Hi,

Well – the Netgear RTP change was a short lived success – my box was back to skipping and crashing within 12 hours, though it has to be said, miles better than before using RTP.  It was still unstable, skipping tracks, and so on, just not so bad.  However – I have now got it completely stable …. and the solution cost me pennies.  If you are handy with a soldering iron, read on – if you are not, then this message is not for you.

For the first time, I ran the box with the lid off, and I then ‘went nuclear’ in my approach to this POS – kill or cure.

I noticed that the Marvell Wireless chip and the MP3 decoder chip were getting warm to the touch when playing songs (though cool when not).  Ordinarily, I would not have considered either to be getting hot enough to be a problem, but given the heat problems I know some Netgear routers have been having (just Google for it), it is just possible …….

1)       So I took a passive heatsink off an old graphics card (who would want a PCI Riva TNT 2 these days ….), and epoxied it so that it spanned the Marvell, the MP3 and the tops of the flash rom and the DRAM chips (hard to describe, but the approx 1.5 inch square heatsink really does cover 75% of all four of those chips and doesn’t foul anything at board level).  Saved me having to cut down the heatsink.

I noticed that there are two positions on the right hand edge of the board (looking from the front panel) where power supply smoothing capacitors seemed to have been deleted from the production mainboard (though you can bet they were present in the prototype and development boards).  One of them looks to be intended to be smoothing the power supply to the components inside the metal RF shield, and the other possibly for power supply to the Microcontroller in the socket.

2)       So I soldered two 470 uF 16v electrolytics (the standard ones fitted to this board are 1000 uF 10v, but 470’s were the largest I had to hand… and something has got to be better than nothing) into the two ‘empty’ spaces on the mainboard.  Those caps *must* have been originally in the design for a reason, and so putting them back has to do some good methinks.  Perhaps removed to save money in the final design?

I noticed that if I held my finger over the solder pads where the aerial joins the mainboard (an old RF engineers trick to add a mix of capacitive and resistive load…), the signal strength stabilized according to the Link Info on the MP101.  I looked closely at the board and noticed that two SMT capacitors had been deleted from the mainboard around where my finger was poking – one of which (if present) would capacitively couple the antenna inner wire to what looks like a power supply chip ground pin.

3)       So I soldered a 100 pF (just a guess) capacitor between the antenna inner wire and earth on the mainboard (thus replacing one of the two deleted capacitors).  This helped a little, but when I replaced it with a 47 pF capacitor instead, things got a *lot* healthier looking.  This could be a missing part of an intended LC resonance circuit, with the capacitor deleted from the production version because it relies on the capacitance of the PCB (or just deleted by mistake) … just a hunch – my MP101 seems to like having the capacitor in place, and clearly the PCB made provision for one being in that location ……..  I didn’t bother with the other deleted capacitor, as it looked to be a part of a low pass filter (to block out RF higher than 2.4 GHz).  I have nothing transmitting higher than 2.4 GHz, and so didn’t see the point.

Guess what.  My MP101 is now *completely* stable.  It finds the wireless network first time, it doesn’t crash, skip tracks, freeze, or do any of the other ugly things it used to do to irritate me.  As bad as it gets now is that it sometimes takes two goes to find the server (Twonky).

My MP101 is now so stable that I have reinstated the original antenna, and taken off the high gain one I added as an experiment – so the box looks pretty much as it did when it left the factory – until you take the lid off that is, and notice the new heatsink, extra electrolytics, and ceramic disk capacitor J

Hope this helps someone.  It has saved my MP101 from the trash, and I believe returned the circuit to near ‘pre-production’, where it must obviously have worked well.

Thanks,

Andrew.