Looks pretty obvious that Dave is using IC components.
Dave is a careful listener and I am sure he tried all kinds of values of resistors and caps (not sure if there are caps in the circuit) to arrive at what he liked.
Nonetheless, I bet a discrete one will likely be superior to an IC. Board space probably had a lot to do with his choice ALONG with the likely fact the discrete option will require measuring/selection of components. Dave's board is a small production run product, not made individually.
Whatever component he is using he sure is using a bunch of them. They must be used for various tasks. As I have mentioned before, if you read the articles Dave has written, he is a current source partisan when it comes to power supplies. He has to be using this same device to power the DAC, the input chip and all of that other stuff required for an SPDIF input and maybe some signal conditioning for the output, as part of supply regulation.
Maybe ZEN MOD, a master of circuit pathology and component identification could tell something from the picture? He is the only person I know who could see what is there and interpolate from a picture of a PCB.
I know I would rather use a ZM circuit than an IC. What a great conundrum and the proverbial embarrassment of riches to have a choice between a ZM idea and a Lowe idea. Both need to be explored.
Doede, thanks for the correction.
considering small voltage window , there is not many possibilities , all including usage of small voltage range demanding parts ; be it CC diode ( nothing else than sole small Jfet , in most cases ) or discrete version ....... where any cascoding technique is pretty close to voltage starve of entire shebang
can you point me to exact picture(s)/page ?
however - sonic difference between various CCSink iterations for pin 20 - need to be evaluated ; it can happen easily that there is no difference between CC diode and discrete solution of any complexity
whatever we put there , it'll be many times greater dynamic impedance than 10K-ish resistor ........ and - then , it's also worth to investigate various currents from pin 20 , in reasonable range
Thanks for these. Looking at your diagram I notices the +8V is connected to ground which seems odd, also the DDDAC does have 8V feed/regulators. Why would we need a seperate feed?
Look here:
K & K Audio - Lundahl Transformers, audio DIY kits and more
I sent Dave money for one of his boards but he was not exactly willing to take the chance of finding out whether it would work with the WaveIO so I ended up refusing the package with everyone's approval. I would like to have heard what it can do and likely would have been perfectly happy with it.
I think the problem was that both WaveIO and his DAC use those isolators (with the amusing name) and even though WaveIO has "non-isolated" outputs it was made clear if it did not work I would be stuck with a rather expensive board. Who wants to go backwards when it comes to USB to I2S and the board Dave offers (on close out) is outdated technology.
I will use Dave's line stage for my I/V stage as I have for years. Previous, though it is closing on two years since I had a working digital system, I used Dave's DACs and the "line stage", I/V stage.
So, I do not feel like I, or we, are doing anything untoward. Plus this picture is very low information!
K & K Audio - Lundahl Transformers, audio DIY kits and more
I sent Dave money for one of his boards but he was not exactly willing to take the chance of finding out whether it would work with the WaveIO so I ended up refusing the package with everyone's approval. I would like to have heard what it can do and likely would have been perfectly happy with it.
I think the problem was that both WaveIO and his DAC use those isolators (with the amusing name) and even though WaveIO has "non-isolated" outputs it was made clear if it did not work I would be stuck with a rather expensive board. Who wants to go backwards when it comes to USB to I2S and the board Dave offers (on close out) is outdated technology.
I will use Dave's line stage for my I/V stage as I have for years. Previous, though it is closing on two years since I had a working digital system, I used Dave's DACs and the "line stage", I/V stage.
So, I do not feel like I, or we, are doing anything untoward. Plus this picture is very low information!
well - if you ask me - judging by picture at http://www.kandkaudio.com/images/rakk_dac_2013.png (deliberately not puttin' picture here) and - what's written :
I can tell few things :
CCSink is CCSink is CCSink ........ if all of them are good enough ; considering needed current - under 500uA , there are myriad possible iterations , all of them having oomphteen kiloohms of dynamic impedance ; so - what's in quote is sweet talk oriented to paying customers ears
however - what's important is not in CCSink confiiguration , but can be in exact (chosen) value of programmed current
I'm writing all that with disclaimer - I still didn't have sole minute of experience with PCM1794 and importance of pin 20 either current or voltage ;
so , if we are talking strictly about CCSinks , one can use good programmed diode or good chip or Wardenclyffe Tower (good by default) ..... there can't be catastrophic difference between them
isn't that common way of connecting negative rail to gnd ?
I had to look that one up. Right, that's a new kind of "magic" for me.
Fascinating, I C we'd need a seperate 8V PS..
no need to look up
if you take independent (common electronic term is floating ) PSU of 8V , it's having positive and negative side
so , while floating in space and time and continuum ...... neither side can be called GND
if you connect it's negative side to general GND , then positive side is representing positive PSU rail in system
if you connect it's positive side to general GND , then negative side is representing negative PSU rail in system
if you take independent (common electronic term is floating ) PSU of 8V , it's having positive and negative side
so , while floating in space and time and continuum ...... neither side can be called GND
if you connect it's negative side to general GND , then positive side is representing positive PSU rail in system
if you connect it's positive side to general GND , then negative side is representing negative PSU rail in system
CCS Circuit Dynamic Testing
I did dynamic testing of the 2 transistor CCS circuit I submitted to the forum by injecting a 0.5vrms signal to the "pin 20" source point on my breadboard. Sine wave and square wave signals up to 100 KHz were tested. Most observations were done with a 10 KHz square wave. I connected the 825 ohm JFET source resistor to my oscilloscope and learned the following:
1. At the most sensitive setting on my scope, there is only a 0.2MV high signal that looks like a clean small replica of the signal at the "pin 20" location. The 0.4ma signal varies by less than 0.1%
2. The 22uf cap at the ZTX450 base is not needed. Doede, looks like a component reduction here!
3. Removing either the ZTX450 base resistor or the 2SK170 gate resistor results in a visible leading and trailing edge overshoot spike that is 3 times higher than the observed square wave signal. There was no oscillation from removing these resistors.
