I am looking to build a modern DAC with USB as well as optical and/or coaxial input and the ability to handle a wide range of input standards - but primarily 44kHz 16-bit for now. My aim is to play with better power supply, clock and output stages. I don't want to spend too much on my first attempt but I am aiming at better than just good sound quality. Ideally I wanted current-output DACs so that I could do my own I/V stage (discrete or quality transformers). Most are voltage output except the ES9038.
HIFI CS4398 CS8416 24bit/192KHZ DAC board kit audio decoder USB optical Input | eBay
ES9038Q2M XMOS USB Optical fiber coaxial input Stereo Mobile Audio DAC | eBay
WM8740 + DIR9001 DAC Board Support Coaxial And USB Input | eBay
DAC KIT CS8416+CS4398 DAC board ( USB+coaxial DAC 192K/24BIT Board ) | eBay
CS4398 DAC DIY Kit with USB Coaxial 32K-192K/24BIT Decoder Supplies Kit AC 15V | eBay
https://www.ebay.co.uk/itm/114749894910 (is it feasible to add USB input to this one?)
If anyone has experience of any of these boards then I would like to hear about it. Buying from China on eBay always runs the risk of fake and/or QC-fail parts being used, hence my interest in real world experience.
Any recommendations for current-output DACs that I haven't yet been able to find on eBay also welcomed. These seem pretty hard to find now.
Thanks in advance.
M
HIFI CS4398 CS8416 24bit/192KHZ DAC board kit audio decoder USB optical Input | eBay
ES9038Q2M XMOS USB Optical fiber coaxial input Stereo Mobile Audio DAC | eBay
WM8740 + DIR9001 DAC Board Support Coaxial And USB Input | eBay
DAC KIT CS8416+CS4398 DAC board ( USB+coaxial DAC 192K/24BIT Board ) | eBay
CS4398 DAC DIY Kit with USB Coaxial 32K-192K/24BIT Decoder Supplies Kit AC 15V | eBay
https://www.ebay.co.uk/itm/114749894910 (is it feasible to add USB input to this one?)
If anyone has experience of any of these boards then I would like to hear about it. Buying from China on eBay always runs the risk of fake and/or QC-fail parts being used, hence my interest in real world experience.
Any recommendations for current-output DACs that I haven't yet been able to find on eBay also welcomed. These seem pretty hard to find now.
Thanks in advance.
M
No direct experience of this board but its using a current-output DAC and the PCB layout doesn't look like a complete dog's dinner.
ZEROZONE Assembeld Hifi Mini PCM1792 DSD DAC Decoder board support Double mode L11 35|Amplifier| - AliExpress
The other current out chip I'd say was worth your time is the AD1955. However the only board I've been able to find using it looks a bit of a risk layout-wise.
ZEROZONE Assembeld Hifi Mini PCM1792 DSD DAC Decoder board support Double mode L11 35|Amplifier| - AliExpress
The other current out chip I'd say was worth your time is the AD1955. However the only board I've been able to find using it looks a bit of a risk layout-wise.
> HIFI CS4398 CS8416 24bit/192KHZ DAC board kit audio decoder USB optical Input | eBay
Uses CM102, adaptive USB so high jitter, no onboard clock, garbage.
> ES9038Q2M XMOS USB Optical fiber coaxial input Stereo Mobile Audio DAC | eBay
There's only one opamp which means it uses voltage mode, not current mode, so expect high THD.
> WM8740 + DIR9001 DAC Board Support Coaxial And USB Input | eBay
DIR9001 so no jitter rejection, not clicking on the link.
> DAC KIT CS8416+CS4398 DAC board ( USB+coaxial DAC 192K/24BIT Board ) | eBay
CS8416 so no jitter rejection, not clicking on the link.
> CS4398 DAC DIY Kit with USB Coaxial 32K-192K/24BIT Decoder Supplies Kit AC 15V | eBay
No crystal oscillator on board, CM102 adaptive mode, high jitter.
> https://www.ebay.co.uk/itm/114749894910 (is it feasible to add USB input to this one?)
Uses CS8416 so no.
Much nicer than the others but it has split ground plane, would have to sand off the soldermask to fix that, lots of work...
