Is gas the roasting nirvana?

I realize I’m preaching to the choir here but would be interested if any of you came to the Bullet by way of a gas fired or fluid bed roaster and could comment on some talk I have seen in other forums about gas roasting coffee being the only way to bring out the best. Obviously some of that may be due to the lag of standard electric element roasters responding to changes. I’ve seen a few fluid bed roasters claiming they have the better tasting coffee because well, no smoke.

But is this really just like the BBQ community where the old school stick burners square off against the pellet grill or propane guys ? Is it just machismo that real coffee roasters deal with flaming machines to go along with their beards and tattoos ? I’m sometimes feeling seriously under dressed as I have neither :wink:

Would be interested to hear before I go poke some ant hills…

If I don’t post in a while, you know they got me.


A guy I worked for decades ago advised that picking a fight with a zealot was like wrestling with pigs: you’ll only get dirty and the pigs love a fight on their terms. I took that lesson to heart and would advise you ignore the bait! :slight_smile:


1 Like

The BBQ guys are funny. Full disclosure I migrated to a pellet grill . I also use a washing machine instead of a Rock in my local river. I used to drive stick but now have an eight speed auto. I’m used to dogmatism. I was genuinely curious if there was any real data or science like gas oven afficianados say their cakes are moister because when gas burns it generates moisture. So would you want that in coffee beans you just dried ? Seems fake science…I think like a rumbling V8 these folks like the roar of a wide open gas burner. But I have no beard or tattoos to back up that claim!

Thanks for having my back Bruce :joy:

1 Like

Gotta fight for what’s right… at a safe social distance!! :mask:


I got a real chuckle out of this thread with the beards, manual transmissions and masked emojis …read it as my Bullet was cooling down just now.

I’ve never roasted with a gas roaster, but had plenty of coffee from them. I don’t think I’ve ever had better coffee than I roast myself (being totally objective, of course). Or maybe I just don’t have educated taste buds.

My real roaster is a Diedrich IR-1 roaster that runs on natural gas. I do have many years worth of experience roasting on that machine and results are predictably consistent, and what I would call excellent.

My experience with the Bullet is quite limited. My results on the Bullet, to date, are very inconsistent and my best results are a pale shadow of what I get regularly on my Diedrich.

I am able to easily reproduce the temperature profiles I get on the Diedrich, with the Bullet. What I can’t reproduce is the final result, e.g. roasted coffee quality. And this is with only 1lb/454g charges. I shudder to think what I would get with larger coffee doses.

For me, the jury is out on the Bullet, being entirely honest.

Hi Ken,

Appreciate your comments - I’m trying to find out what objective evidence there would be for “gas is better”. My grandmother swore her cake making went downhill after her gas stove was replaced with an electric one and I know some BBQ pitmasters put a tray of water in with their meat for “moisture” but given part of the coffee making process is drying the beans, that seems counter productive.

So is it the rapid change in power/heat that gas affords in order to change course that electric elements can’t match ? Possibly. Is it the over BTU reserve that some gas roasters can tap into ? I’ve heard some tales that you should only roast to 90% of the stated capacity of most machines but on the other hand someone did BTU calculations on one machine and established it was techinically overpowered by 100%. So does a V8 engine work less hard than a V6 ? I mean they can both get to their destination but…

Just trying to see past any smoke and mirrors and also the self justification filter - I wouldn’t spend Diedrich money on a home setup. So in that sense is the Bullet “good enough” compared to someone who uses a heat gun in bread machine and thinks I’m crazy spending the money I did on the Bullet, again because their coffee is “good enough” or “best I ever drank”. Love that last phrase, and maybe that’s the key…

In the interests of being quasi scientific (and for my own peace of mind) I still from time to time buy one of the “better” grocery brands that are local to me and have a roasted on date within 30 days to see if “I’m all that” with my own output and so far, I feel I hold my own.

But maybe we’re comparing apples with fish - does a profile (or even the Rao Rules) apply well or equally to the more diminutive mass (thermal and otherwise) of the Bullet with the induction heating. It doesn’t necessarily have to be built hulky to deal with the onslaught of a gas burner.

I don’t pretend to have any answers, I just thought it was an interesting line of discussion as a change from “does my curve look good in this”.

