Need for an audio interface with fader capabilities

A lot depends on your intentions.
 
If it's simply mixing with faders - you can do that with an inexpensive live mixer - there are lot's to choose from behringer, mackie, and others. In this scenario, Sonar becomes nothing more than a stereo recorder, which is overkill. You could use something like Audacity much more economically.
 
If you are looking to mix within Sonar, than you will need both an audio interface and a midi controller with faders. There are several maker's of interfaces - I'm partial to Motu, but interfaces with 8 pre's aren't exactly cheap. The most inexpensive controller with fader's is Behringer's BCF2000, the market goes up from there.
 
There's nothing that I know of that supports both multi-track recording and midi flying fader control that is even remotely inexpensive.

You may have 2x BCF2000 (someone can confirm ?).. but not Tascam or most inexpensive interface.
 
Only some models can be expanded. Those that I know are :
- Roland Studio Capture
- some Motu model
- others with ADAT (Scarlet, RME, etc.. )

The following applies to audio interface.. as I've no experience with the BCF unit yet.
 
Basically, you install the interface driver and set the driver mode in Sonar correctly. Each interface may require different setup (turn off, install driver, turn on..etc..) Best read the manual first. For audio interface you will most likely use the ASIO mode (rather than WDM, MME, etc..) but it depends on your system. Some system gets better performance by using other modes, but generally ASIO will be the better choice.
 
Sonar works best with Win 7 or 8 and a current PC spec.. (Tascam website mentioned the US 1800 is tested on Win 8 too), go fully with 64 bits if possible. Common problem is the pops and clicks during playback, which can be diagnosed by running a system latency test :
 
http://www.thesycon.de/deu/latency_check.shtml
 
And most of the time, it's due to wifi - turning it off solves most of the problem, or turn it off at least during Sonar session.
 
Play around with the audio interface buffer size - most defaults will work, but may not give you the best real time response. Not a problem if you don't have to monitor the audio real time, especially with vst effects. For example, if you're recording a band and everyone has their amp and you're just miking or lining in the audio - leaving the buffer higher should be fine and not having Sonar to echo the output.
 
If you're doing one track at a time, such as listening to Sonar playback while recording the track, echoing it's output with sound effects, then the buffer size needs to be setup just right, so you will get the lowest possible latency before pops and clicks set in. For this, you're probably targeting around 12 msec or lower round trip latency.
 
The other consideration is making sure your hard disk spin at 7200 rpm and not lower (green drive).
 
Those would be most of what I can think off at the moment .. at least some of my own experience when I got first got started.
 

 
 
 

Perhaps knowing the difference between direct monitoring and low latency monitoring will help a lot in your decision.
 
Direct monitoring : to be able to hear the audio as you play it. You can even achieve this for units that don't have this option at the front panel, by simply splitting the audio via a DI box, one line goes to the amp, the other goes to the interface for recording (one way). Ideally, you would want to record the audio dry and only apply effects for the line that goes to the amp, sort of a cue-mix for the performer. This is the common way to record when PC / audio interface performance is not great. Latency can be set higher to compensate.
 
Low latency monitoring : If the system is capable, you can save lots of space / hardware by having the DAW to provide vst effects and channel back the processed audio to the performer (two-way), while still recording dry signal. The processed audio can be routed to the main mix or a separate line out, depending how many extra outs the interface has. The cue-mix can sound the same, as it's using the same vst effects. 
 
Some better interface provides a separate cue-mix/effects option to improve latency. Audio is going to the DAW one way, dry, but the interface provide effects for the performer, bypassing the need for DAW to provide the effects during recording.
 
Low latency monitoring is most important if you're going to record midi, using a midi controller to trigger software synthesizer, building up one track at a time - dance music, orchestra arrangement,etc.. If this is your main recording mode, you'll want to have a good PC spec, and a better interface, perhaps just 2 or 4 ins/outs to keep cost low.
 
At the price point of Tascam US 2000, I would personally go with the Focusrite Scarlett 18i20.
 
From the previous forum discussion regarding Tascam interface (search the forum, you'll get many insights), the US 1800 is preferred over US 2000. At the price point of US 1800, there is no competition that has that many inputs.
 
The only issue the reviewer at SOS mentioned that can be solved is the driver setting, running at it's lowest latency, which he mentioned Tascam would have an update by now. The other issues are hardware design that can't be changed, such as the way of handling direct monitoring via an on/off switch for each track that does not allow individual track level mixing. And I suspect this is not an issue if you either do direct monitoring on a single track at a time, or you'll do a band recording and you do not need direct monitoring (from Tascam unit) since the band would have amps / PA and you're just DI-ing all the audio to the Tascam.
 
Mixer based audio interface may also be a solution for live band recording / mixing (such as Presonus StudioLive, Makie Onyx..) as it provides real time mixing / monitoring. There are inexpensive units from Phonic Helix board and Alesis Multimix, but the Alesis MultiMix is no longer in production and the newer Alesis MultiMix FX is USB 1.1 based, backward (only support stereo mix to the DAW).