Maximizing convection in the Bullet

Hey everyone,

I’m looking to maximize the convection in the Bullet and have been considering blocking the rear inlet to see if it would direct more hot air through the roaster. Has anyone tried this?


Link to airflow in the Bullet:

I read a post from Aillo that F1-F3 would increase the roast temp by picking up more heat from the drum. Starting at F4 you start to get a cooling effect. Hear’s a similar video I posted on air flow cutaway.

Aillio has responded to some other posts about blocking the air intake vents and says it is a bad idea because it will interfere with the proper cooling of some of the Bullet components.
I was looking for a way to increase convective heating too, so I did a little experimenting and measuring the exhaust air temperature. Here is a post I did with the result.
Bullet Behavior - Convective Heating
Even at F3, convective heating occurs for only about 3 minutes. At this point the IBTS shows the bean temperature above the air temperature inside the roaster. This would mean that the airflow through the roaster is now drawing heat away from the beans. Faster fan speeds cause more cooling. I haven’t tried F1 or F2 with this, but I suspect that any increase in internal air temperature wouldn’t make a big difference. So, I decided to accept that the Bullet is predominently a conductive roaster that cannot be easily changed. If you want a bit of cooling, then airflow is your friend.

Thank you for measuring that. It seems like there would need to be a lot more heat in the air for convection past 3 min.

Did you also measure it with different fan settings?

You’re welcome. You can see a fan change to F4 toward the end of the roast I posted. The ExT drops about 5F and then keeps on rising again. I looked back at some other roasts where I had a change from F3 to F4 and then again from F4 to F5. It shows about a 5F down step at each one. Bear in mind that my ambient temperature was 63F which is fairly cool. This might make bigger steps than say an ambient temperature of 75F. You’re right in that the inside air temperature would have to be a lot hotter to contribute to the bean temperature. On the sample I posted, maximum ExT was about 375F which wouldn’t get you even close to first crack temperature. I haven’t tried any really big fan changes to see what happens. The example is for a 450g batch size. A much larger batch size would probably drive the peak ExT up some, but I think it will always lag behind the bean temperature and thus provide a cooling effect. By the way. When I preheat, I watch the ExT and the BT readings. For a PH of 437F the BT will level off at about 300F and the ExT about 325F. This makes sense because the drum heats the air and the air heats the front plate and BT probe. The front plate loses heat to the outside air and probably won’t ever get up to the inside air temperature. I use a fan change mostly about 90sec after first crack in order to move the steam and CO2 out quickly as well as to get more chaff out.

Interesting measurement and observations, and yet there must be a level of convective heat transfer happening to explain why roasting at D9 results in faster roasts which tends to imply that there is convective heat transfer happening when the beans are in contact with the hot air in the roast chamber for longer.

A way to maximise convective heat transfer in the bullet is to limit the roast to smaller batches and only roast at D9.

Smaller bean batches have less heat capacity and hence suck less heat from whatever hot air is in the chamber while D9 ensures maximum agitation and maximum bean to air contact for max convective heat transfer.

If the Bullet is an inefficient convective heat transfer machine, larger roast batches will only make this worse by quickly sucking out what ever convective heat capacity the machine is capable of.

For what it is worth, I limit my roasts to 300g for this reason and also because it is what we tend to consume in a week.

I could be very wrong of course but these are just my thoughts :blush:

Some good points. Thanks for adding to the thread. Let me pose some alternate thoughts. Then, I think I’ll measure a few more roasts with higher drum speed to see what turns up. I mostly roast at D8.

I don’t think roasts at D9 are faster because of increased convection. I have seen discussions by (I think) Rob Hoos about higher drum speed lofting beans higher, resulting in increased exposure to hot air. But he is probably talking about a more classic drum roaster that can keep the internal air temperature significantly hotter than the bean temperature. The Bullet cannot do this. Remember, once the bean temperature exceeds the internal air temperature, there will be convective cooling and not heating. So, increased lofting would actually cause a bit more cooling.

Increased drum speed also increases the contact time of the beans with the drum. Since the Bullet is majorly conductive, I think that it is the increased contact that imparts more heat to the beans and speeds up the roast.

I also don’t think that batch size has the effect that you describe. According to my measurements (batch size 450g), the beans will absorb heat from the internal air for only about 3 minutes. At this point, the beans have gotten enough heat from the drum to exceed the internal air temperature. Heat flows from hotter to cooler, so the internal air is now receiving heat from the beans as well as from the drum. You can see this in the graph of ExT because the trace is sloping upward, that is, getting hotter. Larger batches have more heat capacity. The mass of air inside the drum is pretty fixed. It seems to me that larger batches, containing more overall heat energy, would actually heat the air more than smaller batches. This could also tend to minimize the cooling effect because the heat capacity of the larger batch is affected less.

I may already have some data to see if this is true, but will need to pull it together in order to post it.



Thanks Jim, you make good points too. For what it is worth, if according to your observations, the bullet is capable of a maximum of 3 minutes of convective heat transfer, that is almost 40% of the total roast time which is material for small batches where the average batch roast time is between 7 to 8 minutes.

I am mindful that after a certain point n the roast, the chemical reactions turn exothermic and the beans internal temperature rises by itself. From that point on, cooling the roast helps manage this reaction not going out of control.

When considering this, 3 minutes of convective heat transfer at the start of the roast is fairly significant especially for smaller batches.

Just my thoughts :blush:

Pushing hot air into the roaster and reducing the cold air intake improves flavor…. But it will destroy all the plastic bits, and there is a lot of plastic in the airways. And yes, you are breathing that if you roast indoors.

Yeah. Good observation. The way I see it, the convective heat early on helps get the roast started well. Since a lot of airflow isn’t needed yet, a slow fan speed should help extend the convection zone a bit. Once the roast is under way, the drum has already responded to additional power and a little more fan would help to purge out the steam from the G/Y phase. Into the MAI phase, the most important thing to me is adjusting the RoR to a good level for first crack. Here power changes and fan changes can be used. I tend to use power reductions ala Morten Munchow. Once first crack starts, purging steam and hot gases is important. Also, the beans are more sensitive to heat so more airflow can help control the post first crack temperature rebound.

I did look at the ExT behavior of a 600g batch size and didn’t see much difference in the maximum ExT achieved,

As an aside, one way to increase convective heat might be to use a higher PH temperature. This should raise the initial internal air temperature some. The downside is, of course, increased risk of tipping or scorching.

Have you seen any good discussions of the pros/cons of conductive vs convective heat with regard to flavor? Not that we could do much about it with the Bullet, but it would be good to know.


Thanks for sharing Jim. I attended an online workshop by Scott Rao a while ago and he mentioned that you need convective heat transfer to get the delicate floral/fruity notes in the roast.
He seemed to suggest that it is harder to get these delicate notes with conductive heat transfer roasters for some reason.