Wave Bubble 2010

[Mictronics]

I'm doing a rework of the Wave Bubble hardware.

Schematics and layout for the USB/Batt and main board are already done. Since a prototype is not made yet there is still space for changes and improvements.

I changed a couple of things so far:

- using a ROS-1300 and ROS-2700-1819 to cover the whole range from 400MHz up to 2700MHz
- the voltage for the VCO is selectable by jumper, either 5V or 12V to keep compatibility to the original VCOs ROS-1200 & ROS-2500
- using a GALI-84 or SGA-6589Z
- input voltage for the TPS79333 is selectable by jumper, either Vbatt or 5V
- using the low batt voltage detection
- there will be no internal board antennas since I have real doubts that they are that efficient

Size of both PCBs is 77mmx52mm, means 5mm shorter than the original, of cause battery size remainded the same 65mmx35mm.

USB/Battery Charger board is finished:

USB/Battery Charger board, Top

usb_batt_top.jpg (155.84 KiB) Viewed 10929 times

USB/Battery Charger board, Bottom

usb_batt_bottom.jpg (126.85 KiB) Viewed 10929 times

A LTC1730-4.2 is used for battery charging. Here the charge current is limited by the power supply/wall adapter. The overcurrent protection is set to 2A. An NTC is used for thermal protection.

Main board, top layer

Main board, Top

IMG_83512.jpg (175.76 KiB) Viewed 9392 times

See Wave Bubble wiki for additional details and downloads: http://ladyada.net/wiki/wavebubble/wave_bubble_2010

[vitaminmoo]

I'm interested.

Things I have been pondering changing on the design:
- Removing unused pads and built-in antennas, and trying to get it down to one board (PCB is awfully expensive to use as a case material).
- Remove the DC power jack, and add support for high-power cell chargers (TX and RX are shorted at <200ohms to tell the charging device that it's ok to pull up to 1.8A). This would also likely line up with changing the USB connector to micro-B instead of mini-B, as micro-B is the new standard for most cell manufacturer's equipment that supports the previously mentioned high-power charging mode.
- Changing the power button to something more indicative of its status
- Changing the battery pads to pads designed for direct-mount of the batteryspace protection board (I've already done this, and I'm happy to send you the Eagle part)
- Making the AVR programming port right-angle and available from the side (unless the top PCB is gone), so you don't have to file it down, and it's available for re-programming without disassembling the device.

Things I'd like to see:
- Active and/or community maintenance, so errata can be fixed instead of scattered throughout the forums.

I'd be happy to give input/help as soon as you make the files available.

[Mictronics]

PCB antennas will not be included anyway and the RF section is already optimized.

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

What about an illuminated power switch? Or better changes to a push button with extra components to have some ON/OFF logic controlled by the AVR?

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

I will change the ISP connector, that's a good point.

But everything on one board isn't that easy. I think not archivable without changing the size of the board.

[TheFallen]

To be fair, PCB may be expensive as a case material, but it's cool as hell.

As the internal antennas are pretty rubbish removing them is lovely. As a side effect you could use lasercut acrylic/plexiglass etc for a cover.
The power switch is very hard to find, try to use a more common one.
Make it a bit fatter so it can easily use the Sparkfun 2000mAh batteries, or 3 AAs.
Upgrade the ATMega88 to a ATMega328 for super-awesome!

As for mainainance, Use Google Code to maintain the source files for uC/PCB?
For ISP connector use an edge-mount pad design, that way there's no need for 2.54mm header pins.

I love the Mini-B idea, I've got a Nokia with the new charger, so that'd suit me perfectly.

I was already redoing the design as a bit of a mental exercise, but instead of that messy 555 I was going to use a pair of DACs to control the VCOs.

I was also looking at this: http://www.hittite.com/products/view.ht ... MC822LP6CE but it seems like it only covers 3 of the 4 mobile bands, and only those bands with gap in between them.

[Mictronics]

Using two DACs or a dual DAC would be nice.
We would have full control of the VCO and could save some more parts like the 555 circuit, one LM358D, the AD8402, L5/6, C22/23, R8/9 and R17/R18.

What about an MCP4922? Dual DAC, 12bit, SPI interface and setup time of 4.5us which means maximum of around 200KHz change rate, single 5V supply.

By the way, the ATmeag48/88/168/328 are fully compatible in package and pinout, so you can put in whatever you want.

[TheFallen]

The only downside with a DAC is that you'd need to update it at a higher frequency than what you want the output to be.

e.g. the current Wavebubble has a maximum sweep frequency of 200kHz. To recreate that we'd need to have the DAC sweep the output voltage over the range in 5uSec thats 4096 12 bit values in 5microseconds. Of course we can skip steps, say every 4 step but that still 1024 steps in 5uSec. This then assumes that the microcontroller can do this and that the VCOs will output this exactly, no PLL to make corrections. Add in a PLL and you'd need something with a bit more speed, like an ARM processor.


