Hi -

I have a problem involving buck/boost converters on which I would appreciate advice.

I make small numbers of devices which contain high-power LEDs. The present device is illustrated by the "Current Version" in the attached diagram. A set of rechargeable batteries giving around 12V is dropped to 5V to run an MCU which controls the operation of an LED driver which feeds a constant current to the LEDs (actually two separate LED drivers which feed the two LEDs independently). This works reliably.

For various reasons I am seeking to build an 'improved' model which will run from one of the many 'powerbanks' currently used to recharge mobile phones. This is illustrated by the "Revised Version" in the diagram. The 5V from the powerbank is used to run the MCU and is also fed to a boost converter to bring it up to 12V. This feeds an LED driver which drive three LEDs in series with 350mA. When I run this from a bench PSU, it works perfectly. However, I get varying symptoms with different powerbanks, from irregular flashing of the LEDs, to the LEDs not starting up at all.

I am wondering if the 'cascaded' boost/buck converters are causing this problem which is why I am seeking advice from those that know a lot more about these things than I do!
If this is thought to be the problem, what would be the recommended solution?

 

It might be better to run the 12V boost converter directly from the battery, rather than from the 5V.
That way you only have the inefficiency of one converter, not two, for the 12V.

How much voltage do the LEDs require?
What about running an LED switching current driver directly from the battery, which could be the most efficient solution.

 

It might be better to run the 12V boost converter directly from the battery, rather than from the 5V.
That way you only have the inefficiency of one converter, not two, for the 12V.

Not really possible as the idea is to use an off-the-shelf powerbank which just has 5V USB-type outputs. I can't get at the 3.7V battery.

How much voltage do the LEDs require?

Each drops around 3.5V (it varies with temperature) so I need around 12V to drive the 350mA LED current.

What about running an LED switching current driver directly from the battery, which could be the most efficient solution.

See first item above! I think I will need to scope the various voltages to see what is happening. Unfortunately, the prototype is with a friend for testing at the moment so that'll have to wait! I was just wondering whether anyone had any experience of cascading these devices and whether this induced any instability.

 

Each drops around 3.5V (it varies with temperature) so I need around 12V to drive the 350mA LED current.

So how many LEDs are in series?

 

So how many LEDs are in series?

As in original post: "This feeds an LED driver which drive three LEDs in series with 350mA. "

 

any experience of cascading these devices and whether this induced any instability.

No direct experience, but a switching regulator has a negative input impedance (input current decrease with increasing input voltage), and that could lead to instability if one is driving another.

A large capacitor (e.g. a mf or more) at the output of the 5V supply to common may help.

 

If you need 700mA for the two LED strings, then you need more than 1.7A from your power bank. Quite possibly over 2A. What is the output current rating of the power bank?

 

A couple of switching current sources come to mind, these are not cascaded supplies but switching type current sources:

Change R1 to change the current.
More on this circuit at http://cappels.org/dproj/1 watt white LED power supply circuit/1 watt white LED power supply circuit.shtml


Below is a switching type supercap charger that can also be and LED driver.

Change the resistor between MC34063 pins 6 and 7 to control the peak current in the inductor, and thereby the average current across the load. In your case, the load would be the LED with a capacitor across it. Change the resistor divider (1k + 1k here) to adjust the voltage limit. Figure 13 in this See this MC34063 datasheet https://www.onsemi.com/pdf/datasheet/mc34063a-d.pdf


Charge-up test, horizontal axis is seconds, vertical is 10 mv/count. If the capacitor is 100F the charge current is 700 ma. voltage rise is 7 mv/second and time from zero to 2.5 volts would be six minutes. For your use, this only shows that the current was constant over a range of voltages (about 0.75V to 2.5 volts in this case).

 

If you need 700mA for the two LED strings, then you need more than 1.7A from your power bank. Quite possibly over 2A. What is the output current rating of the power bank?

No - there's only one LED string of three, running at 350mA. Draw from the powerbank is 800mA.

 

The requirements and capabilities of a "powerbank" device intended for phone charging are totally different from what is needed to power LED lighting packages. The powerbanks are simply not suitable supplies. Rach voltage conversion is less than 100% efficient, and probably none are fast enough responding to load changes. Plus, power out is always less than the power in.
. Or better yet, just get a battery pack with three of the 3.7volt batteries in series an use that 11.1 volt pack instead of the 12 volt source.

 

The requirements and capabilities of a "powerbank" device intended for phone charging are totally different from what is needed to power LED lighting packages. The powerbanks are simply not suitable supplies. Rach voltage conversion is less than 100% efficient, and probably none are fast enough responding to load changes. Plus, power out is always less than the power in.
. Or better yet, just get a battery pack with three of the 3.7volt batteries in series an use that 11.1 volt pack instead of the 12 volt source.

