Flight controller -Electrical checking

the rocketry show
The Rocketry Forum

Help Support The Rocketry Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Here is a much nicer looking setup using high-side control:
pyro6-1.jpg

It needs less resistors, plus the op amps will not need rail to rail inputs (which neither the OP7xx or CS3014 had). It turned out I actually couldn't find a single op amp that suited the high side-controlled design, so this will probably be better.
 
Thats brilliant! Thanks for that :)

And for this running on a 9V battery without a cap, It would be able to fire most ematches I take it?

Is the current only limited by the FET's?

EDIT: There seems to be no SMT version of that FET? I guess it doesn't matter, but I'm trying to keep as much as possible SMT :)

Which features should I look for in a P-Channel FET to replace the ESS92's on the schematic? Is it the same as N-Channel?
 
Originally posted by basil4j
Thats brilliant! Thanks for that :)

And for this running on a 9V battery without a cap, It would be able to fire most ematches I take it?

Is the current only limited by the FET's?
Click to expand...
9V batteries will fire off ematches easily, as well as many igniters. The fets will not really limit the current. The battery and the resistance of the load will. You still have to make sure the FET max current rating is not surpassed (this should be pretty high, though).

EDIT: There seems to be no SMT version of that FET? I guess it doesn't matter, but I'm trying to keep as much as possible SMT :)

Which features should I look for in a P-Channel FET to replace the ESS92's on the schematic? Is it the same as N-Channel?
Click to expand...
Oh, those are just some random ones I picked out of eagle's library. For specs, it's generally the same. Low on state resistance, good power dissipation (2 watts or more), good max current (over 10 amps), and good price. Unlike the N channels, you don't need a small threshold voltage, since you will be driving it with ground and the source is at 9V, giving a vgs pf -9V. I recommend fairchild semiconductor for FETs.
 
Originally posted by mtwieg
9V batteries will fire off ematches easily, as well as many igniters. The fets will not really limit the current. The battery and the resistance of the load will. You still have to make sure the FET max current rating is not surpassed (this should be pretty high, though).
Click to expand...


Oh no problem then. Im looking for an FET whose current rating is close to 20A :)

Originally posted by mtwieg
Oh, those are just some random ones I picked out of eagle's library. For specs, it's generally the same. Low on state resistance, good power dissipation (2 watts or more), good max current (over 10 amps), and good price. Unlike the N channels, you don't need a small threshold voltage, since you will be driving it with ground and the source is at 9V, giving a vgs pf -9V. I recommend fairchild semiconductor for FETs.
Click to expand...

Kind of like the Fairchild FDS6681Z?
Rds(on) ~4.6mohm
Id 20A,
Vdss 30V
2.5W power dis.

Added:

Also, for the OP-amp, would the Analog Devices AD8630 do the trick? I like the fact it has 4 op-amp's in one package!
 
The FDS6681Z should be fine. The AD8630 won't work. Whatever you use, it will have to tolerate input voltages at least as great as your firing voltage (9V).

The OP747 is a quad package that should work. And it's cheaper.
 
Originally posted by mtwieg
The FDS6681Z should be fine.
Click to expand...

Originally posted by mtwieg
The AD8630 won't work. Whatever you use, it will have to tolerate input voltages at least as great as your firing voltage (9V).

The OP747 is a quad package that should work. And it's cheaper.
Click to expand...

Oh ok, I was looking for a super low offset op-amp :) The OP747 has the same footprint as AD8630 so thats not a big re-design :p

Thanks
 
A little off topic, but do you use Eagle for your PCB layout? If so, do you know how to add thermals for the ground plane?! Its frustrating me cause I can't figure it out!!

I know how to make a polygon and all that but can't turn any through hole or via's into pads :\

Grrrr
 
Beats me. I've never used thermals. Are you actually expecting to be dissipating lots of heat? Using heavier copper usually works better.

And 1uV is overkill. 100uV should be fine. And conveniently, most op amps have the same pin configuration.
 
Originally posted by mtwieg
Beats me. I've never used thermals. Are you actually expecting to be dissipating lots of heat? Using heavier copper usually works better.
Click to expand...

Oh, is there another way to connect via's to the ground plane?
 
If the trace is connected to ground in the schematic, it should allow you to connect vias between them wherever you want.

I'm definitely not the best person to ask about this. Try looking at eagle forums. I think there are a few out there.
 
