While waiting for the eParrot PCB’s I ordered in China a question was asked on a Dutch forum how to automate a still. Members of this forum were already experimenting with a stepper operated needle valve. I remembered seeing a couple of years ago a nice solution on the HD forum using a peristaltic pump to control the distillate flow from the still (
https://homedistiller.org/forum/viewtopic.php?f=17&t=43172&p=7122379) but the prices of these pumps were rather high at the time this was presented. Today you can buy various types of these pumps on AliExpress for affordable prices.
So I order two different ones to experiment with, a GAB55 head from GRO-THEN, which I combined with a NEMA 17 stepper motor (in total about 25 euro) and a INTLLAB RS385-635 that costed 5 euro.
The GAB55 head was a failure, not enough flow and mismatching and out of center motor shaft profile in the roller assembly.
The cheap INTLLAB pump was a success, enough flow, low noise and low power. The vendor send a second pump (unpaid) under the same tracking number. Nice! Thank you!
At the same time another member of the Dutch forum did a test with his Franzmaster automatic still to see the influence of the stabilizer vessel right under his column (basically a bubble ball). The Franzmaster computer can log to a SD card and the resulting graph showed that with the stabilizer vessel the still is stable, and without unstable. The reason of this is the ABV concentration dependency of the volatility of the fusel oils, at low ABV the volatility of the fusel oils is higher then that of ethanol, at high ABV it’s lower, effectively trapping the fusel oils in the column. This reduced the effective length of the column and can trigger hiccups that result in high ABV ethanol ejected from the top of the condenser. (See also this threat on the HD forum (
https://homedistiller.org/forum/viewtopic.php?f=1&t=42227)
So ketel3 and I made preparations to test these pumps and to see if the removal of the fusel oil at the bottom of the column made any difference in the behavior of the still and the quality of the output.
I made a test box that could control two pumps, record the temperature value in the middle of the column, and record the flow and volume of the fusel oil and the distillate. To measure the last two parameters little magnets were mounted inside the rollers of the pumps and Hall sensors were used to detect the passing of those magnets while the pump was running. An Arduino was used to display the quantity and flowrate of the distillate and fusel oil en to show the temperature in the mid of the column. The sensor used was a SMT172 TO18 with a resolution of 0.01 ºC The collected data was stored on a SD card. At the start of the test the fusel oil was withdrawn with about 10 cc/min, later this was lowered to 5 cc/min.
After the test the conclusion was that the operation of the still was stable, the pumps operated flawlessly and the removal of the fusel oil at the bottom of the column gave a huge distillate quality improvement. Even the last jar that we squeezed out of the boiler was of acceptable quality. There was no influence noticed from the withdrawal rate of the fusel oil on the temperature mid column.
One of ketel3’s girlfriends works in the perfume industry and has a very good nose. Se sampled all of the distillate and fusel oil jars:
Wash was based on a of 11% sugar wash with a couple of cans of grape concentrate:
Distillate (5536 cc total ABV > 92%):
jar one: nail polish
jar two: not bad
jar three: sweet and useful
rest of the jars: excellent
last jar: faint hint of something but still acceptable. (200cc jar containing the very last of the ethanol in the boiler, ABV probably below 90%).
This batch was much better then previous ones distilled without the fusel oil side streaming.
Fusel oil (866cc total about 12% ABV)
All of the fusel jars had a very disagreeable odor, first one smelled like cognac, and one (jar 5 or so) had a rum caramel smell. Most of the jars made her nearly faint. Last fusel oil jar was tasteless water.
Only after seeing the graph’s of the logged data the very slow reaction of the vapor temperature at the top of the column on the withdrawal rate became clear, something that puzzled me during the test.
To be continued in the next post…….
