Modified Firestarter
Modified Firestarter
Hi all, just posting this here looking for some feedback.
I tried the original Firestarter and had fairly good results with it. I changed the fan speeds to 16100 and 15600 to achieve quick cool down without blowing beans into the chaff collector.
I thought I'd see if I could do better by adding boosts and a corner but apart from being quicker I don't think it's better flavour-wise. Can anyone tell why that might be from looking at these logs? 14 is the original Firestarter and 17 is my modified one.
In both cases two days rest seems to be optimal.
I tried the original Firestarter and had fairly good results with it. I changed the fan speeds to 16100 and 15600 to achieve quick cool down without blowing beans into the chaff collector.
I thought I'd see if I could do better by adding boosts and a corner but apart from being quicker I don't think it's better flavour-wise. Can anyone tell why that might be from looking at these logs? 14 is the original Firestarter and 17 is my modified one.
In both cases two days rest seems to be optimal.
- Attachments
-
- log0014.klog
- (90.44 KiB) Downloaded 1123 times
-
- log0017.klog
- (85.86 KiB) Downloaded 847 times
- kaffelogic
- Site Admin
- Posts: 554
- Joined: Thu 18 Oct, 2018 9:38 am
- x 81
- Contact:
Re: Modified Firestarter
Hi Simon,
Thanks for sharing your modified fan speeds. It is good to have your findings and I am sure others will appreciate the ability to roast without so many beans among the chaff.
I believe that by adding boosts and a corner you have indeed improved the curve following of this profile. You can see from the log that it follows the line better, and that it suffers less oscillations. The early boost adds energy to the roast and this would be consistent with bringing first crack earlier.
However, you are ending the roast after a lengthy development time, and I would expect this later part of the roast to influence the flavour more than the early part. In other words, caramelisation and roast notes can be expected to dominate. Your changes have not really made much difference to the later part of the roast.
If you were stopping the roast closer to level 3 you might be more likely to detect a difference in flavour as your changes have made a difference to the onset of first crack when volatiles are being vaporised and the acid/sugar balance is changing rapidly.
Thanks for sharing your modified fan speeds. It is good to have your findings and I am sure others will appreciate the ability to roast without so many beans among the chaff.
I believe that by adding boosts and a corner you have indeed improved the curve following of this profile. You can see from the log that it follows the line better, and that it suffers less oscillations. The early boost adds energy to the roast and this would be consistent with bringing first crack earlier.
However, you are ending the roast after a lengthy development time, and I would expect this later part of the roast to influence the flavour more than the early part. In other words, caramelisation and roast notes can be expected to dominate. Your changes have not really made much difference to the later part of the roast.
If you were stopping the roast closer to level 3 you might be more likely to detect a difference in flavour as your changes have made a difference to the onset of first crack when volatiles are being vaporised and the acid/sugar balance is changing rapidly.
Re: Modified Firestarter
Thanks Chris. The results are very consistent thanks to the real-time control.
I've been working my way down the darkness scale, with some improvement. I also have a different blend of green beans to try next!
I've been working my way down the darkness scale, with some improvement. I also have a different blend of green beans to try next!
Re: Modified Firestarter
I'm keen to try this profile. Do you guys think this profile does a better job than the default Kaffelogic one for espresso? If so, what roast level have you found to give the best result? Any bean recommendation?
Re: Modified Firestarter
Actually I've found a modified soft brew profile is giving very good results. The mod is essentially to drop the rate of rise right down...
That and keeping the roasted beans in an airtight glass container. Seems to work well for most blends.
That and keeping the roasted beans in an airtight glass container. Seems to work well for most blends.
- kaffelogic
- Site Admin
- Posts: 554
- Joined: Thu 18 Oct, 2018 9:38 am
- x 81
- Contact:
Re: Modified Firestarter
Yes, and you don't need to be afraid of dropping the rate of rise down to zero. The temperature probe is influenced by air temperature as well as bean temperature. The actual bean temperature will continue to rise for a time after the probe temperature RoR drops to zero.
Re: Modified Firestarter
Hi Chris,
Yes I thought it might. The air temperature oscillates quite a lot as the RoR approaches zero. Can you think of any settings tweaks that might reduce the oscillation? In PID terms, perhaps it's underdamped.
Yes I thought it might. The air temperature oscillates quite a lot as the RoR approaches zero. Can you think of any settings tweaks that might reduce the oscillation? In PID terms, perhaps it's underdamped.
- kaffelogic
- Site Admin
- Posts: 554
- Joined: Thu 18 Oct, 2018 9:38 am
- x 81
- Contact:
Re: Modified Firestarter
Hi Simon,
You may be able to get some improvement from better tuning. If you switch to engineer mode you can set zone multipliers for Kp and Kd. You could experiment with increasing Kd, perhaps multiplying by 2.0 initially? (There is no multiplier for the integral term because experiments have shown that the PID algorithm we use is destabilized by any amount of integral term.)
There are other factors to consider that might come in to play. The bean mass oscillates in a slow cycle between full fluidisation and glass flow at the airspeed normally used in the Nano 7. Increasing air speed might even out some of the oscillations, although you can only do this if you have some spare heater power available and that in turn depends on supply voltage being close to 240V. It may also be that these oscillations are detected at the probe due to bean mass oscillation, but are not as great at actual bean surface temperature. This is to say that 1. you might not be able to get rid of the oscillations by PID tuning, and 2. they might not have a great adverse effect on flavour because the beans do not experience the same oscillation as the probe does. To a certain extent my statements are speculative and you are, once you have activated engineer mode, free to experiment with quite fundamental changes to the settings. Getting a 'better' line in the log also might not manifest itself as better flavour: lots of cupping is required.
You may be able to get some improvement from better tuning. If you switch to engineer mode you can set zone multipliers for Kp and Kd. You could experiment with increasing Kd, perhaps multiplying by 2.0 initially? (There is no multiplier for the integral term because experiments have shown that the PID algorithm we use is destabilized by any amount of integral term.)
There are other factors to consider that might come in to play. The bean mass oscillates in a slow cycle between full fluidisation and glass flow at the airspeed normally used in the Nano 7. Increasing air speed might even out some of the oscillations, although you can only do this if you have some spare heater power available and that in turn depends on supply voltage being close to 240V. It may also be that these oscillations are detected at the probe due to bean mass oscillation, but are not as great at actual bean surface temperature. This is to say that 1. you might not be able to get rid of the oscillations by PID tuning, and 2. they might not have a great adverse effect on flavour because the beans do not experience the same oscillation as the probe does. To a certain extent my statements are speculative and you are, once you have activated engineer mode, free to experiment with quite fundamental changes to the settings. Getting a 'better' line in the log also might not manifest itself as better flavour: lots of cupping is required.
Re: Modified Firestarter
Thanks Chris. The slow oscillation is interesting and would certainly account for what's going on. I'll try increasing the Kd multiplier first, and if that doesn't have the desired effect I'll try increasing fan speed. Might have to roast at about 3 in the morning to be confident in getting close to 240V though!