Sounds awesome...

 

sounds good to me too...what if i ship you my dcu and have you convert it into a dci???

 

Modifry:

I'd recommend against it myself. I doubt that you can really meet the $50 mark for this device, even if you do use a lower power processor. There are many other things to take into consideration with this as far as price is concerned, but I'm sure you are aware of that.

For the DCU, it cost $1500 to have PCBs made in small quantity so you'll have to invest quite a bit just to get started. If you make them yourself or try to use protoboards, you can't wave solder because you have to have a solder mask.

You'll need to have them wave soldered so you'll have to pay for assembly, and this means production PCBs. If you don't they may fail in six months after 180 days of temperature/humidity cycles. The soldering will cost a minimum of $5/ea no matter how few parts you have.

Then there is a fee for testing, a fee for programming, the cost of packaging and shipping, etc... Of course you can try to do all of this yourself, but it will take a LOT of time (I did it). Also don't forget that a few units will fail and people will expect you to replace them. If you don't build enough extras, people will get really upset. If you don't wire them all up to the stereo harness, many customers will wire them poorly and complain when it doesn't work. You'll end up giving several refunds to people who don't know how to do wiring.

Don't forget that UPS will lose one. You'll replace it and decide to switch to priority mail, and then U.S. mail will lose one and claim that they have no information about the tracking number so you won't be able to get an insurance refund (this happened). You'll need to replace that one too. (Tedsters)

Make sure you build a good test rig too (you'll need to buy a $300 stereo control from Honda) and have a dedicated Sony stereo for testing because your test car will get rear ended by an idiot, pushed into another car and spend 3 months in the shop. (This happened)

One other thing: You'll place an order for some of the parts and they'll ship you one of the wrong IC's. Of course you won't notice that the number on the chip is 4152 instead of 4052 until after they are assembled. These units will smoke during testing and will have to be replaced before you can ship them. (This happened)

This is only small sample of the problems that can (and have been) encountered. Every one of them costs money and soon you are paying for it all out of your own pocket. If you do decide to do it, charge 3 x what you think it will cost and if you are lucky you'll break even. If I ever decided to sell the DCU again (which I wouldn't, or any other electronic item for that matter), I would'nt do it for less than $250/ea.

Also: You cannot do 'multiple button' presses with the circuitry on the S2000. I thought of that too. The buttons are so that each button press removes the rest of the buttons from the circuit. There is a schematic of it in the electrical book, take a look and you'll see it isn't possible. If it was, the DCU would be able to be programmed to use them.

 

modifry, I'd buy one and keep it as a backup in case my current DCU fails completely. As it is, I don't think I can give up controlling vol up/dn, seek +/-, presets +/-, source switching, display switching and muting. The other option for me is to wait and see how that "other" DCU holds up in terms of reliability. I'm not at all thrilled with having just "basic" functions.

s2kman, thanks for the time and effort getting the DCU into production. I came into the group buy knowing that maybe there would be reliability problems. I figured that was the chance I was willing to take and have no regrets doing it.

 

I'll save space by not displaying your whole message. I appreciate your concern, but I have done this before too. I have not done anything as complex as your DCU, and I try for as simplistic a design as possible, because in my mind, simple is better. But...

Am I to understand that before wave soldering, there was no such thing as a PCB that lasted more than 3 months? I think it's possible to solder by hand and have it work, and I agree that there will be more failures with hand soldering, but I'm not making 10,000 of these things, maybe 20, maybe 50.

I just made 15 "Redline Shift Modules" in my basement this past weekend, and shipped out 9 of them, all hand soldered. I made the pcb's myself, but am looking to get them made for under $5 each, in quantity's of ONE. I guess it depends on where you get them.

Since I put them together myself, I test myself, so no cost there, just time. And for my current projects, I did make a simple test rig. The DCI will be more complex, but not that bad. I'll probably make another micro to test the target output resistances, so I won't have to hook each one up to my car.

I've been pretty lucky so far with shipping, but accidents will happen. I've been debating switching to some method that is unsured, just to cover myself. Postal is easy and cheap, and I like being able to keep the cost down.

Now, for the "multiple-button press" comment. YES, I have looked at the schematic, but I can't see any DPST switches, only form A SPST switches. I see no way to disconnecdt the other buttons when one button is pressed. And even if it did, I can still do multiple button presses - it just depends on how you do it.

Each button press puts a different resistor into the circuit, IN ADDITION to any buttons already pressed. I don't remember the exact values for all the buttons, but I definitely remember that Volume UP and Volume Down are the lowest resistances, around 300 ohms and 47 ohms. The other buttons are much higher resistances. So the two conditions below are different, and can be sensed as different by the micro, and can be programmed to produce different results:

condition 1 - resistance goes from 220k to 300 ohms (press Volume UP)
condition 2 - resistance goes from 220k to 10k to 300 ohms (press Mode, hold, press Volume Up)

 

OK, my two cents...

Hand-soldered boards can be just as reliable as wave-soldered in small quantities because (hopefully) you're paying more attention to what the joints look like. Since these boards are so small, you could actually make one design, copy it multiple times to a 8"x8" board (or choose some other size) and have any of the standard proto shops cut you two for $150. That would cut down on time, and would probably cost you <$10/board.

Insurance costs whould be up to the consumer...if they choose to ship without insurance, it's their potential loss, not the sellers, and that should be obvious. If they want it, they get charged more, just like all other businesses out there.