Would be nice to re-jig the 555 design though, use separate output buffers for the VCOs control voltages, individual op-amps that would let you set the op-amp at either 5V, 12V or 28V it would then scale the 555/PWM output for best results from that individual VCO.

[Mictronics]

From where you ttok the 200KHz value for the Wave Bubble?
The schematic and the design guide says something about 20KHz max.

But you are right, bottle neck is the DAC in this case. but you will hardly find a cheap and simple DAC with settling time less than 1us, the 4.5us is already a good value compared to the price for the DAC.

[TheFallen]

Ah yes, my bad, it is 20kHz, so that 50uSec. My point was not that a capable DAC would be difficult, there are TI DACs that will do 1gigaupdates/sec at 16bits. No, the problem we have is communicating fast enough.

One thing I'd love in a Wavebubble design would be to reduce the number of voltage rails.At the moment there are:
+28V
+12V
+5V
+3V3
+2V5

I think the best way to do this is to find a freely available +3V3 PLL. That gets rid of the LMX2433 which is the most expensive/difficult part to get. I'm looking at £20 per chip. It'd be nice to be able to drive the cost of the Wavebubble down.

[Mictronics]

I think a 5V PLL would be even better. Together with two 5V VCO there is only need for 5V and 28V supply. This would save a lot of space and parts.

[TheFallen]

a 5V PLL would be hard to find, but it would it would make life easier, at the moment it's teh Digital Pot and 555 timer which use the +3V3 rail.

One thing to remember is that it's easier/more efficient to go down in voltage. I'd love a +3V3 Arduino but that would mean we could only run at 8MHz.

[Mictronics]

Another thing came in mind:

The T attenuator between the VCO and amp should protect the PLL and gainblock from high input levels. Especially the PLL is rated with a maximum of 0dBm.

While I was calulating values for the VCOs I'm going to use I found that the attenuator will work for the gainblock, may not for the PLL since it's connected inbetween the T pad and not on the attenuator output.
A simulation showed that the level feed into the PLL would be higher than 0dBm if the attenuator is setup for 0dBm output.

Example:
Attenuator: 22Ohm Rs and 47Ohm Rp
8dBm attenuator input
-0.18dBm attenuator output to gainblock
3.0dBm PLL output

The simulation is maybe not correct since I found nothing about the input impedance of the PLL pins FinIF and FinRF.

Any ideas?

Souldn't be the PLL connected on the output of the attenuator? In this case the PLL and gainblock gets a 0dBm signal which is fine for both.

[c.b.]

Hi Michael,

why do you want to feed the gain stage with 0dBm in the first place? As far as a newbie can read in the datasheet, SXA-389 is rated for 100 mW input power, BGA6589 has a "maximum drive power" of 15dBm. A direct feed of the gain stage by the VCO's should squeeze a little more power out of the wavebubble, shouldn't it?

[Mictronics]

The maximum drive power is the max limit the device can handle on the input.

But the output power is not increasing linear with input power . At a certain point of input level the device goes in saturation which means the ouput power is constant or even drops.

In all datasheets you will find the parameter "Load power at 1dB gain compression". That's the point where the input power causes a 1dB drop in the linear gain due to device saturation. Means you can not get more power out of the gain stage.

In the GALI-84 datasheet the saturation power is noted to be 22.6dBm @ 1GHz, you will not squeeze out more.

Anyway, your limiting factor is the PLL with 0dBm.

[Edit]

I found the spec for the PLL input impedance. But for the bandwidth we will use the impedance varies so much it's impossible to make an attenuation pad.

On the other hand I neither want to fry my gain stage nor the PLL, so sticking with the 0dBm version will be the safe way.

[c.b.]

Thanks for the explanation! I'm absolutely not satisfied with the range of my wavebubble (about 10cm in the DCS band). I should remove the PLL feed resistor and skip the attenuator.

@TheFallen: you mentionened the HMC822LP6CE and pointed out that there are gaps in the range of that device. What about National's LMX2541? It can generate frequencies as low as 31 MHz. My only concern is the SMT package I can't solder.

[TheFallen]

From what I can tell of the data sheet the LMX2541 has a bandwidth of 104MHz. It may be able to generate signals down to 31MHz, but it then only can generate up to 135MHz. To make it generate 400MHz to 2700MHz I'd have to spend considerably more than an equivalent VCO design.

I'm rather taken with the idea of a ruggidised jammer using a DAC and the VCOs (ROS-1300+ and ROS-2700-1819+) and a theoretical 5V dual PLL with a output of 1Watt on the high and low bands. Of course that'd going to take a lot to design, but hey, it'll be FUN.