I haven't been concerned about phase shift at all when designing my filters. That's for the simple reason that I know I can correct for phase distortion digitally if indeed it is an issue.
As to which images are the biggest problem, I don't know. If the issues of audibility are about IMD then the higher the frequency the worse the IMD's going to be. But it might not be only about quantity of IMD, it could be a matter of where in the spectrum it occurs - having tones close together (say 20kHz and 24.1kHz) puts a difference tone theoretically at 4.1kHz which is in a region of high sensitivity for the ear.
As to which images are the biggest problem, I don't know. If the issues of audibility are about IMD then the higher the frequency the worse the IMD's going to be. But it might not be only about quantity of IMD, it could be a matter of where in the spectrum it occurs - having tones close together (say 20kHz and 24.1kHz) puts a difference tone theoretically at 4.1kHz which is in a region of high sensitivity for the ear.
I've often wondered about the audibility of phase shift as well. My friend Hifiamps swears that the TDA1387 prototype he recently built--with no filter whatsoever--is the best DAC he's had to date. I've listened to TDA1387 with a few different analog filters, but there were always other differences in the implementation each time, so no direct comparison.
The swappable daugterboard filters for Deca DAC are genius. If you are willing to accept some NOS droop in exchange for phase performance, you could indeed drop in a filter with a higher cutoff, as you suggested. I also wonder about the viability of a Bessel filter, which touts a near-perfect phase response.
Kind of. After growing to dislike the effects of OS and digital volume control with NOS DACs the idea of remaining 'bit-perfect' started to appeal to me, which is only really possible with a NOS DAC, it would feel wrong to add in some other form of DSP at this point, it is also kind of wrong to lump all DSP in in together in pursuit of 'bit-perfectness' but addressing it in analogue domain like this would be what I'd lean towards.
I also prefer that it wouldn't have to rely on software.
DSP would be anything that changes the original data, a fixed delay wouldn't count I dont think.
I also prefer that it wouldn't have to rely on software.
DSP would be anything that changes the original data, a fixed delay wouldn't count I dont think.
I got it I think. So it seems you'd be cool with a transversal filter method of oversampling where none of the digital data is changed but the tap weights are implemented in an analog manner. Which is what I implemented in my 17 or 19 tap 'tower DAC' many years ago. I don't think I'm going to revisit that design though as its not very flexible.
I liked the sound but it didn't sound as good as a simple DAC with an LC filter after it. So I put it to one side as a curiosity while I worked on LC filtered NOS DACs. Now I wonder if a non-bit-perfect variant of it mightn't be a better approach as that will be compatible with my passive filters.
Hello Richard
Hope you are doing good.
I recently got a older version of the phidac along with the 3rd order filter board when one of the members from Italy was selling it. After receiving and powering up using my simple LM317 psu board set to 14.50vdc and initially with PCM2706 i2s dac, only one of the channels was working. So replaced the pins on the output side and directly soldered the wires. The left channel works but the right channel does not make any music. So when I check the DC offset, the left channel has 55mV and right channel around 0.6-0.0mV. Unable to figure out whats the issue with the right channel. I then changed the i2s dacs to XMOS working fine with my other 1862 dac got from diyinhk.hk and then also an Amanero clone. It is the same issue but curiously if I test it using a 2RCA to 3.5mm headphone output cable, then inserting the headphone jack only a bit inside I get the both the channels working. I even replaced the RCA jacks but still the issue persists.
The member agreed to get back the dac and refund me but before doing that want to figure out if I can fix it but the SMD parts and my poor eye sight makes it difficult
Here are some pics for your reference.
Hope you are doing good.
