Amp Calibration in SW

[arc2v]

Hey fellow Speaker Workshop users:

Attached is my amplifier calibration file. I never claimed it was a good amp, but does this look right or is it too noisy?

I ran it in the wizard, then converted the pulse to a frequency. Uniform Windowing, Use Channel Calibration, Use Entire Range for time range.

Anyway, I searched for some calibration files to compare to and came up empty (many of the links to images on the Audua site are broken now).

Thanks for any advice.

A

 

[Vek]

That looks way too noisy! It should be an uncluttered line ...
It looks like averaging is missing to smooth out the curve.
I would go ahead and setup the MLS repeat count (preferences/measurements) to at least 4, that should do the trick.

Cheers

 

[arc2v]

Hmm, that's what I thought.

However, I have the MLS repeat to 10 and it still does that!

Any other ideas? Could it be a warmup or sampling rate error?

Thanks again.

A

 

[arc2v]

Well, I have to get ready for work, so I can't continue testing this morning, but here's what I came up with:

Change MLS warmup to 200 -- no change
Change MLS # samples to 10 -- no change
Change Sample Freq from 48 to 44.1 -- very slight change, but still very noisy
Change system level to 20k instead of 25k on the Vu meter -- no change.

So I think that's what my stupid amp/SC combo measures at.

I"m wondering if this is a MOSFET amp sort of thing.

Tonight after work, I will set up the rig using my Outlaw amp and retest. If I get a nice smooth curve, I know it's the amp, if it's still noisy, I know it's the sound card.

 

[Vek]

Other possibilities
- overdriven sound card inputs (but not likely, as you say you have lowered the level to no success)
- bad duplex sample accuracy (iirc there were issues with some creative sound cards)
- ?

So how does your sound card perform o/p to i/p without amp?
It shouldn't matter if the amp is MOSFET or something else ...

I remember having had similar problems when I started using SW, but somehow with fiddling here and there I finally got sorted this out. Biggest step forward in my case was setting MLS repeat to >>1.

In the end I was able to get something like the attached example ...

 

[y8s]

change your graph scaling to 40 db spread and it wont look so fuzzy.

 

[arc2v]

God I'm an idiot!

I widened the spread and it actually looks fine. I had not realized how zoomed in it was -- just that it was fuzzy. I'll have to apologize to my amp, why did I ever doubt it.

See attached. 40 dB scale. Looks much better.

 

[y8s]

sometimes it's very easy to forget the simple things when you're solving complex problems!

also you shouldn't need to use any repeats when doing the calibration. they're there for ambient noise rejection during acoustic testing and you're testing a closed system...