Wave Bubble 2010 Moderators: adafruit_support_bill, adafruit

[c.b.]

I just looked again into the datasheet - you're right, the LMX2541 is not suitable for a wave bubble design. Although you can program a couple of frequency dividers thus obtaining a range of 50MHz to 4GHz, there are a lot of gaps in the frequency range. Quite the same thing as HMC822.

[Mictronics]

I made another main board version with two separte T attenuators for the PLL.

The attenuators are matched to 50Ohm output an then connected to the PLL. Here the input impedance varies from 250Ohm @ 300MHz to 10Ohm @ 2650MHz.
The attenuators will be calculated to have 0dBm ouptut level. With the 50Ohm load and the PLL impedance in parallel, the PLL will "see" a signal ranging from -1dBm to around -10dBm over the full frequency range. And this is well within the spec.

But thats all theory... ;)

[Mictronics]

@c.b.
What type of amplifier do you use in your wave bubble? And how are the attenuator pads configured?

I thought about the gain stage and would say the SXA-389B is the wrong choice for this purpose.

I made a comparsion between possible replacements.
About the parameters:
Gain = Small signal gain of the amplifier
P1dB = Maximum output power of the amplifier before saturation
Input = Required input signal level to get maximum output power = P1dB - Gain

Parameters taken for 1950MHz

SXA-389B
Gain = 13.6dB
P1dB = 25dBm
Input = 11.4dBm

SGA-6589
Gain = 20dB
P1dB = 19dBm
Input = -1dBm

BGA-6589
Gain = 17dB
P1dB = 20dBm
Input = 3dBm

GALI-84
Gain = 19.2dB
P1dB = 21.2dBm
Input = 2dBm

SGA-7489
Gain = 18.5dB
P1dB = 20dBm
Input = 1.5dBm

ADL-5320
Gain = 16.9dB @ 880MHz
P1dB = 25,4dBm @ 880MHz
Input = 8.5dBm @ 880MHz

As you can see the SXA requires a high input signal to get maximum output power.
If the attenuator of the original wave bubble is trimmed for 0dBm (to protect the PLL) then the SXA will not output more than 13.6dBm@1950MHz or 18dBm@850MHz which is 20mW and 63mW. Possible reason for a poor range.

The GALI-84 or SGA-7489 would be the best choice so far, a high gain paired with a high P1dB, but still requires 1.5-2dBm input signal.
This leads again to the solution with separate attenuator for the PLL and amplifier.

[c.b.]

Hi Michael,

I use a GALI-84 in the DCS band. What do you mean with attenuator pads? I removed all resistors in the T-networks and the PLL feed resistors and replaced them with a copper wire directly from the VCO to the gain stage capacitor.

Regarding the gain stage replacement parts, could you please add the ADL5320? I use it on the GSM band and it works quite well.

[Mictronics]

Attenuator pads = T-network

So you don't use the PLL at all?

Hmm, the ADL requires different external components for various frequency ranges.

What components do you use on your wave bubble?

[vitaminmoo]

Can you provide some specs or links were I can read about the high power cell chargers?

The bottom line is that if the data pins in a USB connector are shorted at <200 Ohms, the port is detected as a "Dedicated Charging Port", and can supply maximum current (generally 1.5A or 1.8A)

See
http://www.usb.org/developers/devclass_ ... ng_1_1.zip
for details.

You may attach the Eagle drawing here and I will see if I can include this as well.

I've changed my mind about the battery situation - most battery vendors sell the LiPoly packs with the limiting circuit already built in, and those have leads pre-prepared. Given that soldering directly to LiPoly tabs scares the hell out of me, and the price difference isn't huge, I've gone this route. The way I'd do it is to have some connector so the battery is removable without a soldering iron.