I recently got a older version of the phidac along with the 3rd order filter board when one of the members from Italy was selling it. After receiving and powering up using my simple LM317 psu board set to 14.50vdc and initially with PCM2706 i2s dac, only one of the channels was working. So replaced the pins on the output side and directly soldered the wires. The left channel works but the right channel does not make any music. So when I check the DC offset, the left channel has 55mV and right channel around 0.6-0.0mV. Unable to figure out whats the issue with the right channel. I then changed the i2s dacs to XMOS working fine with my other 1862 dac got from diyinhk.hk and then also an Amanero clone. It is the same issue but curiously if I test it using a 2RCA to 3.5mm headphone output cable, then inserting the headphone jack only a bit inside I get the both the channels working. I even replaced the RCA jacks but still the issue persists.
The member agreed to get back the dac and refund me but before doing that want to figure out if I can fix it but the SMD parts and my poor eye sight makes it difficult
Here are some pics for your reference.
Last edited:
Hi Richard
Here are the voltages of TP1-9 with respect to the 0v (GND).
TP1 = 5.47v
TP2 = 8.27v
TP3 = 9.97v
TP4 = 4.65v
TP5 = 2.48v
TP6 = 7.80v
TP7 = 7.81v
TP8 = 2.37v
TP9 = 1.87v
TP9 seems to be too low compared to the TP8. Let me know what else do I need to check or change as these are SMD I can hardly see even through magnifying glass with my poor eye sight
Some close up pics for your reference.
Thanks very much
Last edited:
Hi manniraj - thanks for posting up those numbers. The problem area looks to be around U8 (AD744) as its output seems stuck at the lower power rail. I will take a look at the pics to see if I notice anything but one possibility is U8 has failed.
I had a look at your pics, didn't notice any obviously bad joints around U8, you might add a little bit of extra solder to all those U8-local joints to be sure. Then if it still doesn't work the next step is checking voltages on the pins of U8.
I had a look at your pics, didn't notice any obviously bad joints around U8, you might add a little bit of extra solder to all those U8-local joints to be sure. Then if it still doesn't work the next step is checking voltages on the pins of U8.
Last edited:
Hi Richard
You were absolutely right about the U8 (AD744), I somehow managed to reflow the legs and now I get TP9 to exactly 2.36v. Both the channels working fine now. I touched with the hot iron the filter board mounting plastic socket using magnifying glass
but managed to keep the socket holes open and could mount the board and it plays fine.Thanks again for the help as always
Hi Richard,
First of all, I wanted to say thank you. Shipping was quick and the kits were well packed. When I first saw how small the chips are, I thought it was going to take me quite a while (they are smaller than they appear in the pics). Turns out it went very quickly though.
I could use some help troubleshooting. I followed the soldering guide, and measured the test points after completing each step, they were spot on. For some reason I am only getting output from the left channel, right channel is nothing. I tried reflowing the solder on the LT1028's and the ISL 28210, but no change. Any recommendation where to start next?
First of all, I wanted to say thank you. Shipping was quick and the kits were well packed. When I first saw how small the chips are, I thought it was going to take me quite a while (they are smaller than they appear in the pics). Turns out it went very quickly though.
I could use some help troubleshooting. I followed the soldering guide, and measured the test points after completing each step, they were spot on. For some reason I am only getting output from the left channel, right channel is nothing. I tried reflowing the solder on the LT1028's and the ISL 28210, but no change. Any recommendation where to start next?
My TP8=3.6v and TP6=10.1v as shown in the guide in post #304. TP1 is 6.9v though. I get maybe .01v across R50, nothing across R31. But also not getting any more than .01v across R26, which appears to be the equivalent on the working channel.
Hi Ben - those TP6 and TP8 voltages are for when no filter has been plugged in. I wanted to know what you get with the filter in-circuit. For the resistors, on mine I get 16mV for R50 and 12mV for R31. Yours sound close enough.
<afterthought> If you're getting the voltages you just reported with the filter in it means the filter isn't being connected correctly - check its not misaligned with the 20way sockets (not too hard to do) and also that the 20way sockets are fully soldered.
<afterthought> If you're getting the voltages you just reported with the filter in it means the filter isn't being connected correctly - check its not misaligned with the 20way sockets (not too hard to do) and also that the 20way sockets are fully soldered.
Last edited: