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

[Bubar]

Nice. I'm building current prototypes out of mason jars. It works fine with the controls remote, but I'm switching to putting everything inside.It'll work for your verson, but it'll be much harder for the lcd version. I'm gonna have to epoxy the lcd to the outside. This is too much work really, but it'll be cool to have one prototype. I have ceramic fiber insulation to protect the circuitry. Have you measured the temp inside your round version?

 

[Hippie Dickie]

no inside temperature measurements yet.

i got some ceramic tape (2" wide by 1/8" thick) several years ago. i'm going to use it as "stuffing" under the leather cover, on the outside of the stainless steel. Should make for a very soft, cushiony feel in the hand.

one thing i have learned is all matter absorbs heat, so i make sure to have an air gap between the heat generators and everything else. Also, i use an aluminum foil reflector to block the radiated heat.

 

[Hippie Dickie]

Bubar - have you looked at the math in the duty cycle algorithm yet? There is a lot of improvement possible here.

For example, after the initial temperature peak (that is, the start up from room temperature), the coil temp drops away too far and takes about 30 to 45 seconds too long to get back to setpoint.

The algorithm immediately tries to get to the steady-state duty cycle value, but i think conditions are still too dynamic during the initial warm up to vape temp.

The software is where this vape will REALLY be fun ... once i can get to a stable, reliable, buildable physical configuration.

 

[Bubar]

I haven't looked through your code at all. I only code in C and don't plan on learning assembly. But I use PID control in my version. Before my control was tuned poorly, but PID is pretty much standard for heater control. I have some new constants that I haven't tried yet, but should theoretically work well.

Also it's pretty funny that I started the other way around. Where I built up the software to be polished, but didn't have a set physical build.

 

[Hippie Dickie]

yeah, the software is a recent thought for me. i was convinced for 7 years that Ohm's Law was a law, instead it's just a suggestion. So i kept trying to get the physical device right.

15 minute warm up, 50F degree temperature drop on the first toke ... hey, that's just the way it was, that was the LAW forchristsake.

WRONG! 70 amps of current flow and computer control makes all the difference.

Edit: Thanks for the comment about PID -- i don't know about that -- there's a nice Wiki article about it.

Edit2: Wow. Finding out about PID is like finding out about PWM about a year ago -- which started me working with the PIC and the MAX. Wow.

 

[Bubar]

Yeah... I mentioned PID earlier in the thread, but w/e. The constants I'm gonna try next are P=200, I=.008, and D=.006.

 

[Hippie Dickie]

well, i either didn't notice, or didn't have time to Wiki the PID acronym.

So after reading about PID i see i don't have the integrative portion of the PID in my algorithm. And i can see how that would help a lot.

But i'm also intrigued by this comment about PID:

Note that the use of the PID algorithm for control does not guarantee optimal control of the system or system stability.

and this part:

Replacing the integral function by a model based part ... also achieves a control error of zero in the steady-state but avoids integral windup and can give a significantly improved control action compared to an optimized PID controller.

This is what i've been trying to get to, because the action of the nichrome wire heater coil (function of temperature vs current over time) can be very precisely estimated -- it's just an exponential -- with a lookup from a table of values.

Also, i'm thinking that the overshoot should be eliminated with a soft landing at the set point temperature. This will take longer to achieve the set point, but will be gentler on the components.

Then, the only variable in the system is the amount of battery charge, which determines the actual current delivered to the coil. So measuring the time to reach set point -- or alternatively, the maximum change in temperature between two MAX6675 readings -- is a measure of the amount of battery charge. This has a peak with a fixed maximum current -- starts slow, has a max of about 14F between samples for fully charged batteries, then drops off to about 10F, as the coil nears set point.

Well, there's an effect from the amount of bud and the cure (% water), but i'm guessing that will not be as significant a factor.

 

[Bubar]

Hey. Tomorrow I'll have some pictures of my prototype. I may solder up everything to a track board and stick it in. Lugging it around in breadboard has not been helpful. Every time I run it I gotta search for loose connections. Lol. My 5V model is working approximately the same for amperage so PID constants we should be able to use similar constants as long as we use the same algorithm. I use pretty simple PID Code.