Originally posted by basil4j
Oh, is there another way to connect via's to the ground plane?
Click to expand...

A "thermal" connection to a ground (or power) plane is designed to restrict the flow of heat.

This is useful if you have a through hole component and are soldering it to a via/pad with a ground connection. Otherwise the connection to that big chunk of copper acts as a heat sink making soldering difficult.
 
Hmmm yeah. I guessed that was thier purpose, what I want to know is how to connect a via/pin to the ground plane.

I think ill take the suggestion and ask on the Eagle forum :)

Thanks. :)
 
Creation of a ground plane, connected to the GND net:

Click Button: Polygon
Enter in Command Line: ‘GND’ [ENTER]
Draw Polygon
Click Button: Ratsnest (To show ground plane)

Creation of a via, connected to the GND net:
Click Button: Via
Enter in Command Line: ‘GND’ [ENTER]
Position Via
Click Button: Ratsnest (To update ground plane, if present)

Of course, its possible to substitute GND with any other net name. When setting vias, take care to enter the net name WITH the apostrophes, otherwise you will receive an error message. Its not necessary to do this when creating a ground plane.
EAGLE does create thermals for pads by default, but doesn’t create them for vias. This can be changed in the design rules. This can be useful when creating pseudo double layer layouts, where the top layer only contains wire jumpers.

Reinhard
 
Hi All again :)

Regarding the accelerometer.

I am now using the ADXL278 2 axis accelerometer.
Questions about interfacing it to the ADC...

The datasheet documents the decoupling well so thats good.
I noticed though, that the full range voltage swing is +- 0.25V.
Isn't this range a bit small to not need an op-amp?

I am using and MCP3208 by microchip.
VDD = 5V
VREF = 5V

From what I understand, would I not need to amplify the input from the ADXL278 by 20times? (or is it 10...)

If so, can anyone suggest an op-amp circuit to use :) Preferably using OP07's or OP777's :)

Cheers!

Alec
 
I don't know where you're getting the +/- .25V output swing... are you sure you're not referring to supply voltage? This device should function the same way as the ADXL78. That brings up the question of why you are switching to the 278 anyways. An extra axis of sensitivity doesn't matter at all.
 
Originally posted by mtwieg
I don't know where you're getting the +/- .25V output swing... are you sure you're not referring to supply voltage? This device should function the same way as the ADXL78. That brings up the question of why you are switching to the 278 anyways. An extra axis of sensitivity doesn't matter at all.
Click to expand...

The 0.25V Voltage swing is from page 3, 'Output Voltage Swing'. I guess im looking at the wrong thing lol

I want to use the 2 axis accelerometer to give me little more accuracy when determining VERTICAL height rather than just the distance the rocket travels (which could be at an angle). Having a 'Y' axis will allow me to calculate tilt by detecting acceleration due to gravity in the y axis as the rocket leans. Obvioulsy I would need 3 axis' to determine tilt in the other direction but 2 is a good start!

I would like to use 3 axis but there is no high-G 3 axis available. I may use a single axis, low g accelerometer added to a small board at right angles to the main board to give the z axis.
 
Ah, the the voltage swing isn't +/- 0.25V. The swing ranges from 0.25V to Vdd - 0.25V. So basically it can't output voltages within 0.25 volts of the supplies. your total output range will then be Vdd - 0.5V. This is normal, and should be no problem.

Tilt can only be determined by accelerometers when the body is not accelerating. Even 3 axes will not give any useable tilt data. That requires gyroscopes or some other sensors.

Ideally, you could determine tilt if you could assume that all your acceleration is along one of the sensor axes, which should be the case in rocketry. However, the cross axis sensitivity will probably corrupt the other axis reading so much that it will be unuseable. The best you'll get is a measure of roll rate.
 
It would be a good attempt to "straighten" an accelerometer controled flight path to isolate the vertical component.

You would need to sample fast enough to oversample, capture and calculate-out any spin on the rocket - unless you can insure an uncorrellated sub-sampling to alias the effect.

Tricky.....go for it!
 
Thanks for both those inputs :)

I should mention that to start with, I will only be using 1 axis in my code until things are working.
Once I have passed that stage and this device is working, I will begin figuring out the 3 axis software :)


My main question about the ADXL278 has been answered though, thanks mtwieg :)
I took a good read of the data sheet when I got home last night and came to the same conclusion about the Voltage swing.