Right now if you have metal hardware, and you set the WB on a conductive surface, the battery is directly shorted (through the standoffs). This seems sub-optimal.
Given the antennas are gone, it seems as though removing the electrical uses of the standoffs would make sense, if the traces and four pin header can take the current. If the four pin header (power + serial) was also made so both sides were surface mount of some type, and the outside vias were solder masked, the general risk of shorting out the device would drop drastically. Not having a hybrid electrical, chemical and metal fire in your pocket seems worth it.

Regarding the power button - The other thing that bothers me about the current one is the loud click when enabling. It's rather conspicuous to cycle through several programs covertly. The button itself also has a rather huge footprint.

One thing that might be nifty is to have the power button be a right-angle tactile button that is recessed inside of the device just enough to prevent accidental activation. On the original lm3622 battery board, the coordinates 2850, 12 would be a good position. Right-handed people could hit it with their thumb, while lefties could use an index finger. If the switch-program logic was moved to a gpio attached to another switch, it could be symmetrically placed on the other side of the device. I doubt this description is clear - Let me know if you want a mockup to illustrate.

[Mictronics]

AS far as I understand the datasheet you provided the "Dedicated Charging Port" is related to the USB host or charger device, like a wall adapter.
But if you want to be compliant with the spec the portable device needs some kind of intelligent charging logic where D+ and D- are also handled to find out what type of host or charger is connected.

About the standoffs I was not even thinking to connect the battery there. This is just crazy, especially running the device on LiPo's.
So forget about it. ;)

A different board to board connector is possible but like for the switch there are many solutions.
The conenctor should be flexible some kind to cover a variable board to board space. depends on the battery people want to use. Maybe a single LiPo cell, two stacked cells in parallel or like me two round LiPo cells.

The last thing is the current which must be handled by the connector. For now power traces on the boards are designed to handle 1A minimum. That's needed for the connector as well.

I thought also about the switch but there are so many types just looking at Digikey as main supplier. May you can recomment something suitable.
What about these from Digikey CKN9629-ND, CKC5121-ND, CKC5124-ND? I would place such type of switch directly on the USB/Batt board.

Also your description is now completely clear to me, especially the coordinates, since I'm not very familiar with Eagle. :)
So some kind of illustration would be nice.

Anyway, thanks for the input so far.

[Edit]

I placed a right angle SMD tactile switch on the main boards lower side just below the right angle ISP port, line up with the edge of the board. This switch can be used to cycle through the programs.

About the power switch I'm not sure, the canditates mentioned above are way to big in size. :/

[TheFallen]

So, did we decide where to stick this stuff up? I'm thinking google code would be perfect in theory.

I've been looking into the VCOs and the calibration tables for my version are:

Code: Select all | TOGGLE FULL SIZE

// first row is frequency in 50khz steps
// second row is the DAC bit value that matches it
unsigned int calibrationLow[2][20] = {6930,7386,9642,10736,11840,12988,14194,15410,16606,17748,18820,19848,20838,21806,22752,23682,24578,25438,26244,27658
0x0000,0x0031,0x0148,0x01EC,0x028F,0x0333,0x03D7,0x047B,0x051F,0x05C3,0x0666,0x070A,0x07AE,0x0852,0x08F6,0x099A,0x0A3D,0x0AE1,0x0B85,0x0CCD}; // ROS-1300+
unsigned int calibrationHigh[2][20] = {24334,24596,27728,29308,30804,34218,37690,39506,41346,44832,47930,49440,50756,51886,52920,53830,54862,55454,56844,58050
0x0000,0x0019,0x0148,0x01EC,0x028F,0x03D7,0x051F,0x05C3,0x0666,0x07AE,0x08F6,0x099A,0x0A3D,0x0AE1,0x0B85,0x0C29,0x0CCD,0x0D71,0x0EB8,0x1000}; // ROS-2700-1819+
// it is assumed that between these points the VCO is linear

Basically, you get the frequency you want in 50kHz steps, e.g. 400MHz is 8000, find the next biggest one, in this case 9642 then map between 7386 and 9642 to get a value to feed to the DAC. The PLL would then correct the output until you got close enough.