Can you be more specific about why "powerbanks are simply not suitable supplies"? Not disagreeing - just interested! Efficiency is not a problem here - the powerbanks I am intending to use are more than enough, capacity-wise, for the application, so 95% efficiency is fine.

Yes, I could just use a battery pack, but then I need protection circuitry, charging circuitry etc, and being able to buy a powerbank off-the-shelf is a huge saving in effort -and probably cost - for me.

 

Post #7 mentioned this, but a boost converter will require much more initial input current to startup, even without a load, than when operating. Its likely the battery cannot support the initial required input current at startup and causes the battery voltage to drop.

 

The difference is that charging a battery via a mini or micro USB connection is quite a different set of requirements from powering an LED light system. The voltage regulation that is good for charging a battery is probably not fast enough to keep a constant 5 volts for the MCU system while also powering that 12 volt boost converter. If that boost converter is 100% efficirnt, the input power is 4.2 watts. 4,2 watts at 5 volts is 0.84 amps from the 5 volt source. So far I have seen no mention of the capability of the powerbank device. That leaves only 160mAfor the digital control part, if the powerbank regulator can deliver one amp. (these calculations presume 100% efficiency, which is rather optimistic, but easier to do. So that is why I suggested that the powerbank was not adequate.
What might work is two powerbank devices in parallel, depending on their current ratings , which have not been mentioned that I recall.

 

True - I was a bit vague about the powerbank!

The claimed figures are Capacity: 20,000 mAH; Output: USBC-5V 3A, USBA-5V 4.5A

So even if those figures are pretty optimistic, there should be plenty of overhead. The MCU only takes 70mA, by the way. Once I get the unit back, I will run some tests with the 'scope.

 

Another option is to delay startup of the second boost converter until the first converter is in run mode (has stabilized).

 

True - I was a bit vague about the powerbank!

The claimed figures are Capacity: 20,000 mAH; Output: USBC-5V 3A, USBA-5V 4.5A

So even if those figures are pretty optimistic, there should be plenty of overhead. The MCU only takes 70mA, by the way. Once I get the unit back, I will run some tests with the 'scope.

While the current capability for battery charging may be adequate, the response time of the regulator part may not be nearly fast enough. Battery charging does not require fast regulator response times. For the next test, connect a fast voltmeter across the five volt pair, towards the load end of the connection.

 

Another option is to delay startup of the second boost converter until the first converter is in run mode (has stabilized).

Yes - this actually happens as part of the process. The unit only lights up when darkness falls. So the processor etc are all on and running for quite a time before the light level drops enough to turn on the LEDs. I also plan to modify the software so that there is a soft-start for the LEDs, rather than a sudden hit.

Mind you, it's an interesting thought that the second boost converter does actually start running as soon as the unit is powered up, though with no load. That shouldn't be a problem, but...!

 

Mind you, it's an interesting thought that the second boost converter does actually start running as soon as the unit is powered up, though with no load. That shouldn't be a problem, but...!

Hmm...I think that is a problem because the first converter can't supply the required current to the second converter until after it has stabilized. It doesn't matter that there is no load.

As a test, try disabling the second converter first, and see if the first converter startup normally. If it does, then manually enable the second converter without the MCU. See if it starts up normally.

 

Hmm...I think that is a problem because the first converter can't supply the required current to the second converter until after it has stabilized. It doesn't matter that there is no load.

As a test, try disabling the second converter first, and see if the first converter startup normally. If it does, then manually enable the second converter without the MCU. See if it starts up normally.

Good idea! I'll update when I get the chance to try this.

 

Update as promised...

Interesting one this. If I apply the powerbank output to a passive load (a 12V halogen bulb which draws around 0.75A) and monitor the voltage and current, they are absolutely rock solid for as long as the powerbank lasts.

If I apply the output to my circuit with the converters, it runs fine for a short while, then dies. The current monitor shows a reverse spike at that point. So I put a Schottky diode in series. This improved matters a bit. Then I put a large (2200μF cap) across the supply which improved it a lot, but...

The attached picture shows what is happening now. The driven device flashes a small LED about once every 1.25 seconds, which are the peaks of current that you can see (about 15mA). Around every 30-40 seconds, the power spontaneously dies, down to a low level, then recovers and continues. The MCU keeps running, however.

When the light level drops to the point at which the main LEDs should switch on (this is a moth-attracting light that comes on when it gets dark!), they come on as they should - now consuming around 750mA - and the current and voltage remain completely steady!

I'm lost on this one!