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

[THC]

Just awesome.

But not a good idea to try to decipher those schematics after vaping

Sorry to say though, the "bud bomb" name is taken. I've got a metal pipe that looks like a little torpedo called the bud bomb.

http://www.everyonedoesit.com/img/products/budbomb_delux_pipe_gold_1.jpg

I Have a portable jet flame with the same design as the BB but it has pumice in the top above a gauze to stop burning when u use a jet flame when u get the knack good one hitter

 

[Hippie Dickie]

all i can say is ... just say NO to butane ... lol

On another topic, i've been considering the division problem again ... had to wait for a car service ... and i decided the easiest thing to do is multiply by 64ths. This gives me a multiplier from 0.015 to .984, which should be more than adequate. Dividing by 64 is just a couple of nibble swaps (i.e. rotate right by 4 bits), and the digital multiply portion is trivial (i.e. already coded). And i can ignore having to deal with binary fractions.

Just as a reminder, this has to do with applying a gain factor to the Sum of Errors (the "I" term of the PID).

 

[THC]

Btw I think the Bud Toaster is a great name - even 'Food Toaster' but i think bud is better. Kind what your trying to do atm I think activate the herb and decarboxylat the latent THCA to THC is that right? according to Dale Gieringer Ph.D. then the 'Idea vape temp is 355'f-400'f or 180'c-200'c At this point terpenes are completely vapourised 5%-10% of THC vaporises in 60seconds ; 445'f or 230'c THC completely decarboxylates in 5mins at this temp about 30% of THC vapes in 90 seconds.MJ is left dry crisp (often at this temp with a rusty red hue)". After reading this I noticed my best vapes even though alot are fixed 200'c but u can boost it with most but all my good ones produce rusty duff as aposed to dark/black/brown, so I think as ur obv limited bud size wise so get the most out of it by getting the hottest end imho (accidental combustion is not acceptable lol). Seriously great work m8 I take my hat off to u.
does anyone remember the mj maximiser machine in the 80's that applied the then new science of activation of mj to cause it to make the mj much more potent by curing and activating the chems without drying your bud, humidor style - it was so good they pulled it off the market anyone got one?

Coldfinger Extractor/Distiller The Apothecarian The Apothecarian is a hybrid of the home and professional units. It combines the convenience of a sophisticated water cooled condenser with the simplicity and lower cost of the home unit. With two sizes to choose from, this is the obvious choice for most herbalists. Please allow 5-10 weeks for delivery of extraction units as they are custom made to order for you by the glassblowers. Pending extractor orders cannot be cancelled within this time frame. Does not quality for free shipping offer. PRICES: 4 LITER (inner soxhlet basket holds about 400ml or 2-4 ounces of herb - $595.00. $45 of extra shipping will be charged for US orders on this 4L extractor.; 6 LITER (inner soxhlet basket holds up to 1L or a half pound of herb - $895.00 (includes heater) $75 of extra shipping will be charged for US orders on this 6L extractor.

its a made to order reflux!! cool



Oh yea I hear you about butane but its electric (lighter so no flint) and my IOLITE produces enough vape for a very sober mouse LB not bad waiting to try my titanium vape engine thigy but I have feeling it will be like a VG which is cool but hopefully with much thicker (a la SSV) vape thats a portable please I beg to the gods of tech they can make a table top as good as a Verdamper then I want SSV performance with my stealth portable, somebody come and let me know on the couch that I can finally move around lol!!!! Your doing great things m8 ur actually try to fix the prob by making your own - respect that level of ingenuity and I hope it vapes like a ssv and is as big as a rubic cube say and say 2hr recharge batt Ill by 10 a decade!!!And if you produced true stealth portable screw the Volc being a way of printing money if yours did preform that well or better I would pay $750 as there is no alternative the race is on lol! by the way i checked the price of those thing they are insane!!!!!!!!!! for the same price you could by and ssv and 2 da budda s or 2 deluxe verdamper and a vdw for blowing the best bags in the world!!! do the Vols come with a free sex slave? I don't get it????

 

[Hippie Dickie]

i don't think i ever heard of it (but maybe i forgot???) ... quick Google search:

In 1976, Thai Power introduced the mew model ISO2 for $159, but soon other
companies began to vie for the dollars of dopers looking for extraordinary
super-highs. Competition from imitators, such as the Maximizer at $29.95 and
Kik at $69.95, dropped the price of the ISO2 to $99 in 1979.

from this page: http://www.michiganmedicalmarijuana.org/node/8007

 

[THC]

that the fella never tried it wouldn't mind but i activate herbs and have a humidor lol!!!
but that was some fast research m8 lol!
God dam you can still get em!!!

 

[Hippie Dickie]

First Vapors from Vapor Processing Unit #2 at 6:15pm ...

still needs the skin covering, but i just couldn't resist.

i now have 4 types of wood for the top and bottom discs: cherry, white oak, maple, and mahogony. i was thinking white oak would work best with pink leather, but maybe the contrast of mahogony would work better.

This unit has the cherry.

 

[THC]

Seriously bruv you must be as proud as hell that is fucking cool, and you, developed it publicly with every one throwing opinions criticism ideas at you - amazing, how long has it taken you from sketch to this? Just to rain on your parade I already own a herbaire PD in Europe are widely regarded as a vape for people who cant afford grass the zapper its based on i have seen picks of but thats it why would this fill a niche not already filled with a herbalaire or pd (except build quality from what i hear than the pd ) what ever your answer even if you use as a pen holder it still look fucking cool lol!! what are the dimensions can i ask? and what temp does it run at?
Respect! Peace

 

[Hippie Dickie]

Thanks ... and all questions welcome.

This is just the vape i use, i really don't know how it compares to other vaporizers. i think it is good and effective, and other people might like it too, so it's open-source hardware/software.

i made my first all glass vaporizer in 2000. About a year ago i decided to add computer temperature control. i think it is finally good enough to try to produce as a product.

Niche? You tell me ... the vape is small (palm size -- see my avatar -- 1.75" diameter by 3" tall), it has an all glass vapor path, it's battery powered and portable, it goes from room temperature to producing vapor in 90 seconds. Temperature is user controlled and can range from room temperature to 550F (auto shut off above this temperature).

16% of the US economy is spent on health-care, soon to be 20%. The popular media and airwaves are saturated with propaganda to convince ALL americans to consume drugs. Cannabis is legal in 15 states and the baby boomers are aging. The time is finally right for this vape.

This is the iPod of vaporizers. Maybe. imho. It would cost $20 to make in china-volume.

i run it at 400F - that is, when the thermometer reads 440F, the bud is at 400F.

i agree with you --- it does look cool.

 

[Hippie Dickie]

Weekly status update:

The way i was applying the gains was nagging at me and i finally did something about it ... that is, i de-simplified the math a bit -- i decided i had simplified the PID calculation too much, and so i added more resolution to the operation when calculating the new control value. As a result, i finally am getting a feel for the effect of changes in the PID gain values.

So here are today's gains:

vAbove: (-15 * P), (-25/64 * I), (either -10(+) or +5(-) * D)
vBelow: (+4 * P), (+4/64 * I), (either -10(+) or +5(-) * D)

Legend:
vAbove: COILTEMP > SETPOINT
vBelow: COILTEMP < SETPOINT
P, proportional term is (SETPOINT - COILTEMP)
I, integral term is Sum of (SETPOINT-COILTEMP) until there is a zero crossing.
D, derivative term is COILTEMPprevious - COILTEMPnew.

These gains worked pretty well, and then the battery died at the end of the test run. i used the same gains with a fresh battery pack and the accuracy was even better -- a lot of stability at SETPOINT with a 2F wobble.

Now i need to diddle with the gains a bit more to reduce overshoot, and figure out how to compensate as the battery pack expires.

 

[reece]

Single quantity can cost a couple of $ each. Do a google on "borosilicate" or "kimax" or N-51 -- stay away from "pyrex" as that brandname has been sold to china and the formula is now a soda-lime type glass.

Hey, just wanted to let you know, according to the company that purchased the Pyrex brand, World Kitchen, they are a U.S. company and they have not changed the formula for the glass, which has been soda lime for the past 60 years.

They claim it is equal to borosilicate is resistance to breakage due to temperature changes and more resistant to breakage from impact.

http://www.pyrexware.com/thetruthaboutpyrex/index.htm

None of the experts seem to agree about their claims.

It's strange though, from a (very) little research it seems people started having problems with Pyrex dishing after World Kitchen took over and the reports seem to suggest that date of the changing of the formula is unknown.

It's also funny that Pyrex brand manufactured by another company in the UK and Europe uses borosilicate and there is one report of breakage, but three other countries with Pyrex by World Kitchen also have numerous reports of breakage.


If it wasn't for your post I wouldn't have known it wasn't borosilicate. I've seen a couple of places selling Pyrex "borosilicate" tubing and rods. I wonder how they can make that claim?

Thanks.

 

[Hippie Dickie]

well, it's kind of like the first (and only important) lesson about drugs i learned over 40 years ago ... know your chemist.



meanwhile, after a full day of PID diddling, i found a brain-dead program bug (using the wrong variable to stuff the dutycycle value), and after rereading the wiki article on PIDs, i tried a different approach and ... OMFG!!!!

i had assumed that the dutycycle value should be some more or less fixed value that would be adjusted: if the measured temperature is above SETPOINT reduce it, or if below SETPOINT increase it.

After 6 hours of testing every adjustment of PID gains i could think of, i couldn't kill the oscillations. i could reduce it, and see moments of stability, but not the real thing.

So then i tried something different: now i calculate a dutycycle value when below SETPOINT, and then adjust that value when above SETPOINT. And the result for somewhat arbitrary gain values: almost immediate settling to a value that is fixed. OMFG!!! what a win. NOW, i can adjust the gain values to fix a minor offset from SETPOINT.

And what felt like a wasted day has turned into a successful effort. Whew!

Now i just need to figure out how to bring the thermometer probe leads outside the body, and make a bunch of units.

============

Ignore all that ... just more Dickie blather ... after having some food, it occurred to me that if the COILTEMP is at SETPOINT, the dutycycle would be set to ZERO (since the calculated PID control value is a function of the SETPOINT error), which is clearly wrong. So, as a test, i just set the dutycycle to a fixed value (1/4 of full value), and the same stability is observed. WTF?

So it was a wasted day, somewhat.

 

[stark1]

Dickie, are you lookin for PAYING Beta testers.

If so, I am a willing participant. & a good writer, to boot.

PLMK, we need some new, innovative, products.

 

[Bubar]

Yep. The core weakness of the PID when improperly tuned is that at the setpoint the error is 0. You can get around that with integral term. Assuming you haven't overshot too much, the integral will be non-zero and unchanging at the setpoint. So what you need to do is adjust integral. If I were you I would kill derivative and just focus on tuning integral to get steady state worked out. If know the magnitude of the errors and the ideal duty cycle to keep at the setpoint you can calculate what integral gain you want. Once you have integral nicely tuned you can re-introduce derivate if you want to decrease overshooting/ringing.

I haven't gotten any good work done recently, but this week I will actually crank out atleast 2 prototypes. One for myself and one for my brother. Been using your model. The post office has my glass tube cutter, so hopefully I'll have some nice looking oven tubes.

 

[Bubar]

Haha. I got the battery set up going. The only problem is now I gotta make my set up faster responding. Running a .5ohm element with 8v heats up so fast by the time the thermocouple's temperature catches up and it tries to turn it off, the teflon gets melted through and the thermocouple shorts it... Idk how to deal with this. Probably turn it on full power for part of a second to get it to vape temp, then runnign pid algorithm with very limited power, maybe 1/3rd or less. It reaches vape temp in literally seconds. Probably faster, since that is just what the lcd is reporting.

 

[Hippie Dickie]

reID -- in my rig, a dutycycle of 0x0100, causes a 1F per second rise when the temperature is in range of SETPOINT. So if the COILTEMP is below SETPOINT, i just set the dutycycle to this value, and if the COILTEMP is above SETPOINT, i adjust this value by the PID calculation. This works amazingly well -- no oscillations and long points of stability (say, 5 seconds at a time).

i'm going to try using the PID calculation to offset a bias value ... the algorithm should be able to adjust this bias value dynamically.

your battery startup sounds weird. my batteries are 6.75v on full charge, my heater is 0.5ohm (theoretical, not measured), and it takes 30 seconds to reach vape temp. i put in a 12 awg copper wire as a bus bar and temp ramp up is now around 15.5v per second. i would check for a short.

The thermocouple i use is 0.010" leads and has .8 sec response. And i'm sampling every 300 milliseconds.

 

[Bubar]

How did you measure the response time?
My batteries top out at 8v and run down to 6v.
A short is possible, but I thought i had checked it. EDIT: It is only a .3ohm element.... so 24A

Once again on the PID my advice is kill derivative at least temporarily. A lot of PID controllers for heating applications, especially with higher power ones only use PI. Sampling 3.3 times per second can be almost guaranteed to not meet the nyquist theorem. Since derivative is much noisier than integral or proportional it will cause problems.

 

[Hippie Dickie]

The response time (on the K-type probe, you mean?) is the spec from the manufacturer -- so theoretical, not measured.

Okay, 24A is double what i'm pulling. i set the duty cycle to 0x03FF (full value) on startup until about (SETPOINT - 50F). If you're using my code you need to reduce the startup duty cycle value.

I've got a limit on derivative to keep it to a physically realistic value.

Unless i'm reading the PID stuff wrong (distinct possibility), it appears when the error is zero in steady-state, the PID value is zero, which can't be used for a duty cycle value, eh? Since i know the duty cycle needs to be around 0x0100. So, just using the PID value is guaranteed to create instability. No?

 

[Hippie Dickie]

Okay, now i'm getting somewhere.

The fundamental problem i have with understanding the PID theory is that the error can be negative or positive, but the duty cycle must always be a positive number. So i just modified my algorithm to apply the PID calculation to a "bias" value. i picked 0x0080 as the duty cycle bias (DCBIAS) value. This works amazingly well (gee, haven't i said THAT before?!) -- but i just saw long term stability (>15 seconds) for the first time.

However, the DCBIAS is affected by the state of battery charge. For example, with a partially charged pack, a value of 0x0100 shows a 1F/second temperature rise. But with a fully charged pack, even a lower value of 0x0080 shows a 2F/second temperature rise.

This is going to make loop tuning very difficult (impossible?), since the first step (either manual tuning or ZieglerNichols method) is to find a control value that causes oscillation of the temperature --- but that is clearly not possible since it is so affected by battery charge.

Also, the value for DCBIAS will determine how quickly the vape responds - a larger value will want to push the temperature up quicker, and the gains when too hot will need to knock the duty cycle down to zero faster to compensate.

So i need to have the algorithm adjust the DCBIAS dynamically to compensate for amount of battery charge.

Anyway, i'm very encouraged -- finally.

 

[Hippie Dickie]

Status update ...

So i was thinking about Bubar's start up problem, where he is pumping so many amps the vape melts ... i think the consumer protection agency would frown ... and i came up with a solution.

Instead of slamming the heater with full-on duty cycle (0x03FF), i ramp up the duty cycle in small steps (0x0010) until the change in COILTEMP is > maximum-change-value, which is set to 16F per second, and that becomes the maximum duty cycle value to be used in the PID. The ramp up adds about 5 seconds to reaching vaporizer temperature, and the total time to vape temp is about 35 seconds.

i can live with that. It was good to work this issue out, because if i had plugged the Bud Toaster into the car battery (via cigarette lighter outlet), mine would have melted too.

In the process of implementing this nifty "fix", i screwed up some flags and it took me until tonight to realize my programming bug ... but it's resolved now. What's nifty, is that this sets up the coding necessary to characterize the charge on the battery pack, too. And this affects the proper value of DCBIAS, the "neutral" value for the duty cycle that, theoretically, is the steady-state value to maintain SETPOINT (with the help of the PID gains).

So, i should be able to adjust the DCBIAS value as the battery discharges to achieve the best temperature stability -- or if the supply voltage is > 6.5vdc ... this part is still an exercise to be worked out ... but i almost have a first approach figured out.

Also, it has been a conceptual breakthrough for me to set up the PID using the DCBIAS value.

Also, i think i need to add a "Dry The Bud" mode, where the Bud Toaster will only go to 220F -- well below THC vaporization temperature -- for 10 minutes (actual time to be determined) to drive out the water. Then, click one of the buttons to start the vape session.

