Cygnus Module development

Project to send a balloon into lower space with payload
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Cygnus Module development

Post by Daniel Wee »

Given that we need 9V for the RF module's power amplifier and the nominal current draw is 0.35A (probably less), we can now compare whether a linear or switching supply (assuming 82% efficiency) is more efficient.

For the linear supply, taking a nominal input voltage of 11V, the power lost is:-

(11V-9V)*0.35A = 0.7W

The efficiency is 9/11*100% = 81.8% which is a bit lower than the switching supply. This efficiency actually increases, though, as the battery voltage drops further, up to 90% at 10V. The switching supply's efficiency, however, drops with the input voltage. What this means is that for this particular application, a linear 9V, 1A supply may be the better solution than an 82% switching supply. Unless we can get a higher efficiency power supply, it would then be easier to use a linear 9V regulator.

Note that the PTH08080W series is capable of efficiencies around 86% and upwards at 0.35A current draw and 12V input voltage. Unfortunately, it can only output up to 5.5V max.

Potential candidates for 9V LDO linear regs are the ST LF90CDT and the NS LM2940S-9.0.

Dainel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Cygnus Module development

Post by Daniel Wee »

I think I'm going with the LM2842X/Y for the 3.3V switching PSU. I will need a reference voltage source to track stuff.

Daniel
Attachments
LT-LT6656 Voltage Ref 2.5V.pdf
(558.95 KiB) Downloaded 5626 times
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Cygnus Module development

Post by Daniel Wee »

Examination of the LM2842 3.3V power supply shows a roughly 62mV peak-to-peak ripple. Oddly though, even without any power applied this seems to be the ambient noise It turns out that the 10x mode was too high to measure the noise correctly as it was picking up environmental noise. Re-measuring with the 1x probe impedance yielded more realistic results - about 30mV p-p noise, which is quite respectable. A lot of the noise appears to be coming from the MCU activity - giving reason to isolate the I2C bus and the GPS.
PSU Noise 004.PNG
PSU Noise 004.PNG (16.02 KiB) Viewed 26066 times
The actual noise level changes according to the input voltage but is around 40mV nominally. This could be reduced further with some filtering. The addition of the GPS causes a voltage drop of about 0.02V. We should take care of this by separating the GPS filtering from the rest of the devices.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Cygnus Module development

Post by Daniel Wee »

Is there a need to drive any servos on the control boards? Say for camera control or for balloon release? What is the release mechanism? If we are using a hot-wire to melt the retainer, will the nichrome wire get hot enough? How will we drive the wire - MOSFET? This will probably require about 3A through the wire to melt plastic. 2A will melt foam.

A servo draws very little current when stationary, particularly when it is not receiving PWM position updates. A medium-large servo can draw up to 250mA when fighting a load to hold a position. In view of this, it is probably okay to put a 5V linear regulator to power the servos if necessary.

Daniel
Rris
Posts: 66
Joined: Mon 12 Oct 12 2009 9:17 pm

Re: Cygnus Module development

Post by Rris »

I dont think we need a balloon release mechanism. It will just pop when it gets high enough.

Personally, i don't believe we need pan on the camera, tilt could be interesting though so you can get a down facing shot of the ground and a shot of the horizon. The only question is, how will servos (specifically the lubricant in them) work at 30km? If the lube freezes, servos just going to strip some gears.
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Cygnus Module development

Post by Daniel Wee »

One way to mitigate such a condition would be to monitor the current draw when executing a tilt. If the current exceeds a certain number, the move can be aborted. The problem, though would be the need to provide a sufficiently beefy 5V supply.

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Cygnus Module development

Post by Daniel Wee »

Hmm... some interesting developments and information have been ensuing from other high-altitude balloon attempts. Of greatest interest are the facts that the TMP102 was used to measure temperatures down to -50C, and the BMP085 down to around 9kPa (rated to 30kPa). We also have some information about the LS20031 module.

http://www.sparkfun.com/news/389
http://www.sparkfun.com/news/503

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Cygnus Module development

Post by Daniel Wee »

It may make sense to run the I2C sensors off a 3V rail that we regulate more tightly. Most of the sensors appear to work at 3V. In fact the Melexis IR sensor specifically needs 3V (although we may no longer need this if the TMP100 proves to work at such low temperatures.)

Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Cygnus Module development

Post by Daniel Wee »

Preliminary board layout complete - sized about 100x55mm. I still need to work out the mounts, the power connector and the RF impedance controlled traces. This is a 4-layer board.

Also need to check PCBCART's drill and width limits.

Daniel
Attachments
Cygnus1.jpg
Cygnus1.jpg (101.54 KiB) Viewed 26052 times
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Cygnus Module development

Post by Daniel Wee »

For the antenna connector (may need to use IE):-

http://chemandy.com/calculators/microst ... ulator.htm

It looks like using a thinner PCB will help with lowering the microstrip impedance. I'm wondering if using a co-planar trace style will be a better approach though since microstrip calls for fairly wide traces.

I should note that since our working frequency is in the 70cm (700mm) wavelength, and our trace length is probably less than a hundredth of this, it probably isn't very critical to get the impedance right. Nevertheless, in the name of good engineering, one might as well get it right while at it.

Daniel
Post Reply