Your desire to have really good s/w for control and for logging is not unreasonable. That is why I embarked on almost 2 years of development which resulted in my computer controller roaster system. I have achieved it and the performance is flawless. but the price is not what you would expect. You are unrealistic about the development costs. A few weeks ago I met with the owners of the company which makes the HotTop and they said that they are only interested in their own in-house development of simple microcontroller low cost, modest i-roast level performance home roasters. They are very concerned with the cost and therefore did not even want to see my system operate! To give you some calibration of the real costs associated with the development of my roaster system, here are just a few of the more serious costs: 1. 1 man-year of my labor when calculated full-time 2. $6000 for the software development system plus an additional $600 for a technical support contract and $1000 for the compiler and licenses to distribute the compiled software and runtimes. 3. $1200 for a precision thermocouple welder 4. $7000 for a precision thermocouple system calibrator 5. $700 this month for minor repair and complete recalibration of the calibrator. If you figure the above costs divided by the possible number of sales you will easily see why this is expensive and will never meet your priceline. PS: I will be putting my control system on a professional gas fired sample roaster in a few months followed by a full roaster later this year. Regards, Jeffrey Pawlan www.computercontrolledroaster.com
* On **Sat Apr 22 00:50:01 2006 Jeffrey Pawlan wrote:* .....They are very concerned with the cost and therefore did not even want to see my system operate! To give you some calibration of the real costs associated with the developm= ent of my roaster system, here are just a few of the more serious costs: 1. 1 man-year of my labor when calculated full-time 2. $6000 for the software development system plus an additional $600 for a technical support contract and $1000 for the compiler and licenses to distribute the compiled software and runtimes. 3. $1200 for a precision thermocouple welder 4. $7000 for a precision thermocouple system calibrator 5. $700 this month for minor repair and complete recalibration of the calibrator. If you figure the above costs divided by the possible number of sales you w= ill easily see why this is expensive and will never meet your priceline...... Hey, thanks for your reply. Wow, that's some serious change. However, I don't know, it seems like you've got the toughest part behind you now. Why not start devising plans to sell just your software only- prepare a user's guide, add some user-friendly features, and work on making compatable with = a wide range of available thermocouple usb dataloggers? I mean, I would be happy with just the software. You could leave hardware aquisition and setup to the individual users. If you haven't already, check out Ambex's site www.ambexroasters.com. They sell their profiling technology to small business roasters for only about 1200 -1600 dollars- that includes software AND hardware. I would imagine that includes more than just thermocouple for temp, but probably also humidity, and such as well, not to mention programmability of roast process probably- all this for the "big boy" roasters too. That seems pretty cheap to me. After I figure that I could do without the programmability feature for now, but be content to manually control roasts as I monitor them; and that I already have thermocouple datalogger with usb; and I won't go crying to anyone to install the datalogger into the roast chamber, I figure $500 at most for top notch software. Believe it or not I would be willing to pay this much. Figure in the smaller size and greater ease of installation in home size roaster, and we are now down to about $300- $400. But here's where my even lower price analysis comes in: If you were to make your software so attractive, so complete, so user friendly, AND the option to continuously upgrade (for a price) to more and more features, I think you could turn a profit, atleast break even for the services and man-labor you put in- which still means money in the pocket, but more importantly homeroasters' rejoice. The reason is that a much greater percentage of home roasters would buy software at $70 base price, and then upgrade over the next few years, than the 5 or 10 lonely souls that would put down 500 or 600 dollars like me. And with software, once it's made, its made. It only takes 50 cents to burn a disc- that's atleast 50 dollars or more of money made each order to pay off expenses and turn a profit. Keep customer support constrained to a website that takes replies, and work hard on that for the first few months or so after release, then taper off that while you develop a newer edition. Let people know in advance the cheap price it at the expense of amazing customer service. I think the whole key to making it profitable would be to make it both easy to use, easy to buy, easy to tell people how to set up on their own (home coffee roasters are handy by design), cheap, and very well publicized (so much so that people would think they are missing out on something by not taking that next leap in home roasting- and they would be!). Furthermore, it would have to have more features than just datalogging. Another major selling point might be t= o include a good enough roast log in the program so as to make the software tempting to people who don't even ever plan to find out what a thermocouple is. I realize that things such as this don't just fall off of trees..., an= d it is much more complicated than I make it out to be. I am just really frustrated, because I feel as though if I knew how to write software I coul= d get it done. I feel ambitious but I lack any knowledge to move forward. The thing is is that I love engineering and I love the idea and the challenge and rewards of computer programming, and I love to learn new things in thes= e areas as well. I just don't have any time right now to learn, because I am = a med student. If I were you, though, I would work on getting your software out there, and at the very least earn back some money for your hard won investment so far. Getting the software out there would also mean you would have the chance to find out what people like about it, don't like about, wish it had, etc... It's hard for a dedicated home roaster to turn down tried and tested solid homeroasting software. Another major selling point may be to include a good enough roast log in the program so as to make the software tempting to people who don't even ever plan to find out what a thermocouple is.... Anyway keep trying, forget about Hottop, and thanks for the reply. I would love to see your software on SM website someday. I would pay double the price just to see it happen! But, in the meantime, don't go into debt...... Sincerely, Jeremy.
On 4/22/06, Jeffrey Pawlan wrote: <Snip> Your insight would be great to have in terms of how you did the instrumentation part. I am trying to think of a low-cost COTS solution here, with maybe a microcontroller or something as a go-between between the sensors and the computer, maybe by USB or serial interface... -- Steven Hay hay.steve -AT- gmail.com Barry Paradox: Consider k to be the greatest element of the set of natural numbers whose description require maximum of 50 words: "(k+1) is a natural number which requires more than 50 words to describe it."
Steven, The quandary for Jeffrey here is, you are asking him to give for free what has taken him a considerable amount of experience, time and money to attain, so he can create some business. Intellectual property is his most valuable asset, not the commercial hardware / software which anybody with some cash can get. Now if he at some point ended his goal of marketing a roaster, he might then be in a reasonable position to share that intellectual property :-).
I don't want to minimize Jeffrey's work and rights to his intellectual property. I've spent a career in process control (software side) including working and managing R&D before moving towards marketing so I know the costs involved. Jeffrey is underestimating what it would cost to have an actual product. However, there are other ways that are perhaps more suitable for some of us home roasters who want to fool around and make stuff. Here are a few thoughts on what would be the simplest way to proceed. 1. PC-based. Many people have access to a laptop that could be used. Informal open source would allow many to adapt to their needs and share their results. I think our volumes are too small and our interest in evolving the software are too great for a microcontroller solution to be practical. 2. Cheap IO. There are many possibilities here. Personally I bought a LabJack which is overkill for automating a popper. Recently some even cheaper USB temperature loggers have been discussed which would eliminate the need for the thermocouple amps I had to build.. For heater control, I've bought SSR's on eBay but haven't tried them yet. There has been recent discussion of SCR's as being an alternative way to go. The software must be well layered so that different IO solutions can be easily added to the core application. 3. Temperature precision. I think that in practice, accurate calibration to temperature is not needed. The differences between thermocouple placement in different roasters will be larger than the calibration error. Repeatability in a given raoster is what is important. 4. I'm using Excel's VBA (Visual Basic for Applications). It comes with Excel which is quite common. No extra development system to buy. Lots of people are familiar with how to use it. You can use the charting stuff Excel already has making the operator interface coding pretty simple. Windows and VBA are not a good real time environment but, for a computer dedicated to running a roaster during a roast, they are good enough. So far I have got the temperature logging going and building a graph. Some ideas of what a simple user interface could be. The next step would have been programming heater duty cycle control, But I got sidetracked building a manual control box using variacs. If anybody's interested, I'll send the file. Bear in mind, it needs the Labjack add-in to compile and it it definitely a work in progress. Perhaps we should start a thread over at homeroasters.org to continue this discussion. Dave
I can see something like Jeffry's' program working well for a commercial roaster (large mass of beans that one doesn't want to "loose", and a roaster with a lot of control over heat / cooling). I don't see how much extra value it would add for something like a HotTop in a home environment though. A couple of reasons: The heating coil in the Hot Top is electric, and although it can add a lot of heat, it takes a good bit of time for it to go from temp 1 to temp 2. Think about an eclectic burner on a stove trying to heat up to change the temperature of a cup of water on the burner. I'm using that analogy as a cup of water plus one ounce weights about 250 grams. You can "command" the heating element on your stove to take the cup of water from 100 to 200 in 30 seconds but in reality, that's not going to happen. I've found the element in the HotTop can add about 25-30 degree of heat per minute to a 250g mass of beans - given you haven't pre-heated the burner and chamber to a higher start temp. The same for cooling - if the element is red hot and the roast chamber is heat saturated to 450 (matching the heating element) if you tell it to go to 400 in one minute (to slow the roll into second crack for example), again, its not going to cool down that fast. The basic limitations of the HotTop for heating is about 1 pound of coffee can complete "normally" in 16-17 minutes. 