You'll find the usual XO and EQ tools such as delay, IIR, FIR and gain. Dynamics processing will be included, but might come only with the second release depending on time. Custom filters for serious tweekers are also included (i.e. the loading of a transfer function from a file).
I had written in a previous post that processing blocks are organized into "themes", which are collections of processing strategies. Within these 'themes' there are processing blocks that are specific to loudspeaker management, which is something that I haven't seen anywhere else and that's why I'm not going to develop on this topic until the boards are here
Hi Indemini,
Thanks for that, that's much appreciated!
Two big tips for the CS3318:
1. The hardware mute line may not be left floating! It will pick up logic noise and will degrade the ouput quality. I discovered this on my protoboard. Everytime there was activity on the control bus, the output distorted.
You tie it up logic level or ground ( the last option requiers a different register setting)
2 Bypass caps for the analog voltage rails. In the datasheet they use one 100nf and a 10uf over the + and - rail...... Not so effective, should be two of them to analog ground!
1. The hardware mute line may not be left floating! It will pick up logic noise and will degrade the ouput quality. I discovered this on my protoboard. Everytime there was activity on the control bus, the output distorted.
You tie it up logic level or ground ( the last option requiers a different register setting)
2 Bypass caps for the analog voltage rails. In the datasheet they use one 100nf and a 10uf over the + and - rail...... Not so effective, should be two of them to analog ground!
I always wonder why people do that. Is that for cost or is it an improvement ? I would have thought that it was much less effective for a single ended or half bridge output stage.
A decoupling cap between the the rails is doing nothing, I think it is a bad habbit from logic designers who do "some" audio.
but then you are just coupling the noise onto the opposing rail and relying on the power supply to shunt it to ground. The large loop impedance and output impedance of the supply makes this a dubious practice.
maybe for single rail opamps it will work for obvious reasons but for split rail I just cannot see it doing anything better than having two good caps grounded right at the opamp.
Perhaps it's assuming there is a poor ground connection on the board or is it trying to force a star grounded connection right at the power supply ? I saw them so this on the Silicon Chip DAC project which only uses a double sided board so there is no proper ground plane so this practice appears to be widespread.
Last edited:
Yes but 'shunt to ground' depends on the ground impedance. There's no perfect ground, so some of the noise current now appears as a voltage on the ground. In this situation, the cap rail-to-rail is not to reduce the noise on the supplies - we rely on the PSRR for that - rather its to provide low impedance power for the IC its feedling.
I'm not following you here.
I don't use a rail-to-rail cap on its own as I've found opamps can still oscillate when I do only that. What I do is include smallish caps to ground, but only through small series resistors (say 1R - 10R, depending on opamp) so as to attenuate any induced RF voltage in the ground. The alternative is to ensure extremely well regulated supplies (discrete super regs or something similar, with good RF rejection). Seeing as I'm a cheapskate I try to get decent sound for lowest cost
<edit> At RF, all grounds are poor grounds because wire inductance dominates over resistance even at the top end of the audio band.
Last edited:
Yes but 'shunt to ground' depends on the ground impedance. There's no perfect ground, so some of the noise current now appears as a voltage on the ground. In this situation, the cap rail-to-rail is not to reduce the noise on the supplies - we rely on the PSRR for that - rather its to provide low impedance power for the IC its feedling.
I'm not following you here.
I don't use a rail-to-rail cap on its own as I've found opamps can still oscillate when I do only that. What I do is include smallish caps to ground, but only through small series resistors (say 1R - 10R, depending on opamp) so as to attenuate any induced RF voltage in the ground. The alternative is to ensure extremely well regulated supplies (discrete super regs or something similar, with good RF rejection). Seeing as I'm a cheapskate I try to get decent sound for lowest cost
<edit> At RF, all grounds are poor grounds because wire inductance dominates over resistance even at the top end of the audio band.
but you are not dealing with RF. You are dealing with audio.
follow the current flow. If you use an opamp with class B or class AB output and the positive side is conducting then the current flows through the load back to the power supply common point. The cap across the split supply is doing a lot less to provide a low impedance point to ground as it is now in series with negative supply impedance. Even if it is used to shunt noise between the +/- rails it is still of no concern to the signal driving the load. Similarly for a negative going output signal. You are relying on the supply line and power supply impedance to be low which may not be the case in a simple single or double layered board with no dedicated ground plane.
