IoT water tank telemetry project – part 4

Boost converter

The 4-20mA loop used for the water sensor needs a source of 24VDC. This is obtained from the battery (3.6-4.3V) using a DC-DC boost converter which runs only when a measurement is made.

This article discusses related issues.

B6287K boost converter

This converter is available on eBay for ~$1, and looks deceptively simple and would be a good candidate if it worked. In fact visual analysis gives a strong hint about its suitability.

I have been unable to find an English datasheet for the IC used on this converter, and specifications for the module vary from seller to seller, a common but frustrating situation with Chinese product.

This particular implementation was difficult to adjust, and it did not make the maximum output voltage specified by the seller, no surprises there. When adjusted to 24V output, it worked down to 2.5V input.

The application is for a 24V supply for a 4-20mA instrument loop, and it will be powered from as little as 3.3V, needs to start quickly, and produce a clean stable output.

Initial tests were that the output is noisy, and little surprise as there is no output filter capacitor. A 10µF 50V electrolytic was added to the output to act as a filter / reservoir capacitor. It improved the noise considerably, so the cheats have built the board without necessary output filtering.

Above is a capture of the on/off control waveform (cyan), and the Vout from the converter powered from a 5V regulated power supply. It works ok, starting quickly, but the output is a little noisy.

Now let's try it on 3.3V.

Above is a capture of the on/off control waveform (cyan), and the Vout from the converter powered from a 3.3V regulated power supply. Output load current is about 12mA, less than the 20mA that may be demanded. The output is low (levelling off at 15V), ripply and unsuitable.

Above is a capture of converter on a 5V supply, on/off control (yellow), Vout (cyan) and Vload (magenta) which is the input voltage ‘read' by the MCU. There is a lot of ringing on Vload and it seems to be induced by the ripple on the converter Vout. It is worse on 3.3V supply, so dismissed as unsuitable.

MC34063 boost converter

This converter came from Sure electronics, it has a schematic and the well known chip has a datasheet. These were purchased more than 10 years ago and I doubt the modules are currently available.

Above is the converter. These are available in buck and boost configurations, this one is a boost.

Above is a capture of converter on a 3.3V supply, on/off control (yellow), Vout (cyan) and Vload (magenta) which is the input voltage ‘read' by the MCU. Vload settles at 1.75V, and so loop current stabilises at 11.7mA. To allow for slower start at 20mA loop current, the firmware is configured for 500ms measurement delay.

Importantly, both Vout and Vload are relatively clean without ringing or spikes.

Above is the same scenario with supply voltage increased to 4.2V (the high end of the battery range), and it is evident that the converter starts much quicker.

Above is a run with 3.3V supply and a 4-20mA simulator set to 20mA (evidenced by Vload=3.0V, Iloop=3/150=20mA). Vout has stabilised in 200ms, and apart from some early glitches caused by the simulator, voltages are relatively clean (considering the breadboard lash up).

LM2577 auto boost/buck converter

Above is an auto boost/buck LM2577 converter.

Above is the response on a 3.7V power supply.

LM2577 boost converter

Above is a LM2577 boost converter.

Above is the response on a 3.7V power supply.

XL6009 boost converter

The XL6009 appears to be a Chinese attempt to produce a competitor to the LM2577 chip, and it is very hard to find LM2577 modules on eBay, they are almost all Xl6009 though often deceptively presented as LM2577.

Specs 5-32V in, when adjusted for 24V out at 5V in, powered up at 3V in produced 50V out. Chinese junk, cannot be used.

Conclusions

There is a range in quality of component, and more importantly, design of small DC-DC converters on eBay.

Many are advertised under a headline that includes chips other than what they use, the Chinese are masters at the ‘bait and switch' fraud.

Most use chips that are very difficult to find English language datasheets for verification of claims.

The MC34063 and LM2577 boost converters are quite suitable to the project, and have withstood extended trials.