Maybe the dog bowl heat gun guys have it right - if it tastes good to you that’s all that matters.

Disclaimer: The previous musings were not powered by caffeine. I’m going to fix that now with another Control Brew of Cuvee El Salvador compared to my El Salvador from earlier this week.

I doubt it is as simple as “gas is better;” it may have little to do with the actual fuel source, other than the obvious observation one has with most gas ranges or cooktops vs. electric ones; it’s easier to fine tune gas and it comes on strong really quickly, compared to electric. At first blush I’d say that the differences are due to the thermal mass of the roasters and their drums, combined with the thermometry that each roaster offers. In the Diedrich the drum temperature may be the same or nearly the same as the air temperature measured by the probe, and the drum is so heavy and carries so much heat that the presence of the beans doesn’t reduce the drum temp very much. This may not be the case in the Bullet and in the earlier phases of the roast the drum may be cooler than the air temp, helping to make the bean temp less than the measured air temp that is imputed to be the bean temperature. I’m guessing here.

It could be that using other parameters than the reported bean temp would lead to a better roast product.

I should add that I go between two houses, one my primary residence in the mountains, and the other in the big city where my work takes me. The Diedrich is at the primary residence, and when I am there I tend to have a lot more free time, hence can use a roaster fairly easily. The Bullet is at my work residence and when I use it my time is limited hence I don’t do a lot of roasts and don’t do them very often.

Two of my most recent roasts might be illustrative. Both were with an extremely fruity, small beaned, DP Ethiopian I bought from SM/Coffee Shrub, Humbela Buku (?). I have produced delicious and enticing results on my Diedrich but nothing good on the Bullet with this bean. Both of the Bullet roasts I am describing had essentially the same profile, other than that one of them was carried to a higher final temperature. The parameters are below:

(1) Drum preheated to about 180 C (~355F).

(2) Steady rise to first crack at about 9:30 minutes at around 375F, which was largely a guess since this bean’s first crack is just about inaudible. On my Diedrich I hit onset FC a bit earlier, maybe 8:45 to 9:00, but this is hard to accomplish on the Bullet.

(3) I aim to end the roast just before 2nd crack commences, within 4 to 4.5 minutes after onset FC, so total roast time is on the order of 13:30 to 14:10: These parameters are achieved by adjusting the heating element intensity and the fan on the fly, mostly by increasing the heating element intensity to near maximum and the fan to near minimum before onset FC, and thereafter moderating temps and increasing fan speed so as to draw out the time between onset FC and end of the roast.

(4) the first roast finished at around 419 F (215C) and looked quite light in color for what it should have been, showed no oil, and lacked flavor and was hard to grind 3-10 days later in actual use. My go to grinder is an HG-1 hand grinder with Mazzer Robur 83 mm burrs in it, and you can tell how hard a bean is when you grind it by hand! Weight loss from roasting was about 16.5%, which is similar to what I get on the Diedrich. The difference was that the beans in no way resemble the beans I roast in the Diedrich that lose 16.5% of their weight in the roasting process, they seemed very under-roasted.

(5) I took the second roast 6 degrees F higher, to 425F (218.3C) finishing bean temp, on account of the results I got on the earlier batch. I never roast this dark normally, this was for educational purposes given the results above. The beans came out very much darker than I am used to, and by day 3 had some oil spotting. Grinding them was like grinding charcoal, and the taste was similar :slight_smile: Weight loss from roasting was a whopping 19.5%.

These are only two of the roasts I have done but they illustrate the results I am getting. There is no way that accurate thermometry would show a 6 degree F difference between these two batches, with weight loss percentages of 16.5 vs. 19.5 %. There is no way that observed thermometry and roast timing throughout the roasts would be similar, other than for 6 degrees F, and maybe 30 seconds in overall time difference, and in one case produce beans that were very undeveloped, and in the other that were like mini-charcoal briquettes.

That’s where I am at the moment.

1 Like

Ken, do you have the IBTS on the Bullet ? From your preheat temp I’d say not…?

My bullet is brand new, just received in the last batch from SMs, so I think YES. I have only used the “bean temp” so far. I am finding the Roast Time Software to be very lame, although I have been controlling the roaster through my PC to this point (would prefer to stop doing that, actually).

That would make me think it has the IBTS. Mine was a retrofit and came with notes that the preheat temperature should be increased compared to the old. You didn’t mention your batch size - I’ve switched (back) to Artisan so attaching a PDF from a recent 900g Ethiopian. My charge temp was 310C. As you know those Ethiopian suckers can be pretty dense. My usual aim is DE around 4-4:30 with a FC of 8. My Bullet is raised to accommodate the cooling tray from CoffeeTea so I find that puts the bottom of the door at around ear level and since then, hearing first cracks (even on this one) have greatly improved. I would say FC is usually around 195C.

After a couple of days to chill, the fruity Ethiopian seems on point…

Ethiopia Banti Balo_20-05-05_1053.pdf (59.8 KB)

Ken, I sure don’t have the experience you have, so am reluctant to suggest something to you…but on the Bullet, I preheat a 454g charge at 230C for most roasts and think it helps a great deal…takes about 23 minutes to preheat the roaster to this level on the first roast of the day…maybe this would help overcome the smaller drum mass that is one of your thoughts. I roast a lot of Ethiopia DP.

Thanks very much for the suggestion. I was taught to preheat the Diedrich to above 400F (like 410-415, which is a little less than your recommendation). With the Diedrich you then turn off the heat on the drum for a minute or two and let the temp drop to 400 F, then charge the roaster. I was going by the recommendations in the Bullet manual for preheating by charge weight. Perhaps this is the issue. It’s worth a try and I’ll try to do a batch today. I won’t be staying at this house long enough to try the coffee, so I’ll have to take it with me to my primary residence and try it there.

Thanks again.

You’re trying to emulate a roast profile from a roaster that is scaled differently because of its mass. So with less mass in the drum, that means that, for a given Preheat temp, less heat is stored for when the drum is charged. To compensate, a higher preheat temp would make sense as the drum temp will drop quicker when the cold beans hit the metal than it would in the Diedrich. Bit of a balancing act- trying to store the same amount of heat in less mass requires a higher temperature.


I found some time today to roast on the Bullet, using 454g batches and preheating the drum to 230C as suggested above. I had intended to do only one batch, but since the results were so disappointing I did two others. The first batch, unfortunately, was the beans I was hoping to learn how to roast on this roaster. These are Ethiopian DP Humbula Beku (?SP) recently sourced from SM and Coffeeshrub (may still be on their list, I don’t know). The first crack, at least, is basically inaudible so I went with my most usual onset 1st crack temperature from my Diedrich IR-1 Roaster, which is around 375 F. I took the beans to 419F, which where I usually end my roasts, which on the Diedrich, at least with the beans I use and at my altitude for the Diedrich is around 5850 feet of elevation (1783 meters). This is normally at the end of 1st crack but before onset of 2nd.

Ethiopian DP Humbela Buku 454g 5/24/2020

The end product weighed 371g, for a loss in excess of 18%

I should add at this point that I have never had a good roast product (to my taste) where the weight loss from roasting has been less than 15%, or more than 17%, or 16% +/- 1%. That doesn’t mean that every roast with a loss in this range is good, but rather that a weight loss of 16% +/-1% is a necessary but not sufficient requirement, at least for espresso. Perhaps if I roasted coffee for use other than as espresso, or if I preferred really darkly roasted beans, I’d have a different set of parameters.

After this experience I decided to tack on a couple of extra roasts. I have an unrelated real estate business relationship with a fellow who owns a coffee roasting business that I have never frequented and about which I have no opinion. I happened to meet up with him this past week and he gave me a tour of his roasting plant and once he knew that I roast my own coffee, he gave me a couple of pounds of 3 different green coffees in his inventory. I know nothing about these particular coffees and have never tasted them roasted by his hands. So I have no expectations, good or bad, about what I might accomplish with these beans. I decided to roast two of them, if for no other reason than to get a little bit more experience with the Bullet.