I have to admit the more I look at this DAC version the more I admire the simplicity of Lady Ada's original Wave Bubble.

For my version I was going to use about 12V of lithiums to give me a nice run time at 1Watt.

[Mictronics]

Main problem with the DAC is the limited resolution.
With the PWM driven tuning voltage generation for the VCO we have a theoretical resolution of 65536 steps, means you need a 16bit DAC to be equal.

The 555 is needed anyway since the DAC is to slow while generating a sawtooth.

So a DAC would be usefull only for tuning of the centre frequency, all other components are still required.

Compared to 2 capacitors and 2 resistors for the PWM low pass filter, the DAC version would be much more complicated, 1 or 2 more IC's, additional components but the main problem, we are short on pins at the ATmega.

Also sweeping the whole frequency range of the VCO will be much faster with the PWM output than with a DAC. Just depends on the timer1 clock speed.

I will stick with the default design so far.

Just check the maximum input voltage for the DC-DC converters, the TPS79333 is rated for 6V only and the LT1301 for 10V input.
I think input voltage is not critical but capacity of the LiPo should be as high as possible. I was thinking of two round LiPo's in parallel, each 3.6V/2150mAh. Size of one cell is 65mm long 18mm diameter. They will fit on the USB/Batt board, even on my shorter version.

To replace the power switch I found a simple solution using a 2.5V digital MOSFET, a second tactile switch and some additional parts.
Both switchs are placed symmetrical on the lower side of the main board, left and right. Easy to handle with thumb and index finger for left and right hand people. See first post of this thread.

Also this solution includes now the auto power OFF feature on low batt.

[TheFallen]

Oh it's a complete redesign from the ground up.

The DAC has a higher resolution, but yes it's slow, but only because the Arduino/ATMega is.

[TheFallen]

I think the MCP4922 is probably not such a great DAC, going with my preference for Texas Instruments I've found the TLV5638. 12bit, 1uSec settling time, 2 channel and all in an 8pin SOIC package. Perfect. That gives me 10 steps at 20kHz, 1000 at 200Hz, add a proper low pass filter and I'm sorted.

The DAC increases the theoretical complexity, but in reality it makes the design much easier. I don't need to work out where to put the digital pot, where the NE555 is or use the PWM to generate a poor analogue voltage. I mean if you look at the wave bubble's schematics, the LMX2433 and the digital pot are both SPI controlled devices, but are not connected to the same pins, let alone the actual SPI pins. Trying to rewrite the code would be next to impossible.

As I said I'm doing a complete redesign hopefully producing a rugged jammer that can inhibit frequencies 400MHz - 2.7GHz effectively, it's nowhere near as complex as a military grade systems, but it'll still give most communications systems a real headache including civilian personal radios.

[Mictronics]

The TLV5638 seems to be indeed a good choice.

Running the PLL on hardware SPI might work but there is no dedicated CS pin. Not sure if the LE pin works like a chip select.
What I understood from the datasheet, as long as there is no transition from LO to HI on LE the data is held in the shift register only. So keeping LE low while writing the DAC might work.

[robcole]

Hi All,
In the original design Wave Bubble schematic(cigpack_main rc1a) has one side of
R90 connected to C36 - however the other side of R90 is open?????

[TheFallen]

I wouldn't have said that this is the best topic for an answer. In fact what has this to do with a discussion on upgrading the Wavebubble design?

I would suggest using the search function, which is located at the top of each sub forum to find your answer.

*HINT* try typing in the words next to the resistor.

Also, more question marks do not make the precious sentence more questiony.

[TheFallen]

Right back to teh topic at hand, I'm looking to upgrade the ATmega168 to an XMega32A4. This gives me 2 12 bit DACs in the same package, more pins, 32MHz at +3v3 and many other advantages. Coupling this with a +3v3 PLL ADF4212L I should be able to make a respectable Wavebubble.