And now I have one more for you ;) On the schematic you posted at the top of this page, there are 3 Bias Resistor Transistors.
I have never even seen one of those before until I saw them here! Is the MUN2211T1 on the schematic ok to use?
 
Originally posted by basil4j
And now I have one more for you ;) On the schematic you posted at the top of this page, there are 3 Bias Resistor Transistors.
I have never even seen one of those before until I saw them here! Is the MUN2211T1 on the schematic ok to use?
Click to expand...
If you can get them, fine. They work well for high volume customers who can get priority with the semiconductor manufacturers.

But, if availability is a concern, you can roll your own easily enough. They consist of just a transistor with two built-in resistors. Not hard to work around.

Doug
 
Originally posted by doug_man_sams
If you can get them, fine. They work well for high volume customers who can get priority with the semiconductor manufacturers.

But, if availability is a concern, you can roll your own easily enough. They consist of just a transistor with two built-in resistors. Not hard to work around.

Doug
Click to expand...

Thanks for that :)

Fortunatly, the company who I am getting to prototype this board is one of those companies who has priority with semi manufacturers ;)

I would like to go with the Bias Trans as it will cut down on components and board space :)

Would the MUN2211 do the trick do you think?
 
Originally posted by basil4j
Thanks for that :)

Fortunatly, the company who I am getting to prototype this board is one of those companies who has priority with semi manufacturers ;)

I would like to go with the Bias Trans as it will cut down on components and board space :)

Would the MUN2211 do the trick do you think?
Click to expand...
I've spent a career advising customers on components selection. The basic rule, unless you're IBM, Intel, Sony, etc, is to never use a sole sourced part unless it's mission critical. You take sole sourced risks on processors, key transducers, and similar items. You don't take those risks on pop-corn items. You don't spec in triple-output voltage regulators and other gee-whiz items just 'cause they're cool. That's how you end up with line-down production situations. When an LM7805 voltage regulator will do the trick, use it and not some exotic, low-drop-out thing. Similarly, if you can make a 2N3904 and two resistors work instead of the MUN2211, then you know you can source that with what, 25 different semiconductor companies? Now, how many sources are there for MUN2211's?

Normally, components engineers handle this type of issue, but every designer needs to do the basic components research up front when doing a board. Don't spec it in if you can do without it.

Doug
 
Hmmm. I never thought about it that way.

Do you think the bias resistors are needed? Or could I use just a plain transistor (such as the 2N3904 you mentioned)
As you can probably tell by the length of this thread, I am learning about transistors quite a bit!
 
ON semiconductor and Rohm are your best bets for bias transistors. I'd go with whatever part is most easily replaceable (sot23 packages). I picked that part only because it was the only bias transistor I could find in the eagle library.

You will some resistors to limit the base current and pull down the gate. If you're comfotable with using very small resistor packages (like smaller than 0603) then having them external won't be a big deal. I don't see what's wrong with using prebiased, though.
 
Hmmm ON Semi I can do.

I think for now I will go with the Pre-Biased ones as they seem to not be a product which will be discontinued.

What sort of R1/R2 ratio should I be looking for? I would assume 1:1 with 1K/1K resistors? but you know how all my assumtions so far have turned out!
 
For driving FETs, you only need a tiny amount of current, so pretty much any ratio will work. You would usually want the base-emmiter resistor to be bigger, though, so I would go with the MMUN2233LT1 or something like it. Either way, the transistor is most likely to be saturated.
 
Originally posted by mtwieg
For driving FETs, you only need a tiny amount of current, so pretty much any ratio will work. You would usually want the base-emmiter resistor to be bigger, though, so I would go with the MMUN2233LT1 or something like it. Either way, the transistor is most likely to be saturated.
Click to expand...

Well that is easier than expected :)

Next questions about board layout ;)

I am doing the board layout at the moment and am wondering if its good practice to run tracks underneath SMT components on the same layer. (E.g. The ADC)
Would there be any sort of interference doing this?
 
High frequency traces (like clock sources and serial data paths, which the ADC will have) are the biggest sources of EMI, so yes, there will some interference. You don't necesarilly need to avoid putting traces under them, though. It depends how vital the stability of the traces are. Analog and other clocked traces are bad to expose to EMI. Same with traces that control mosfets. If the trace powers big, bulky components like LEDs or buzzers or whatever, then a little interference on them probably won't matter.