Test jigs do not need to cost $300 because you bought it from Honda...I would have made a test jig from a few extra resistors I had sitting on the desk for a total price of $0. Sorry, but spending $300 to test a handful of $50 items is silliness, plain and simple.

While catastrophic accidents DO happen (car getting rear-ended), does it need to put a project on hold for 3 months? I should hope not!

Should you check all major components before installing them? You bet! This is like the "measure twice, cut once" rule...check the values, it'll save you a lot of heartache in the end.

It seems like there was a goodhearted attempt to make the DCU, but it also seems like there were some unfortunate decisions made along the way (unnecessarily expensive boards, unnecessarily expensive test jig, unnecessarily time-consuming wait for test radio, unnecessarily bad shipping methods, and failure to check received components). It's also unfortunate that someone else designed the circuit, so we can't get out hands on that (although we could easily look at the board and create the schemayic from that...unless it's multi-layer). We should be able to get a look at the code, but hopefully it's fully commented.

I can think of an easier way than sampling like the PAC does, and would be essentially foolproof once set up. Hardcode all possible codes in memory, just as is normally done with most generic remotes. You enter a code, and from then on the remote works with brand X component. For this case, we could use a few DIP switches, the customer sets the code pertaining to their head unit, and all that's left to learn is the resistance ranges.

Now, looking at the resistance ranges changing with temperature problem. Place a known-value, cheapy resistor on the board and sample it everytime you read something from the dash controls. Use any change in the known-value resistance to modify the read dash control value. It only costs you a resistor, and extra A/D line, and some code space.

Oh crap, my fingers are tired, and I have to get some work done today...I'll finish up some more comments for a new unit when I get home. Besides, this alone should give you guys enough to make comments on.

 

Just came across this...3 free PCBs for new customers. Basically you're getting a $150 order for free. You have to stick with their default choices of board material, etc., but trust me, this is NOT a problem for this type of project.

http://www.4pcb.com/Free/eet/free.asp

 

[QUOTE]Originally posted by MacGyver
[B]Just came across this...3 free PCBs for new customers.

 

If you have the circuit designed already, one "simple" way may be to install a 30-day trial version of some of the better packages, quickly learn the basics (I'm a fan of Protel myself...easy to learn and get up and running quickly with their tutorials), enter the schematic, and output the necessary files. If you search the newsgroups, you'll also find <cough cough> cracks <cough cough> for the time limit.

The temp change is only a percent or two with pretty wide temp ranges, but I was thinking of future operability with panels containing more switches (why limit yourself to just S2000s?). Also, take a hint from DJ Qubes setup...don't hardwire a rmeote in, but use the multi-position hat switches.

I am talking about an IR interface, but without the slowdown and/or delay others are seeing with the PAC (even I notice it...takes me 1/3 the time to go from full volume to mute with the HU's remote than with the PAC). Again, this opens up your market to more than just Sony units, but pretty much ANY unit with a remote and has a documented IR code stream. Most companies (Sony, Pioneer, Alpine, Jensen, etc. etc.) use documented code streams. So, you don't have to store a map for every make and model, you only store the different TYPES of streams and how certain companies MAP those streams to each function. Look at a list of codes for, say, a TV remote...you'll see a list of probably more than 100 codes. However, there are only a handful of actual stream types since several companies use the same stream type. The amount of memory required is actually a few kB, easily fitted into a modest PIC chip. Downside is, as you said, more development time on reading in the DIP switch code (trivial), switching between stream types based upon the DIP switches (again, trvial), and adding in the code for several stream types and actual stream creation (long and tedious, but certainly a trivial job with the specs in front of you).

The PAC samples (most likely, I have no proof of this) by grabbing a "window" of samples from your remote and playing that "window" over and over again. Since the "window" is usually long enough to hold several copies of the desired command, you get what you want. The reason why the newer remotes by Sony don't work is that the bit rate is significantly higher (about 10 times higher), so the PAC is sampling the "window" at too slow a rate to catch all of the bit changes.

 

The beauty of designing parts for the S2000 is that most larger companies don't bother making similar products, since it is not as mass produced as say, a civic or mustang. The dashboard audio controls definitely fall into this category. I've thought about expanding my market to other vehicles, but there are already companies that do similar things for larger production vehicles.

I'm already using the hardwired remote as the IR interface that you are describing. Why reinvent the wheel, right?

A "universal" solution (all brands) style remote would be ideal, and I'd love for this to be available, but for very specific application on such a low production vehicle, I don't think the return on investment would be very high... Like many of the innovators of products on this board, you may even lose money (s2kman included ). I'm not trying to discourage anyone from being creative here, just make sure to be realistic financially.

I have also toyed with the idea of using a programmable, learning remote with my solution. You can find one for $20 nowadays (yes, a LEARNING remote for $20!). But then you have to teach people how to program it, have people publicly blast your product because the remote lost its memory, etc. For me, it wasn't worth the trouble. Hopefully, you guys could develop an easier way to do it (dip switches, like Mac says).

If you couldn't tell from the simplicity of my design, I am not an EE. I'm sure my design could be much more robust, but I don't have the money to invest, nor the EE background. I'd love to work with you guys, once you're finished with the brainstorming phase. I don't plan on making my product for the rest of my life, and would love to see a solution that could make everyone happy, even if I'm not making any $$$ off it. I'd probably buy it myself!