Now to build a pair of mini boards without the caps.
Today I upgraded my oldest DDDAC board with changes I already have on my newest DAC, both of which already have Bellesons on the digital circuit. I will be testing the CCS boards on this unit. The changes are:
1. 330 ohm load resistors were installed on the DAC board Bellesons to increase the current output to be 15ma.
2. I replaced the C1 470uf with a Panasonic 1000uf FC cap. I also replaced the C3 47uf cap with a 100uf OSCON. All the other large caps on the digital circuit were already OSCONs.
The audible improvement was immediate with the music being a bit tighter and clearer sounding with no negatives. Anyone who is using Bellesons on the digital circuits must make all the other changes and additions to get the full benefit.
I did dynamic testing of the 2 transistor CCS circuit I submitted to the forum by injecting a 0.5vrms signal to the "pin 20" source point on my breadboard. Sine wave and square wave signals up to 100 KHz were tested. Most observations were done with a 10 KHz square wave. I connected the 825 ohm JFET source resistor to my oscilloscope and learned the following:
1. At the most sensitive setting on my scope, there is only a 0.2MV high signal that looks like a clean small replica of the signal at the "pin 20" location. The 0.4ma signal varies by less than 0.1%
2. The 22uf cap at the ZTX450 base is not needed. Doede, looks like a component reduction here!
3. Removing either the ZTX450 base resistor or the 2SK170 gate resistor results in a visible leading and trailing edge overshoot spike that is 3 times higher than the observed square wave signal. There was no oscillation from removing these resistors.
Now to build a pair of mini boards without the caps.
Today I upgraded my oldest DDDAC board with changes I already have on my newest DAC, both of which already have Bellesons on the digital circuit. I will be testing the CCS boards on this unit. The changes are:
1. 330 ohm load resistors were installed on the DAC board Bellesons to increase the current output to be 15ma.
2. I replaced the C1 470uf with a Panasonic 1000uf FC cap. I also replaced the C3 47uf cap with a 100uf OSCON. All the other large caps on the digital circuit were already OSCONs.
The audible improvement was immediate with the music being a bit tighter and clearer sounding with no negatives. Anyone who is using Bellesons on the digital circuits must make all the other changes and additions to get the full benefit.
no need to look up
if you take independent (common electronic term is floating ) PSU of 8V , it's having positive and negative side
so , while floating in space and time and continuum ...... neither side can be called GND
if you connect it's negative side to general GND , then positive side is representing positive PSU rail in system
if you connect it's positive side to general GND , then negative side is representing negative PSU rail in system
Ok to test this with battery source vs the diodes I assume?
black and white (it says in dutch) cut short and let them float
Value is not critical. 10k - 15k was reported to be good, but you can try if it does something to your ears. No limits in value though. You can leave it out as well
I have just wired up the Cinemags, and I get quite a bad hum from the speakers. Would this be de to me not putting a resistor between the yellow and orange wires.
The sound is also quite thin in the mid range, would this also be due to the lack of resistor or do the Cinemags need running in a bit?
Thanks!!
Cool i should have all the necessaries
Do I need to be at 8v or do I have a little bit of a range?
Nice to see some good developments here lately 
Did anyone manage to properly test the effects of different types of caps for the vcom yet? I see a some have gone for OSCON there, but not all...
I see it's already been looked at, but for what it's worth, when I was searching for a good pic of the mkIV RAKK dac last week, this was the best one I found.
http://www.hifisentralen.no/forumet...70782174-bilder-av-diy-dubeditter-rakk4_1.jpg
Did anyone manage to properly test the effects of different types of caps for the vcom yet? I see a some have gone for OSCON there, but not all...
I see it's already been looked at, but for what it's worth, when I was searching for a good pic of the mkIV RAKK dac last week, this was the best one I found.
http://www.hifisentralen.no/forumet...70782174-bilder-av-diy-dubeditter-rakk4_1.jpg
I recommend not to do this. the bias curent in combination with the 8 volt supply and the Rload resistors are a well balanced combination of DC bias, maximum output swing at 0dBFS and THD...
higher or lower bias currents at pin 20 wil disturb this equilibrium and hence can produce distorted sound at higher outputs, both at top or bottom which have different distortion signature.
as long as you measure 2,7 volt DC at the POS or NEG pin, the bias is ok....
when using a in tests a variable CCS this is the best way to trim the CCS....
Nice to see some good developments here lately
Did anyone manage to properly test the effects of different types of caps for the vcom yet? I see a some have gone for OSCON there, but not all...
I see it's already been looked at, but for what it's worth, when I was searching for a good pic of the mkIV RAKK dac last week, this was the best one I found.
http://www.hifisentralen.no/forumet...70782174-bilder-av-diy-dubeditter-rakk4_1.jpg
I see in the picture that this a version 2. there is no CCS on this one, just a resistor connected to pin 20.
anything between 4-12V
scale accordingly resistor for T2 base ( ref to sch in #2637 )
bjts - any NPN uni type , BC546C , BC 550C for chaps in EU
ZTX450 for US guys
Last edited: