The signal integrity of SD cards can be marginal - the signals have to come off the SoC, down the PCB traces, through the socket, and into the SD card. SD cards, especially fast cards, can be clocked at up to 100MHz. MicroSD adaptors are often designed to be as cheap as possible, and may make the signal integrity worse. For example, I have a $1 USB card reader that will read MicroSD in its own dedicated MicroSD slot, but not in an adaptor in the normal SD slot (which reads full SD cards just fine). If you open up an adaptor, you find it's just a series of metal pins in plastic - no shielding or any other signal protection.
Now the 3ds isn't a $1 card reader, but the same issues may apply - the more connectors you have, the higher chance you have of pushing it over the edge in high speed modes when playing games. If you look at dmesg and see SD command errors, that's where you'll see problems if they occur. It's more likely if you're thrashing the card rather than people who just use them for moving photos about. In addition, with an adaptor, the 3ds has to go through the adaptor, through the pins and then reach the microSD card. This can generate a lot of heat if you're playing for long periods of time and has potential to "burn out" your microsd whereas when you're using an SD card you will have a lower risk.
Multiple contact points and trace length, as it relates to HEAT ONLY are only impacted by line resistance (not total line impedance in this case). Although in high current applications you can make an argument about multiple gold wipe connections, in this case it has no validity because the current levels are so low. The current in the power and ground pin are very small and the current on the data lines are extremely low as they are high impedance single ended simple data lines. Note that the connections on a usd card or card adapter are going to be on the other of about 10-30mohms per contact.
Incidentally it's not the clock rate of 100MHz that causes concern, but rather the rise and fall time that determines ultimate analog signal bandwidth from which transmission line concerns are derived from. In fact at 100MHz in FR4 PCB material you are looking at a wavelenth of about 175mm which is awfully long, but the true bandwidth could be 10x or 17mm. (note: going longer than the wavelength of the highest frequency is generally the rule of thumb in EE for signal reflections having a travel time such that they can reflect and interfere with the original intended signal, but in reality it's a lot more complex than that). In your case though, arguing that the sd card reader is blanketly a culprit is unlikely as the total data distance traveled in an adapter vs a true sd card are going to be comparable. Think about it fundamentally. You really think samsung is going to ship high data rate microsd cards with an adapter that has a very high likelihood of not working? (not that was rhetorical not looking for an argument of "yes they are that dumb and it happened to me once so it must be statistically significant"
Also keep in mind that SD cards are a 4 bit bus and the new SD specs (though likely not in use) support DDR (data clocked on high and low clock signal). So most people's cards are going to be running around 10-25MHz at typical data rates (especially if they are not using a USB3 or other high bandwidth card reader). BUT, we can speculate all day on what speed and clock the cards are running at on average. I'd venture to guess though that the 3ds itself runs at a much lower SD clock speed that the recent specifications, as is typical of embedded devices.
You won't find "shielding" in devices like this. It's not really relevant. The value of carefully laid out trace and planes is of course some shielding but mostly balanced impedance for differential signals. SD card signals are single ended. In fact too much "shielding" can increase the line capacitance which will cause integrity issues if too high. So it can easily be a BAD thing.
Also keep in mind that SD card data transmissions use a CRC-7 so although you can deteriorate the data integrity such that data can't get through. You are unlikely to have "corrupted" data wiping out your card (as was the original concern in this thread which you responded to.)
Yes, I will grant you that if you get a crappy reader, and a crapper microsd card and a crappy adapter you are going to inject a finite likelihood of having a crappy experience, so in that sense you are absolutely correct in your assertions, but your base blanket argument that adapters are bad and will in general harm the performance lacks technical accuracy.
Sorry guys, didn't mean for this to turn into two guys soap boxing about a marginally related topic trying to prove who has the largest proverbial male "egg plant"
