Wheresthedug - I am impressed! 20mS of latency is 40 ticks at 125bpm. Or to put this in real world terms, that's a perceptible snare flam or like someone clapping slightly out of time.
Here's another experiment for you to try:
Set up a midi drum loop at 125bpm. Then move ONLY the Snare backwards (or forwards) 40 ticks (=20ms) and try playing along with this. Alternatively, move ONLY the Kick (backwards or forwards 20mS). Or if you prefer leave the Kick and Snare and only move the Hi-hat 20mS. Then try playing along with any of these permutations.
Another variation is to have a click going and have the entire drum midi either +/-20mS and try tracking with both click and offset midi drums.
If you can cope, as I said earlier you're blessed. I can't!
People say to me: "what the heck, 20mS is 1/50th of second". I reply, "yes". But equally, at 125bpm a semiquaver (1/16th note for US friends) lasts 120mS. So 20ms is 1/6th of a semiquaver or, put another way, 20mS of latency is like having semi-quavers played approx. 17% out of time. When I think of latency this way, I feel less bad about my inability to deal with it!
Here's what some other people - Yamaha - say about latency:
http://www.yamahaproaudio.com/…ter5/05_absolute_latency/
In summary, to save the hassle of booting up the link:
Signal path latency for in-ear monitor system:
1.15 - 2 ms Playable without any big problem.
2 - 5 ms Playable, however tone colour is changed.
5 - 10 ms Playing starts to become difficult. Latency is noticeable.
>10 ms Impossible to play, the delay is too obvious.
Kemper's latency, according to the manual, is 3.5mS or (I believe) if you hit the constant latency button 4.5mS. And this is plugging a guitar directly into Kemper and listening on headphones also plugged directly into Kemper. If one then goes via an audio interface, even with direct monitoring another cycle of ADC / DAC conversion is added and you're up to the circa. 5mS where, according to the Yamaha-latency-Beaufort-scale, latency starts to become a problem. This is Kemper.
For comparison, the declared latency of AmpliTube 4 is a mere 0.2mS. However, the problem with AT4 (and its ilk) is that to use it you need to software monitor. And for my set-up (probably not untypical) round-trip latency @64 samples I/O buffer is 5.8mS (even at 32 samples it's 4.3mS). Add the 0.2mS AT4 latency and (at 64 samples) you're up to 6mS cumulative latency and into storm-brewing-Yamaha-Beaufort-scale latency.
The only solution I've found is to split my guitar signal (analogue-ly) and send one split to my 1998 POD and the other to the Kemper. Each split then goes into the Apogee on Direct Monitoring and I monitor only the POD sound and record the Kemper. POD is not latency-free, but it's better than Kemper. And this produces monitoring sound which is (to my ears) in time with my playing. I can then re-position the Kemper track to compensate for latency afterwards. OR I can take it one stage further, and monitor with POD and record a dry DI and re-amp the latter (with AT4, GR5, Waves GTR3, Kemper or real amp). But, at the end of the day this is 1) a hassle, and 2) sobering that latest technology Kemper has an effective cumulative latency performance that is worse than a 20 year old POD.
I know I seem to have a gene in me which objects to latency, and this is a cross I must bear! But, equally, the above monkey maths are the facts.
More subjective is that, in dusting down my old POD for the above operation, I was struck by the other problems mentioned above in relation to the Kemper sound: lack of attack transient, flat sound, too much bass, too little harmonics and so on. POD may be 20 years old, but it stands up remarkably well. It it not perfect by any means - for example, there is a weird notch EQ dip about 1.3kHz on some Marshall presets that you don't get on the real thing or AT4 / GR5 or indeed Kemper Marshalls. This is a bit like that 750Hz notch on a bass SansAmp Bass DI which - to some tastes - mangles the sound of bass (and the new SA Bass DI now has a mid control added). Strange.
Anyway, thanks for all the posts. It seems like a) I am in fact not doing anything fundamentally wrong, b) there is a latency problem with Kemper if, like me and Yamaha, you notice these things, and c) the "sound issues" are subjective.
On c), if anyone knows of profiles that are recorded with just an SM57, or SM57/MD421 pair or similar i.e. the old fashioned way (no ribbon mics, or room condensors etc.) please let me know. Also, it would be nice if the SM57 wasn't completely jammed up against the cab grill (check out the SM57 manual - it recommends 6-12 inches distant for mic-ing up a guitar amp) and without the amp bass control on 11. Just a few decent Marshalls at sensible levels off gain, an AC30 from clean to flat out and maybe some clean Fenders for my occasional forays into funk playing - that's all I really need. Any and all such profile suggestions gratefully received.
And then I can use Kemper as intended when I bought it, which is to record guitar silently with headphones, and without wires, mess, hum and buzz... ...and without annoying my wife or disturbing my kids' studies!
I won't be around for the next few days, as I'm off to visit my Dad who has just had a knee replacement operation. I am going to enquire about getting my latency aversion gene removed, but I'm not sure they do this on the NHS!