Dear Jens,
congratulations to your excellent design and efforts to make this test equip to happen !
I would like to ask your kind help around AK4490. I'm planning to add this DAC with
OPA1622 to our DIY amp based on TI's TAS5508+TAS5531B as a headphone solution.
Currently we use the dedicated HP PWM outputs of TAS5508, but I want to have a more
general HP output with 4 Vrms output and >115 dB-ish DNR. I need this DNR to cover
16..600 ohm cans with digital volume control.
The problem I encountered is that AKM doc set doesn't cover deep enough what signal
source quality is needed to feed VREFx and VDDx pins to reach the target specs. I would
looked for VREFx current load + PSSR data and VDDx PSSR but I couldn't find any info
on these.
Can you please give us some hints about your experience, what sources are really
needed for these pins?
Best regards, Endre
congratulations to your excellent design and efforts to make this test equip to happen !
I would like to ask your kind help around AK4490. I'm planning to add this DAC with
OPA1622 to our DIY amp based on TI's TAS5508+TAS5531B as a headphone solution.
Currently we use the dedicated HP PWM outputs of TAS5508, but I want to have a more
general HP output with 4 Vrms output and >115 dB-ish DNR. I need this DNR to cover
16..600 ohm cans with digital volume control.
The problem I encountered is that AKM doc set doesn't cover deep enough what signal
source quality is needed to feed VREFx and VDDx pins to reach the target specs. I would
looked for VREFx current load + PSSR data and VDDx PSSR but I couldn't find any info
on these.
Can you please give us some hints about your experience, what sources are really
needed for these pins?
Best regards, Endre
First, I would suggest analog gain settings to handle the range needed for headphones. in ear headphones can be quite sensitive although the new EU standard is effectively 94 dB for 75 mV. Having 4V available could permanently damage someone hearing if not managed carefully. Still changing the analog gain after the DAC will optimize the SNR better than using the DAC's volume control. The datasheet info will get you the stated performance and the chip doesn't need heroic magic to work. No hidden registers to turn on and off special features. Its all in the docs. Check out the AKD4490 for the details: https://www.akm.com/akm/en/file/ev-board-manual/AK4490EQ.pdf The active supplies they use are on pg. 44.
Hi,
thanks the quick response !
I already chewed the datasheet + eva board docs few times, but honestly don't
find these data in them. Eva board doc is also incomplete, they doesn't specify few component types, like regulators for VDDx, diodes in the VREF regs.
I also don't quite like their VREF regs. One they cost more then the DAC chip itself.
On the other hand SPICE simulation shows that it might not be the most optimum solution especially if you consider Fig 9. that demands additional huge caps on VREFx.
If I swap AD817 for OPA1602 for much less system cost both noise and low freq line regulation are dramatically enhanced. Maybe no need for huge caps neither. No wonder why, lower noise and higher open loop gain in the low freq band.
So that started me thinking if I knew what is going on those pins I could design a more cost effective infrastructure for AK4490.
Regarding HPA levels, it is not a commercial product, my goal is to cover Beyer T1, Senn HD800 on one side, Senn Ie80 on the other. Maybe driving level can be lowered.
Amp will have two HP jacks one 6.3 mm and one 3.5 mm. CPU detects which
connector is used and applies its own volume settings. CPU stores 3 volume settings,
one for the speakers and two for the HP jacks. I think it is good enough safety for a DIY kit. I would be happy with >95 DNR when IE80 sings.
thanks the quick response !
I already chewed the datasheet + eva board docs few times, but honestly don't
find these data in them. Eva board doc is also incomplete, they doesn't specify few component types, like regulators for VDDx, diodes in the VREF regs.
I also don't quite like their VREF regs. One they cost more then the DAC chip itself.
On the other hand SPICE simulation shows that it might not be the most optimum solution especially if you consider Fig 9. that demands additional huge caps on VREFx.
If I swap AD817 for OPA1602 for much less system cost both noise and low freq line regulation are dramatically enhanced. Maybe no need for huge caps neither. No wonder why, lower noise and higher open loop gain in the low freq band.
So that started me thinking if I knew what is going on those pins I could design a more cost effective infrastructure for AK4490.
Regarding HPA levels, it is not a commercial product, my goal is to cover Beyer T1, Senn HD800 on one side, Senn Ie80 on the other. Maybe driving level can be lowered.
Amp will have two HP jacks one 6.3 mm and one 3.5 mm. CPU detects which
connector is used and applies its own volume settings. CPU stores 3 volume settings,
one for the speakers and two for the HP jacks. I think it is good enough safety for a DIY kit. I would be happy with >95 DNR when IE80 sings.
Moved from this thread:
http://www.diyaudio.com/forums/equi...alyzer-ak5397-ak5394a-ak4490.html#post4405373
http://www.diyaudio.com/forums/equi...alyzer-ak5397-ak5394a-ak4490.html#post4405373
I think that the reason for not specifying the PSRR is that there isn't really any suppression of noise! On the initial AK4399 based design I made, before the AK4490 was available, I found that the noise on the output was very dependent on the noise on the analog supplies. Having previously designed a DAC based on the Cirrus Logic CS4399 this came as a bit of surprise. The CS4398 has a specified PSRR of 60 dB at 1 kHz.
For the AK4490 design I started out with a design similar to the one used on the EVB. I replaced the op-amp with an OPA209. This worked very well.
Later I changed the design to an LT3042. This also works fine. I think the performance of the two supplies are very similar in terms of noise.
The LT3042 is not DIY-friendly though. But for me that wasn't an issue since the boards would be mounted on a pick-and-place machine anyway.
For the AK4490 design I started out with a design similar to the one used on the EVB. I replaced the op-amp with an OPA209. This worked very well.
Later I changed the design to an LT3042. This also works fine. I think the performance of the two supplies are very similar in terms of noise.
The LT3042 is not DIY-friendly though. But for me that wasn't an issue since the boards would be mounted on a pick-and-place machine anyway.
JensH, thanks your comments! LT3042 seems to be an excellent part with reasonable price compared to what it does. I guess one of this reg could do all VEFx and VDDx pins with proper PCB layout and no need for huge caps! Never knew this part before, but it makes discrete LDO idea obsolete.
- Status
- Not open for further replies.