A little gem - The Yaorea YRG01 (Review & Mod)

What I am referring to are the spikes that you can see in the graph below. This is caused by the output filter capacitor basically dying. Traditional linear power supplies simply used AC to transform the voltage down, rectified to DC and then smoothed with the output filter capacitor taking the 50Hz lumps and bumps out of the DC output. Low frequency and could be large values, so gives a super nice smooth output. Modern SMPS use high frequency DC pulses into a small transformer, then the DC pulses are run through half wave rectification (only half is needed as this is pulsed DC) and then smoothed with a much smaller (than a linear PSU) capacitor that is tuned to the frequency of the switching pulses. As the frequency is high (a few KHz) these capacitors get hot, then eventually dry out, reduce capacitance and cause the DC pulses to make their way through to whatever it is powering. Add more current to the mix, the capacitor gets hotter, lasts less and when the DC pulses make their way through they are often more powerful than if a small current is pulled. Generally, they are pretty crappy for running sensitive equipment, unless you spend a lot of money on one. I have some good SMPS's. One is just like any normal inline power supply. It's 18v at about an amp. It is medical grade and cost me about £70 IIRC. Another is a Traco 5v 3a metal cage internal chassis type power supply. That came in a weather station display I bought for modification. Again that is an expensive SMPS, maybe around £50-60. Both of these offer excellent filtering and isolation from mains, hence their price. A cheap £6.99 wall wart (most USB chargers for example) use the minimum of components as they are designed to just charge a battery. Using these types on sensitive equipment is a backwards step. And not all SMPS are equal, by a long shot, so best to check the ripple as you want close to 0mV. ~10mV should be acceptable. 30mV, I'd throw it away.

In the graph you see all the little spikelets that are not coming from the geiger tube, but from the power rail, due to a bad PSU. It had been in service for around 8 years, which is really good. It was a reasonable quality PSU too from some HDMI over CAT6 converters I installed some time back and were surplus. Eventually though it's capacitor dried out a bit, wasn't smoothing properly, and those DC pulses made their way into the Geiger counter (not a cheap one either - a NetIO GC10.) If you look at the bottom graph, the long term trend, you can see at the right the counts have elevated from the normal. This was simply due to the power supply dying. So that is why I suggest a good quality PSU, or batteries. 

 

 

I have tested the power consumption of this counter in pretty much all configurations. There is a little bit of ambiguity between having a battery fitted and full charged and not having one fitted. I think this is due to some current draw from the battery as well as the 5v input, and why it occasionally tops up the battery. Also enabling clicks appears to add more current draw in certain configurations than others. I did not test any consumption with the disco LEDs on. Here are the results:


During LPM the BLE and ESP are always off. When running on 5v USB with a full battery, it will occasionally draw up to an additional 22mA to top up the battery.
The BLE when enabled adds another 4~5mA across all configurations, and the ESP adds another ~95mA across all configurations.

5v USB input - No battery fitted
LPM ~10mA / ~10mA (with clicks enabled)
RPM ~55mA / ~60mA (with clicks enabled)

5v - full battery
LPM ~4mA / ~4mA (with clicks enabled)
RPM ~60mA / ~78mA (with clicks enabled)

5v - empty battery
LPM ~450mA / ~452mA (with clicks enabled)
RPM ~478mA / ~480mA (with clicks enabled)

Battery Only 4.2v Full
LPM ~6mA / ~9mA (with clicks enabled)
RPM ~57mA / ~77mA (with clicks enabled)

Running on a full (4.2v) battery, with everything enabled it is going to draw ~177mA, so for uploading to radmon.org, or MQTT or something else it would be prudent to run this on a 5v supply permanently. Keeping the battery in place offers a nice ~14 hours of backup runtime. Turning everything off and letting it drop into LPM consumes around 6mA, so should theoretically run for around 18 days on a full charge and whilst won't be uploading to the internet, it will be logging the counts and storing them on the device itself.