Review: BrewPi Spark

[molc]

As discussed elsewhere on the forums, here is a review of the BrewPi Spark that might be interesting for the front-page. There is very little information out there on these devices and I couldn't find a build guide anywhere, so I'll focus on that information, along with a comparison of data with an STC fermentation in the same environment.

What is it
Brewpi Spark is the next hardware version of the popular BrewPi system, as previously reviewed in http://www.nationalhomebrewclub.com/wordpress/review-brewpi-fermentation-temperature-controller/. Now, instead of the Arduino shield setup, a spark core is used for the control algorithm, along with a new colour touchscreen and custom enclosure as one package. The previous hardware is no longer supplied and the developers are switching over to this version as the standard configuration.
The new hardware will still allow for controlling fermentation, but will also have support for controlling your herms or multiple fermentation chambers, though the software is not at that point yet. If you buy one of these devices at the moment, you are in the alpha customer group, so be forewarned.

What do you get
You can buy the enclosure and optionally the spark core. It's a pretty simple device, much smaller than you expect from the images on the website. The version that was shipped had a simple hardware test firmware loaded, though I believe now they are shipping with the full control interface enabled.
It's worth mentioning, as the site doesn't make it clear, that the enclosure also comes with three connectors for the outputs, so you can just add in basic wiring to connect these up.
The software at this point is worth a mention. This is a very early alpha version and still needs quite a bit of work. The flashing procedure, while recently simplified, still caused a lot of issues and needed a bit of debugging to get going. Also, the Spark, while being sold as wireless enabled, must still be physically connected to your brewPi server, so bear this in mind. I had a raspberry pi lying around, so I used that instead of my house server which I had planned on using instead.


What do you need
The online store is quite complete and will sell you all the extra components you need. At a minimum, you'll need three onewire temperature probes with RJ11 connectors, as well as two SSR's to control your heating an cooling circuits. Also, they recommend you measure the beer temperature with a thermowell in your fermentor, as they feel a probe on the side is not accurate enough to get a true reading. You can also put the probe directly in your liquid, but the plastic will eventually absorb some of your wort and probably lead to an infection over time.
For my setup, pictured below, I picked up a 60W tubular heater and a standard thermowell that I fitted to my lid using some silicon tube and an o-ring to keep a clean seal.


Control Circuit Build
This part is actually quite poorly documented. I sketched out a circuit that you can build which involves no modification of the fridge and instead just creates two plugs that are controlled by the SSR, all coming from a single power source. In the diagram, the branch point of the wires is handled by a simple coupler that you can pickup in any DIY shop. The enclosure is a simple extra deep plug socket, with a blank panel. The SSR's I got were rated to 10A, so all the wiring and connections are made to the same level. Note, the input labels for the Spark are wrong, as this diagram started with a herms idea, but the idea is the same for  any two control circuits sharing a common power source


Here's the final unit built. Quite a neat package and I haven't noticed it getting warm in usage so far, so it looks like the SSR's are dissipating heat correctly. Unlike my STC setup, everything can be unplugged from the spark core, which makes it a breeze to move pieces around as required. On brew day, I have separate temperature probes that I use in my mash tun and liquor tank, which makes things a real breeze.


Logging
Once you get upgraded to version 0.28, this all works as expected. Do not use a lower version, as the number of bugs that were present with the touch screen etc will drive you bonkers and interfere with logging if the spark or pi restarts. The interface has been documented in the past, so it needs no more said in this review.

Comparison with STC1000
This is the whole reason for setting up this device. While my STC should have done the job correctly, I was seeing some odd results and getting off flavours which pointed to issues with temperature control. In a previous brew, I could see that the fridge and the heater were fighting each other, which lead to some crazy temperature spikes.
In both charts, the green and red lines are the ones to pay close attention to, as these are the beer temperature.

This is how my STC looked, using a brewbelt and a probe stuck to the side. You can see the temperature ramps by about 1C everytime it heats and quite the zigzag on the data. I can see almost a 2C swing in heating, though the fridge kept initial fermentation under fairly tight control, which should reduce off flavours a lot. Also, to stop the STC fighting itself, I had to plug out the heater for initial fermentation and the fridge for the later, otherwise it was just going nuts. Lastly there are a good few gaps in the data, where the 0.27 version of the firmware popped up a calibration screen and stopped logging. That version of the software just was not ready for proper use, hence the next test.

Now, slot in the brewpi and do a more complicated ramp profile:

Here, the data says it all - accuracy to within .02C at almost all times. This is only a fermentor filled with water and also I'm not using the brewbelt anymore, but still, the accuracy of the system is just astounding. Also, this is the latest release of the software and I've found it is working without a hitch. If the power drops at any point, logging and control resumes automatically.
I hope to do my first beer with this next week, with a belgian golden ale, so I'll post up some more graphs of how that went with a exothermic fermentation rather than water.

