The main problem with a kana transcription of Spanish (let's call it KTS) is the complex syllable structure. KTS would have to be able to adequately render two types of syllables that cannot be found in Japanese: 1) closed syllables (except those ending with a nasal), 2) consonant clusters in the onset.
For the closed syllables, KTS will be acceptable (in the sense of producing a phonetic result that would be easy for the Spanish hearer to recognize) if the coda consists of an unvoiced consonant such as /p/ or /s/ and it's followed by another syllable that begins with an unvoiced consonant, since short Japanese u between unvoiced consonants tends to be devoiced and to Spanish ears it practically disappears, so for example パスタ pasuta would be correctly heard as pasta. With voiced consonants the following vowel would not be devoiced and the result would be off.
The consonant clusters in the syllable onset are a bigger problem, because these consist of a stop or fricative plus a (voiced) liquid, so no devoicing would take place in the Japanese reading, and the vowel would be perceived by the Spanish hearer.
The liquids will be a problem. A Japanese speaker can be trained to distinguish /l/ from /r/ (trill) and from /ɾ/ (flap) but there's no way to distinguish those in standard kana (maybe a modified R series —ラ with a dakuten for /la/, for example— could do the trick). The Spanish ll on the other hand should not be much of a problem, since in most Spanish dialects the sound of ll has merged with the one represented by y (yeísmo).
The Spanish /f/ could also be a problem, but Japanese has for the most part adopted a way to represent /f/ in loanwords with the H series kana.
The dental pronunciation of z and c is not representable in kana, but it could arguably be left out without much of a problem. The majority of today's Spanish speakers do not employ that sound, and its merger with /s/ only rarely produces ambiguity.