Its NDA, in case that wasn't made clear already. However, I used to use torriodal pulse transformers at work that went from about 1Hz up to a few MHz. They were expensive, thousands each. It can be done, is the only point.
Aware of this NDA for detailed specifics, but is it too much to find out about FR?
My own DAC modules with a customer designing bespoke equipment uses a special custom transformer. Also costs in the thousands... BW 50KHz, absolute sublime performance for all inputs. Won’t reveal other details though, I/P
My own DAC modules with a customer designing bespoke equipment uses a special custom transformer. Also costs in the thousands... BW 50KHz, absolute sublime performance for all inputs. Won’t reveal other details though, I/P
Last edited:
I just told you about FR. Public information will be released at some point. That's all I can say.
I should have stated in a better way… when this 150k BW transformer and other passives are used with Marcel’s DAC, the effective output BW as analysed by Marcel in post#1229 is less than 50kHz. As, I understand it we are hearing the system as put together.
Acko,
There is some HF roll-off, all things considered. However, that's not the biggest issue at the moment. Easy to fix that with a little DSP, at least until I can get inductors I like. The really hardest thing to get right are depth cues in the soundstage. Very difficult to excel at that. A little mild FR droop is easily factored out.
We're just not at the point of talking about such things as depth yet. We need to get through the first stuff first.
There is some HF roll-off, all things considered. However, that's not the biggest issue at the moment. Easy to fix that with a little DSP, at least until I can get inductors I like. The really hardest thing to get right are depth cues in the soundstage. Very difficult to excel at that. A little mild FR droop is easily factored out.
We're just not at the point of talking about such things as depth yet. We need to get through the first stuff first.
.. So, possibly there is room for improvements?
And, I would still like to see the FFT results. I beginning to feel that we are just ‘hearing’ things, excuse the pun..
And, I would still like to see the FFT results. I beginning to feel that we are just ‘hearing’ things, excuse the pun..
Last edited:
Reclocker board is hooked up and running. Filter board has been configured to operate as originally designed. Dac is still warming up after having been off a few days, so no comment on sound yet. Will give it some time, then report back.
Okay, back with a few brief comments. Mostly it sounds like transients have tightened up some. Before when something like a snare drum was hit, the sound of the stick hitting he drum had a little more of a "thud" sound to it whereas now its starting to sound like more of a "Crack" sound, more snappy/crisp/precise on the attack. Similarly for other sounds, mostly noticing more precise transients. That said, with the filter board in its current configuration there is some blurring and loss of low level musical details. Just wanted to try this way first.
That’s a neat job with all the coax cabling.
So this board comes in between the Fifo and Marcel’s Dac ?
It is supposed to isolate and reclock but I don’t see a crystal. Maybe it’s on the backside or is using the isolated incoming MCLK ?
Hans
So this board comes in between the Fifo and Marcel’s Dac ?
It is supposed to isolate and reclock but I don’t see a crystal. Maybe it’s on the backside or is using the isolated incoming MCLK ?
Hans
The crystals are in the Acko Labs AKX-22 clocks. They clock Andrea's FPGA-based FIFO buffer board. The FPGA outputs the clock and data signals to drive Marcel's dac. However, the FPGA outputs are not isolated and reclocked, as Andrea does those things on his dac boards. So what this board I just added does is isolate and reclock the FPGA outputs. It also allows Marcel's dac to be fed over very short length coax cables.
Also, just configured the filter board to slow down the common mode rejection. I like the filter board sound a lot more this way. Still a little blur and a little loss of detail, but it sounds cleaner (less veiled from EMI/RFI type noise) than before the isolator/reclocker was added, and more precise in time in a good way. Some people might think this is the best dac they ever heard at this point, although seems possible it might get a little better if we try a few more things. We'll see.
Will probably remove the filter board next and try the passive filter approach.
Also, just configured the filter board to slow down the common mode rejection. I like the filter board sound a lot more this way. Still a little blur and a little loss of detail, but it sounds cleaner (less veiled from EMI/RFI type noise) than before the isolator/reclocker was added, and more precise in time in a good way. Some people might think this is the best dac they ever heard at this point, although seems possible it might get a little better if we try a few more things. We'll see.
Will probably remove the filter board next and try the passive filter approach.
Last edited:
However, the FPGA outputs are not isolated and reclocked, as Andrea does those things on his dac boards. So what this board I just added does is isolate and reclock the FPGA outputs.
Hi Mark,
Can you understand I’m lost ?
It does not isolate and reclock, but it isolates and reclocks 🤣
Hans
Hi Mark,
Can you understand I’m lost ?
It does not isolate and reclock, but it isolates and reclocks 🤣
Hans
Hans,
Here's pic of when I first hooked up one of Andrea's PCM dacs. He included the pic in the FIFO buffer manual and labeled some of the parts:
Anyway, some of the boards and modules are labeled with Andrea's part numbers as follows:
TWSAFB-LT is the FIFO Buffer board. The display in the middle of the pic is used with the FIFO Buffer.
