I almost used a uC. Didn't for two reasons: uC don't have enough allowed VCC range and LDO are notoriously unreliable IMHO (and expensive). 2nd, I'd need two uC since both delays must be HW independent for safety reasons. Like the navigation computers on large airplanes.
-- Regards, Joerg http://www.analogconsultants.com
On a sunny day (Sat, 20 Jan 2007 18:00:48 GMT) it happened Joerg wrote in :
PIC 12F629 (8 pin DIL) is specified to work between 2V and 5.5 V. It has an internal oscillator you can software calibrate, many more features. And 1nA standby current. The 12F675 also has a 10 bit AD. I have used ten recently and so far no problems. The 12F629 costs 0.67 $ in volume.
As you referred to 4060 I think this beats it every way, except price perhaps. For things like prof equipment (like i nairplanes) the 67 cent x 2 cannot possibly be a problem?
Add to that an LDO and low-ESR cap for 40c, comes to $1.74. I've got to deal with a 4V to 9V range. Now the two 4060 chips plus two resistors and one cap each add to a grand total of 35c, no LDO needed. Also, there is only one LDO (TPS71xxx series) under 2.5uA and even that is a bit painful in this app. Plus I had some not so good experiences with those.
The delays is all I need from them and the 4060 does this just fine. Using a uC would have added NRE time and cost for the client while the
4060 just gets pulled from the CAD library and that's pretty much it. Ok, I also need a reference here but a uC with that on board pops right through the 80c barrier. Ye olde TLV431 for 8c or so will be fine.This one isn't for an airplane but a semi-disposable and it is very important to demonstrate the economic side of things. Two 67c micros would freak everyone out :-)
Even if it was for an airplane there is another side of this equation: False alarm rate and stuff like that. Simplicity of your circuit is rather essential and a uC doesn't exactly score that high. If you have a solution that can live without code and with parts that can be backed up with 30+ years of reliability data why not use it?
What could work would be two 4-bit micros as long as they are under 15c.
-- Regards, Joerg http://www.analogconsultants.com
On a sunny day (Sat, 20 Jan 2007 20:37:07 GMT) it happened Joerg wrote in :
OK, but I do not see why 4 bit would be OK and 8 not. One thing is very true, we do not know how long flash memory will last (if I had to guarantee it), we will probably see many electronic devices fail after 25 years... maybe sooner. That would be bad for an airplane, OTOH airplanes are taken apart completely quite often for maintenance, replacing micros after 10 years should be made mandatory (but can you still get these). Old Apollo spacecraft had core memory IIRC. But the mars Rovers have Xilinx FPGAs IIRC, lot of cosmic high energy rays.... Have fun.
The usual reason: $$$
Many times even a lowly 8-bit micro is just too expensive for a certain job. Probably half a dozen times I thought that it'll be time to roll one of my older designs to a uC solution. Called the client, asked for the current cost and was blown away by the low Dollar number they replied with. It's amazing what you can do with discretes these days.
Our here it isn't customary to take a whole airplane apart after 10 years. What is taken apart after 2000hrs or whatever their TBO rating would be the engines.
If it ain't broken why fix it? The design of the pump for the pool sweep system I am restoring right now probably dates back to the 40's. Works fine.
But there aren't a few hundred people on board that could die if something in the electronics croaks.
Actually, I am. Doing analog stuff is still fun. But not today. Got to work the honey-do list, repairing some freeze damage (global warming, ha ha). That's less fun than circuit design but someone's got to do it. Then in the evening, barbeque time :-)
-- Regards, Joerg http://www.analogconsultants.com
Piston engines do need a major look-see and overhaul every 1K to 2K hours. Lotsa pounding there. The cylinders have to be removed, stripped with boiling lye, dipped in red dye, rolled in white powder, and inspected for microscopic cracks.
Jet engines by comparison can run almost forever, only needing periodic crack inspections by boroscope and X-Ray. Some of the parts on the hot end do need periodic replacement as they anneal and corrode.
Planes are made of aluminum mostly, and aluminum gets tired after so many flexings, so planes do have routine inspections, every xxx hours or yyy trips for some parts, xxxx hours or yyyy trips for less stressed parts. That was much easier when smoking was allowed, as even minor cracks would be outlined in brown. Now they hafta run an eddy-current probe over every dang square inch.
Every 15K flight hours or so they need a major inspection, which takes many weeks to months of downtime as the wings have to be completely opened up, the inside of the fuselage skin has to be inspected, and worse.
The new jets will have mostly carbon fiber and epoxy composite skin, which hopefully won't need the huge downtimes for inspections. No fatigue cracking it seems. I hope they're right.
AIUI the US military wants new air frames designed for 100year lives.
Cheers Terry
"Terry Given" schrieb im Newsbeitrag news:1169334614.667521@ftpsrv1... | AIUI the US military wants new air frames designed for 100year lives.
With or w/o missile attacks included?
- Henry
--
The Motorola MC14060B data sheet does give consistent frequencies, between a calculation based on the sum given on Fig3 and and readings off the graph in Fig5, (for C=200pf and R=56K). So the CD4060 data sheet is misleading, (or wrong), cause unknown.
-- Tony Williams.
-- Well, no shit Sherlock?
Hi Jörg,
Sorry all its latenight here so I'll do it quickly in german, Jörg will understand it ;-)
Die Umschaltschwellen liegen bei 2/3 der Versorgungsspannung bzw 1/3. Eine Ladephase dauert also 67/63 der Zeitkonstante (63% ist 1/e) also 1,1 RC. Mit Entladephase also 2,2 RC. Warum das TI-Datenblatt andere Zahlen nennt, wird deren Geheimnis bleiben :-(
Marte
Translation of the last sentence: "As to why the TI datasheet offers other numbers will remain a mystery"
Well, that's exactly why I posted. Maybe I'll ask TI but it's just a scanned Harris sheet. Last time I found something like this with another mfg the response was something like "Oh drat, thanks, we'll look into it".
-- Regards, Joerg http://www.analogconsultants.com
Yep.
The factor of two is the key. It isn't that the output is coming from a divider, it is that the "T=2.2 RxCx" formula with Fig 12 on page 3-160 isn't backed by much text to make it clear... T has to be the time constant of the RC circuit (allowing for when through the cycle the gate switches), so it is the time for each *half* of the cycle, not the period of the oscillation. So the formula for the frequency should be:
0.5/(2.2 RxCx),plus a bit of adjustment for supply voltage etc. The datasheet really should have made this clearer.
The numbers work out in the HEF4060B datasheet and figure 8.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.