I am sorry to insist, but I only saw instructions and comments about 5 kb fragments. Is there anything about fragments sized 20 kb or more (give my needs, this is quite critical)?
Never mind, I found it! (https://www.neb.com/~/media/Catalog/All-Products/0B8A6512699A4CA4A1C3C767D6DC8D5F/Datacards%20or%20Manuals/M0535S-0011110.pdf)
if you wish to test something w/o buying on the spot (and you happen to have Pfu and Taq at hand!) you can mix Taq and Pfu for either long-range or high fidelity mixes and run your PCR as if it was a Taq PCR (for long-range make a mix 1:4 [Pfu:Taq]).
I've successfully used Sigma AccuTaq LA to make 30kb products from E. coli genomic DNA. It's a mixture of Taq and a proof reading polymerase (as Joachim suggested).
I would definately look at the enzymes from Kappa Biosystems as was mentioned in a earlier post - we have found these ezymes to be excellent. Their standard enzyme easily amplifies up to 15 kb.
Brand like promega should have all those kind of products. depending on what your objectives are, you may choose a specific taq polymerae with specific function from Promega, USA.
20-30 kb is quiet a lot. Maybe the phage phi 29 DNA-Polymerase will work. Available from NEB or other companies. See Bianco et al. 1989, JBC 264: 8935-8940.
Thermo Scientific Long PCR Enzyme Mix is optimized for generation of very long amplicons: up to 47 kb with viral DNA and up to 21 kb with genomic DNA templates. The specially formulated Long PCR Buffer protects DNA from depurination and nicking during long thermal cycling. The products generated with the Long PCR Enzyme Mix are mostly 3'-dA tailed. Long PCR Enzyme Mix is also used for efficient amplification of GC-rich DNA regions.
Thermo Scientific DyNAzyme EXT DNA Polymerase is an optimized mixture of DyNAzyme II DNA Polymerase and a proofreading enzyme, capable of amplifying templates up to 40 kb in long PCR reactions.