Re: [Oberon] Updated RiskFive FOM schematics
Hello Wojciech,
I added a pin connector for plugging in the NRF24L01 wireless module into the FPGA On Module (FOM). The updated schematics is available on the Schematics page on RiskFive.com. The schematics is posted for review.
Thanks. These days, my first concerns with any new electronic machine is reliability and repairability. Too many gadgets are built with poor quality, barely in spec. components, surface mounted. After a few years a board fails. There is no straigthforward way to identify the failed component. Replacing a surface mount component is tedious and error prone. Replacing a many-pin device just to find whether it is faulty is too time consuming. Conclusion: a board with components mounted in sockets and through hole soldered would have much more appeal than fragile components surface mounted. Perhaps simply not feasible in our world and I should forget modern electronics. No offense intended. Just an frank comment. Best Regards ... Peter E. -- 123456789 123456789 123456789 123456789 123456789 123456789 123456789 Tel: +1 360 639 0202 Pender Is.: +1 250 629 3757 http://easthope.ca/Peter.html Bcc: peter at easthope. ca
Hi Peter,
These days, my first concerns with any new electronic machine is reliability and repairability. Too many gadgets are built with poor quality, barely in spec. components, surface mounted. After a few years a board fails. There is no straigthforward way to identify the failed component. Replacing a surface mount component is tedious and error prone. Replacing a many-pin device just to find whether it is faulty is too time consuming.
Conclusion: a board with components mounted in sockets and through hole soldered would have much more appeal than fragile components surface mounted. Perhaps simply not feasible in our world and I should forget modern electronics. No offense intended. Just an frank comment.
I both agree and disagree with you. :) Obviously the intention with surface-mounted devices (SMD) is that boards can be produced so cheaply that you just throw them away - neither you nor I like that wastefulness, but it's certainly cheaper than the time to diagnose it (time being money). Also, sockets have their own unreliability. And SMD means you can lay out a lot more on a board - if the board is so big it flexes a lot, that's another unreliability. But when manufacturing in the hundreds of thousands, the temptation is to use "only just" enough solder, and too much is often problematic in automatic processes - so this may be giving the principle of SMD a bad rap. On the other hand, I've watched with interest and admiration as what is now called the "maker" community grapples with the unavailability of through-hole parts. They even have great success hand-soldering ball-grid array chips these days. Maybe the way forward is to use SMD, but design for hand-re-workability wherever possible. Just my two .5mm-spaced pins :) Paul
Paul: apart from automated manufacturing, another reason for SMD is bandwidth. An SMD part which is sitting close to a ground plane has much lower inductance than the leaded part. Lower inductance translates into a higher toggling rate and lower electromagnetic emission. Wojtek
Paul Reed wrote:
On the other hand, I've watched with interest and admiration as what is now called the "maker" community grapples with the unavailability of through-hole parts. They even have great success hand-soldering ball-grid array chips these days. Maybe the way forward is to use SMD, but design for hand-re-workability wherever possible.
I assembled most of my prototypes by hand, including QFN parts with 0.5 mm spacing and a ground pad hidden under the chip. An interesting exercise was soldering power nano modules whose pads are entirely under the chip (part number LMZ10500SIL, take a look at the data sheet). Another interesting exercise was AD9650, which is a QFN part costing about $150 a piece. With time and practice I became pretty good at this kind of assembly, though reworking 0201 parts gives me goosebumps. If you ever looked at these then you will know what I mean. I would never assemble a BGA myself. I witnessed a couple horror stories in the university lab equipped with professional tools, but without proper infrastructure. After watching the failures I know where to stop tinkering and when to turn to professional services. Some reflow processes require nitrogen gas flow, for example. I would not try this at home using a toaster oven. I would not try BGAs with a chinese reflow oven, which I have used for the QFN assembly. It is a fine device which I used more than once. But one has to know where to stop. I have had my share of disappointments with some American assembly houses, including one from Silicon Valley whose name I do not even want to mention. I have had a horrific experience with the original manufacturer of the BeagleBone. They went bankrupt soon after they screwed up my boards. RiskFive boards will be assembled professionally. I have had good experience with chinese companies. Their work is reasonably priced and they are very professional. There are number of fine companies out there who focus on short runs. I have had good experience with these. I will use their services again. W.
Peter: thank you for the comment.
These days, my first concerns with any new electronic machine is reliability and repairability. Too many gadgets are built with poor quality, barely in spec. components, surface mounted.
Use a reputable manufacturer with proven track record. Manufacturers are not equal. Some do lousy work, some do excellent work. A well assembled board is very reliable.
After a few years a board fails.
While anything can fail, this statement is generally not true. It is only true statistically.
There is no straigthforward way to identify the failed component. Replacing a surface mount component is tedious and error prone. Replacing a many-pin device just to find whether it is faulty is too time consuming.
High reliability boards can provide board diagnostics. For example, my most recent 40-channel digitizers monitor all the currents and voltages on board. We will record a database of monitoring results over the course of our underground experiment, which is five years. Less critical boards, like RiskFive, do not provide such extensive monitoring features. However, you can monitor internal FPGA voltages and the temperature, using its built-in ADC. The appropriate software needs be written, of course. You can then project an imminent board failure, if the FPGA temperature raises for no good reason.
Conclusion: a board with components mounted in sockets and through hole soldered would have much more appeal than fragile components surface mounted. Perhaps simply not feasible in our world and I should forget modern electronics. No offense intended. Just an frank comment.
This conclusion seems logical, but it is wrong. Sockets do not provide the bandwidth which is required by modern interfaces. A standard ball grid array (BGA) pin can operate at over 1 gigabit per second per pin. Special mutligigabit transceiver pins (MGT, which I am not using on RiskFive) can operate at gigabits per second, as the name suggests. A socket pin can provide a couple hundred megabits per second at the best. The modern surface mount technology was developed for a reason. I suggest reading the following reference to clear the confusion. It is available from Amazon. "High Speed Signal Propagation: Advanced Black Magic" by Howard Johnson and Martin Graham. Thank you, Wojtek
16 jan. 2018 kl. 17:21 skrev <peter@easthope.ca> <peter@easthope.ca>: […] Conclusion: a board with components mounted in sockets and through hole soldered would have much more appeal than fragile components surface mounted. Perhaps simply not feasible in our world and I should forget modern electronics. No offense intended. Just an frank comment.
Best Regards ... Peter E.
+1 … for what it’s worth from a lurker ( with 50+ years worth of experience from ’embedded’.) —
participants (4)
-
Bjarne Bäckström -
Paul Reed -
peter＠easthope.ca -
Skulski, Wojciech