The next bean I roasted was his Guatemalan Mira Linda, which was obviously wet processed. I decided to take it to a lower temperature based upon the weight loss I got from the Ethiopian DP coffee above. I terminated the roast at 415 F, which is only 4 degrees F short of what I took the first Ethiopian to. This coffee lost 13.6% of it’s weight (starting weight 454g, ending weight 392g), which was similarly unsatisfactory in my view. Here’s the roast:

It was kind of astonishing to me that 4 degrees F, perhaps 30 seconds of a roast at the end, could change the weight loss from roasting so much. Of course, it is a different bean, so we may be comparing apples to oranges.

The third roast was an Ethiopian Sidamo, not otherwise specified, which was obviously WP by appearance. I decided to split the difference in final roast temperature, so I took it to about 417 F. This coffee roast product weighed 385g, or a weight loss from roasting of 15.2%, which would put it into my self-defined potentially acceptable range. Once again, it’s a different bean and comparisons among these 3 roasts might be fraught with potential error.

Here’s the roast profile:

I haven’t done this sort of an experiment with my Diedrich, but I would be surprised if a difference of 4 degrees F among 3 roasts, from 415F to 419 F, which would take 30 seconds, maybe, would produce a range of weight losses from roasting between 13.6 and 18.3%. That just seems huge to me.

What seems much more likely to me is that the thermometry I’m observing is not reflecting the reality of what is going on in these coffee beans I am roasting. Perhaps the bean temp measurement is not accurate for the way that I roast. Perhaps I am being too “hands on” in my roasting, trying to control the heating element and the fan with fine control that looks like I’m accomplishing something but in fact it is all observational noise, i.e. the signal to noise ratio is not good.

One thing that was obvious however is that preheating the drum to 230C definitely made the roasts go faster than they did before when i preheated to around 180 C. These roasts all finished about a minute to 90 seconds faster than my typical roasts on the Diedrich, which doesn’t necessarily make them bad, just different.

I will start to taste these coffees on Day 2, 2 days from now. My assumption is that the first and 2nd will be nearly undrinkable, and that the last might be drinkable if the coffee used was worth roasting, something I don’t know at this point.

If anyone has any comments on this flight of ideas, they are welcomed, and thanks in advance!

Hi Ken,

We’ve had a lot of discussion recently about whether the Rao RoR curve best suits the hardware/scale/technology of the Bullet compared to the hulky machines :wink:

But from my perspective (and I don’t hold that out as an authority view) common to all your profiles is a huge loss of momentum just after a flick around FC. It looks like the combination of dropping power and ramping the fan way up may be to blame. Again maybe this is something that differs in the hardware. Do you need to pilot a yacht different to a speedboat ? Does a Cessna handle like a 747 ?

I would suggest cutting back on the maximum fan speed and drop the power a little before FC to reduce the drastic cliff. Then after FC, in combination with a less drastic fan increase, don’t drop the power out as low (P2). Induction is more responsive that plain electric elements and the thermal mass of the Bullet is lower so it probably recovers a little quicker but it may not be as fast as your Diedrich gas. So try and think a minute ahead. I would also separate out the power drop and fan increase so they don’t coincide. Give them 20-30 seconds breathing room.

As I think someone else already said, it’s probably best to develop a modified roasting style for the Bullet. Certainly it was something I had to reset after spending some time coming from a Behmor and then a HotTop. I burned through some coffee learning/observing the what ifs. And just because you have lots of things to twiddle, doesn’t mean you need to. You may want to drop the drum speed a tad to D7 or D8 with the 450g charge.

As they said on that great coffee show Galaxy Quest, never give up, never surrender…

Hi Stuart,

I think what it comes down to is whether or not the temperatures one gets on the Bullet (I’ve been using “bean temp,” but there is also IBTS) are accurate, or if not accurate, at least consistent. If the temperatures are in fact accurate and/or consistent, it shouldn’t matter at all how you get them. There are a limited number of parameters you can change with any roaster, not all of which are present in every roaster.

These are:

(1) Charge Temperature;
(2) Charge Weight;
(3) Heat Input;
(4) Ventilation.

Some of these may be fixed in some more limited roasting setups, especially the ventilation.

Other than for such things as chaff elimination, it shouldn’t matter one whiff whether you effect a change on the temperature in the roasting chamber via increasing/decreasing ventilation, vs. decreasing/increasing heat input. The only change I can think of where this might matter would be in a hypothetical situation where smoke isn’t being evacuated from the roasting chamber and that the excess smoke might effect the taste of the resulting beans. I have no knowledge that this is true in any roaster, but I guess it is possible.

