It is already long time, when I upgraded my Motorola SB4200 modem to SB5100.

So when sometimes happen that after my ISP cable line went down and the modem crash (one just cannot enter the modem menu, available at http://192.168.100.1 ) and I can't rest it over the net - well - I started suspect that the caps might not be as good, as they can.
And when recently I noticed too much packet loss, that even it can by my ISP fault (as it usually is, just check there for graphic display how my ISP sometimes "rule": http://trodas.wz.cz/index.php?act=ST&f=8&t=151&s= ) I still decided to replace the caps in my cablemodem.

Motorola SB 5100
–----------------------
1x 470uF 25V d10 4PZ            -> 13.85V, 209mV ripple                    -> 470uF 16V GC d8
1x 100uF 25V d8 4QK              -> 13.96V, 200mV ripple                    -> 220uF 25V GK d6.3
3x 330uF 10V d8 4LY              -> 5V all, 0.7 - 0.9mV ripple all          -> 470uF 6.3V GD d6.3
1x 470uF ???V d8 (r4, 470J, UD)  -> 2.516V, 0.6mV ripple                    -> 1000uF 6.3V GC d8
1x 150UF 10V d6.3 4JZ            -> 3.346V, 0.6mV ripple                    -> 330uF 6.3V FM d6
1x 47uF 16V d6.3 49K            -> 2.519V, 0.6mV ripple                    -> 150uF 6.3V FM d5
2X 22uF ???v d4 (22, r6.3)      -> 5.16V, 1mV ripple + 1.202V, 0.6mV ripple -> 68uF 6.3V FC d4

The catch? Well, all these Chemi-con caps are SMD ones, while the replacements are radial.

Luckily, there is not too many of them:

So, I just bend their legs to the sides, cut them to match their SMD soldering place size and solder them like that:

It is maybe not best-looking solution, but it works and the net seems now faster and less packets seems to be lost - but that can be just a coincidence, or I imagine it 😛

Regardless, at least I now know that fault can't be in my modem 😄