I'm afraid they have some protection system that detects scraping and after one or two downloads that work ok server starts to reject requests.
You could try the recipe from some other IP address and placing this in your code:
Code:
simultaneous_downloads = 1
delay = 4