Help with testing - the numbers look right but my ears tell me otherwise!

[PointOfGravity]

Hiya,

I've been stuck on this problem for a while now.  I'm in the middle of building this modular digital effects pedal using chips from https://www.dream.fr/. I'm at the point where I'm worried about tone going through the DSP being sucked up by the ADC/ buffer so I ran some tests. My testing methodology is this:

Sweep tone generator 20 - 20kHz -> Pedal -> Saffire 6 USB Interface -> analyse in Audacity.

I tried passing a tone both with true bypass and through the DSP, with a "bypass" program written (where it only passes audio, no effects implemented).
The results I got were this:

In "Bypass Mode" (True Bypass):



Through my DSP's "Bypass Program":





Yes, the frequency responses look pretty much the same (apart from the low end bump, which doesn't look that substantial) . But when I play my guitar through it, I've noticed that the DSP signal is missing all the highs! There's no evidence that the high frequencies in my tests would would suggest a high end cut in the DSP signal. The original signal sounds much richer, and fuller.

So, I want to ask some questions:

• Is there anything crucial I am forgetting in my testing methodology? Why does it sound so different compared to what's on paper?
• Has anyone else encountered such a problem?

Thanks in advance!

[Digital Larry]

Don't take this the wrong way... but once I was doing some testing and my guitar was missing all its highs.  Ten minutes later I tried opening up the tone knob which I had bumped somehow, and there they were.  That slope on the entire response is a little odd though... like you have a 6 dB/octave low pass at a really low frequency (even though diff between 100 Hz and 200 Hz is more like 3 dB, regular analog filters don't come in " 3 dB/octave" flavors, only multiples of 6).

[slacker]

Have you got a schematic or a link to the circuit you're using? If you look up the datasheet for the AK4556, used in the reference designs, the input impedance is only about 30K Ohms which is far too low for passive guitars, so if your circuit is similar that is where your treble loss is coming from. Your tone generator probably has a strong enough output to drive 30k without any signal loss which is why it looks ok using that. The solution would be to add a high impedance buffer in front of the circuit.

[ElectricDruid]

Yeah, like Slacker, my first suspicion would be an input/output impedance problem of some kind. The circuit seems fine when driven with something can can cope with a low impedance input, but then sound horrendous with something that doesn't have the same drive power and needs to see a high impedance.

HTH,
Tom

[PointOfGravity]

Don't take this the wrong way... but once I was doing some testing and my guitar was missing all its highs.  Ten minutes later I tried opening up the tone knob which I had bumped somehow, and there they were.  That slope on the entire response is a little odd though... like you have a 6 dB/octave low pass at a really low frequency (even though diff between 100 Hz and 200 Hz is more like 3 dB, regular analog filters don't come in " 3 dB/octave" flavors, only multiples of 6).

Haha, well I checked my guitar and my tone knob was turned all the way up. I even took my board to a friends place and tried on his amp and guitar and the results were same. As slacker said below, it may be to do something with the ADC and input impedance of it.

[PointOfGravity]

Have you got a schematic or a link to the circuit you're using? If you look up the datasheet for the AK4556, used in the reference designs, the input impedance is only about 30K Ohms which is far too low for passive guitars, so if your circuit is similar that is where your treble loss is coming from. Your tone generator probably has a strong enough output to drive 30k without any signal loss which is why it looks ok using that. The solution would be to add a high impedance buffer in front of the circuit.

I am using a circuit pretty much similar to their 5504FX-EK board, which is using an AK4556. The buffer solution might be on the money, I'll do some research on it, thanks!

[Digital Larry]

Don't take this the wrong way... but once I was doing some testing and my guitar was missing all its highs.  Ten minutes later I tried opening up the tone knob which I had bumped somehow, and there they were.  That slope on the entire response is a little odd though... like you have a 6 dB/octave low pass at a really low frequency (even though diff between 100 Hz and 200 Hz is more like 3 dB, regular analog filters don't come in " 3 dB/octave" flavors, only multiples of 6).

Haha, well I checked my guitar and my tone knob was turned all the way up. I even took my board to a friends place and tried on his and guitar and the results were same. As slacker said below, it may be to do something with the ADC and input impedance of it.

I'm with Slacker all the way on this one.  I had the same problem with my FV-1 board until I used an external buffer.  For some reason I thought about my stupid maneuver with the tone knob and decided to mention that.   

[PointOfGravity]

Hey, sorry I haven't came back with a reply! I decided it was too much hassle to reconfigure the circuit, I found I was able to get pretty damn close to the analog dry signal by programming in a software EQ and adjusting it so it matched the dry-thru signal. Sounds near enough to me, and it's what I need anyway.  Thanks for the pointers anyway!

[Hatredman]

But a buffer is just one transistor...

Scarlett Johansson uses a Burst Box with her Telecaster.

[bloxstompboxes]

But a buffer is just one transistor...

Scarlett Johansson uses a Burst Box with her Telecaster.

Mmmmmm. Scarlet Johansson.