250 grams can be done in as little as 7 minutes with a high pre-heat and always on element. From a heat control standpoint, the HotTop is more of a "slow and steady". If I wanted fast and flexible, I'd be PID'ing a air roaster where one can literally shutdown the element and drop the chamber temperature like a rock. Another thought: a standard PID like the one I use allows 8 ramp and 8 soak points - I'm only using 3 or 4 of them on the HotTop. The PID follows a "reasonable" heat curve very closely, so I don't see the need for a laptop for trying to control the heat ramp/curve to a more detailed level as the PID is matching the heat curve on second by second basis. Your "roast graph" = "your PID programming curve" if the curve is within the limits of the Hot Tops heating element and bean mass. And it only takes a few seconds to change the heating set points and times of the ramp. Give the cost of a PC / USB controller / PID RS 485 to RS 232 interface, I can't see the value. I have it and don't use it by the way...
Here is something that might work. It uses a PC's game port. http://www.download.com/Thermometer-Plus/3000-2130_4-10126851.html?tag=lst-0-1Dan
On 4/23/06, Dan Bollinger wrote: <Snip> lst-0-1 <Snip> All of these are great and very interesting! I've learned quite a bit abou= t TCs in the past few days.. Actually, the thing that looks like it would be the hardest to do is the real time connection via USB. If it were, say, a parallel port, it would b= e a lot more managable. However, I am nervous about using a parallel port, because they are kind of going obsolete for home computers. Not sure I understand why a microcontroller setup wouldn't work here. Microcontrollers themselves are pretty cheap, aren't they? And usually hav= e serial/usb/a to d/etc already built on them for use? I'm trying to find because I know I've seen it, solid state temperature sensors that are +/-1F that I would want to use for to establish a baseline for the TC signal. I guess TCs can only do difference in temperature measurements, so the way to do it is to have some other device act as a baseline. This device only has to be accurate at near-room temperature vice the TC which gets very hot. Still thinking of the best way to go about making something like this. I'm going to start keeping notes and make incremental progress, but since this is my first DITY project, it might be a while before I get up the guts to actually do anything. I'm also trying to keep the price down, which looks like it will be a challenge on a few levels. I'm wondering if an RTD senso= r might be a reasonable way to go, although I have no doubt it would probably be bigger. What kind of accuracy do you think is needed on something like this. Thinking about it, I'd say +/- 5F is probably okay, but would want to shoot for +/-2F as a design goal. Steve -- Steven Hay hay.steve -AT- gmail.com Barry Paradox: Consider k to be the greatest element of the set of natural numbers whose description require maximum of 50 words: "(k+1) is a natural number which requires more than 50 words to describe it."
This is a multi-part message in MIME format. By microcontroller do you mean a PID? Many PIDs can be ordered with an = output. Analog, I think. Dunno how this would work, but it could be = the basis for a data logger. Dan Not sure I understand why a microcontroller setup wouldn't work here. = Microcontrollers themselves are pretty cheap, aren't they? And usually = have serial/usb/a to d/etc already built on them for use? I'm trying to = find because I know I've seen it, solid state temperature sensors that = are +/-1F that I would want to use for to establish a baseline for the = TC signal. I guess TCs can only do difference in temperature = measurements, so the way to do it is to have some other device act as a = baseline. This device only has to be accurate at near-room temperature = vice the TC which gets very hot.
