Testing the canonical variable boost circuit using the MT3608 High Efficiency Step Up Converter.
The MT3608 (parts also paroduced as the B6286) is a very efficient boost converter that can deliver up to 24V at 4A. It requires only 6 external passive components, and is readily available as a complete module for as little as $0.40.
In this project, I wanted to build up the standard variable converter circuit from scratch and verify its performance.
- 2V to 24V Input Voltage
- 1.2MHz Fixed Switching Frequency
- Internal 4A Switch Current Limit
- Internal Compensation
- Up to 28V Output Voltage
- Automatic Pulse Frequency Modulation Mode at Light Loads
- up to 97% Efficiency
The M3608 comes in a miniscule SOT23-6 package. If I was building this circuit for a real application, I would design an SMD custom PCB. I might still do that, but it would very much be purely for education & fun, as the total cost would probably be over 50 times commercially available modules.
So for this build I’m going ugly style on a hand-prepared copper PCB.
The datasheet recommends a 4.7µH to 22µH inductor with low core loss at 1.2MHz. I’m using a 22µH CDRH104R SMD power inductor.
22µF input and output ceramic capacitors are recommended. I don’t have any ceramics of that capacity on hand, so I’m using through-hole electrolytics. They are rated for 50V.
A low forward-voltage schottky diode is recommended. Seems like a 1N5819 would be a good choice, but I don’t have any available right now. I’m using a 1N4148 instead - not ideal, but satisfactory as reverse breakdown voltage is suffificiently high at 75V.
The feedback voltage-divider with two resistors establishes the output voltage level where Vref is 0.6V:
Vout = Vref * (1 + R1/R2)
The lowest output voltage is in fact not that low. It should be limited to around 1 diode drop less than the input voltage.
Enable Pin Connection
The enable pin is active high. If the chip should always be on, the pin can be connected directly to the input voltage. Since I want to test the enable functionality, I connect it with a 10kΩ pull-up resistor, so I can gound the enable pin to turn off the output.
When running with
Vout=16V, the enable pin is pulled-up to Vin with a 10kΩ resistor.
Grounding (pulling down) the enable pin disables the converter. But this does not cut output. It just removes the voltage boost, and output passes directly through the inductor and diode to output (less a diode drop).
Vin=4.97V, Vout drops to 4.40V when disabled (given my choice of and diode).
Note: one of the disadvantages of most commercial modules is that they tie the enable pin to Vin so the enable functionality is not available (without cutting and patching the board traces).
Performance - Build #1
I’m not getting efficiencies anywhere near what the datasheet claims, but I suspect this is mainly due to the low currents I am testing at. My component selection might be responsible for most of the losses: the forward voltage of the diode, and I’m not sure how well the electolytic capacitors perform at 1.2MHz.
The module under test with
Vout=12V and a 10kΩ load.
Most of the current is drawn to power the voltmeter I have attached on Vout:
Performance - Build #2
I received some more appropriate components (1N5819 diode and 22µF ceramic caps) and built a new board. Interestingly, results are approximately the same, although the diode does allow for a lower voltage drop.
The maximum voltage I’m getting (38.2V with a 10kΩ load) is way over spec - SW Voltage maximum is 30V. I still ran the test and everything seemed to survive the short over-voltage.
Performance - Commercial Module
I compared the performance of a commercial MT3608 module from a seller on aliexpress. The module has very similar parts selection to my DIY builds:
- 100kΩ pot and 2.2kΩ for the feedback voltage divider
- SS34 SMD schottky diode
- Enable pin tied directly to Vin
- 22µH SMD inductor
- ceramic caps - apepars ~15µF on the output and the input cap appears to be higher, but I can’t measure it reliably in-circuit.
Performance is slightly better. This may be due to the layout which follows the guidelines in the datasheet very closely.
Here’s a notional breadboard layout:
Build #1 - Ugly-style on Copper Stock
- 1N4148 diode
- 22µH CDRH104R SMD power inductor
- 22µF through-hole 50V electrolytics
Layout for an ugly-style build on some copper PCB stock with mainly SMD components:
Build #2 - protoboard with Better(?) Components
- 1N5819 diode
- 22µH CDRH104R SMD power inductor
- 22µF ceramic SMD capacitors