Discussion in 'Rocketry Electronics and Software' started by DexterLB, Feb 23, 2009.
lol! I don't want that rocket landing in my back yard, for sure!
Ok, timer almost ready. All i need to do now is finish the pc-communication board and solder the communication connector. And test it. (So far LED blinks correctly, but I can't test the "timing" because I can't set the time variables without USART.)
is there any particullar reason that you used 78L05 regulator? If I were you I'd use any other, low consumption one, there are chips at Linear Technology which draw less than 10 microA (!!) and deliver some 80 mA at output, which is more than enough for a microcontroller.
Not correct. It really doesn't matter which regulator you use, they (linear regulators) will all DISSIPATE or waste the same energy based on the load the current.
If you're talking about quiescent current, the quiescent current on a 78xx is about 5mA nominal which is negligible when compared to the load current. You're going to dump way more than that in pull-up resistors, uProc, and other loads.
So, unless you're running from a watch battery or planning on running for weeks at a time, a 78xx is fine. The only thing is that it is pretty big, however you can get 78xx series in small TO-92 (3-pin transistor) packages.
Yes... but: go ahead and callulate energy losses with quiescent current of 78L05 (5mA) and 10 microA (in the same time). The current that device uses on load may be high however, but it takes very short time.
10 microA * 3 seconds =30 microA seconds (low quiescent current regulator)
5mA * 3 seconds = 15.000 microA seconds (500 times more!) 78L05
1.2A * 0.1 second = 120.000 microA seconds (load current)
I supposed that timer was in function (in the flight) 3 seconds, but, having in mind that you turn timer on maybe 1 minute or 2 minutes before flight:
10 microA * 120 seconds =1.200 microA seconds
5mA * 120 seconds = 600.000 microA seconds = 0.6 Amp second.
These are currents of only regulators. If regulator uses that much energy, less of it remains for other electronics.
I really do not see a reason why not use less energy and use low quiescent current regulator instead of 78L05?
Again, it doesn't matter if you're using a 9V battery.
A 9V battery can easily provide 5mA of continuous power for at least 100 hours.
Unless you're running with a very small battery (i.e. watch cell), it makes little difference for the durations altimeters are typically run at. Now if you were building something battery powered that was operating for months or even years at a time, than sure, your statement would make more sense.
The other advantage if using a 'modern' regulator is low drop out, which is a major plus in avoiding a MCU brownout condition, especially if the battery is of questionable freshness, when firing the e-matches. it might be nice to have the option of running on a watch battery too, so why not. its certainly not a matter of cost. the point is thee is no disadvantage in using a better regulator and no advantage in using a poorer one.
Well, thats why you filter your Uproc Vcc input, so you don't have to worry about the Vcc drooping during an ematch fire . . .
Oh wait . . . didn't we already have this discussion ? ? ?
That's what I did! Look @ the pic - 78L05 in TO-92!
Well I could replace the 78L05 with an LDO, because this may happen to be a test board - the USART doesn't work in proteus on this pins, but I had already made the PCB when I simulated it , so I decided to test it anyway.
Yup, we've had this discussion... That's why i've put the diode and the 100uF capacitor (On the PCB it's 2.2uF - I've put the wrong size! That's another reason I'll have to re-make the board....). But an LDO won't make things worse, right?
2/3 of the USART communications board is ready. When I finish it i'll post the results
Quoting myself here:
"the point is there is no disadvantage in using a better regulator and no advantage in using a poorer one."
A disadvantage of Low Dropout Regulators is less stability--they are more likely to oscillate if the bypass capacitors freeze due to extreme cold. They also need larger capacitances for stability, which have more mass. An oscillating regulator can result in unreliable circuit operation.
OK now that explains everything in a few words. Thanks!
O! Once I blew an AT-tiny2313 i think with an oscillating regulator! No question about it - if it starts to oscillate due to small capacitance, things go bad...
It depends on the particular LDO. All linear regulators have an internal control loop which is used for regulation. Older style regulators such as the 78xx series use BJTs for the pass element while new LDOs use both N-Channel and P-Channel MOSFETs for the pass elements. Their low RDS on characteristic is what makes them "low drop out" vs. the relatively high Vsat of BJTs used in say a 78xx.
Also, its not really the size of the capacitor, its the internal resistance (ESR) of the capacitor which is the major factor. N-Channel based LDOs utilize a source-follower arrangement, in which the gate-source is isolated from the input control loop and are more stable. They also have a low output impedance which requires a simple ceramic low ESR capacitance. P-Channel based LDOs on the otherhand are less stable and also due to high output impedance require a large ESR type capacitor to be stable (i.e. tantalum or electrolytic). However, again, its not the size (or mass as you called it) but rather the series ESR element which is required for the stability.
The temp of the capacitors also has little to do with the stability problems as you mentioned. Cold temps can damage the capacitors, but they aren't going to affect the characteristics of a capacitor enough to cause problems unless you're at 40 degrees below zero.
a) Not always - that's mostly a problem of the past- use a good modern one and "forgeddaboudit"
b) if your expecting freezing cold, then don't use crappy caps. duh!
c) Come on- how much mass and capacitance are we really talking about? 0.1uF 1uF vs 10uF? i can get either in a 0603 ceramic at $ 0.008, in X7R material that is sufficiently temperature stable, very low ESR, and either weighs 0.01g - i.e a non-issue.
I use 100nF for, umm, $0.009, and they work excellent above 10deg. C. And, I don't think I'm gonna launch a rocket in 0K!