This is a multi-part message in MIME format. The same PID's that we use to hot rod the Silvia can be ordered with 8 segments, 2 alarms, and RS 485 output. From: homeroast-admin [mailto:homeroast-admin] On Behalf Of Dan Bollinger Sent: Sunday, April 23, 2006 5:46 PM To: homeroast Subject: Re: +New HotTop Roaster-Control Panel Adaptability AND Datalogging By microcontroller do you mean a PID? Many PIDs can be ordered with an output. Analog, I think. Dunno how this would work, but it could be the basis for a data logger. Dan
Steve, TCs (even really cheap ones) hooked up correctly are well withing the + or - 2° F criteria you mention. The required reference junction is allready within most all of the current crop of TC measureing instrumentation and chips designed for TC monitoring. The level of accuracy is well beyond the requirements of roasting, in my opinion. My feelings echo previous posts identifying the fact that TC (or other sensor) location in the roast chamber by even a small amount produces different temperature indication for exactly the same measurement environment. Again, in my opinion - and also previously posted, the important requirement of temperature measurement instrumentation is repeatability. TC instrumentation will provide this repeatability to a level of precision beyond that required for our purposes. The ability to identify roast levels and events to within a couple of degrees provides the data necessary for repeatability of roasts. It also provides a reference with which to compare between others results and between your own results while experimenting with different profiles. If I haven't made it clear - this is all "in my opinion". YMMV Mike (just plain)
On 4/23/06, Dan Bollinger wrote: <Snip> Hmm.. That is interesting, but I was actually thinking about the microcontroller itself, which is probably what the PID uses. A microcontroller is a small computer with input and output capability, usually through some status bits, analog out, serial out, usb out, etc.. They make a bunch of different kinds with different levels of capability, and they can be used for the brains of things like keypads, controllers, garage door openers, microwaves, whatever.. -- Steven Hay hay.steve -AT- gmail.com Barry Paradox: Consider k to be the greatest element of the set of natural numbers whose description require maximum of 50 words: "(k+1) is a natural number which requires more than 50 words to describe it."
On 4/23/06, Michael Dhabolt wrote: <Snip> I need to research this point more, I've been dinkering around with it and have found digital thermometers that are +/-2F and precision amplifiers for thermocouples, but the ones I've found would require some kind of reference and I've not put it all together yet. Still, it is heartening to hear that TCs are within my basic needs. For the wide range, I'm also looking into selectable gain amplification to keep the accuracy good throughout the scal= e of the instrument (keeping the input to the A/D in the top half can do this pretty well.) -- Steven Hay hay.steve -AT- gmail.com Barry Paradox: Consider k to be the greatest element of the set of natural numbers whose description require maximum of 50 words: "(k+1) is a natural number which requires more than 50 words to describe it."
RGFuLAoKPk1hbnkgUElEcyBjYW4gYmUgb3JkZXJlZCB3aXRoIGFuIG91dHB1dC4gIEFuYWxvZywg SSB0aGluay4gPAoKVmlydHVhbGx5IGFsbCBjdXJyZW50bHkgcHJvZHVjZWQgUElEcyBjYW4gYmUg b2J0YWluZWQgd2l0aCBNb2RidXMsIFJTCjIzMiBhbmQvb3IgUlMgNDg1IGNvbW11bmljYXRpb25z IGFzIGEgbG93IGNvc3Qgb3B0aW9uLiAgVGhlIGFuYWxvZwpvdXRwdXQgaXMgbm9ybWFsbHkgdXNl ZCBmb3IgU0NSIHR5cGUgYW5hbG9nIGVxdWlwbWVudCBjb250cm9sLiAgVGhlCmNvbW11bmljYXRp b24gb3B0aW9uIGFscmVhZHkgcHJvdmlkZXMgdGhlIGFiaWxpdHkgdG8gcG9sbCB0aGUgUElEIG9u IGEKMS8yIHRvIDIgc2Vjb25kIGludGVydmFsIChkZXBlbmRpbmcgb24gdGhlIFBJRCkgZm9yIHRl bXBlcmF0dXJlCmluZGljYXRpb24gYWxvbmcgd2l0aCBhbGwgdGhlIG90aGVyIHBhcmFtZXRlcnMg b2YgdGhlIFBJRCBhcyB3ZWxsIGFzCmNvbnRyb2wgb2YgdGhvc2UgcGFyYW1ldGVycy4gIFRoZSBz b2Z0d2FyZSAobWluaW1hbCkgdG8gcmVhZCB0aGlzCmluZm9ybWF0aW9uIGlzIHVzdWFsbHkgcHJv dmlkZWQgYXMgYSBmcmVlYmllIHdpdGggUElEcyBzdWNoIGFzIHRoZQpwb3B1bGFyIEZ1amkgbW9k ZWxzLgoKTWFyYyBDb2hlbiwgYSBpbmZyZXF1ZW50IHBvc3RlciBoZXJlLCBidWlsdCBhIFJvc3Rv IGJhc2VkIHJvYXN0ZXIgd2l0aApGdWppIFBYUjMgKHdpdGggY29tbXVuaWNhdGlvbnMpIFBJRCBj b250cm9sIGFuZCBwcm9kdWNlZCBhbiBleGNlbGxlbnQKc21hbGwgY29mZmVlIHJvYXN0aW5nIHNv ZnR3YXJlIHBhY2thZ2UgdGhhdCBwcm92aWRlcyBtb3JlCnNvcGhpc3RpY2F0ZWQgUElEIGNvbnRy b2wgYW5kIGFjY2VzcyB0byB0aGUgUElEIGluZm9ybWF0aW9uIChhbG9uZwp3aXRoIGRhdGEgbG9n Z2luZyBjYXBhYmlsaXRpZXMpLiAgVGhlIGxhc3QgSSBoZWFyZCwgaGUgc3RpbGwgZmlndXJlcwp0 aGUgc29mdHdhcmUgaXMgaW4g4mV0YSBjb25maWd1cmF0aW9uIGFuZCBoZSBpcyB1cGRhdGVpbmcg aXQgb24gYQpwcmV0dHkgcmVndWxhciBiYXNpcy4KCk1pa2UgKGp1c3QgcGxhaW4pCg==
Mike, That could be a great solution. Not only for data logging, but for fast and easy altering of the ramp profiles, too. We need to hear from our buddy Marc on this! Dan <Snip>
Dan, <Snip> I've been planning on harrassing him into participating when I put together the article on PID controlled P1 roaster. Mike (just plain)
So, what do you guys think? Is it worth the trip to have a USB interface o= r would parallel port be fine enough? -- Steven Hay hay.steve -AT- gmail.com Barry Paradox: Consider k to be the greatest element of the set of natural numbers whose description require maximum of 50 words: "(k+1) is a natural number which requires more than 50 words to describe it."
This is a multi-part message in MIME format. Steve, Aren't you up around AK, OH? If so are a short drive from a FrankenTop.... From: homeroast-admin [mailto:homeroast-admin] On Behalf Of Steve Hay Sent: Sunday, April 23, 2006 9:29 PM To: homeroast Subject: Re: +New HotTop Roaster-Control Panel Adaptability AND Datalogging So, what do you guys think? Is it worth the trip to have a USB interface or would parallel port be fine enough? -- Steven Hay hay.steve -AT- gmail.com Barry Paradox: Consider k to be the greatest element of the set of natural numbers whose description require maximum of 50 words: "(k+1) is a natural number which requires more than 50 words to describe it."