Obviously, you do not want to stall the roast, although I have yet to read a precise definition of what constitutes “stalling a roast.” The idea is that you want the temperature to continue to increase throughout the roast; if you do that, you aren’t stalling the roast. Now you might “flatten” the coffee by taking too long to roast it, otherwise called “baking” the coffee. You also might roast the coffee too quickly, which can certainly make it unsuitable for use in espresso, and at some point unusable for any brewing method.

What I am doing with the fan and the heating element is very much intentional, because I am trying to slow down the roasting process once first crack commences, from what it would normally be if you were to proceed with the same heat input and ventilation you used to get to the beginning of first crack. I’ll link to an ancient online post I participated in eons ago that explain the philosophy at the end of this overly long post. There are others but they all state the same idea.

The funny thing about the comment on needing to anticipate what the roaster is doing in advance is almost a joke (to me) when it comes to the Bullet; in comparison to what I am used to, driving the Bullet is like driving a speedboat; roasting on my Diedrich, is more akin to piloting an oil tanker. You need to think 2 minutes ahead, minimum, with the Diedrich because it has so much thermal mass and so much power that there is a long delay between your input changes and actual changes made to the roast progression. With the Bullet, I find that you can stop it on a dime; the fan and heating element are very responsive compared to what I am used to.

The reason why I try to slow down the roast once 1st Crack commences, is that I want to get 4-5 minutes more roasting time in before the roast is terminated, and I do not (intentionally) roast beyond the end of first crack. This is because blind tasting experiments I did with another forum participant (Jim Schulman) on HB were absolutely definitive in the effect that this has on roasting coffee for espresso. I have been responsible for a number of blind tasting experiments related to coffee back in the days when I used to post online on coffee sites (not for the last 6 or more years). Most blind tasting experiments are rather confusing and often hard to pick out differences between the conditions being tested. In this case, the other taster and I could pick out coffee roasted with intervals of 2.5 minutes after onset FC, vs. 4 minutes, 100% of the time. That’s absolutely astounding. Just compare it to any blind tasting you might have done at some point with wine. It’s incredible. It’s a real difference in taste.

I can’t find the original post we made on this experiment, but it is discussed on this page:

Ancient HB Thread

So that’s why I’m doing what I am doing after onset FC. I’ve successfully done it for a number of years on an old 1lb knockoff Probat sample roaster, and another 9+ years on my Diedrich. I can’t think of any reason why you couldn’t do this on the Bullet as well.

For me it comes down to whether I can believe the thermometry on the Bullet, and if so it shouldn’t matter how I get the profiles I do. When I first got my original 1lb sample roaster, I had to hack it repeatedly to get a thermocouple into the right spot to be both reliable and repeatable; once accomplished, I could do whatever I wanted with the heat input and the results followed the thermometry, not the heat input.

If I can’t believe the thermometry on the Bullet, then I need to find some formulaic manner of roasting with it and just consider the thermometry to be an imprecise indication of where I am in the roast. I haven’t yet figured out which of these things is the truth.

Hi Ken

We’ve come a long way in this thread from the starting point “is gas better” haven’t we.

I don’t disagree that IMHO the Bullet responds very quickly and more nimbly than your average electric roaster, like the aforementioned Behmor or HotTop. I appreciate compared to the average production roaster those things appear like toys. But we all got to start somewhere. Mine was a FreshRoast SR500 about five years ago because an Amazon Gift card was burning a hole in my wallet.

So I’d suggest that one of the reasons the Bullet can stop on a dime is partly the induction heating and partly the fact that it isn’t lugging an oil tankers worth of thermal mass around. So that said it’s no surprise to me that if you throw out a boat anchor that is P2 at the same time as cranking the fan up to F7, the graph is going to show what it shows. You jammed on the brakes when a squirrel ran out on the road. You didn’t slow anything down that I can see, you came to a dead stop - driving the 18 wheeler, the squirrel would be an ex-squirrel. But the Bullet doesn’t carry that level of thermal mass AFAIK.

Now I agree, does that necessarily mean the coffee tastes bad ? I don’t know. You say so.
But you have repeated the same experiment a few times and so far I think you said the coffee was less than drinkable. The fact that your graph shows similar signatures each time you repeated the same behaviour, to me implies the Bullet is recording consistently. They all show the same “signature”. So you now know what bad coffee looks like. Change one or two variables and see how the coffee tastes then look at the graph. I wasn’t saying that getting a Rao RoR 10 was going to fix your bad coffee. There are some who would argue a perfect curve equals the best coffee QED. Maybe such a perfect curve on the Dietrich gives you the coffee you seek. I wouldn’t know. If you’re able to share a graph from Artisan on a good Dietrich roast, I’d be very interested to see it.

Full disclosure, I don’t roast for espresso, I’m a pour over guy.

@jacob do you want to throw in some observations and rationale on the IBTS vs the rest of the world ?

Hi Stuart,

My Diedrich IR-1 was one of the first few produced, and for a short while they did not require the purchase of the electronics package with the roaster, which would have greatly increased its cost. I did not get the electronics package and have no way to easily log my roasts. I operate the roaster entirely manually, adjusting the gas input and the damper, in response to the displayed bean temp from the TC in the middle of the drum; the damper controls where the airflow goes, either 20:80 to drum:cooling tray, 50:50 to the drum:cooling tray, or 80:20 to the drum:cooling tray.

When I roast on the Diedrich I do more or less exactly what my logged roasts on the Bullet show, with the caveat that one has to anticipate far out with the Diedrich, hence the gas input is declining over the last minute before onset FC, and then massively reduced once the FC starts. The temps flatten out a bit more slowly on the Diedrich, but the roasting temp curves are fairly similar to what I’ve posted here with the Bullet.

Part of my observed results probably have to do with the specific Ethiopian Bean I have been roasting (mostly) in the Bullet. The beans are tiny, the smallest I’ve ever roasted from any origin, the FC is entirely inaudible to my hearing on either roaster, and the flavors in the roasted bean are quite unique, extremely fruity and complex. When roasted well, which I have accomplished every time on my Diedrich, the espresso shots are among the best I have ever had. I did roast another Ethiopian DP coffee in the Bullet once and the results were actually pretty good. The roasting profile would have been more or less the same as what I have posted here for the other coffees. Unfortunately, I have had intermittent problems with the Roast Time software, and that particular roast was not saved. I should probably roast a few others in my stash of Ethiopians I know to see if it is just this particular bean (Humbela Buku) I’m having trouble with.

Hi Ken -

Since I don’t have anything close to your experience with large scale roasting I wasn’t going to jump in here, but we do have one thing in common: a new V2 Bullet. Mine is from the spring 2020 batch from SM. So here are a couple of suggestions, take or leave as you see fit:

I’ve learned that drum and fan speeds don’t necessarily correlate to older Bullets, so take care when comparing and trying to reproduce other’s profiles. My calibrated F3 fan shows 1080 rpm on the Info panel, which corresponds to about F6 uncalibrated, which you might get with older machines. I initially made the mistake of assuming I was using calibrated fan settings, but the display in menu showed “—” not “PPd” so I had run the calibration procedure but not activated it. Similarly, my D9 drum speed is 74 rpm (measured by marking the drum shaft and counting with a stopwatch).

The point of this is that a very cool feature of the Bullet is that we’re supposed to be able to share roast profiles. Mostly by luck, I managed a truly fantastic roast of the Ethiopian Organic Dry Process Sidama Keramo from SM which you can see here:

If you are lucky enough to have any of those wonderful beans, in theory, you should be able to download that profile, see it appear in your Roast Time list, and replay it on your machine.

Since you mentioned Hambela, I also roasted a pretty tasty Ethiopian Dry Process Hambela Buku 13 Screen from SM here (although next time around I’d drop it a few degrees C earlier):

The only other suggestion I’d dare offer is that the IRBT temperature sensor seems very valuable. First crack temperatures seem very repeatable, even across different batch sizes where the bean probe temps don’t match. So I’ve been focusing on IRBT when anticipating FC events and setting recipe actions.

Hope this helps!

  • Brad