I have only been roasting for a short amount of time so I am looking to the pros for some insights. Here's a bit of background on my journey thus far:
Since I was looking to only roast about half a pound a week I decided to use a FreshRoast SR700 as the base for this project.
In order to accomplish my desired charge rate I had to modify the following: Fan
- The stock fan only allowed for a setting of 1-9. This was not ideal as each jump from each step was too large. In order to help create a stepless fan for better control and incremental options which helps create a smooth ROR. Heating element-
The stock heating element configuration was not able to heat the additional load quickly enough so my initial roast was very long or very quick to start and long to finish due to the fan settings needing to be lower to get the heat rising faster. Less airflow was giving me the temperature I needed but I was sacrificing the ability for a high fan setting at the beginning of the roast to extend my DE. I just recently figured out a few solutions to this issue this weekend. After some experimenting, I realized that by removing the built-in fuse and the bi-metallic switch combined with disconnecting the heater coil from the fan, (which was then re-routed back to the primary heater coil) so they can now both run at the same time. By doing so, I was able to increase output from 1520 watts (with major oscillations) up to 1580 at high and full fan before bypassing the controls. Now at 100%, I was able to get a higher range of 1600 to 1750 watts (with little to no oscillations). This modification was key to allowing me to extend my DE and increase the heat on demand when needed for FC. The issue now was that the stock BT temperature that the machine was outputting was so far off I needed another solution to control the temperature accurately. Phidget setup
- By using a Phidget hub (HUB0000_0 - VINT Hub Phidget), TMP1101_0 - 4x Thermocouple Phidget and TMP1000_0 - Temperature Phidget I was now able to get an accurate reading of the BT and the ambient temperature of my roasting environment. I could now use the roasters heating element "BT" (stock BT reading from the FreshRoast) and set that as my ET via Artisan. By doing so I am now able to use the ET curve to build a BT curve via the derivative of those functions in real-time (more to come on this below). Roast chamber
- The stock roast chamber allows for ~80g roast with a 90g max depending on how dense the bean and how long you want to roast. By adding a longer glass insert into the stock glass roast chamber I was able to roast 160g consistently with the modifications above and now with the proper airflow.
I am having a larger 500mm glass roast chamber made for me at the moment but I am currently using a few 3D printed parts to allow for up to ~350g roast with 260g being the sweet spot. This nets ~.5lb per roast after the weight loss. The issue now was the stock chaff collector top cap could not handle all the additional chaff. External chaff collector
- By adding a cyclone and a few 3D printed parts for high temp I was able to build an external chaff collector system that has removed close to 100% of the chaff during the roasting:
The cyclone on top of the roaster has been cut in half to become a "top" for the 3D printed part. This allows me to use the same connection on both sides and not have to fabricate another connecting piece:
I was able to start to produce some really great tasting roast with this setup. Now that I have had some time to roast more and read a few books I have been looking to see what makes one profile taste better than another (similar) profile.
My first few profiles were short to start and long to finish so I have adjusted the phases of the roast via software and some server-side processing on my local machine to help to automate the process of alarms. A combination of alarms (milestones) with log10/20 math for curve targeting has allowed me to plot points on a graph and see a theoretical curve of based on all the roast factors.
It's currenlty theoretical and not 100% accurate as of yet. This is mostly because the program is modulating heater output and the fan speed together in real time which then adjusts the Phidget BT temp that is needed, to the time of the total roast to achieve the desired RoR.
There's a lot more going on here to get all this setup and working correctly but for the sake of this post I wanted to give some contextual background so hopefully, now we can focus on the profiles and less on the modifications for the SR700. That will be another thread once I get everything dialed in 100%. If you have any questions about trying to modify an SR700 please PM me and keep this thread for general roast discussion. Here's an example from a roast yesterday:
One-second sample time, zero smoothing:
With optimized smoothing:
Here are a few of my initial questions:
1. Most say a 45/35/20 split for phase % is perfect. Why is this?
2. What is an acceptable RoR max?
3. What do the evaluation stats mean? Leathery, Grassy, Toasty, Acidic OK, etc..
4. Most say the last 20% of the roast should be 2.5 mins after the first crack why is this?
5. If you could design a perfect profile what would it look like?