The Bud Toaster - (currently: Model 14, version 3)

...better when the glass "bowl" area has been preheated. I don't think this produces vapor by conduction nearly as much as it reduces heat loss through the walls of the chamber...

Click to expand...

yes, i agree -- and i think that's a good way to express what is going on.

i find the operation/performance of this version Bud Toaster far different from what i had before when it was a more typical log vape -- fixed resistance/fixed voltage/15 minutes to get to temp). And my understanding of the physics is certainly no better.

For example, the Bud Toaster can hit vape temperature of 385F in 20 seconds (with fresh batteries) to 40 seconds (last session on a charge), but it takes another 60 seconds before a serious vapor cloud is produced. Then from 90 seconds on to the end of the session, the vapor density is the same.

But there's a prickly issue that i've been trying to understand:

The temperature setpoint is higher than what the thermometer displays until the final minute of the session. Specifically, when the setpoint is 388F, the thermometer will start at 383F (when the green LED starts to flash, which indicates the computer is reading at or above setpoint) and rise over the next 5 minutes to rest at 388F. At which point the vapor is pretty tasteless and much lighter, though still visible. If i extend the session the temp will stay at 388F and only light wispy vapor is produced.

So i always adjust the temperature, if needed, during the first part of a session. Different dank need different settings. Normally i never have to change the setpoint.

So heat loss has been stabilized as the glassware gets heat saturated inside the metal heat shield. Also, i think heat is being removed as the trichomes evaporate their THC goo -- evaporative cooling, or heat of vaporization.

But here's the dilemma -- the customer is going to complain that the temperature is not stable. And there's no way i can measure the difference unless the computer can "see" the thermometer reading. Because the green LED is blinking on and off during the entire session, meaning the computer thinks it is dancing on both sides of the setpoint temperature.

So i've got to figure out how to spin this as a useful, even desirable, feature. Or, fix it in the code. Somehow.

Hippie Dickie
But there's a prickly issue that i've been trying to understand:
The temperature setpoint is higher than what the thermometer displays until the final minute of the session. Specifically, when the setpoint is 388F, the thermometer will start at 383F (when the green LED starts to flash, which indicates the computer is reading at or above setpoint) and rise over the next 5 minutes to rest at 388F. At which point the vapor is pretty tasteless and much lighter, though still visible. If i extend the session the temp will stay at 388F and only light wispy vapor is produced.

Click to expand...

Hi Hippie Dickie,

I haven't followed the whole thread, but I find your project very intersting and would like to help if I can.
My understanding is that you built a PID temperature controller out of a 12f683 processor.
The controller input is from a k-type thermocouple combined with a max6675 (A/D converter).
The controller output in pulse widht modulation mode goes to a switching device (Q2) which drives the heater coil.
Both the thermocouple and thermometer probe are installed close together on the outside wall of the vial.

Is this correct?

That is correct.

The bare k-type thermocouple (bare) is touching a loop in the middle of the heater coil -- separated by a layer of polyimide, while the thermometer probe (insulated) is between two loops of the heater coil and touching the glass of the oven tube.

My basic premise for the PID algorithm is "there is a number" -- basically the duty cycle -- that will maintain a setpoint temperature. This number is dynamically calculated 3 times per second (fastest sampling rate of the MAX6675).

This number will need to increase slightly as the batteries loose charge. However, the number will also decrease (during a session) since the evaporative cooling effect lessens as the trichomes are vaporized.

It occurred to me earlier that i need to interrupt the session at several points to read the current number value to see if this is actually happening. If it is, i may be able to adjust the number in the algorithm to counteract the heat rise.

So if I understand correctly, the thermometer probe is a stand alone temperature indicator only and the thermocouple is actually the input to the controller?

Say you would like to adjust the controller setpoint at 388F, how would you do this?

the thermometer probe is a stand alone temperature indicator only and the thermocouple is actually the input to the controller?