From here:http://www.rabbit.com/downloads/files/PID.C

The only change is that I limit the values of all the coefficients. Right now it's not implemented, but I will limit each to 256 the highest number my compiler takes for pwm (i lknow the hardware is capable of 1024, but I don't feel like setting up the custom pwm stuff).

typedef struct PID {
double SetPoint; // Desired Value

double Proportion; // Proportional Const
double Integral; // Integral Const
double Derivative; // Derivative Const

double LastError; // Error[-1]
double PrevError; // Error[-2]
double SumError; // Sums of Errors
}
PID;

double PIDCalc ( PID *pp, double NextPoint) {
double dError, Error;

pp->SumError += (Error = pp->SetPoint - NextPoint);
dError = pp->LastError - pp->PrevError;
pp->PrevError = pp->LastError;
pp->LastError = Error;
return ( pp->Proportion * Error + pp->Integral * pp->SumError + pp->Derivative * dError );
}

 

[Hippie Dickie]

re: limit duty cycle to 255 (out of 1023) ... that may work because i think the steady state value is around 0x00C0 (192.), but it will probably take a lot longer to get to set point.

 

[Bubar]

I don't mean only 256 out of 1023. My compiler scales a value of 0-256 (to 0-1023) to be the duty cycle. It just means it's 2 bits less resolution on the pwm. I didn't take the pictures today, but I will tomorrow.

 

[Hippie Dickie]

ah so ... crank it up!

 

[Hippie Dickie]

Plans for my working weekend got snafu'd, so i made a little progress on M13v5. The three pcbs will be stacked in the bottom of the cylinder:

Already, this is being a pain in the ass to assemble. Maybe if i stand the pcbs on edge, i can cut some grooves in the base and they'll just slide in ... or maybe ...

 

[Hippie Dickie]

New Direction!!!

Okay, that last implementation just pisses me off ... three PCBs and a vertical mounting tower. Proof of concept, maybe, but not worth pursuing further.

But as i gazed at its physicalness, two things dawned on me:

(1) the need for 3 PCBs is dictated solely by the need to have the 8-pin DIP version of the 12F683, so that it can be unplugged from the vape and plugged into the programmer board, and

(2) instead of having to build my own in-circuit programmer, it should be trivially possible to slave a SOIC version into the programmer board ... a couple of jumpers and a DIP plug ... and i had all the pieces.

So, it is time to embrace my inner SMT (surface mount technology) and SOIC (small outline integrated circuit) demons and rise above my perceived obstacles.

So, i ordered ten 12F683 in the SOIC package (RoHS and elevated temperature tolerant) and redesigned the three PCBs into a single board, double sided.

i also decided to drop the snubber diode (saves a lot of space on the PCB) since the MOSFET has one built in.

Also, the MOSFET -- the only heat dissipating component in the circuit -- can be mounted ABOVE the pcb to keep the heat it generates away from the other components.

There has been an ongoing problem of mounting the push buttons so they register with the drilled wood disc ... requires too much attention to soldering detail. So it occurred to me to search out a right-angle version of the switch.

Well, it's still a work in progress, but i am giddy with excitement about this latest turn in the design.

The major plus is that the single PCB will be mounted on edge ... it will simply slip into a channel cut into the bottom wood disc ... no fastener will be needed.

In fact, there will be no fasteners at all. The whole vape can be taken apart without using tools. In my mind this makes the ultimate user servicable vape. And, based on personal experience, when the vape ain't running, the hippie ain't happy.

Here's some pictures of the progress over the last three days ...

only have the PIC 12F683 and the buttons mounted on the pcb ...

the buttons will be in the bottom of the vape ...

Here's the jury-rigged programmer connected to the PIC ... tonight i got the PIC programmed with the control algorithm ... ready to vape when i get the rest of the components mounted ...

 

[B.]

 

[Bubar]

Gotta say, the stacked pcb are really awesome.

Have you considered replacing the buttons with one knob and a power switch? If the vape just starts automatically when it's switched on, you could get the same functionality, but it's one less hole to deal with when you're making them. Maybe slightly less safe, but you could just make the switch hard to accidentally hit.

I got some scrap copper board in little segments, and I soldered up a connector pads for drain and source. Copper really fixed the problems I have been having with overheating. It's even been fine stuck in the jar with the heating coil.

I use alligator clips on 14awg copper wire as connectors and the hardest part of everything is getting the damn clips to cooperate with me inside of my casing. I've previously stuck the alligator clips on the nichrome instead and that actually seems to be the better option.

 

[Hippie Dickie]

awesome perhaps (thank-you), but i can't make a thousand of them.

i have a built-in bias against a knob. Those analog potentiometers will always fail -- develop a null spot (which will be at the perfect vape temp) or become "noisy" (different readings for the same setting) -- with as much use as a vape gets (at least for me).