When doing the layout, I generally design in this order:
1. Power supplies. "Star" layouts are generally good at reducing EMI.
2. High di/dt paths. For you, these are the traces that power the pyros. Keep them straight, short, fat, and very far from vital traces.
3. "Vital" traces. Make sure sensor outputs, FET gate controllers, and high frequency signals are short and direct. High (in your case, clocked data) and low (sensor outputs and mosfet drivers) frequency traces should not be too close, and should not cross. Also, give your oscillator a good ground plane around it.
4. Everything else.

Keep in mind, I have no idea how noisy your parts are. I can afford to bend the rules a lot, since I know my AVRs produce very little noise, even when running traces directly under the MCU with no ground plane. There's a good chance that EMI won't be an issue for you. I can't say, though.

That document I linked you should be better at explaining it than I can.

Also, even if you want to eventually make the jump to 3 axes of acceleration, I don't recommend the 278. This is because it works out much nicer to use one Hi-g single axis sensor for your vertical direction and a second low G 2 axis sensor for the X and Y. Three axis sensors are very limited in variety, and I doubt one exists that fits this application. Stick with the ADXL78 for now, then adding on the other axes is a matter of adding a second cheap sensor instead of replacing it with a much more expensive (and less precise on the X and Y axes) sensor.
 
Originally posted by mtwieg
High frequency traces (like clock sources and serial data paths, which the ADC will have) are the biggest sources of EMI, so yes, there will some interference. You don't necesarilly need to avoid putting traces under them, though. It depends how vital the stability of the traces are. Analog and other clocked traces are bad to expose to EMI. Same with traces that control mosfets. If the trace powers big, bulky components like LEDs or buzzers or whatever, then a little interference on them probably won't matter.

When doing the layout, I generally design in this order:
1. Power supplies. "Star" layouts are generally good at reducing EMI.
2. High di/dt paths. For you, these are the traces that power the pyros. Keep them straight, short, fat, and very far from vital traces.
3. "Vital" traces. Make sure sensor outputs, FET gate controllers, and high frequency signals are short and direct. High (in your case, clocked data) and low (sensor outputs and mosfet drivers) frequency traces should not be too close, and should not cross. Also, give your oscillator a good ground plane around it.
4. Everything else.

Keep in mind, I have no idea how noisy your parts are. I can afford to bend the rules a lot, since I know my AVRs produce very little noise, even when running traces directly under the MCU with no ground plane. There's a good chance that EMI won't be an issue for you. I can't say, though.

That document I linked you should be better at explaining it than I can.
Click to expand...

Thanks for those tips. I have the groud plane sorted so thats no prob.

There looks to be a few changes I need to make then with the high freq traces.

I have a quick questions about the cont test part of the circuit diagram you attached earlier (which I am blatantly plagiarising ;) ).

Just trying to get my head around it so that I learn as I go :)

1) Does the cont test op-amp bit, get its voltage reading from what is effectively a voltage divider with R1 being the 47K resistors I circled in the diagram, and R2 being the load/ignitors?

2) Also, the safe/arm switch. What is the purpose of the 1k resistor? It looks as though it allows a constant, but limited, current to flow through for the cont test. It is obviously bypassed when the arm switch is armed.

3) Last question :) How would the safe/arm input going to the MCU ever go high?
If the arm switch is off, then the MCU input would be grounded by the 1k resistor, as the resistance to +9v is a fair bit higher. If the arm switch is on, then it would be a direct ground and hence no change.

Thanks :)

Alec
 

Similar threads

brockrwood
LPR Idea: 12V Rocket Launch System Schematic Using Transistors Configured as Darlington Pair; Also MOSFET-based Circuit
2 3
Replies
88
Views
4K
brockrwood
brockrwood
F
I2C Rocket Starter Controller
Replies
8
Views
1K
HyperSpeed
HyperSpeed
Adrian A
Altimeter MOSFET deployment switches turning on when power is applied
2
Replies
30
Views
1K
rocket_troy
R
eJar0k
Composite Nose Cone Mold from 3D-Printed Parts
Replies
0
Views
344
eJar0k
eJar0k
novahobbies
Bigger Daddy / accidental BMS school rocket clone
Replies
14
Views
774
BABAR
B

Latest posts

Back
Top