If you had a fully differential configuration which did not reference ground then the cap across the +/- supply rail would make perfect sense but for a half bridge configuration with ground reference it doesn't make sense at all.
Maybe I am missing something but it sounds like a kludge that is trying to solve some other deficiency such as a poor board layout.
Last edited:
Yes but 'shunt to ground' depends on the ground impedance. There's no perfect ground, so some of the noise current now appears as a voltage on the ground. In this situation, the cap rail-to-rail is not to reduce the noise on the supplies - we rely on the PSRR for that - rather its to provide low impedance power for the IC its feedling.
I'm not following you here.
I don't use a rail-to-rail cap on its own as I've found opamps can still oscillate when I do only that. What I do is include smallish caps to ground, but only through small series resistors (say 1R - 10R, depending on opamp) so as to attenuate any induced RF voltage in the ground. The alternative is to ensure extremely well regulated supplies (discrete super regs or something similar, with good RF rejection). Seeing as I'm a cheapskate I try to get decent sound for lowest cost
<edit> At RF, all grounds are poor grounds because wire inductance dominates over resistance even at the top end of the audio band.
Better to (re)read the Wikipedia artikel:
"A decoupling capacitor is a capacitor used to decouple one part of an electrical network (circuit) from another. Noise caused by other circuit elements is shunted through the capacitor, reducing the effect they have on the rest of the circuit."
Yes, and what is the load? I don't have a resistor to ground as the load. Most opamps (I'm talking about active XOs here) in the circuit are driving just the next opamp. The output opamp is a special case - it does drive current which returns to ground (assuming an unbalanced output, not my preference normally). Given my preference for inverting opamp configurations, the load current never touches the 0V, it always returns from one or other of the rails.
I do indeed prefer fully differential for a whole host of reasons, including this one.
You are indeed missing something, but it is also a kind of kludge to solve poor board layout. Perfect board layout is quite impossible, there's always going to be a tradeoff. Optimising that tradeoff is what engineering is all about.
If it's a non-inverting opamp amplifier circuit then it is referencing ground. A Sallen and Key or Infinite gain active filter will have some elements connected to ground. A two resistor inverting circuit will have it's non-inverting input either connected directly to ground or be referenced to ground through a low impedance path such as a capacitor that has one side connected to ground.
You can't avoid the ground point that's why the decoupling between the two power rails doesn't quite make sense.
Indeed, but in my last post I said this:
Given my preference for inverting opamp configurations
I prefer to use MFB, it has the +ve opamp input as the connection to ground - this is a signal ground, not a power ground. Decoupling is for power.
I can't see how. A two resistor inverting config just has +ve input to 0V. No capacitor is needed, though if a bias current compensation resistor is used then a cap might be placed in parallel with it. Only sub microamp currents are going to flow here.
I'm not avoiding the ground point, rather I'm avoiding putting current into it. Significant difference.
I don't mind at all, and no apology required - but are you going to decouple your CS3318?
Chances are I have
Hi ds23man,
Thanks for your advises.
The mute pin is connected to the logic and can be set low or high on demand, no need to worry!
Cheers
Pin 4 usage is a nice omission in the datasheet, you can be tempted not to use it if you do all the muting with Mute all register. It is presented as a hardware mute..... Fact is it has to be terminated, it must be high or low all the time. If you use a switch to control this pin you have to use a pull up or down configuration. If you don't **** happens.........
Hi all...
I would be interested but have some questions, just to see if I get this right ...
sorry for bad english...
Now i have active 3 way loudspeakers and useing behringer dcx2496 as crossover
So with this project I could replace it and have better sound ?
I could use it as ;
Source > DSP Xover (which have onboard,6chanell, DAC and volume control) > amplifiers > drivers
I would not nead preamp ?
Amps are Pass Aleph series clones.
Thanks
I would be interested but have some questions, just to see if I get this right ...
sorry for bad english...
Now i have active 3 way loudspeakers and useing behringer dcx2496 as crossover
So with this project I could replace it and have better sound ?
I could use it as ;
Source > DSP Xover (which have onboard,6chanell, DAC and volume control) > amplifiers > drivers
I would not nead preamp ?
Amps are Pass Aleph series clones.
Thanks
Last edited: