I generally use somewhere between 35 - 60 ms. I keep this fairly standard front-to-back but I alter according to taste (i.e. try modifying the amount and listening again).

I got this from the book Mixing Audio where the author reminds us that samples and convolution reverb do not behave like reality (and so trying to model them that way often causes problems). For instance Early Reflections from a convolution reverb "interfere" with the original signal whereas in real life where should (in a good ambient space) blend with the original signal.

So ... I treat them like signals with reverbs and delays and use those rules to guide me. Anything much less than 30ms and you can get phasing problems, anything above 80 ms and it will start sounding like a separate signal (i.e. like a delay rather than a reverb).

In reality the predelays should be shorter but as I said, ERs from convolution reverbs will cause comb-filtering that wouldn't happen in the real world.

I hope this is useful but I'm sure others will disagree with me on this and have a completely different way of doing things.

Hi Robert

Theory

As you know, the "predelay" is the time between the dry signal and the arriving of the very

first reflected signal of a wall, the ceiling, or what ever.

Our ears and our brain tell us:

The longer the delay time (difference between the direct signal and the first reflections)

the larger the room, the hall.

So this parameter "predelay" allows us to extend the room impression a bit. 

A side effect is, that we can enhance the "depth impression" as well even if we

really adjust the distance between "us" and the closest wall so to say. 

If the pre delay time is > 60 - 80ms we can make out echos - and thats no more

a natural reverb.

BTW: 80ms means, that the closest wall is in a distance of 26m > that's very far... 

Really depths you mainly will get by balancing the dry/wet paramenter.

The result( depth) of all these adjustments can only checked with speakers.

Further: It all depends on the selected Hall or IR-Impulse. So it is difficult to give

exact figures in general here.

The only advice: Try to find your "distance" by adjusting all the parameters and by

keeping the predelay within 0 - 60 ms.

Have also a look in the free trial part of my tutorial 

Regards

Beat Kaufmann

 Thanks Dietz and Beat.  Let me try flipping my PD on the busses to 'hear' the results.

All the best,

Rob