TWTMC-DRIXO is one of the two clock modules.
TWSDAC-LT is the PCM dac.
My I2SoverUSB board is shown on its side just below the FIFO Buffer board in the pic.
Everything else is power supply stuff.
Marcel's dac is being driven from the FIFO Buffer board in the pic. To do that I have to disconnect Andrea's DSD dac board (not shown in this pic).
The FIFO board does not have isolated and reclocked FPGA outputs. However, Andrea's dac board as isolated and reclocked inputs, so all is good in his case.
Now Marcel's dac also has isolated and reclocked digital inputs.
I might also add that there are a total of 7 power transformers in use to get the sound out of Marcel's dac I am now getting. 5 transformers are R-core, 1 is a torroid, and 1 is split bobbin. Probably better if all were split bobbin that's not what we have here at the moment. Thus you can probably see that most of the cost here is circuitry surrounding Marcel's dac. So far almost all the changes made up until this point have been in the surrounding circuitry. The only changes to the dac itself are clean power for the filter board, bypassing the last stage of the filter board, and slowing down the common-mode attenuation op amp. Cost of the surrounding circuitry and shielding cases is few thousand dollars, but not several thousand.
All that having been said, cost could be reduced a lot if all the power supplies were integrated into the dac board and simplified as much as possible. That's the challenge in making a pretty good high end commercial dac. How can you get the build cost for one unit down to 1/6 of what the dac is targeted to sell for?
Here's pic of when I first hooked up one of Andrea's PCM dacs. He included the pic in the FIFO buffer manual and labeled some of the parts:
Anyway, some of the boards and modules are labeled with Andrea's part numbers as follows:
TWSAFB-LT is the FIFO Buffer board. The display in the middle of the pic is used with the FIFO Buffer.
TWTMC-DRIXO is one of the two clock modules.
TWSDAC-LT is the PCM dac.
My I2SoverUSB board is shown on its side just below the FIFO Buffer board in the pic.
Everything else is power supply stuff.
Marcel's dac is being driven from the FIFO Buffer board in the pic. To do that I have to disconnect Andrea's DSD dac board (not shown in this pic).
The FIFO board does not have isolated and reclocked FPGA outputs. However, Andrea's dac board as isolated and reclocked inputs, so all is good in his case.
Now Marcel's dac also has isolated and reclocked digital inputs.
I might also add that there are a total of 7 power transformers in use to get the sound out of Marcel's dac I am now getting. 5 transformers are R-core, 1 is a torroid, and 1 is split bobbin. Probably better if all were split bobbin that's not what we have here at the moment. Thus you can probably see that most of the cost here is circuitry surrounding Marcel's dac. So far almost all the changes made up until this point have been in the surrounding circuitry. The only changes to the dac itself are clean power for the filter board, bypassing the last stage of the filter board, and slowing down the common-mode attenuation op amp. Cost of the surrounding circuitry and shielding cases is few thousand dollars, but not several thousand.
All that having been said, cost could be reduced a lot if all the power supplies were integrated into the dac board and simplified as much as possible. That's the challenge in making a pretty good high end commercial dac. How can you get the build cost for one unit down to 1/6 of what the dac is targeted to sell for?
Last edited:
Proceeding along, the filter board has been removed and the passive filter hooked up. Results are more musical details and precision, but bass is maybe a little weak. Or maybe it just seems that way because the overall sound is bit bright. Now that I can hear deeper into the music, I can also hear some opportunities for possible further experiments. One thing we talked about trying before was to drive Marcel's shift register arrays so that RTZ can be done by the master clock. For that purpose there is a another little board from Andrea that would go between the reclocker/isolator board and the shift registers. That may take a little while to implement so I'll leave the dac like this for a day or two so other people around here have a chance to listen its sound at this point.
Last edited:
There is no correlation between good sound and the number of transformers. It is possible to get excellent sound from Marcel's dac with just one transformer (or one battery).
Yeah, I know.
The problem in this case is that its not all integrated. When regulators are designed into the dac board and carefully laid out and routed, then one transformer can be better than many.
I still don't get it. When the FIFO board drives my DAC, what has Andrea's PCM DAC got to do with it?
My experience with the valve DAC is that DIY audio enthusiast like to change the power supplies anyway, so it would have been a waste of effort and board area to include them. I've made an exception for the really critical ones: two reference regulators and the regulator for the clock processing circuitry.
Nothing. I was trying to explain to Hans what was in the picture to start with. Conceptually your dac takes the place of Andrea's dac in the picture, so just imagine Andrea's PCM dac being replaced by your dac combined with the reclocker board hooked up as it is now.
Anyway, the idea was to illustrate the architecture of Andrea's system of boards, how they interconnect with each other, where the reclocking and isolation needs to take place, etc.
Last edited: