In general, if running a traceroute out to an external address doesn't show 192.168.98.1, then you won't be able to coerce Ping Plotter into reporting on that specific router in the same instance that you're tracing to the external address.

Now, let's have a look at possible problems based on the information you gave me here (which is a bit light, but I can probably point out some ideas for you).

First off, if you really want to trace 192.168.98.1, then you can set up an instance of Ping Plotter tracing to that address - and another instance tracing to another address. This will give you the packet loss results you're after. In a lot of cases, tracing to your default gateway isn't a reliable way to troubleshoot, though - as many, many cable modems won't respond to ICMP echo requests on their default gateways.

What you probably want to do is to set up a trace to a very close Comcast address (like the comcast internal SMPT or POP server, news server, or some similar customer facing, but hosted-by-Comcast address). If you get poor results tracing to this, then that's a smoking gun that there's a problem - as everything you're using for network there will be owned by Comcast.

If you don't have problems with an mail/pop/news type .comcast.com address, then you'll want to trace out of their network to a reliable destination (ie: tucows.com / yahoo.com / or similar) and see where the times start to decay. If it's at a border between comcast and another provider (ie: Cable & Wireless, AT&T, etc), then it still may be Comcast's fault - as the border might be undersized - or there may be some similar problem with the connection between Comcast and the border.

If you're getting packet loss at hop 1 (or hop 2 if hop 1 doesn't respond) all the time - to all destinations, then the problem is possibly with your physical cable modem line (ie: something between your computer and the comcast headend in their local office).

There are a lot of possibilities. If you want to post (or email to [email]support@pingplotter.com)[/email] a picture of what you're seeing (with a bunch of samples collected - 20 at a minimum, 500 would be better), then we can have a look at it and offer advice.

- Pete