Click to expand...

that is correct.

The "factory default setting" is a setpoint of 385F. Each click of Button1 adds 5F, each click of Button2 subtracts 4F.

So click Button1 + click Button2 adds one degree to setpoint.

There is a startup procedure (hold Button1 for 2 seconds until green LED comes on) that resets the setpoint to the factory default setting.

Ok, I see.

When you start a session from ambiant or room temperature, how long does it take to reach the setpoint (green led flashing)?

And once the setpoint is reached, does the green led keeps flashing?

Okay, there's overshoot and then the temperature settles to setpoint. 30 seconds to pass setpoint, but a total of 90 seconds before the green LED is blinking on and off.

i know when setpoint is reached because the green LED blinks. When i look at the values stored in eePROM, the measured temperature is bouncing about 4 to 8 units -- that is, the MAX6675 value is bouncing by 4 to 8 units.

Ok, I see. This seems like a reasonable control. The overshoot is pretty common in a feedback loop (as long as it does'nt start to oscillate all over the place) and since the geen led blinks shortly after indicating that we are at setpoint and keeping it at setpoint.

I think that the initial temperature difference between the thermocouple and the thermometer probe can be explained by the fact that the insulated temperature probe will need more time to reach the same temperature as the bare wire tc, also by the fact that both probes are not at the same measuring point. The bare wire tc will be much faster since it is directly touching the heating element while the temperature probe touches the glass(this adds a delay because the glass will also need more time to reach the coil temperature. Also the technology used for the thermometer probe might be different and this could cause an additional delay.

So if you add all these delays this could be what you are seeing in term of the temperature probe is ramping up till it to equilibrium.

What happens to the thermometer probe and the green led when you take fast a large hit? ... also when you take long slow hits?

So if you add all these delays this could be what you are seeing in term of the temperature probe is ramping up till it [is] to equilibrium.

Click to expand...

i don't think so -- this (the temperature probe ramping up) is a very short term effect, say, 10 seconds, not the 5 minutes of the drift up.

Overshoot does not cause oscillation -- i'm just trying to reach setpoint as fast as possible, so it's a tradeoff of overshoot vs time. i may damp it down a bit -- i.e. cutoff the max power on at startup by about 20F sooner.

But, yes, the thermometer is much slower responding to the temperature changes. i think of it as showing a thermal average, whereas the processor is seeing instanteous temperature (of the coil).

What happens to the thermometer probe and the green led when you take fast a large hit? ... also when you take long slow hits?

Click to expand...

fast hit or slow hit -- the green LED continues its normal blink rate. Being able to pump 12 amps into the coil can compensate for any instanteous cooling effect.

Just a taught, if you could read and display the actual thermocouple temperature, you could do without the thermometer. Maybe it's not possible. Anyway keep up the good work.

edit: didn't see the last 5 posts, this one is directed in reply to FlowersInTheAttic:

I'd say most of the power would be transferred by infrared actually. that's probably what 'seems' like conduction (the IR only heats the herb around the outside) it'd also be absorbed by the glass which would produce a bit of direct conduction.

Infrared is pretty neat, if you sit in a room with the walls at 30C, all the objects will eventually absorb enough IR and become the same temperature. all objects also emit IR (black body radiation) so they reach an equilibrium of absorbing and emitting IR at 30C.
Same thing happens in space, average temperature is a fraction above absolute zero, so any object sitting in dark will just loose heat by IR until it reaches the same temperature as the average.. there are no 'room temperature' objects around to emit IR back at you. that is why its cold in space even though there is no air to conduct heat away
You sit by an open window on a cold day and can feel the 'cold' from outside, its actually just a lack of IR
House insulation uses a layer of reflective film to reflect IR back the way it came, makes a big difference.

As for the preheating-the-bowl effect, as above, if you preheat the bowl it just means less energy is lost via the bowl.. by conduction and IR.

Yet another code re-write ...