Uses CM102, adaptive USB so high jitter, no onboard clock, garbage.
> ES9038Q2M XMOS USB Optical fiber coaxial input Stereo Mobile Audio DAC | eBay
There's only one opamp which means it uses voltage mode, not current mode, so expect high THD.
> WM8740 + DIR9001 DAC Board Support Coaxial And USB Input | eBay
DIR9001 so no jitter rejection, not clicking on the link.
> DAC KIT CS8416+CS4398 DAC board ( USB+coaxial DAC 192K/24BIT Board ) | eBay
CS8416 so no jitter rejection, not clicking on the link.
> CS4398 DAC DIY Kit with USB Coaxial 32K-192K/24BIT Decoder Supplies Kit AC 15V | eBay
No crystal oscillator on board, CM102 adaptive mode, high jitter.
> https://www.ebay.co.uk/itm/114749894910 (is it feasible to add USB input to this one?)
Uses CS8416 so no.
Much nicer than the others but it has split ground plane, would have to sand off the soldermask to fix that, lots of work...
Thanks guys. Sounds like a "No" to all of those then!
The board that abraxalito suggested would appear to need at least an input board (for USB, optical, coax etc), possibly also a clock? Are these easy to add without worrying about synergy of the combinations? If so then I like the idea of choosing good examples of these too. Please excuse if this is a silly question but digital electronics is not my strength.
@abraxalito I see from some old posts that you were a fan of transformers for I/V at one point. Do you still like to do this or have discrete/opamp circuits taken over for you? I am having some positive experiments with transformers for both I/V and differential voltage output. Quality transformers have surprisingly low distortion even at low frequencies and (importantly) IMD. You get no DAC I/V TIM/Phase modulation in the Hawksford sense, galvanic isolation, good CMR for differential outputs and 'free antialiasing filters' (though you are at the mercy of the particular design as to Fc and slope - you can always add passive pre- and post-filtering filtering though).
M
The board that abraxalito suggested would appear to need at least an input board (for USB, optical, coax etc), possibly also a clock? Are these easy to add without worrying about synergy of the combinations? If so then I like the idea of choosing good examples of these too. Please excuse if this is a silly question but digital electronics is not my strength.
@abraxalito I see from some old posts that you were a fan of transformers for I/V at one point. Do you still like to do this or have discrete/opamp circuits taken over for you? I am having some positive experiments with transformers for both I/V and differential voltage output. Quality transformers have surprisingly low distortion even at low frequencies and (importantly) IMD. You get no DAC I/V TIM/Phase modulation in the Hawksford sense, galvanic isolation, good CMR for differential outputs and 'free antialiasing filters' (though you are at the mercy of the particular design as to Fc and slope - you can always add passive pre- and post-filtering filtering though).
M
As far as I saw (I only looked briefly) that PCM1792 had I2S input so that would mean any interface (coax/optical/USB) that outputted I2S would be a suitable match. I rather favour using an SDcard player for DAC development myself, the kind that spits out I2S.
As regards transformers, yes still a fan of them though I've curtailed winding activities of late and have been gradually getting into sourcing off-the-shelf ones. I'm currently working on a revision of my 'GrossDAC' design (so called because it started life with 144 paralleled DAC chips) and that's an attempt to use a moderately cheap Taobao-sourced nickel-cored trafo. Opamps are good for when small size, low weight is wanted though I don't used them 'naked' on a DAC output, only after LC filtering.
As regards transformers, yes still a fan of them though I've curtailed winding activities of late and have been gradually getting into sourcing off-the-shelf ones. I'm currently working on a revision of my 'GrossDAC' design (so called because it started life with 144 paralleled DAC chips) and that's an attempt to use a moderately cheap Taobao-sourced nickel-cored trafo. Opamps are good for when small size, low weight is wanted though I don't used them 'naked' on a DAC output, only after LC filtering.
Some modern current-out dacs have such low output impedance that transformers and discrete circuits are not the best way. Better results can be had using opamp I/V. Some or most of the older current-out dac chips are higher output Z and thus more suited what you say you are interested in doing. The new Rohm BD34301EKV may be that way too.
Regardless of everything else, none of the low-cost (<$100) Chinese dac boards I have seen use 4-layer PCBs, which IMHO is necessary for a good dac. Also, the cheap boards typically have a genuine dac chip, but all the passive components are cheap as possible. Sometimes caps are made to look quality name-brand parts, but are often fakes. That's how they make a profit selling low-cost boards. If you want to hone your skills with one, I would suggest to look for one with as much of a continuous, uninterrupted ground plane as possible. At least you have a chance of modifying it into something better, and the ground plane side can be a location to mount additional components such as local dedicated LDO voltage regulators for the various low voltage loads (except maybe not LDOs for dac chip analog reference voltage pins -- e.g. AVCC).
Regardless of everything else, none of the low-cost (<$100) Chinese dac boards I have seen use 4-layer PCBs, which IMHO is necessary for a good dac. Also, the cheap boards typically have a genuine dac chip, but all the passive components are cheap as possible. Sometimes caps are made to look quality name-brand parts, but are often fakes. That's how they make a profit selling low-cost boards. If you want to hone your skills with one, I would suggest to look for one with as much of a continuous, uninterrupted ground plane as possible. At least you have a chance of modifying it into something better, and the ground plane side can be a location to mount additional components such as local dedicated LDO voltage regulators for the various low voltage loads (except maybe not LDOs for dac chip analog reference voltage pins -- e.g. AVCC).
Last edited:
@abraxalito I'm getting good results with Carnhill 9072 which is a quality DI TX with a large mumetal core. I'm using it backwards with 1:6 step-up though it can go to 1:24 if you're careful with termination and have a nice high source impedance to work into and good shielding, as those high impedances do tend to pick up all the electronic hash that is flying around.
After reading the Hawksford paper and because I am using a DAC with high output Z as mentioned by @Markw4, I no longer use virtual earth I/V with opamps; instead getting the voltage across a smallish resistor to ground (8.2 ohms for the TDA1541) with a capacitor across it (~0.22µF) to get one LP pole before the I/V opamp, and then gain - though far less than is customary at the first I/V stage. As a result the neg feedback FR and pulses are very much reduced. Combining this with a fast opamp (was using LT1028 until I killed one with a soldering iron from +ve to o/p, now thinking about trying OPA1611/1656/827 but in the meantime OPA134), I'm hearing less electronic glare, more body/presence and in particular cymbals that sound less synthetic. Still prefer the TX though. I would like to try the Hawksford nested differential FB scheme using opamps from his paper but I can't get a sim in LTSpice to make sense and can't see what I did wrong either! Having said that, I'm a novice with that software.
After reading the Hawksford paper and because I am using a DAC with high output Z as mentioned by @Markw4, I no longer use virtual earth I/V with opamps; instead getting the voltage across a smallish resistor to ground (8.2 ohms for the TDA1541) with a capacitor across it (~0.22µF) to get one LP pole before the I/V opamp, and then gain - though far less than is customary at the first I/V stage. As a result the neg feedback FR and pulses are very much reduced. Combining this with a fast opamp (was using LT1028 until I killed one with a soldering iron from +ve to o/p, now thinking about trying OPA1611/1656/827 but in the meantime OPA134), I'm hearing less electronic glare, more body/presence and in particular cymbals that sound less synthetic. Still prefer the TX though. I would like to try the Hawksford nested differential FB scheme using opamps from his paper but I can't get a sim in LTSpice to make sense and can't see what I did wrong either! Having said that, I'm a novice with that software.
@Markw4 I would like to use a current o/p DAC to allow experimentation with the I/V stage which does seem key to sound quality but I also recognise that most DACs now are voltage-out and some of these do seem to be highly rated for SQ (e.g. WM874*, CS439*) and deliver you line level and good drive capability without much extra work to do, and sometimes I do wonder if the extra work is worth it.
Everyone has mentioned PCB quality which I had not taken that seriously but it sounds like I should. Not sure how to confirm that in advance of buying something.
I'm not worried about the passive components as anything important I would be replacing anyway. Would also strongly prefer to avoid SMC as at my age the eyes and the hands struggle enough with normal components!
Everyone has mentioned PCB quality which I had not taken that seriously but it sounds like I should. Not sure how to confirm that in advance of buying something.
