Russ White said:2) DSD mode does not utilize the custom FIR filters. So if you are thinking of using the ES9018 to do an active filter DSD input is not an option.
Interesting bit of info here! So the ES9018 supports user downloadable FIR filter coefficients? Are the coefficient sets independent for each channel? How many taps are possible per channel?
Best regards,
A curious prospective TwistedPear customer who would like to use the Buffalo for a stereo 4-way active XO, and therefore would like board #2 🙂
ojg said:
1) Interesting bit of info here! So the ES9018 supports user downloadable FIR filter coefficients?
2) Are the coefficient sets independent for each channel?
3) How many taps are possible per channel?
Best regards,
A curious prospective TwistedPear customer who would like to use the Buffalo for a stereo 4-way active XO, and therefore would like board #2 🙂
1) Yes.
2) Yes.
3) I can't say at the moment, not because I don't know, but I will tell you the filter operates in two modes. Dustin has promised an app note for people wishing to implement an active filter. I will let him speak to that. I don't want to get it wrong in any case. 🙂 I am still new to the DSP aspects.
Cheers!
Russ
Yayyayyay!
I feel like being a kid waiting for Christmas eve, only worse because now we don't know which date Christmas is at
I feel like being a kid waiting for Christmas eve, only worse because now we don't know which date Christmas is at

Well, let me correct myself. That's a provisional yay...
Whether the ES9018 will be usable as a speaker crossover engine is dependent on the number of FIR taps available.
To illustrate this let's use a 4th order Linkwitz-Riley filter as an example. To approximate a 4th order LR lowpass at 2000Hz with a FIR filter running at Fs=96kHz you need approximately 128 taps.
A 4th order LR lowpass at 300Hz will need ~512 taps.
A 4th order LR lowpass at 60Hz will need ~2048 taps.
So depending on the number of taps available this chip may or may not be usable as a digital crossover for other than mid/tweeter crossovers. If that is the case then much of the market for a type #2 board disappears don't you think? Then it would only be interesting for multichannel surround sound applications.
A bit of reverse engineering applied to page 18 of the ES9008 datasheet indicates to me that it has a 128 tap half-band FIR filter. Am I close?
Whether the ES9018 will be usable as a speaker crossover engine is dependent on the number of FIR taps available.
To illustrate this let's use a 4th order Linkwitz-Riley filter as an example. To approximate a 4th order LR lowpass at 2000Hz with a FIR filter running at Fs=96kHz you need approximately 128 taps.
A 4th order LR lowpass at 300Hz will need ~512 taps.
A 4th order LR lowpass at 60Hz will need ~2048 taps.
So depending on the number of taps available this chip may or may not be usable as a digital crossover for other than mid/tweeter crossovers. If that is the case then much of the market for a type #2 board disappears don't you think? Then it would only be interesting for multichannel surround sound applications.
A bit of reverse engineering applied to page 18 of the ES9008 datasheet indicates to me that it has a 128 tap half-band FIR filter. Am I close?
ojg said:Well, let me correct myself. That's a provisional yay...
Whether the ES9018 will be usable as a speaker crossover engine is dependent on the number of FIR taps available.
A bit of reverse engineering applied to page 18 of the ES9008 datasheet indicates to me that it has a 128 tap half-band FIR filter. Am I close?
I will see if I can get you more information, but here is what I know:
There are two filter mode to which you can apply custom coefficients:
1) Slow roll-off which allows 64taps. The first is always 0.
2) fast roll-off which employs an interpolated filter at 2x and 4x making it an 8x oversampling filter.
The best thing I can suggest for now is to read up on interpolated filters:
http://www.mathworks.com/access/hel...m=0&ct=result&cd=1&q=interpolated+FIR&spell=1
To be honest I don't know enough on the subject yet to say how well it will work as an active filter, but I do know that ESS plans to incorporate that application in their demo board UI.
Cheers!
Russ
4th order may not be necessary for successful crossover applications. IMHO 3rd order - what I'm using now w/ IIR - can be quite effective and even a really good 2nd order filter would satisfy alot of ears.
Frank - Mpls.
Frank - Mpls.
Hmm, that's what I feared. The custom FIR filters are not meant for crossover use but are meant for those who wish to design their own upsampling filters. That way ESS can compete well with the Wolfson DACs and their choice of upsampling filters.
Going to 3rd, 2nd or 1st order filters do not reduce the requirements for number of filter taps very much. In this case it's the ratio of sampling frequency to crossover frequency that is the major factor. The higher this ratio the more taps you need.
Going to 3rd, 2nd or 1st order filters do not reduce the requirements for number of filter taps very much. In this case it's the ratio of sampling frequency to crossover frequency that is the major factor. The higher this ratio the more taps you need.
Right. Hence my original question (and the doubt in my voice at the time).ojg said:Hmm, that's what I feared. The custom FIR filters are not meant for crossover use but are meant for those who wish to design their own upsampling filters. That way ESS can compete well with the Wolfson DACs and their choice of upsampling filters.
Going to 3rd, 2nd or 1st order filters do not reduce the requirements for number of filter taps very much. In this case it's the ratio of sampling frequency to crossover frequency that is the major factor. The higher this ratio the more taps you need.
Ok. Back to finding another solution.
peter
Use of custom coeffs as active filter.
I would not be surprised if it works at low freqs, or if it does not.
I have not looked at it close enough to judge yet.
I do know that Dustin seemed to be indicating the interpolated filter effectively solves the issue withe low number of taps. But I am completely out of my element here. The question is best posed to him and not me. 🙂
I do know, they plan on incorporating the featire in their GUI.
Cheers!
Russ
I would not be surprised if it works at low freqs, or if it does not.
I have not looked at it close enough to judge yet.
I do know that Dustin seemed to be indicating the interpolated filter effectively solves the issue withe low number of taps. But I am completely out of my element here. The question is best posed to him and not me. 🙂
I do know, they plan on incorporating the featire in their GUI.
Cheers!
Russ
Thanks Russ that helps a lot.
It is clearly an excellent platform for experimenting with different types of upsampling filters. Unfortunately it is not usable for crossover duty.
So now my vote easily goes to board #1.
It is clearly an excellent platform for experimenting with different types of upsampling filters. Unfortunately it is not usable for crossover duty.
So now my vote easily goes to board #1.
Russ White said:
There are two filter mode to which you can apply custom coefficients:
1) Slow roll-off which allows 64taps. The first is always 0.
2) fast roll-off which employs an interpolated filter at 2x and 4x making it an 8x oversampling filter.
Can the digital filter be disabled altogether? 🙂
Re: Buffala32 thoughts and design considerations
Board #1 for me.
Which is the ETA?
Russ White said:
Here is the skinny. We are planning two new boards.
1) A simple stereo (or mono if you use a controller) Buffalo replacement designed to work primarily in hardware mode while supporting I2C input for control options including Volumite and Femto. This board would only be able to support a single SPDIF and I2S stereo input because of the constraints of using the chip in hardware mode. The 4 differential DACs per side would be in parallel In Short DSD mode would not be available on this board because of the way the chip supports hardware mode. Its a design compromise in favor of simplicity and ease of use over complexity and flexibility.
Board #1 for me.
Which is the ETA?
We will hopefully get the boards orders in the next few days. I have chips on hand already (thanks Shaw/ESS!). Probably four or five weeks. We will also be posting more details about the board(s) very soon.
Is that pre-order in a couple days with delivery in 4-5 weeks, or pre-order in 4-5 weeks? Sorry the anticipation is keeping me up at night.... Cold sweats... lack of concentration... j/k...
How will the new board handle both I2S and S/PDIF. I hear it can detect which is connected without the need for a switch. I want to have both connected always and the ability to choose which is used, is this feasible?
mono playback
hi everybody.
i just finished my buffalo and it sounds great but no imaging.
i tried a test cd and i found that it plays mono.
both speakers work but they play the same things.
currently i am using the buffalo without output stage straight
to the preamp through a coupling cap and a resistor.
any ideas what can be the problem?
thanks in advance
george.
hi everybody.
i just finished my buffalo and it sounds great but no imaging.
i tried a test cd and i found that it plays mono.
both speakers work but they play the same things.
currently i am using the buffalo without output stage straight
to the preamp through a coupling cap and a resistor.
any ideas what can be the problem?
thanks in advance
george.
Re: mono playback
My best guess would be that you are using opposite phases and it sounds like mono.
You will not get stellar performance without a proper output stage (filtering), and certainly not if you are not utilizing the DAC's differential outputs.
back said:hi everybody.
i just finished my buffalo and it sounds great but no imaging.
i tried a test cd and i found that it plays mono.
both speakers work but they play the same things.
currently i am using the buffalo without output stage straight
to the preamp through a coupling cap and a resistor.
any ideas what can be the problem?
thanks in advance
george.
My best guess would be that you are using opposite phases and it sounds like mono.
You will not get stellar performance without a proper output stage (filtering), and certainly not if you are not utilizing the DAC's differential outputs.
It is feasible, as long as your S/PDIF input is TTL level. If it is consumer TTL (0.5Vp-p), you will need to run it through the comparator and switch like on the first generation Buffalo. Alternatively, you could do the level shifting before the Buffalo and get what you want.
Incidentally, our S/PDIF MUX outputs TTL level S/PDIF. 🙂
Incidentally, our S/PDIF MUX outputs TTL level S/PDIF. 🙂
- Status
- Not open for further replies.
- Home
- More Vendors...
- Twisted Pear
- Buffalo DAC (ESS Sabre 9008)