In fact, the first pwm i bought (from eBay) and tried in a prior vape DID develop a null spot ... most annoying, to say the least -- with a prototype (and being inescapably catholic) i always assume i'm the cause of all failures, so it's both reassuring and annoying when a cheap ass component is the problem.

Besides, i'm a digital kind of guy.

Bubar, making a pcb is trivial if you get the right kind of laser printer release paper for the layout mask. i made some with standard copier paper, but the right stuff (about $1 per letter-size sheet) is sweet -- the paper backing drops right off as soon as it gets wet. i haven't figured out the correct iron temperature/duration yet as there are drop outs in the mask, but going over the traces with a sharpie is adequate.

i try to maximize the contact area between the nichrome and the copper -- i think alligator clips may develop hot spots at the points of contact. That much current flow seems to accentuate that kind of problem. Like the prototype i'm using every day, the new board will use a square copper washer with a 4-40 hold-down screw to get about 1/4" of solid contact between the nichrome and the copper.

 

[Bubar]

It's the economics of the pcb's. I priced out a small board for a commercial service. I can get parts for 12-15vapes for $60 ($4-5 per unit). If I have to buy etchant, copper boards, etc it's getting near the same price once you factor in time. Also, when a commercial service makes a hole, they plate the inside of it too. My lcd wiring is very convoluded and would be hard to do myself without using the via's of commercial services

Yeah, Alligator clips suck. But I have a bunch laying around.

 

[Hippie Dickie]

Radio Shack - double sided board is $4, 16oz of etchant is $10. But if helps if you have tools (Dremel, drill press, etc) and a shop to work in.

here's my front-side trace pattern (from MS Publisher):

This is 1.25" wide by 1.00" high.

And for component placement on the top side:

 

[Bubar]

Hi. I just finished up midterms, so I'm getting back into this. I got pretty good PID constants now, the vape hovers about 5C over. I will test later with the integral term turned down. I haven't gotten around to taking a pic, but the prototype is quite the contraption. lol. All kinds of wires and alligator clips stuck inside a mason jar. It now heats in roughly 30-40s. I will probably also lower the resistance of the heating element to be .3-.5 down from .75. I have a few coils I will probably test.

 

[Hippie Dickie]

i'm looking forward to seeing your setup.

i've made a little progress myself ...

Here are three pcb blanks to make this next unit:

each board is 1.25" wide x 1" high (standing vertically) ... i already made one, it wasn't done well enough - for some reason the traces got too wide, maybe too much pressure with the iron or a smear on the toner, anyway, the second try got a very nice board.

Here is the almost fully populated board:

Front side (sorry for the out of focus):

Back side:

now i need to add the MOSFET and the nichrome heater coil and oven tube.

It will go into an enclosure like this earlier version, but will be 3/4" shorter:

 

[B.]

Damn, that thing looks awesome. I like that look way better than the square version. Cant wait til it's for sale.

 

[stinkmeaner]

I second that. It really has come together nice, is there any way smaller batteries can be attached?

 

[Hippie Dickie]

thank-you B and stinker. i have finally come to realize the round shape is the best form-factor. That all-in-one concept, which devolved into that square shape, just overwhelmed me for several months. i think i'm finally past that phase for the next several cycles.

And yes there are an infinite number of battery choices - any high current batteries from 5vdc to 30vdc will work just fine -- like a car battery, DeWalt/Mikita/B&D power tool. The vape uses a "standard" Deans male connector (standard for radio control models):

(how rude ... they blocked that jpg)

Prewired female connectors, that could be soldered to a different battery/pack, are $4USD at most hobby websites, e.g.
http://www3.towerhobbies.com/cgi-bin/wti0001p?&I=LXKX43&P=M

i really like these batteries -- 15 minute recharge and one charge lasts all day (6 or 7 sessions).

 

[Lo]

I love watching this thread!! I don't understand a lot of it, but enjoy and also look forward to seeing it through production Very cool!

 

[Hippie Dickie]

i attached the heater coil and the k-type thermocouple probe.

Here is the first measured heat-up test:

um, kind of inconvenient to hold that tiny pcb ... there is 400F very close to my finger tips ... so i need at least the bottom disc to hold the sucker to take the first vapor hit. GF is dragging me out of my digital bunker this evening, so it will be late tonight before i get that first hit.