I'm not worried about the passive components as anything important I would be replacing anyway. Would also strongly prefer to avoid SMC as at my age the eyes and the hands struggle enough with normal components!
Most of us around here seem to be older, have less than ideal vision, and sometimes shaky hands. There are ways of dealing with those things though, for example: how to hold a soldering Iron steady
Regarding a good PCB for TDA5141, there is probably someone around here who knows the answer. For modern low output-Z dac chips, maybe its time to learn how to design a board. Kicad is free and getting pretty good, so some say.
As to whether or not current out is worth the trouble, depends on what sound you want. If you want micro detail, low distortion, and genuine warmth all at the same time that can take a lot of work, but it can be done.
Regarding a good PCB for TDA5141, there is probably someone around here who knows the answer. For modern low output-Z dac chips, maybe its time to learn how to design a board. Kicad is free and getting pretty good, so some say.
As to whether or not current out is worth the trouble, depends on what sound you want. If you want micro detail, low distortion, and genuine warmth all at the same time that can take a lot of work, but it can be done.
Last edited:
I had to look up Carnhill, not a name I've come across before. My trafos in the latest incarnation of my DAC are also ~1:6.
What led to your choice of 8.2ohm? The compliance spec (50mV total swing) would indicate you could use 12ohm ISTM. I rather suspect you'll be getting a bit too much noise even from an LT1028 with such a low source resistance. A trafo would be able to do better, noise-wise and subjectively the lower noise level (even though notionally already below that on the recording) would give you a wider soundstage with more 'air' I reckon.
It seems that Carnhill, Sowter and OEP are now all part of the same group. I spoke to John Hall who is the MD at Carnhill recently to ask him for some measurements on these TXs so that I can try to build accurate lumped parameter models for them (including hysteresis using the Chan model, I hope).
8.2 ohm because I have read in various places that the voltage compliance for TDA1541 was 25mV and with ±2mA swing that is already close, and I happened to have 2x in my rather limited parts bin. I have used test tones with full CD level: THD at 20Hz, 1kHz and various IMD (SMPTE, DIN, AES TDFD, CCIF) and there is no evidence of gross distortion anywhere. In fact, I'm very surprised at how good the distortion performance is, especially at 20Hz where most TXs are struggling - I guess the AC input level is very low at <25mV which avoids overloading the TX. I may try a larger resistor value to fight noise and seek more transparency when I get hold of some. I am also putting a cap across the I/V resistor to get one LP pole just outside the audio band.
I'm now not sure if the LT1028 is damaged or just oscillates on one channel of the CD94 o/p board. Apparently it is rather highly-strung. I'm about to build the Hawksford NFB I/V using OPA134 and LT1022 (or 1028 if I can get it to be stable) and am very curious how much difference this makes. The pulse behaviour looks very good in LTSpice so it should be a test of what opamps are capable of in this position.
8.2 ohm because I have read in various places that the voltage compliance for TDA1541 was 25mV and with ±2mA swing that is already close, and I happened to have 2x in my rather limited parts bin. I have used test tones with full CD level: THD at 20Hz, 1kHz and various IMD (SMPTE, DIN, AES TDFD, CCIF) and there is no evidence of gross distortion anywhere. In fact, I'm very surprised at how good the distortion performance is, especially at 20Hz where most TXs are struggling - I guess the AC input level is very low at <25mV which avoids overloading the TX. I may try a larger resistor value to fight noise and seek more transparency when I get hold of some. I am also putting a cap across the I/V resistor to get one LP pole just outside the audio band.
I'm now not sure if the LT1028 is damaged or just oscillates on one channel of the CD94 o/p board. Apparently it is rather highly-strung. I'm about to build the Hawksford NFB I/V using OPA134 and LT1022 (or 1028 if I can get it to be stable) and am very curious how much difference this makes. The pulse behaviour looks very good in LTSpice so it should be a test of what opamps are capable of in this position.
In most cases, it you don't want to diy to point of assembling a board, it may be better to go with something like a Topping or SMSL that fits your budget. It will probably sound better per dollar spent than if you try to assemble some low-cost preassembled boards into a finished dac with power supplies, a USB board, a case, etc.