So, last night i finally implemented a change to the PID algorithm that i've been thinking about doing for months, viz. to make the PID gains come from a table, instead of trying to come up with some lame-ass calculation based on the error.

i also thought a bit more about the error data being fed to the PID algorithm and realized, updating at 3 times a second, the error factors are really pretty small numbers -- again, ideal for table lookup.

(i also try to understand why i get so stopped by an approach, once i've come up with it -- and only occasionally do i realize that all ideas are just steps on the path to the perfect vape, no single one is worth crap) << but that's another topic

Anyway, after two tests (and a couple bug fixes) the table-lookup PID algorith is working much better than before. (Duh!). Plus, it will be super trivial to tune up.

For example, a hit drops the temp about 1F for about 3 seconds. And there's a 2F overshoot. And, damn, is there a clear path forward on this aspect of the design.

The code has been updated on my SkyDrive for those following along at home.

Great to hear there is more progress to the vape..I really like complicated stuff,algorithms,tables... science stuff // I've actually already started a fund which is what i meant to do long ago in order to buy a cloud.. but then my interests were redirected into the Bud Toaster .. and now i am even starting to think ,that it will get out before the Cloud does..
By no means i compare both in functions or appearance.. quite diferent breeds.. no doubt bout that, but with all read by now on both threads.. i firmly stand that i will rather shoot 500 $ on the BT rather than the VXC..

Didn't read much at all in the thread. Looked to computer related lol but anyways, so is this vape being sold? If not, when?

@damienm420 -- that is understandable, this thread is really a development diary and probably doesn't appeal to very many people.

And to answer your questions: no, it is not being sold yet (by me), and it won't be offered for sale until development is complete.

And the answer to the unasked question: the price will be $500 USD. Shipping is free.

weekend project: shows just how powerful a prototyping tool the arduino is
code: http://pastebin.com/TjJNJypZ (I used libraries to do all the hard stuff )
This is really just a alpha prototype made with parts I had lying around (Arduino, max6675, nokia LCD, a FET from the parts box and some nichrome wire out of a busted heat gun). It does work very well but has a couple of huge disadvantages due to the glassware used: It is way too hard to load up, the herb has to go between the heating test tube and the stem. There is also not a lot of room in there, thus the dosage is very low. I am still sourcing glassware which is closer to Hippies prototypes.
following images show some testing I did. using damiana for testing. (has been tested with other herbs and works as well )

The thermocouple is in direct contact with the herb.. which has the advantage of better temp control and no need for teflon tubes or high temp tape/glue. has disadvantages obviously, if the draw tube is removed it will overheat, its not very user friendly at this point..

I used some stainless mesh for filtering and allowing airflow around the bottom of the tube

temp profile showing warm up (about 30 sec) and a couple of tokes (time scale is 30 sec/division)

Hippie Dickie said:

@damienm420 -- that is understandable, this thread is really a development diary and probably doesn't appeal to very many people.

And to answer your questions: no, it is not being sold yet (by me), and it won't be offered for sale until development is complete.

And the answer to the unasked question: the price will be $500 USD. Shipping is free.

Click to expand...

Oh nice. When do you estimate you'll be near shipping?

@yayvape - nice work!

couple of comments:

-- whenever i see temperature swings like that i find my gains are too large

-- those are very large changes in the duty cycle value, my duty cycle value just ranges around a bias value of 8.5%, with a max of 9.5% and min of 6.5% - not those wild swings shown in your graph. Is your PID calculation based on zero? That assumes heating and cooling are symmetric, which they aren't.

-- you have a very well stocked parts bin!

@damienm420 - hopefully by 3rd quarter, but i really don't know.

-- whenever i see temperature swings like that i find my gains are too large

Click to expand...

I think its mainly because the thermocouple is in the actual airflow somewhat, this sort of peak is unavoidable with that measurement. BTW where did you find teflon tubing? what is it called? you are holding it to the outside of the tube just with pressure from the leads? have you looked into high temp glues? silicone etc

-- those are very large changes in the duty cycle value, my duty cycle value just ranges around a bias value of 8.5%, with a max of 9.5% and min of 6.5% - not those wild swings shown in your graph. Is your PID calculation based on zero? That assumes heating and cooling are symmetric, which they aren't.

Click to expand...

That duty cycle is scaled from 255 to 100, so the max output is about 40%, min is zero. I did this to avoid excessive thermal stress on my already cracked tube.. and 10A though that little coil doesnt go down well
I used the 'below' coefficient that you provided a few pages back, It shouldn't be too hard to add in the 'above' coefficient, I'll have to read up more on it, but I have not come across this as being an issue with heating systems before. What actual difference does it make to the system?

-- you have a very well stocked parts bin!

Click to expand...

lol thanks. mainly old scrapped projects, made a prototyping pcb for the max chip a while back when I was playing with the idea of making a reflow oven.
I do tend to hoard electronic crap too, never know when it will come in use.

BTW where did you find teflon tubing? what is it called? you are holding it to the outside of the tube just with pressure from the leads? have you looked into high temp glues? silicone etc

Click to expand...

do a google on "PTFE tubing". but i'm only using polyimide tubing now, which is good up to 700F. i just tuck the probes under the vertical lead of the coil. There are pictures in the thread.

That duty cycle is scaled from 255 to 100 ... What actual difference does it make to the system?

Click to expand...

well, you've got a +/- 20C temperature swing. That's a lot! i'm seeing a -1F/+1.5F (!) swing.

I do tend to hoard electronic crap too, never know when it will come in use.

Click to expand...

yeah, me too!

Hippie Dickie said:

@damienm420 -- that is understandable, this thread is really a development diary and probably doesn't appeal to very many people.

Click to expand...

Over 1000 replies and more than 132,000 views. You're too modest.

I've read the entire thread because it's fascinating to watch you struggle and overcome various difficulties. I'm cheering you every step of the way. I'm not alone.

pakalolo said:

Hippie Dickie said:

@damienm420 -- that is understandable, this thread is really a development diary and probably doesn't appeal to very many people.

Click to expand...

Over 1000 replies and more than 132,000 views. You're too modest.

I've read the entire thread because it's fascinating to watch you struggle and overcome various difficulties. I'm cheering you every step of the way. I'm not alone.

Click to expand...

You can count me in that club too

Look forward to the next Bud Toaster video as well

my sincere thanks to pakalolo and sm55

Power Pack - version 1 ... one step at a time (can hardly wait for version 5) ...

Well, after weeks and weeks of dozens of ideas and, for each and every one being totally, absolutely sure i had the concept nailed down (bagged, tagged and deep fried -- or something), i at least finally came up with something i was willing to try to fabricate:

Here are the components of the new power pack (remember, only version 1, so pretty crappy ...)



Here is the pack assembled:



And a closer look at the plug end:



and the opposite end:




So here is the latest iteration of an individual power cell:



With the positive end:



And the negative end:



The idea being that the curled wire on the positive end is springy steel and will just "snap" into the wood end piece and just kind of hold everything together. (the physical world is very simple in my simple mind)

And (!) the power pack can just plug into the Bud Toaster for a cordless vaping session:



So this is just a minimalist design to test a few concepts -- a battery without a soldered on Deans connector, copper contacts that just touch the ends of the battery, does the power pack actually work for a vape session, and, how does using it affect my thinking for the next version -- there always reaches a point where i have to fabricate something to be able to move to the next plateau.

Here are the copper contacts -- using copper rivets (3/32" dia by 1/4" long) and some 16ga copper wire:



so i've already come up with some ideas for version 2, but damn, i gotta put all this on the back burner for several days and start doing some real work for real money.

What do you think the expected battery life out of that power pack would be? It's looking good.