8 bit microcontroller market
in [Embedded]
|
Prev: Tiny Bootloader
Next: Link&Locate 86?
From: Didi on 7 Sep 2006 16:33 Tom, > I presume this monitors the exhaust emissions? no-no, this is not automotive. It is a full-blown analytical gamma spectrometry system, using mostly HPGe (that is high-purity_Germanium) detectors etc. Feel free to contact me directly if you need some related info, I'll be glad to help - right now I don't even know what you want to be measuring :-). Dimiter ------------------------------------------------------ Dimiter Popoff Transgalactic Instruments http://www.tgi-sci.com ------------------------------------------------------ Tom Lucas wrote: > "Didi" <dp(a)tgi-sci.com> wrote in message > news:1157605693.406994.62090(a)m79g2000cwm.googlegroups.com... > <snip> > > <snip feature list> > my gamma spectrometry code (which is probably about 1/3 of all that), > > I presume this monitors the exhaust emissions? Which gasses does it > measure? I've been starting to research spectrometry myself recently but > it is pretty involved and if the trick has been done already then I > would be interested to hear more about it, if possible.
From: Jim Granville on 7 Sep 2006 17:45 Joerg wrote: >> Include that, and the 80C51 has a long life ahead of it still. >> > > An so do discrete parts. Look at "modern" TV sets: Many use non-SMT > parts because they wanted to save a few cents on the circuit board and > consequently chose, ahem, (Yuriy, close your eyes now) the good old > phenolic board. Yes, they can punch that, and it's probably single sided. > There is a reason why smart companies like TI brought > their new 430F2xx out in DIP. Hobbyisist are most certainly not the reason. One thing that surprises me, is the number of new PC motherboards that use thru-hole electrolytics. I know SMD ones are expensive, but I'd have expected the motherboard makers had enough clout to shrink the difference. It seems they prefer the leaded ones. -jg
From: Joerg on 7 Sep 2006 17:57 Hello Jim, >>> Include that, and the 80C51 has a long life ahead of it still. >> >> An so do discrete parts. Look at "modern" TV sets: Many use non-SMT >> parts because they wanted to save a few cents on the circuit board and >> consequently chose, ahem, (Yuriy, close your eyes now) the good old >> phenolic board. > > Yes, they can punch that, and it's probably single sided. > I have seen it with traces on both sides. But only if they have to. >> There is a reason why smart companies like TI brought their new >> 430F2xx out in DIP. Hobbyisist are most certainly not the reason. > > One thing that surprises me, is the number of new PC motherboards that > use thru-hole electrolytics. I know SMD ones are expensive, but > I'd have expected the motherboard makers had enough clout to shrink > the difference. It seems they prefer the leaded ones. > Motherboards need through-hole parts anyway for the connectors so they will always have a choice. As long as through-hole caps are only a fraction of a cent less that's what they'll use. I guess computer design is rivaled in cost-sensitivity only by automotive and consumer electronics. But it can be fun to ponder how to replace a FET with a bipolar transistor (yes, Yuriy, here cometh the 2N3904...) just because that saves 1.5 cents off the BOM budget. -- Regards, Joerg http://www.analogconsultants.com
From: Jonathan Kirwan on 7 Sep 2006 18:16 On Thu, 07 Sep 2006 20:32:57 GMT, Joerg <notthisjoergsch(a)removethispacbell.net> wrote: ><snip> >This stuff happens a lot. The nice and fancy solution is not necessarily >the best. Like in your case with the commercial controller it might >actually be an approach that couldn't have worked at all, no matter how >hard they try. Reducing "phase delay" between observation and control is important. Even in trivial (non-real) systems, where the output is simply a phase-delayed copy of the control input, delay wreaks havoc on the effectiveness of PID. Lengthing delay makes things worse, much faster than intuition might suggest. But still worse is letting that delay vary all over the place. If you cannot get the delay short, then at least making repeatable delays is a big help. At least there is then some chance at useful tuning. Control loops are something akin to trying to swing the far end tip of a long, narrow, and flexible bamboo stick to center it about the hole in a distant birdhouse. The length of the bamboo stick is the phase delay. Shorter is easier. Variations in the length of that stick (and distance to the birdhouse hole) is about like variations in the phase delay. It is easy when the stick is short. Almost 2nd nature. It is still a possible thing to work out the details of anticipating flexing of the bamboo pole, when it is very long and you are just controlling the opposite end of it. It's not nearly as easy, but you can learn to get the hang of it and improve your skills by observation and analysis. But then imagine trying to learn to handle fluctuations in the length/distance! It makes the process nearly impossible to anticipate/succeed at. Finding the right 'tuning' pretty much goes out the window. Variation is really bad. And a lot of folks make instruments (and/or controllers) having no idea at all either how they may be used by customers trying to solve problems or that it might have helped things to provide consistent, repeatable outputs at a fixed delay relative to their measurements. In the case I mentioned, I don't think it would have been possible to achieve the desired limitations in ripple along the bool without having carefully selected all the parts along the measurement chain with an idea towards fixed, and short round trip times -- measurement observation to control output. As they experienced, struggling with many prior attempts to "gain control" over their process. Weird thing is, I didn't even know much about these things when I was solving a problem I wasn't even aware of. Getting lucky in that case made me curious and that's when the point settled in more clearly for me. Programmers limited to experiences writing C and feeling that this is all they need to get anything and everything done simply haven't yet had these experiences to teach them otherwise. Although micros are fast today, readily available, and inexpensive, and I frequently hear that there is little or no need for assembly, so it is also usually true that competition has equal access to them and are ever finding smarter ways to do more with less to gain some competitive edge. So having to be smart about the compromises doesn't really just end there. At least, in instrumentation. In "jewelry products" such as iPODs, where appearance sells and where the consuming market isn't buying to achieve solutions to difficult problems, it's probably more about features, convenience, appearance, and you wind up selling sizzle and not meat. A watch, despite the fact that it is supposed to tell the time okay, is really sold as jewelry and if your business doesn't keep this to heart then you won't survive long against other manufacturers who do know that they are selling sizzle and not the ability to tell time. So I suppose different thinking rises to the top, there. Anyway, where the focus isn't on sizzle but on the meat, this isn't some destination we arrive at and say "the day of assembly is over, long live C," because while you are saying that some smart competitor is finding out how to use tight assembly on a 4-bit micro to detect 50/60Hz phase and amplitude and lock their instrument to remove the effects and make it work more conveniently, with 1/10th the power, half the cost, half the size, better specs and drift, and with 2X the profit margin to boot. Well, it sure feels that way at times, anyway. Jon
From: Joerg on 7 Sep 2006 18:32
Hello Jonathan, > Control loops are something akin to trying to swing the far end tip of > a long, narrow, and flexible bamboo stick to center it about the hole > in a distant birdhouse. ... That's a good way of saying it. I always liken it to backing up a truck with a long steered-axle trailer. Takes a whole lot of practice. The best "roll-up-the-sleeves" description of how to tune a PID is IMHO that from Bob Pease. He has a way of explaining things rather directly. > > Programmers limited to experiences writing C and feeling that this is > all they need to get anything and everything done simply haven't yet > had these experiences to teach them otherwise. Although micros are > fast today, readily available, and inexpensive, and I frequently hear > that there is little or no need for assembly, so it is also usually > true that competition has equal access to them and are ever finding > smarter ways to do more with less to gain some competitive edge. So > having to be smart about the compromises doesn't really just end > there. > Yep. He who thinks assembler is all stone age has probably never programmed an engine control or something like that. There might only be milliseconds between loss of loop lock and a destructive detonation, pieces raining out of the sky and all that. -- Regards, Joerg http://www.analogconsultants.com |