Solved: Latencies in AX8 depending on amp?

[Heinzi]

I was just wondering, when I checked the latencies, i recognized that different presets showed different latencies -starting from 2.2 ms up to 10ms like in the Badger 18 Preset and they also vary dependent on the effect blocks - I´m still not on FW 2.02 so I´ll install this and check again, maybe the problem is solved!!

But - I doing something wrong? is that normal or just a bug ? Anybody got similar experiences?

(I used the normal XLR Output)

 

[Smittefar]

How do you check latencies?

 

[Heinzi]

I create a digital spike of 1 sample length and send it through an analog splitter - one side to one sound card input and the other one through the ax8 into a second input and then check the difference in the DAW in ms or bit per second.

 

[Smittefar]

Well - that would do it

I have been thinking about doing the same thing, but I never got around to it

10 ms sounds like a lot - I would like to think, I would have noticed that.

 

[pauly]

Gday,
Yes and no ... A good rule of thumb is that one ms equals around one foot of distance... So the latency difference is there but certainly not bad. Particularly in instances where the number of blocks gets large and the signal is fighting its way through each.

Having said that I must clarify that this is my opinion only!

Thanks
Pauly

 

[Heinzi]

Yes sure, but in this case the latency should not be much more than 2ms

 

[spiderWAW]

Is the same latency on those patches in Axe FX too?

 

[Heinzi]

I checked that a few years ago, and the latency was incredibly low...maybe i should recheck....

 

[pauly]

It'd be interesting to see the comarisons of some of your 'high' latency presets with certain blocks disabled - Particularly the cab.

Thanks
Pauly

 

[Heinzi]

Hi Paul, yes that has an influence - but what made me wonder ist that also diffrent amp types seem to have an influence.

 

[pauly]

Gday Heinzi,
Yup - that also makes sense depending on the emulated circuits complexity I guess. Great stuff eh

Thanks
Pauly

 

[Heinzi]

Yes, but 6.51ms is way, way too high - and I don´t think that FAS intended that.

Ok here´s the Badger 18 without FX - 4.72 ms: "Verzögerung" is Latency in ms:




and here with Comp and Drive - 6.51 ms




Double Verb was lower: 2.38ms without FX and 3ms with FX

 

[yek]

Latency should be around 2 ms. Make sure you don't have Lookahead enabled in the Comp.

 

[atquinn]

Would love to see a comparison with the same presets on a rack unit.
-
Austin

 

[Heinzi]

Latency should be around 2 ms. Make sure you don't have Lookahead enabled in the Comp.

Hi Yek,

Thanks! The Screenshots were made with lookahead off, newest fw on! Maybe FAS could check that again, cause i have to write a review for a magazine and can only explain these values with an error in the unit or a bug in the software- i checked my xl wich was always around 2ms so my measures got to be accurate

 

[spiderWAW]

Fair methodology for this test would be use setup like this:
Guitar with two outs (Godin fro example) or Guitar in Spliter
Guitar to audio interface
Guitar to AX8 to audio interface
Audio interface in to two track in DAW

hit record
hit string
and compare latency/offset between tracks

I use this protocol for measure latency in MIDI converters: http://cyfrowogitarowo.pl/archiwa/56 see pictures.

 

[Heinzi]

That doesn't explain why there would be such a difference to the XL- please before we Continue the discussion-could someone please check with the ax8 and confirm or falsify my Observation?

 

[FractalAudio]

We'll look into this but I doubt there is any problem. More likely the software used to measure the latency is reporting an incorrect result. You can't really automatically measure latency of a nonlinear system (and by nonlinear I mean both frequency and amplitude). I don't know how the software is measuring the latency but you can't do a cross-correlation if the signals are markedly different (which they will be). Automatic latency measurement assumes that the system is linear both in frequency and amplitude.

 

[Admin M@]

I have just repeated the experiment. (In fact we have conducted it here previously.)
I do not get results anywhere near an intolerable 10ms. I find latency across presets in the neighborhood of 2.2 ms.

For the UNALTERED DoubleVerb (factory preset 1-1) with all its effects running, I get a latency in the neighborhood of 2.2 ms.
For the UNALTERED Badger 18 (factory preset 1-4) all its effects running, I get a latency in the neighborhood of 2.2 ms.

To perform this test, I use Logic and my RME Fireface UFX.

1. I set the hardware and project at 96k to ensure high resolution data and prevent the illusion of sample shifting.

2. I place a 96k audio test file on track 1, sending to output 1+2 as a stereo pair.

3. I connect Output 1 directly to Input 5. I mute the monitor for this input.

4. I connect Output 2 to the AX8 Instr. input. I connect the AX8 output 1L 1/4" to Input 6.

5. In Logic, I create a new track and record Input 5+6 as a stereo pair while playing back my test file.

6. I export the recordings and open in Sound Forge for close inspection.

7. By eye, I select the start point of the reference and the start of the AX8 signal. Note: I recognize that at this step there is the possibility of errors in the range of 0.00001 seconds (1/10th ms).

8. Sound Forge displays the duration of my selection in seconds (highlighted in a pink box in the PNG files attached): 0.002240 and 0.002229

 

[Admin M@]

Working with the OP offline so closing this thread.
Another Fractal engineer confirmed my findings (and got even slightly lower results closer to 2.200).

Update: The OP found that the test procedure being used was flawed and has corrected this. Now his measurements reflect reality.