A 78L05 is stable with just one 0.47 uF capacitor close to the input for bypassing--are there any good LDOs that are fully stable without increasing the capacitance? I once debugged someone's 144MHz low noise amplifier--the regulator was oscillating at 1 MHz--you couldn't see it on a voltmeter but easy to see on the spectrum analyzer--adding the capacitor fixed it.
A very old application note that I found quite useful 20 years ago--I wanted my home built radio equipment to still work in winter weather. Commercial gear I used would motorboat in cold weather--the receiver audio would sound like a motorboat engine after the electrolytic capacitors froze.
So far, our club has yet to cancel a launch due to cold. Yes, that means we fly in 10 degree Fahrenheit weather. Cold is easier to deal with than mud.
Yes, tons of them - here's a couple of quick examples, theres plenty more out there you can find with a little web research:
No wonder- an antique regulator series thats infamous for instability issues. We tested them in the early 90's and as a result banned them from use at the company where i was working at the time.
Thanks for the info--I wasn't aware of the newer LDO technology.
Meanwhile, while I wait for the USART board, I decided to make a "tiny" version of the timer (exactly the same, but without the USART connector, which takes a lot of place.) Removing the USART subroutine opened some soft space for increasing accuracy, and adding some more features. Now I'll write a PC programme, in which the user (rocketeer) enters the time values. Then the programme will generate a hex file and will program it via JDM programmer. Then the rocketeer just takes the MCU from the programmer and puts it on the timer. Can it be simpler!?
I had no trouble with the LM2940/41 series--it might have been the most popular chip I used when I was busy designing and building electronic equipment, before having a stroke in 1998. I must have used at least 50 of them.
It was just what I needed for powering microwave oscillators off of car batteries--it could easily handle 1/2A and offered reverse polarity protection--highly important if you are going to loan out equipment. One fellow duct taped my gear to the top of a 100 ft tower for a few weeks--it survived!
I never noticed any sign of instability--powering a VHF crystal oscillator around 90 to 100MHz and multiplying it up to 2, 3, 5 or 10GHz for Ham morse code is a pretty tough test. It takes a good oscillator to generate a pure sounding signal that doesn't warble in an audio bandwidth. The multiplication process does a great job of magnifying problems that you wouldn't notice at VHF. I did tests with the temperature chamber at work, which was great for making sure it would work out in the field at both high and low temperature extremes.
Thats not a very good test for regulator stability. Sure, it will test what kind of noise / ripple you have on the linear regulator and how much rejection you get through the regulator, but really doesn't tell you about stability.
To really test the stability of the regulator (without using a network analyzer or venable, etc... and even with that equipment, you can't access the appropriate circuit nodes to do these types of loop measurements with a packaged regulator device), you need a stepped or transient load condition, (i.e. pulse of current). With a stepped pulse (especially with fast rise / fall times), you induce lots of high frequency harmonics into the regulator and if its not very stable, you can cause an oscillation.
The load of a crystal oscillator is DC, not pulsed. And again, they are pushing factors which will affect the performance of the oscillator, but thats more related to rejection of the regulator itself as opposed to its stability.
Hams have developed an open source Vector Network Analyzer. There is also a commercially available version.
Another use for a dual timer is to release latches for a Rocket Glider. A short delay is needed for some engine classes, namely 1/4A and A, as old short delay motors aren't certified for NAR contests. The first latch would activate the glide configuration. A much longer delay would be useful for releasing a dethermalizer--many NAR RG contests require at least one returned flight. I've been looking into DTs after losing a 1/2A RG--despite carrying the 4g engine casing it still floated away in a thermal--my longest timed flight at 6+ minutes.
Keep your traces on your MOSFET outputs nice and fat! Our ignition system used the thickest copper clad available (and the widest traces that we could fit) and we had the PCB sent out and fabbed for us...
After 200+ firings, the trace had accumulated enough damage from joule heating that the last time it was used the trace vaporized and exploded. Granted, we were running 12V at ~5A through it....
It'd just be smart to do it on a single board but be very conscious of the current limits due to your conductor cross sectional area.
Using PADS to lay it out?
I have that in mind.
P.S. I use Proteus ARES, but I really need new PCB software - it's freakin' me out!
The requirements for an altimeter and ignition system are completely different. In an ignition system, its not common to have a current flowing at very high currents for seconds at a time. In an altimeter, the currents are much more limited (by both the battery, ematch, and Rds of the small MOSFETs being used, as well as the time duration of the firing)
Also, if you're trace vaporized and exploded, you were definitely running more than 5A through it (unless it was a 5mil thick trace), especially if you say you designed it fat and used the thickest copper clad boards available (which i'm guessing means something like 4 or 8oz copper)
If your ignition system was hooked up to a car battery and used a hefty MOSFET, with a shorted igniter, you could easily pull at least 20A through your circuit and perhaps even more.
oh, i also didn't see pull-down resistors on your MOSFET gates. you can easily get away with 10kohm or more...
the point of my previous post was just to make sure you don't settle for 0.5 mm wide traces...
Yes, that is a good point.
Allright! Everything else sorted, just one more thing left: connectors.
At first I wanted to put screw terminals like the ones on the parrot altimeter, but I couldn't find such connectors in the shops here. The smallest I found was DG350 with 3.5mm pitch. I think four of them (two on one side - battery & breakwire, and two on the other - the charges) will do the job. But if you can suggest something better, it'll be great!
Separate names with a comma.