[Garry]

Nice work molc. I reduced the image sizes to fit the screen better. Let me know if you don't like it!

[molc]

Looks good to me. I noticed that as well after posting. It looked perfect on the phone!

[Garry]

Yeah, the phone browsers seem to be a bit cleverer at fitting images to the screen!

[phoenix]

Great review, well done.

I looked at this solution about two months ago but the lack of clear documentation/information put me right off, this answers some questions.

Do you think its the algorithm in the Brewpi software which is giving you better control here?

I went ahead and started to build my own solution for both fermentation and mash control using a Raspberry Pi, DS18B20 temp sensors and an ENER011 Power Management System (to cut out that scary SSR stuff). Ill document it at some point.

[molc]

Do you think its the algorithm in the Brewpi software which is giving you better control here?

There are two parts to the answer, as I changed the environment as well (I need to do a water test with the STC controlling instead to give a true benchmark), but yes, I feel the algorithm is giving great accuracy. The brewbelt only used to heat a portion of the liquid before, whereas now I uniformly bring the chamber to a target temperature. Also the probe is now in the liquid, which again gives better accuracy.
For the algorithm, check out the blue line (fridge temp) and the orange line (control temp) to see how the algorithm is adjusting the fridge temperature. It is much more predictive than the STC, which is simply hitting a target temperature. Since liquid moves slowly, the STC has a real swing in the heating and cooling that you must predict to prevent.

The algorithm is public knowledge and the control variables are visible in brewpi at all stages if you want to implement it yourself.

Also, the SSR's, once you realise they are just simple switches, are not that scary. The problem is the brewPi creators just assume you understand how to wire them into a circuit and a basic knowledge of electrical wiring. That said, my wiring has one flaw - it assumes both control circuits will not be active at the same time. If that were the case, and the draw was at the theoretical maximum of 20A, I would be overloading the plug and wiring. To fix this, you would either wire two inputs or up the rating of all the components and source to match.

In the end of the day though, the brewPi gives you fermentation data. Coupled with a gravity sensor, you can use it to fix problems, which you can't do otherwise. I'd argue even just getting it without temperature control so you can see the temperature profile of your fermentation. Then, if you enter your beer in a competition and they say you got hot alcohol or something, you can quickly pinpoint if that hot summer day or a bad yeast pitch was the reason.

[phoenix]

There are two parts to the answer, as I changed the environment as well (I need to do a water test with the STC controlling instead to give a true benchmark), but yes, I feel the algorithm is giving great accuracy. The brewbelt only used to heat a portion of the liquid before, whereas now I uniformly bring the chamber to a target temperature. Also the probe is now in the liquid, which again gives better accuracy.
For the algorithm, check out the blue line (fridge temp) and the orange line (control temp) to see how the algorithm is adjusting the fridge temperature. It is much more predictive than the STC, which is simply hitting a target temperature. Since liquid moves slowly, the STC has a real swing in the heating and cooling that you must predict to prevent.

The algorithm is public knowledge and the control variables are visible in brewpi at all stages if you want to implement it yourself.

I missed your tubular heater there, great idea, could you link me to where you got it?

Ill have a look at the source code later to see how I will implement this myself.

Whats the third DS18B20 for? Room temp?

Also, the SSR's, once you realise they are just simple switches, are not that scary. The problem is the brewPi creators just assume you understand how to wire them into a circuit and a basic knowledge of electrical wiring. That said, my wiring has one flaw - it assumes both control circuits will not be active at the same time. If that were the case, and the draw was at the theoretical maximum of 20A, I would be overloading the plug and wiring. To fix this, you would either wire two inputs or up the rating of all the components and source to match.

You are still switching mains electricity with lethal voltages here, no? I have zero background in this stuff and feel switching mains power in my rented apartment is not the place to start.

The ENER011 is a nice solution, connect it to the Pi via USB and switch the sockets on/off via the sispmctl app from a python script. I picked one up on Amazon for 40 quid.

http://sispmctl.sourceforge.net/

In the end of the day though, the brewPi gives you fermentation data. Coupled with a gravity sensor, you can use it to fix problems, which you can't do otherwise. I'd argue even just getting it without temperature control so you can see the temperature profile of your fermentation. Then, if you enter your beer in a competition and they say you got hot alcohol or something, you can quickly pinpoint if that hot summer day or a bad yeast pitch was the reason.