On 4/23/06, B. Scott Harroff wrote: <Snip> DC here... Kinda far... I've been sitting here reading about different temperature measurement schemes and I'm kind of impressed by the thermistors that are out there.. They may be much less problematic than TCs from an instrumentation perspective, and I think the range would be fine for roasting.. I'm hearin= g specs like +/- 0.2 F (vs. TC's +/- 2 F) and they are good up to 550 F which should be enough for roasting... and the best part is the signals you get from them aren't in microvolts, always a plus. -- Steven Hay hay.steve -AT- gmail.com Barry Paradox: Consider k to be the greatest element of the set of natural numbers whose description require maximum of 50 words: "(k+1) is a natural number which requires more than 50 words to describe it."
Steve, In my opinion - parallel ports are going to go the way that RS232 serial ports have been going recently......that is - disappearing from low cost home computers and laptops. Which leaves me with the attitude that in the long run setting up to communicate with USB is worth the effort and cost. There are USB to serial (RS232) adapters available that make a lot more sense to me. Parallel port communication has its place, in my opinion that is in an environment where 'many' parameters must be monitored and manipulated quickly....such as CNC machinery with stepper or servo control of several things which must all happen in harmony. What we are dealing with here is pretty low level communication requirements (super low band width)......USB and or serial offers considerably more band width and speed than is necessary (at a considerably lower cost). A USB to serial adapter that I've had good luck with is the 'Keyspan" p/n USA-19HS. For process control the "sort of" standard communication protocol has been for a while RS 485. This is a two wire serial protocol which functions well in 'noisy' industrial environments, works with long wire lead in, and can address quite a number of individual controllers on the same wire pair. RS 232 to RS 485 adapters can be had for $10 or $15. A combination of these two gets me to where I want to be, communications wise. Again I want to make sure that you understand that I am not presenting my conclusions as the final word. I've been playing with this for a while and I'm trying to keep firmly in mind the fact that cost versus the ability to do the needed job is a major concern to the average artisan level roasting individual. Mike (just plain)
On 4/23/06, Michael Dhabolt wrote: <Snip> <Snip> Thanks... My biggest issue with USB is actually the software side though.. I'm thinking at first to stick with Linux because I guess writing windows drivers can become expensive. If you are going from serial->usb->computer, what are you using to read the usb input, and can it be done real-time? -- Steven Hay hay.steve -AT- gmail.com Barry Paradox: Consider k to be the greatest element of the set of natural numbers whose description require maximum of 50 words: "(k+1) is a natural number which requires more than 50 words to describe it."
Steve, The "Keyspan" USB to serial adapter that I mentioned comes with its own drivers - work flawlessly. Yes it is real-time (considerably faster than you would want to poll). The app itself - Fuji gives you a little app that will graph, allow you to manipulate some of the PID parameters and throw the temp indication into a file readable by your spreadsheet software. From there anything is possible. Most other process controllers will accomplish the same things for you. Using something such as a Parrallax Stamp or other PIC is an intriguing prospect - - I always seem to have too many projects to get around to fully addressing the PIC thing. I have a gut level feeling that it would be a worthwhile endeavor. Linux would be the OS of choice unless you want to be able to consider the effort to have community wide (coffee community that is) facility. Mike (just plain
I didn't want to skip over this post without commenting. Sometimes an important post is made that is so important, people don't comment on it. I've had the pleasure of regular updates and techy chats with Jeffrey about his project since he began. Some of you may recall the Jeffrey, myself and a handfull of other list members began on a roaster control project about 3-4 years ago. That fell flat, but Jeffrey persevered. Without a doubt, his CCR is the most precise, accurate, and flexible roaster controller in existence, including those that run commercial roasting plants. The price may be more than we want to pay, but let's face it, its the 'toy' to have. I think commercial roasting operations will be using his unit as soon as next year. I would not be surprised to see Pawlan listed in coffee history books along with other roasting names like Sivetz and Burns. Dan <Snip>
I must have missed the post regarding the cost, what is the expected cost per unit now. ginny thanks in advance <Snip>
I have to take Dans lead here. The results Jeffrey has produced is truly an impressive piece. A no nonsense, super high quality, no compromise piece of work. He definitely deserves Kudos a batch for the effort. I certainly hope he is successful with the commercial venture. Mike (just plain)