Lots of people ask 'how accurate is the MyBrewbot SG estimate?' Well, we thought it was about time we showed the data.
We're working on the next generation of MyBrewbot which will link to the Tilt to both control temperature and record SG data and publish to your Smartphone so we rigged it up and brewed a Raspberry Weiss in a Fermentasurus to collect comparison data.
As it's a pressure ferment we cranked the temperature up to 21C. We wanted to see whether the MyBrewbot algorithms would cope with a fast ferment. The Weiss started with an OG of 1048 and a target of just under 1010.
As you can see from the chart, it was a quick ferment, getting to target in less than 3 days. MyBrewbot has done a good job of estimating the SG (the black plot). The Tilt SG reading is shown on the light blue plot. Both the Tilt and MyBrewbot took readings at 1 minute intervals.
MyBrewbot estimates the SG based on the difference between the ambient temperature (the red plot) and the beer temperature (the darker blue plot). That temperature difference is used to compute the heat generated by the yeast as it ferments the beer and from that the consumption of glucose and the change it makes to the SG.
The screenshot below shows the Tilt data captured by MyBrewbot. We are still working on the interface and plot scaling so that it can be shown relative to the beer's temperature and target
We made sure that MyBrewbot was set up correctly. This is really important to get an accurate estimate.
The ambient probe in 1 litre of water was well away from sources of heat and cold so that it picked up the true temperature inside the fermentation fridge.
The beer probe was attached to the side of the Fermentasaurus with blue tac , insulated with bubble wrap and held in place with a couple of wraps of clingfilm around the whole vessel to minimise any external temperature interference..
The Original Gravity and Target Gravity were both entered into the MyBrewbot interface and all previous data cleared, we were good to go.
I have included some pictures below of the rig with the temperature probe attached and the 1 litre bottle of water with the ambient probe.
In conclusion, we were very happy with the results. We have always had great confidence in the data for a non-pressure ferment but we were delighted to see that the thermic activity algorithms stood up to the fast, pressurised ferment and gave as good accuracy as a normal fermentation.