I have configured three different HP CD-writers for use with EAC. I have used the "auto detect drive features" for all of them and all have shown "no" cache. When I look at the drive database at http://www.ping.be/satcp/eacoffsets01.htm#- two of the drives are reported as rebadged Sonys. (The third isn't in the list) The Sonys however are reported to have audio cache.

Can I trust the auto detect feature of EAC or not?

I have ripped a number of CD:s with the audio cache set as "No". Some of the rips have had correctable and some also uncorrectable (however unhearable to me) errors. Would that mean that the drive in fact doesn't cache audio data?

Detecting cache with EAC should not be considered reliable (for example: it happened that 2 tests gave me 2 different results )
I suggest the following to 'test' your cache setting ('NO disable'):
Rip a few CDs in test & copy mode (action > copy selected tracks). After EAC finished, check the 10th column, which should say 'OK' for all tracks (in fact what it does is comparing CRCs for the test and the copy read)
There shouldn't be any '#' (unless by extreme accident). If you do see '#', then your cache setting was not correct, so switch to the safe but slow 'yes disable cache' setting.

OK, I'll try that. But I know for a fact that I have ripped CD:s (with cache="No")where errors have been corrected (Track quality != 100%). Isn't that proof enough that the drive doesn't cache? If it did cache wouldn't I get Track quality=100% all the time?

I discovered that Feurio! tests the drive caching capabilities (as well as accurate streaming) more thoroughly, compared with EAC ... In addition, Feurio! shows the big advantage of logging the drive features during testing ...

If you check the "Drive Caches Audio Data" box in EAC, you will be on the safe side, though ... and remember that there can be different firmware versions that affect caching ...

No. If I uncheck "Drive caches ..." here (Plex PX-W1210A which caches, C2 off) it still finds errors on scratched CDs but the error correction re-reads are "performed" from cache so the 1st line of red lights (16 re-reads) flashs in < 1/2 second. Nevertheless, track quality is < 100% if pseudo-error correction kicked in. So I guess correct cache setting is most important for error correction while error detection still works if "Drive caches ..." is unchecked erroreously (at least with some drives).

Hmm, OK. But I also had some experience where the error correction failed, no matter how many retries. Would that prove that there is no caching?

Does someone know the correct EAC settings for the Sony DRU-510a
DVD+-RW drive?

Your thought sounds logical, but honestly I can't tell. Maybe it's different on skips / bad scratches where the drive loses sync... . If you don't trust EAC's or Feurio's findings just do what liekloo and JeanLuc suggested.

OK I will test liekloo's suggestion. Thanks everybody for your help!