Lots of new ideas to implement. Cool.

 

[Hippie Dickie]

Gosh ... there sure is a lot i don't understand about temperature control using PID algorithm. my math intuition says one thing, but all the literature i can google says do "trial and error" tuning.

Anyway, here's what works really well for this device -- just use half of the PID algorithm. Specifically, apply the PID only when the COILTEMP is above SETPOINT. When the COILTEMP is below SETPOINT, just apply a constant duty cycle with no attempt at running the PID adjustment. This duty cycle, DCBIAS, needs to be just large enough to drive COILTEMP to SETPOINT at, say, 1F per second. It appears (after weeks of excruciating test runs) that it is not possible to increase the duty cycle in small enough increments without creating instability and temperature oscillations. It appears the PID gains can't accommodate the delay in response from increasing current through the nichrome wire.

my mathematical intuition says this is crazy, but so be it. i'm going to stop fighting it. i can live with 1F stability with this arrangement: the Hippie Dickie temperature control algorithm.

Now i only need to dynamically calculate DCBIAS based on the applied voltage. my two-cell battery pack has a voltage from 5.0v (dead) to 6.75v (fully charged) and a DCBIAS = 0x0080 works just fine. But for a car battery at 12v or a DeWalt battery pack at 18v this would be far too large a value and the algorithm would go to a bang-bang operation and just oscillate. The Bud Toaster allows up to 30v input.

One nice thing i have discovered is how robust the vaporizer is physically ... i have opened and closed it hundreds of times (no tools required) to attach the SOIC clip to reprogram the PIC and there is no damage what so ever to the Bud Toaster. Definitely ready to give it to a girl to use.

 

[Ash]

Lots of new ideas to implement. Cool.

Hey Hippie, don't tinker with this so long that you forget to start selling any vapes. I there have got to be lots of us itching for an official BT.

 

[Hippie Dickie]

thank-you ... i appreciate your comments.

here is the setup i've been using for this last week of testing. i'm putting together a package and made some new batteries. i've been deciding between shrinkwrap tubing and dip-it, a plastic coating, and this round will use dip-it:

i reprogrammed unit #1 and used that this morning. And after the session it is finally hitting me that a 13 month effort to achieve computer temperature control in my Bud Toaster has been achieved. February of 2009 i started programming my first PIC controller. Last night i realized an algorithm that works well enough ... and it probably won't get any unintended acceleration ... (firmware upgrades for units in the field would be nice ... however standard connectors are physically too large ... but i think i've got an approach that will work).

oh, yeah, when i pulled out the bud vial and looked at it i was amazed at how clean it looks even after 100 sessions in the BT:

Here are today's PID gains:

vBelow: duty cycle = DCBIAS (= 0x0080)
(no gains - although DCBIAS will need to be adjusted for the actual voltage input)

vAbove: duty cycle = DCBIAS - (10 * Error) - (0.5 * SumOfError) - (10 * Deriv (if positive)) or + (3 * Deriv (if negative))

i notice that the temperature rises with a toke, this must be the negative Deriv counteracting the P and I terms. This makes toking and passing without waiting for heater recovery a controllable factor. What a win!

This all assumes a 0.5 ohm Nichrome80 16awg heater, and two kickass A123Systems LiFePo4 batteries.

No overshoot, fast settling, and 1F accuracy during a 7 minute session.

 

[Hippie Dickie]

Desktop Accessories ... the mouse and the Pink Lady Bud Toaster:

i finally figured out how to bring the connections for the thermometer probe outside the enclosure ... i happen to have some copper roofing nails, and i found the right location inside out of the way, and a way to plug/unplug the probe when the Lady needs to be undressed and disassembled:

The two copper discs above the 3 buttons provide contact with the leads going to the thermometer body. So, place the Lady Bud on the battery sled (to be determined) and the thermometer will show the temperature of the coil.

Fashion Show!

Front view:

Back view:

She's cookin':

And that's that! Physically, anyway. i still need to mod the code to adjust DCBIAS for higher voltage input, but this baby is ready to ship for beta testing with the 2-cell battery pack.

 

[caseball2051]

can we sign up for the beta testing?

 

[Hippie Dickie]

Sorry, beta units already assigned.