Does mixing dry outside DSP cause comb filter ?

[Vivek]

I'm planning on using an FV-1 to create chorus/reverb/delay

I'm planning that the FV-1 will output stereo wet signals which I will blend with dry signal externally to the FV-1

I understand there might be a millisecond or two of latency caused by the FV-1

Would that cause some odd cancellations when I blend in the dry signal ?

[Digital Larry]

Well, there's going to be that much extra delay that you can't get rid of.  Is there going to be some comb filtering in a flanger?  Yes.  Is this extra bit of delay going to cause you some terrible problem?  I doubt it.  I think people mostly get into trouble with this when their patches aren't explicitly time delay based.  For example, suppose you have a filter and you want to have a fade between filtered and bypassed dry.  This might cause you some problems.

[potul]

As Larry was pointing out, the more similar the signals are (dry & wet), the more comb filtering you will get. In the case of a delay, this is usually not a problem as the wet is delayed already. In reverbs it usually works ok as well, for a similar reason.
In a Chorus you might get some comb filtering, but  that's somehow part of the chorus effect, so I don't think you will notice. Are you planning to have a chorus already coming out of the FV1 or a vibrato and add the dry signal to create the chorus?

[Sweetalk]

Short answer, yes, you will have some sort of comb filter. But as Larry and Potul said depends on the effect if that is going to be a problem or not. You'll get more noticeable comb as much as the signals look alike. You can test this with the internal patch #5 (not the mambo) that's only clean signal but comming to the adc/dac routing. When you blend it with the analog dry you'll notice a small comb filtering.

How do you want to make the stereo? 100% wet in a channel and dry on the other or some more complex mixing?

[ElectricDruid]

I'm planning on using an FV-1 to create chorus/reverb/delay

I'm planning that the FV-1 will output stereo wet signals which I will blend with dry signal externally to the FV-1

I understand there might be a millisecond or two of latency caused by the FV-1

Would that cause some odd cancellations when I blend in the dry signal ?

A "millisecond or two" sounds like a *lot* of latency from the FV-1. Where did you see that? I'd expect more like "a sample or two", which at 32KHz is only 62.5usec - tiny fractions of a msec.

Secondly, if the signal that you're outputting from the FV-1 is delayed anyway, the latency is simply a little tiny bit of extra delay, so it won't matter. That would be the case for chorus, reverb, or delay algorithms. Just output a wet-only signal from the FV-1.

So I would say that you can do the wet/dry mix on the analog side with no worries. You won't get any comb filtering beyond what you'd expect adding a delayed signal.

[Vivek]

Re latency of FV-1

http://www.spinsemi.com/forum/viewtopic.php?f=2&t=896

[ElectricDruid]

Re latency of FV-1

http://www.spinsemi.com/forum/viewtopic.php?f=2&t=896

Thanks. So that says there is about 1 to 1.5msec of delay through the chip itself, even just with a straight feed-through from in to out (which would be the obvious way to test this). That's quite a lot for a DSP latency figure, but it's still not a lot for chorus (where you might be looking at 5-10msecs as a minimum) or a reverb (could easily have 50-100msec before the early reflections) or a delay.
So I don't see that it matters for the specific applications you've got in mind. You'd have much more delay than that between your signals anyway, in which case an extra msec or so isn't going to be noticed (or you could allow for it and reduce the delay RAM requirement by 32 samples!).

Some of the other effects mentioned in the thread (lowpass filter or bitcrusher) don't feature any delay, so doing a analog blend of the wet-only signal from the FV-1 and the input signal *will* produce unwanted comb filtering, since you've suddenly developed a 1.5msec difference between the two signals which wasn't there before.

[octfrank]

So that says there is about 1 to 1.5msec of delay through the chip itself, even just with a straight feed-through from in to out (which would be the obvious way to test this). That's quite a lot for a DSP latency figure

Delay is due to converters not the DSP core and this will exist in all cases. The new AKM 4558 has a 19 sample delay on ADC and 27.8 on DAC so the fastest you can do is 46.8 samples which at 32KHz sample rate is 1.46mS. So no matter what DSP you use with a 4558 you will always have at least a 46.8 sample delay through the system.

You need to always have this in mind and decide if you should use external mixing (fine for reverb, etc.) or internal mixing (necessary for phaser, etc.) where phase of dry and wet are important.

[ElectricDruid]

Thanks Frank!

[Processaurus]

With the chorus and the analog dry, you could always subtract the latency from Frank's spec from the delay line of the chorus, to get the same chorus sound/ delay relationship as with a digital dry/wet blend.  It's interesting with chorus, the change in timbre that comes from changing the delay time on the chorus, like anything 10-30ms is still chorus, just different sounding.

[Vivek]

Thanks everyone

I understood from above regarding blending the dry in analog outside the FV-1 :

A) No problem to blend the dry outside the FV-1 for Reverb, Chorus and Delay, even if they have filters as part of the wet implementation.

I will need bandwidth limiting LPF as part of my Reverb, Chorus and Delay wet implementation

B) There might be a need to adjust the Reverb, Chorus, Delay to account for 1.5ms latency of FV-1. But no need if delay line time is a parameter on a knob.

C) Bigger problems when dry is blended outside the FV-1 for distortion, EQ, filters, phasers (Which I dont need for my pedal)


Did I understand correctly ?


NEXT QUESTION :
The main reason to implement the mixing of dry outside the FV-1 is to have a blend control

So suppose the FV-1 outputs the wet part of a reverb, chorus(vibrato part) and delay in stereo,

and I dont have enough FV-1 pots to adjust the amplitude of each wet individually so they are set at some pleasant value,


Will it make sense to have an analog external pot that adjusts the ratio of

Dry
and
Wet mix of reverb, chorus, delay (all-in-one wet)