This is a review of an inexpensive MH-1210 Chinese bang-bang thermostat that was purchased on eBay for around A$13 complete with thermistor sensor and postage.
Above is the thermostat. Note that the middle terminals, the 230VAC power input terminals are labelled ONLY in Chinese. It is rated at 10A for resistive loads (AC-1), but it is doubtful that AC-3 motor loads of more than 20-30% of that should be used.
The label clearly states this is calibrated for an NTC 10k B=3435 thermistor. More on that later.
Above, a view of the interior.
Note there is no power transformer.
Above, an interior view showing the main MCU and display driver chip on the back of the display daughter board. The MCU is a Chinese Holtek 8-bit MCU with 12bit ADC. The display driver is again Chinese, a TM chip.
Observations
The design is transformerless. This is good and bad, but safety is better for a design that uses a suitable transformer to isolate the power supply from the electronics and external sensor. There is a diode from power supply (4) to NTC (5). Essentially, there is a low impedance path from (4) to (5), so the internals and thermistor need to be insulated for line potential.
Test
The thermostat was tested with a pot containing 1l of water and a 120W immersion heater. The sensor was placed 25mm to the side of the heater element.
Above is the logged temperature of the pot with setpoint=65 and differential=3. There is significant temperature error, but the observed range of 5° fits well with the specified differential, allowing for some thermal delay.
The display was woefully inaccurate, around 6° at 70°. That prompted a check of the thermostat with 10k and 1k0 0.1% resistors in place of the thermostat, and calibration constants calculated (Two point thermistor calibration).
The readings were 25.0° and 87.0 respectively. Thermistors come is common sizes, and 10k/3950 is very common and is probably the correct thermistor for this thermostat, not 10k/3435 as labelled.
If the thermostat was in fact calibrated for 10k/3435 as labelled, the 10k 0.1% resistor should have read 99.5°, so it is 12.5° in error near 100°, way out of spec.
Quality
The calibration failure is symptomatic of cheap Chinese hardware flooding eBay. The defect is overcome with purchase of a 10k/3950 sensor for about A$2.
I cannot say that I am too happy about this particular transformerless implementation. I have others that use a transformer… but you cannot tell when buying them how they are made. There are many products with similar specifications, almost identical appearance, but different internals.
12VDC conversion
I am sufficiently concerned at the safety of the thermostat that I have bypassed the switch mode AC power supply to power it directly from 12VDC.
Above is a minimal conversion. R11 and L2 are removed D1 is bridged and a link from the front side of L2 to the rear end of R11. The original power terminals are + (3) / – (4) 12VDC (no polarity protection). Note that thermistor terminal (5) is connected to -ve (4).
The reworked rating plate.