I used to teach the futility of language acquisition via anatomical instruction, by teaching students (in words) how to produce the ejective [p'], which nobody could so, and then I said "Do this" and pronounced [p'ə], and probably half of them could do it. The point is, you need a preferably interactive model using a real language and a speaker thereof. Hearing and mimicking examples of [h] versus [ħ] in Arabic is the easiest way to get the difference, because there are a fair number of Arabic speakers out there. Plus, Arabic has both [ħ] and [h]. I would even recommend finding a bunch of speakers from different places, so that you can extract the "sameness" that underlies individual differences in pronunciation. (That could backfire, however).
Then you might branch out and get examples from speakers of Somali. However, you are likely to discover that "ħ" is not uniform across speakers and languages, and IPA in fact recognizes two different sounds, ʜ (the epiglottal trill) and ħ (the pharyngeal fricative). There is phonetic literature on the production of the Arabic consonant, and Ladefoged & Maddieson observe that they are actually epiglottals, not pharyngeals. If you can locate a speaker of the Burkikhan dialet of Agul, you ought to be able to get suitable training in the difference between pharyngeal and epiglottal fricatives.
Learning to hear the difference should be easier, by listening to minimal and near-minimal pairs. Since this distinction plagues Arabic language teachers, finding materials, there should be training materials out there somewhere. In the interim, you can use Forvo to get samples to listen to (search for ح and it will return words with ħ somewhere, search for ه and you will get h which unfortunately includes a bunch of final h's that are not pronounced).
