transmitter. The discharge curve is just too flat to take a meaningful capacity reading from the voltage under load. Resting voltage is the only accurate measurement.
I agree with you. I've spent too much time arguing with people why measuring remaining/spent capacity is not ideal. First off, the measurement resets to 100% each time you remove power, so the reading is wrong if you unplug and plug back in. You have to always use 100% charged batteries which you know hold the rated capacity. If using voltage as indication you could at this point tell if you put in a discharged battery, even. And do you always remember if you put in that 1300 mAh or the 1500 mAh after take-off? I speak mostly of racing, where one battery lasts 3 minutes, I find it much more useful to have my transmitter vibrate on Vbat < 14 V (on 4S). If it vibrates once, I had a punchout, if it keeps vibrating, it's time to land.
Actually the components can be quite small. You need to do the calcs to see how much inductor you need. The MOSFET can be in a small package. There is an additional flyback diode shown which is not needed. We used it to generate a small negative voltage to power LM324 op amp negative rails so we could swing to a zero volt output.
paul
I did some more reading on this chip and it seemed to be the standard solution for low cost buck and boost applications for some time. But compare it to the one in my OP, SOT23-6 package and 4 external parts + caps. This is if you're not using it in a SEPIC config. I attached a 3D rendering showing my placements, so you see I don't really have much space left.
-
As the SX1308 is a really cheap, I'd really like to get an update if you ended up testing it and if it worked as intended or if you had to use another solution.
-
Can you use a 3S pack? They're 12.6V fully charged or 12.3V with a conservative 4.1V/cell charge voltage.
-
Thanks for your reply, but my voltage requirements are different than the original OPs, so specific chemistries based on the original post aren't exactly what I'm looking for.
I was mainly interested in experiences with using a cheap boost converter when V_in > V_out.
My specific application is powering a circuit which can only operate between 3.3 and 3.6V from 5V USB and a LiIon Battery (~3V cutoff - 4.2V when charging).
-
My specific application is powering a circuit which can only operate between 3.3 and 3.6V from 5V USB and a LiIon Battery (~3V cutoff - 4.2V when charging).
Just use a 100% duty cycle capable buck converter and either a linear or switching charger depending on efficiency requirements. You won't gain much between 3.3V and 3V.
-
You can also get a boost-LDO for low power applications. Such as some variants of the TI TPS61098x. These have the benefit of always offering an LDO output, even during boost operation.
-
Thanks for the tip, at first glance the TPS61098X chips look really nice. But after reading the datasheet I realized that the LDO output is only supposed to be used for low-power sensors and can only supply <200mA and not the full 500mA of the boost circuit. There is also a significant voltage drop/difference. E.g. the TPS610981DSE supplies 3.3V on the main output (passtrough when V_in>V_out) but only 3.0V on the LDO output (p. 7).
But if you know of any similar devices with higher output currents I'd be stoked. :)
edit: I found the TPS61025 from a very similar series which seems really nice. I has only one output and can apparently also linearly regulate the full 1A it can step-up.
I've also thought about highly integrated solutions of battery charger + boost converter. I've found plenty with a fixed boost voltage of 5V@500mA for USB-OTG, the cheapest (~0,7€) ones beeing the FAN54015 and FAN54005. 500mA are great, but I'd need another buck or LDO to get my 3.3V from the 5V supply. This again leads to increased size and BOM cost while not beeing ideal efficiency wise.
Then there are some with an additional 1.8V output, but I've only come across one (well two extremely similar ones) with 3.3V output (with an internal buck converter + LDO). That would be the TI BQ25120A BQ25122, but they can only supply 300mA via the buck converter. In theory they can spit out another 100mA via the LDO, but until someone corrects me I'd be pessimistic about simply connecting the outputs of both to get up to 400mA.
Anyway, the BGA packages of all mentioned chips are a real "treat" to solder manually.
So my next idea was maybe a SEPIC design behind the charger with something like the TPS61170 (DFN), the LM3481 (more power, sweet VSSOP package) ot the LM2735 in an awesome SOT23 variant. They are a bit more expensive tough. They also need at least 3V, which is fine for my application but would make them unusable to use with NiMH and other chemistries.
If someone knows of the proverbial jack of all trades device or other nifty solutions, let me know! :)
-
Wouldn't using a LDO defeat the point of being able to use the extra last 0.3V of the discharge curve? As in at the 3.7V or so over the majority of the discharge curve, a 3.3V LDO would waste about 11% of the power.
-
Yeah, for now I decided to use a TPS63031 and to stick with the MCP73831 charging IC.
It's the cheapest combination I could find of chips from known manufacturers which are actually available in relevant numbers.
Even though the BOM cost increases because of the second inductor, X7R caps and so on it's still a lot cheaper compared to an all-in-one solution, which are pretty close to unobtainium or can't deliver enough power.
OT: By the way, before I used the MCP1603 and Microchip demands in the datasheet (p.17, lower right corner) that the output capacitor should not exceed 22µF. I guess that's because more capacitance could destabilize the resonant circuit. However I haven't come across this requirement in most other datasheets and the following circuitry explicitly requires several 22µF caps not even counting the C_out of the MCP.
Before deciding to use the TPS I added a lowpass filter with a 4,7µF coil + 47µF inductor to effectively isolate the 2MHz oscillation from whatever sits behind it. Do you think this would have actually been necessary or would it have sufficed to have a few cm of trace behind the output cap, leading to enough phase distortion to not influence the resonant circuit too badly?