We got a call earlier this week letting us know the assembled boards were ready to be picked up!
The build quality looked amazing, and from a first inspection it seems as though most components were placed correctly. The orientation of the LEDs was harder to determine due to their smaller size but those checked out as well.
With high density interconnects, visual inspecting BGAs and fine pitch QFNs is not possible so we had X-ray the boards.
LMS6002D X-Ray, this is a 112DQFN package
Some of the thermal relief VIAs wound up getting plugged, however it is evident that none of the solder leaked outside of the boundaries of the thermal relief pad. The darker spots that can be seen between the two rows of the DQFN are 0201 capacitors and resistors underneath the LMS part.
FX3 X-Ray (0.8mm pitch 121-pin BGA)
We were a little worried about the manufacturability of our boards when the PCB house told us we were getting close to pushing their capabilities, however the X-Ray shows pretty clearly that all of the VIAs underneath the FX3 are well defined and symmetric.
After making sure there were no shorts by checking resistance between the different power supplies and ground, it was time to power up the board. Luckily no sparks came flying out. After checking that all 7 key power domains were at the right voltage, the board was plugged in to a running laptop. The Cypress development tools quickly detected the Cypress USBBootloader that boots from ROM when the SPI flash does not contain a valid image. Within its first few minutes of operation, the Bulk Loopback benchmark utility was running and showing round-trip throughput of 250MB/s. Theoretically, the FX3 should be able to do nearly 400MB/s but Window’s XHCI driver has some catching up to do when compared to Linux 3.5.
Cypress’s development tools talking to the BladeRF
Lastly, builds of this run usually tend to leave a lot of left over components, here’s a few pics of the mess we’ve created:
Left over cyclone 4s
Left over component bags full of tapes, reels, and trays
This is another exciting week, we finally got a call from the PCB manufacturing house letting us know the boards are ready. Although we’re planning on using the same contract manufacturers for assembling all of the boards, we wanted to take a few of the extra boards home and examine their build quality.
As of the moment, there appear to be no shorts between any of the ground or power planes, and most signal traces don’t seem to have shorts with any of their neighbors! This kind of inspection is necessary because flying probes may not catch all these kinds of manufacturing errors.
After meticulously reviewing the schematic and the layout, and correcting all DFM issues, the boards have finally been sent out!
Everything looked good, however the PCB manufacturer did warn us about pushing the limits of their plating capabilities with the 3mil annular rings we used for the 8mil vias. We followed most good engineering practices so we have high hopes the first revision but in the end only plugging it in will tell how well it turned out.
Software defined radios (SDRs) are a popular topic amongst amateurs, novices and professionals in the radio field. They’ve been around for quite some time, and, in the last 10 years, have become significantly less expensive and increasingly available.
Recently, an inexpensive DVB-T / FM radio USB receiver dongle (RTL2382U) was found to supply raw complex samples when put into FM mode. Their software driver was doing the signal processing required to demodulate the signal to produce the audio. The dongle hardware was only being used as a tuner and Analog-to-Digital Converter (ADC). Moreover, the front end of the device, which does the RF tuning and downconversion, was found to cover a much larger frequency range than expected; approximately 60MHz – 1700MHz. It was then apparent that the dongle wasn’t just a DVB-T and FM receiver but one that could receive anything within the sampling bandwidth of the device with the right software. The device is capable to tune and receive FM, GPS, GSM, pagers, ADS-B, AIS, P25 and TETRA just to name a few.
Unfortunately, therein lies the rub. Given the length of time SDRs have been around, there seems to be very little end user software produced. For how popular the RTL-SDR is, very few tutorials and general knowledge transfer happens with regards to how to demodulate and decode these transmissions.
We hope to change that. We hope to ride the experimenters wave of Makers and Hackerspaces to help make software radio significantly more accessible to people who never even knew where to start. We have created bladeRF, a modern and inexpensive software defined radio, to help amateurs, novices and professionals alike. The world is mostly wireless and is becoming more so. We feel obligated to share our knowledge with everyone who wants to learn.
Welcome to your software defined wireless world. I think you’ll like what you find here.