/.
When listening for allophonic variation, it is necessary to try and distance yourself from a language. For the purposes of communication, we need to be able to identify the various forms of /p/ in pot, spot, top and topcat as examples of one and the same phoneme. We thus learn to filter out the gross differences between these sounds. In studying phonetics, we must first learn to attend to these non-phonemic differences. This can be quite difficult.
The series /p,t,k/ can be paired with the series /b,d,g/. Usually, the first is called the voiceless set, and the latter the voiced. Some choose to use the labels fortis and lenis respectively. The reason for this ambiguity is that the distinction between a matched pair like /p,b/ is not as simple as the presence or absence of glottal vibration (phonation or voicing).
In syllable initial position (pat, top, cat) the voiceless stops
are aspirated. This means that pressure builds up
behind the stop while there is occlusion. Once the stop is released,
there is a strong and very brief burst of air (plosive burst) followed
by a short period of voicelessness, while the air escapes. This
period is called aspiration, and because it closely resembles a /h/
sound, an aspirated stop is marked with a superscript: /t/.
The "voiced" stops are not actually voiced in syllable initial position. Rather, there is little or no pressure build up during occlusion, so there is no plosive burst, and voicing starts as soon as the occlusion is released.
Contrast this with French, where "voiced" stops are actually voiced during the closure, while voiceless stops are similar to our /b,d,g/ sounds. Most productions of syllable-initial English /b/ should actually be transcribed [p], but we retain the /p,b/ distinction for convenience.
Contrast these two with Thai. In Thai, there is a three-way distinction between prevoiced /b,d,g/, voiceless unaspirated /p,t,k/ and voiceless aspirated /ph, th, kh/. The stops of English, Thai and French can all be neatly described with reference to the voice onset time (VOT). For prevoiced stops, the voicing starts before the stop is released. For voiceless unaspirated stops, teh voicing starts just as stops are released. For aspirated stops, voicing starts a deal after stop release. (This nice description does not quite capture the 4-way distinction found, e.g. in Hindi. We will return to this later).
The English stops /p,t,k/ can all appear after an /s/ in initial position: spot, stop, skate. Here, there is little or no aspiration. These stops are also shorter than those where no /s/ is present. The stops /b,d,g/ do not appear in this position. So here is your first controversial phonetic issue: why should we call the first stop in the word stop a /t/ rather than a /d/?
In syllable final position, these stops are often not released, especially if the next sound is also a stop. Listen to the /p/ in soap bubble. Is it there? One clue to the identity of an unreleased consonant is the length of the preceding vowel. Compare the vowels in pat - pad, back - bag, and the first vowel in rapid - rapid, camber - camper.
We can mark an unreleased stop thus: 
Can you tell the words rap - rat - rack - rag - rad apart if
the consonants are not released? If so, the vowel must somehow be
telling you not only what vowel it is, but what consonant is coming
up. This is your first example of the phenomenon of
coarticulation. In general, the speech being
produced at any given instant contains information about several
sounds simultaneously. We write the sounds in neat linear order like
beads on a string, but we produce them in a more complex
and interdependent fashion. For another example, find out what
position your lips are in for the 
sound at the start of
sure and sheep. Why are they different?
Now consider the /k/ sound at the start of the words keep and cool. These are made at quite different points in the mouth. For /kip/, the following vowel is articulated at the hard palate, and that is where the stop is produced. In a close transcription, this is [c]. For /kul/, the following vowel has a velar articulation, and so the stop is pronounced much further back. Compare now the /k/ in leak - luke. These are much more similar.
Stops not found in English include those with retroflex articulation,
/
, 
/. These are common in languages of India, and contribute
much to what we perceive as an "Indian" accent.
The uvulars /q, G/ sound to English ears like exaggerated forms of /k, g/, but in many languages (e.g. Arabic), they contrast with the velars.
Finally, the glottal stop, /
/, occurs in all forms of English, though it is not phonemic.
When vowel-initial words are pronounced in isolation,they start with a
glottal stop. It is also an allophone of /t/ in many dialects. This
is very common word finally in American English (cat, hat) and
medially in British or Irish English (better, butter). You may
find it hard to perceive the glottal stop at first, especially in
initial position. It is however phonemic in many many languages.