ghunchu'wI' has already addressed some aspects of this, so I'll try not to double up too much. ghItlhpu' Daniel, jatlh:
While reconstructing the original Klingon lyrics to “You are my sunshine”, I wanted
to express “some nights ago” as a time stamp.
Before addressing your recast, as ghunchu'wI' pointed out, the paq'batlh gives {'opleS} as a term for "someday [in the future]" (paq'yav 2.8, 2.11-12). As such, {'opHu'} "some day in the past" should work, and to specify "some night in the past", {'opHu' ram}. jatlhtaH:
What I came up with was “Hu' puS ram”. Does that make any sense?
For my part, I think {Hu' puS ram} works perfectly well for "a few nights ago"; it carries an implication of recency that {'opHu' ram} doesn't, but whether that's a problem or not is entirely up to you. taH:
julwIj wov SoHbej, latlh jul vIghajbe'.
'eng law' ghajDI' chal, choQuchqa'moH.
bangwI' SoHbej, 'ach chaq not bISovlaH.
julwIj wov yItlhapQo' 'e' vItlhob!
My only critique on the refrain would be on the last line; I don't think an imperative can be the complement of an {'e'} construction like this. Safer would be {julwIj wov DatlhapQo' 'e' vItlhob}. But otherwise, majQa'! taH:
Hu' puS ram bangwI', jIQongtaHvIS jIH,
DeSDu'wIjDaq qa'uch 'e' vInaj.
'ach ghIq jIvemDI', jIQaghlaw'pu' jIH.
nach vI'eSmoH 'ej jISaQchoH.
Purely stylistically, a comma between {Hu' puS ram} and {bangwI'} in the first line would help to clarify that the two aren't intended to be read as part of the same clause: {Hu' puS ram, bangwI'} "a few nights ago, my love". Elsewise, majQa'. taH:
reH qamuSHa' 'ej reH qaQuchqu'moH.
jIHvaD 'e' Dajatlh je vIneHqu'.
'a jIH cholonchugh, 'ej latlh Danejchugh,
'opleS 'e' Dapay 'e' vIHarchu'.
While {nej} "look for" works okay here, for your meaning of "if you find another", {Sam} "find, locate" is better. Elsewise, this is all good too. I was wondering about the possibility of using {'e'} to refer back to a previous {-chugh} clause, but the paq'batlh also furnishes us an example of that: {pop Hevchugh quvwI' 'ej 'e' DaqaSmoHchugh} (paq'raD 16.26-27). taH:
chomuSHa'qu' 'e' wa'logh Daja'pu',
not nuchev vay' 'e' Daja'pu' je.
'ach DaH cholonpu', latlh DamuSHa'qu',
vaj Hoch vItulbogh Daghorpu'.
According to TKD, {'e'} as an object should come after {wa'logh} as an adverbial. The sole canon counterexample, as ghunchu'wI' points out, is problematic in other ways, though I suppose it at least allows an argument based on poetic licence. taH:
reH jInajtaHvIS, cholonqa'law' SoH.
jIvemDI' 'oy'qu' tIqwIj Do'Ha'.
naDev DacheghDI', choQuchqa'moHDI',
Hoch jIqeHHa', 'ej chopIchlaHba'!
According to the rule of {rom}, here {jIqeHHa'} should be {vIqeHHa'} to agree with the object {Hoch}. I've noticed this in a couple of your subsequent translations too - be sure you keep your eye not only on subject-agreement, but also on object-agreement. Overall, though, majQa'. I'm impressed. Have you considered having a poke at a rhyming translation? QeS 'utlh