Behringher reverb/Multifx.

[Jack]

I have both of these, they also have the same board.
They use:
http://www.coolaudio-semicon.com/files/public/V1000_DATASHEET.pdf
And
http://www.atmel.com/dyn/resources/prod_documents/2486S.pdf
A bunch of quad opamps, a dual,  and some other unidentified ones (10, 16 pins)

Now the V1000 has a bunch of different effects, seemingly controlled by a pot and the atmel chip. I want to be able to access the reverse reverb.
But jumping pins 5 and 13 causes no sound to come out, except a clicking when the knobs are turned.
Also I assume I want to lower the bit clock, in order to cause nyquist aliasing, yes I'm that sort
So any ideas?

[Jack]

After a bit more poking around one of the identified ICs is a cirrus logic cs3540zz for analog to digital conversion, it's  slaved to the master clock from
the V1000. The clock generator uses a 12.288 Mhz crystal for the clock generator, I'm a little fresh to this DSP business, but I figure lowering this will lead to a little
nyquist action. But the resonant frequency is a property of the crystal, what do I do?

[Peter Snowberg]

I didn't see a CS3540 at the Cirrus web site, but I'm going to assume it's functionally similar to the Wavefront AL1101. Pretty much every converter used for audio these days is of the sigma-delta variety. In this type of converter, a very fast 1 bit A/D (yes a comparator) is used along with a digital filter to create the final output. For instance the AL1101 samples at 64 times the output frequency rather than the 2 times seen with a successive approximation converter. If you're working with a 12.288MHz master clock, the V1000 (AL3201) will set the work clock at 1/256th the master clock, or 48KHz. In this case the actual sample frequency at the sample-hold is 3.072MHz (!) and the Nyquist frequency for the conversion becomes 1.536MHz. Even a simple RC filter can be used to roll off information above 24KHz because the digital filter takes care of eating the garbage between 24KHz and 1.5MHz. It's hard to say how the filter will sound as you reduce the master clock, but the effect may be a bit different.

The larger problem you'll run into here may be the clock requirements of the V1000. I know the AL3201 uses DRAM for a delay buffer and the refresh of that DRAM is under program control. If the program executes too slowly, the DRAM will start forgetting data which may sound bad, but maybe not in a good way. The lower end of the master clock range for the AL3201 is specified as 6.144MHz.

Generating the master clock at the speeds needed would perhaps best be done with the aid of a frequency synthesizer IC as it's a bit fast for stability with an RC circuit. There are also PLL chips out there with VCOs that will run fast enough, but the good old 4046 will only do about 1MHz at 9V.

[MetalGuy]

Can you please post some pictures of the inside?

[Jack]

Whoops, it was the CS5340 

Can you please post some pictures of the inside?

Is there anything specific you want to see?.
Some of the components are covered in glue, including the ICs. But in a pretty haphazard manner, it looks accidental

[Jack]

A brief update:
I  put a crystal marked 40.6800 (From a remote control car) in and it seems nothing has happend.
Maybe the high end is a little different, but that's probably psychosomatic.
I might go and buy a different crystal, and see if that helps. Maybe 2Mhz or so.

[Peter Snowberg]

The spec sheet shows support for master clocks down to 5.12MHz. An upper limit is not given, but I wouldn't hold my breath above maybe 14MHz.

You may want to try clocking the chip from an external clock, rather than using the built-in oscillator circuit, however you are going to have DRAM refresh issues below 5MHz.

[Jack]

A 2Mhz crystal has made the reverb much longer, with no obvious distortion. There is something subtle going on in the high end though. It sounds much nicer for guitar IMO.

[Peter Snowberg]

Very cool!

I wonder if the DRAM will start "forgetting" at higher temperatures when running that slowly. It's great news in any case that your can run that slowly and still have the thing work. The Alesis/Wavefront SCR/DRE have DRAM based storage and will start to have issues when refreshes get too slow. There was even some tech data about adding additional refresh instructions to your code.  I wonder if the V1000 has the delay buffer implemented as SRAM?

Jack, you may want to try using the oscillator in a 4046 to clock things. The upper limit is slower still (<1.5MHz?), but if it works at 2MHz then maybe it will work at even slower speeds too. If the delay buffer in the V1000 is fabricated in SRAM, the circuit should function down to DC but don't expect the digital filters in the CODEC to behave.

Hmmm.... if the 4046 clock works then the next step would be to add modulation to the 4046 with an LFO.

[puretube]

why DIY?

just re-house this one...

[DrThousand]

 somehow get external access to on chip memory, just disable it & use a static  RAM instead.  I haven't checked the spec sheet yet to see if that's possible, but it'd be nice if it works.In regards to changing clock frequency on these things, pins 3 & 4 are the crystal pins. One is labeled xtal in & one is labeled xtal out.  The one you want to use is xtal in.  If in doubt, use the one with the higher impedance.  (Put a 1 or 2 meg resistor in series with your new clock generator, then measure which pin pulls down the voltage least.  Here you have your clock input pin.  Ta-Da!)

If all you're looking for is Nyquist, do you really need stability?  A little frequency instability might be nice in an application like this.  Start off with some kind of RC oscilator & go from there.  You don't really need a 555 for this clock; 2 transistors might work, or a few inverting sections of whatever logic you have on your board.

Just remember it's a 3 volt circuit & doesn't like high voltage on any of its pins.

About RAM forgetting, I was wracking my brain for an alternare refresh scheme, when I thought- if you can