Absolutely agree, its data is invaluable, ive already used it to find the optimal spot in my tiny apartment to ferment. My solution is to use plot.ly to visualise the data, its powerful stuff. Adding a local weather temp stream etc would be fun too - see the outside temp alongside your sensor data.

http://www.capitalbrewers.org/ciaran/plot.html

After Ive rigged up my fridge this weekend I'll be looking to control my induction plate for designed, stepped mash control. Does the BrewPi Spark do that? I couldnt figure out if it was possible when I last looked at the website.

Also, gravity sensor? Does such a thing exist?

[molc]

I missed your tubular heater there, great idea, could you link me to where you got it?

http://www.ebay.ie/itm/261473462454?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649

Whats the third DS18B20 for? Room temp?

Yes, room temp. It helps the algorithm learn I believe. It's interesting to see when the fridge goes on, how the room temperature goes up:)

Absolutely agree, its data is invaluable, ive already used it to find the optimal spot in my tiny apartment to ferment. My solution is to use plot.ly to visualise the data, its powerful stuff. Adding a local weather temp stream etc would be fun too - see the outside temp alongside your sensor data.
http://www.capitalbrewers.org/ciaran/plot.html

I use highcharts here in work for charting and it's excellent. My whole job is data visualisation and analytics actually, so I love this stuff. The weather stream can be got from openweather.org and easily tied in through a database though.

After Ive rigged up my fridge this weekend I'll be looking to control my induction plate for designed, stepped mash control. Does the BrewPi Spark do that? I couldnt figure out if it was possible when I last looked at the website.

Also, gravity sensor? Does such a thing exist?

Well, you can do timed temperatures, though I've never tried to do it for less than a day. The future versions, which spark was created for, allows you to do stuff like that with the herms setups though I believe.
As for the gravity sensor, it's on their list, but they don't have it currently. I'd say it's a pipe dream for them.

[Shanna]

Hi Molc,

Any hints where you bought or fabricated the thermowell?

Shanna

[molc]

Hi Molc,

Any hints where you bought or fabricated the thermowell?

Shanna

http://www.ebay.ie/itm/231208452163?var=530429124587&ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649

Yes, I got the damn thing shipped from the states, as I couldn't find a supplier here that made one long enough to go in from the top of the fermentor. They are unbelievably hard to source for such a simple component :/

[phoenix]

Wonder would a few be interested in a group buy to at least save on the postage

[molc]

Well if you can find a friendly stainless welder, it's a simple thing to make. It's just a tube sealed at one end. For us, you don't even need a screw connector, as it's going into the fermentor lid anyway.

[Shanna]

Hi Molc,

Any hints where you bought or fabricated the thermowell?

Shanna

http://www.ebay.ie/itm/231208452163?var=530429124587&ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649

Yes, I got the damn thing shipped from the states, as I couldn't find a supplier here that made one long enough to go in from the top of the fermentor. They are unbelievably hard to source for such a simple component :/

Yes I agree it seems pricy/overkill to ship from the states. I have a few ss tubes lying around that I might try crimping and soldering with tin solder to seal.

Shanna

[pob]

Well if you can find a friendly stainless welder, it's a simple thing to make. It's just a tube sealed at one end. For us, you don't even need a screw connector, as it's going into the fermentor lid anyway.

I got a couple of 10mm & 8mm x 600mm s/steel tubes from eBay in the UK for approx €20ea. Got someone to cut into 2, crimp & weld the ends, work perfectly.

I then found a s/steel racking cane on Hop & Grape for £10 (when exchange rate was a lot better). I got that cut in two aswell, it just needed the bend straightened slightly to fit probe it.

However these may be an alternative Thermowell 6-200mm €23.66 for 10 incl shipping = €2.37ea

You just need a suitable bung for ~€2-3

What's really annoying is these 15"-Stopper-Thermowell were only €6ea shipped in December, now shipping is €23!! Which is the same price as the MoreBeer one shipped from US.

[Paul B]

While my STC should have done the job correctly, I was seeing some odd results and getting off flavours which pointed to issues with temperature control. In a previous brew, I could see that the fridge and the heater were fighting each other, which lead to some crazy temperature spikes.
In both charts, the green and red lines are the ones to pay close attention to, as these are the beer temperature.

This is how my STC looked, using a brewbelt and a probe stuck to the side. You can see the temperature ramps by about 1C everytime it heats and quite the zigzag on the data. I can see almost a 2C swing in heating, though the fridge kept initial fermentation under fairly tight control, which should reduce off flavours a lot. Also, to stop the STC fighting itself, I had to plug out the heater for initial fermentation and the fridge for the later, otherwise it was just going nuts.

I'm interested in why you were getting such poor results with your STC setup. Haven't seen any similar complaints so far... any chance this could be improved by tweaking your differential/compressor delay settings?  What do you set these to out of interest?