Simplest delay

P

Primoz Vozelj

Freshman
Apr 21, 2014
49
0
6
I have an active two-way bi-amp speaker and the LF is placed a bit behind HF driver, so the phase response at the crossover region is far from perfect. Ideal solution would be if I could delay the HF driver somewhere in the range of 0.15 - 0.25ms. What are simplest/cheapest solutions? Some DSP kit board? Analog?
My last resort is a cheap DSP like DCX2496, but it would be perfect if there would be something smaller that could be built into the speaker.
Thanks
 
Re: Simplest delay

Primoz Vozelj said:
I have an active two-way bi-amp speaker and the LF is placed a bit behind HF driver, so the phase response at the crossover region is far from perfect. Ideal solution would be if I could delay the HF driver somewhere in the range of 0.15 - 0.25ms. What are simplest/cheapest solutions? Some DSP kit board? Analog?
My last resort is a cheap DSP like DCX2496, but it would be perfect if there would be something smaller that could be built into the speaker.
Thanks
Click to expand...

How did you determine that the phase response has not already be corrected in the crossover?

Mac
 
Re: Simplest delay

mackerr said:
How did you determine that the phase response has not already be corrected in the crossover?

Mac
Click to expand...

I checked schematics in service manual (it's online) and I couldn't see anything there that would look like delay circuit.
And I made a TF measurement, there's steep fall in phase response around crossover freq plus small dip in magnitude response.
 
Re: Simplest delay

Primoz Vozelj said:
I checked schematics in service manual (it's online) and I couldn't see anything there that would look like delay circuit.
And I made a TF measurement, there's steep fall in phase response around crossover freq plus small dip in magnitude response.
Click to expand...

The transfer function is a valid test for phase response, looking at the schematic not so much. The inductors and capacitors that make up a crossover definitely effect the phase response, and in a well designed crossover will correct it for the particular box. Besides the issue of altering the amplifier to get access to the raw signal to apply a delay, the wavelength at your crossover frequency of 2kHz is about 17cm. If the phase error is less than 90º there is good summation. A physical offset of greater than 4cm would be needed to create more than 90º of phase error.

As a science project this may be a useful undertaking, there will be some learning done. As a way improve these particular speakers, buying speakers you don't feel you have to modify is probably a better (and cheaper) choice.
 
Re: Simplest delay

mackerr said:
As a science project this may be a useful undertaking, there will be some learning done. As a way improve these particular speakers, buying speakers you don't feel you have to modify is probably a better (and cheaper) choice.
Click to expand...

That's a basic idea(a science project), with a broad choice of relatively cheap studio monitors I doesn't make a lot of sense to invest money in my Yamahas.

mackerr said:
The inductors and capacitors that make up a crossover definitely effect the phase response, and in a well designed crossover will correct it for the particular box.
Click to expand...

There are no inductors in crossover, it's a biamp speaker with active crossover (R,C&op-amp). I've almost decided for the BBD delay. Are there any obvious drawbacks?
 
Re: Simplest delay

Primoz Vozelj said:
That's a basic idea(a science project), with a broad choice of relatively cheap studio monitors I doesn't make a lot of sense to invest money in my Yamahas.



There are no inductors in crossover, it's a biamp speaker with active crossover (R,C&op-amp). I've almost decided for the BBD delay. Are there any obvious drawbacks?
Click to expand...

Since this will be a science project, why not start with a passive 2 way speaker you can easily change the crossover characteristics and timing of? Your monitors have active crossovers, which also change phase as do all filters, and it would be easier to change more parameters with an external system processor.

Good luck with your experiments.

Mac
 
Re: Simplest delay

Primoz Vozelj said:
I have an active two-way bi-amp speaker and the LF is placed a bit behind HF driver, so the phase response at the crossover region is far from perfect. Ideal solution would be if I could delay the HF driver somewhere in the range of 0.15 - 0.25ms. What are simplest/cheapest solutions? Some DSP kit board? Analog?
My last resort is a cheap DSP like DCX2496, but it would be perfect if there would be something smaller that could be built into the speaker.
Thanks
Click to expand...

Before DSP this is what we did.... http://en.wikipedia.org/wiki/All-pass_filter eg. EV Delta Max
 
Re: Simplest delay

For a simple, cheap lines delay this might do the trick, especially at <$50:

http://www.behringer.com/EN/Products/DSP110.aspx

I keep a couple in the magic bag of tricks and they've come in handy in a pinch.

Added: This is not a quality piece of gear and not an end game answer. But it may help you find your answer, at which point you can source the correct gear for a more permanent solution.
 
Last edited:
Re: Simplest delay

Lee Douglas said:
For a simple, cheap lines delay this might do the trick, especially at <$50:

http://www.behringer.com/EN/Products/DSP110.aspx

I keep a couple in the magic bag of tricks and they've come in handy in a pinch.

Added: This is not a quality piece of gear and not an end game answer. But it may help you find your answer, at which point you can source the correct gear for a more permanent solution.
Click to expand...

What is the actual minimum delay you can do with the Shark, including latency? The OP is looking for a delay in the range of 1/10 to 1/4 of a millisecond.

Mac
 
Re: Simplest delay

Going digital would probably mean using two channels of DSP per speaker.
I found out that BBDs have low dynamic range (60db) and I don't like the idea of companders in listening path.
So far the best option seem to be all-pass filter. I redid measurement today with APL and it looks that I need about 0.35ms of delay. If my calculations are correct, two stages of 2nd order all-pass filter at 30kHz would give 0.37ms of delay. Today I found many different variants of the 2nd order all-pass filter (main difference seems to be if it's using one or two op-amps per filter) so there are still some decisions to be made.
 
Re: Simplest delay

What's the overall phase response look like? Delay isn't always the right answer to correct phase anomalies, and 0.35ms seems like quite a bit of physical offset in a compact studio monitor (corresponding to ~3" of offset, which is unlikely unless a horn is involved)
 
Re: Simplest delay

I don't think there is a problem with your speaker, and I don't think a delay will do what you want. The crossover in the speaker will have some phase shift / delay just because its a crossover.

To fix what you are asking will require FIR processing.

Below is a plot of an 8th order crossover and what it does (no speakers involved). I suspect your speaker may have a 4th order crossover ... i.e. 1/2 the time delay.
 

Attachments

  • 8 th order crossover.jpg
    8 th order crossover.jpg
    57.5 KB · Views: 0
Re: Simplest delay

mackerr said:
What is the actual minimum delay you can do with the Shark, including latency? The OP is looking for a delay in the range of 1/10 to 1/4 of a millisecond.

Mac
Click to expand...

Per the manual; 0 to 2,500 msec, adjustable in time, feet, meters. It, or at least this feature, was designed as line delay, not so much as an effect delay, which is the main reason I keep a couple around. I've always wondered what this device was based on, knowing its manufacturer's past reputation.

Added: Rereading the question, I'm not sure what it's smallest increment of adjustment is as I've only used them for impromptu delay stacks when I wasn't expecting to need. But with an adjustment by distance, I suspect you can get fairly accurate.
 
Last edited:
Re: Simplest delay

Understand that whatever driver offset"correction" you are attempting will only be valid a a single point in the listening field.

If your ms estimates are in range, their effects are de minimus. Typical sound systems have weaker links in the overall chain, and these should be tackles first...
 
Top Bottom