I am looking to use a PCM179x DAC, which has differential current outputs, to feed a power amp that has single ended inputs. For the IV stage I am considering using the Zen IV, but this would result in differential voltage outputs. I see there is the Sen IV for single ended use, but this seems to have single ended output and inputs.
What would you recommend to do for the differential to single ended stage ??
What would you recommend to do for the differential to single ended stage ??
hard not to be amused by people suddenly "getting religion" and wanting minimalist I/V - connected to a good fraction of a Million mostly digital transistors in the delta sigma DAC chip its wired to
having just reread Nelson's Zen IV article its amusing to see he used '70s era op amp examples in showing their "poor" performance for DAC I/V - and admitted he couldn't even compensate them properly
op amps have gotten lots better - especially in the last decade or so with telcom xDSL, medical imaging, other industrial/scientific applications outstripping audio performance demands
its not obvious what's wrong with the datasheet app/measurement circuits - particularly if you assuage your high frequency IMD suspicions with order of magnitude faster, more linear input op amps
with mA offset the I/V op amps are Class A biased so even the shown 5534 will be very good - thats what they claim the numbers were taken with
there are technically better op amps - Analog's "highly linear input" op amps appear to be hiding real innovation applicable to I/V with their rather shy advertising
there are also fet input op amp that should be very good - several recent op amps are targeted at OPA627 but faster, built on modern processes
even if you can source, match the discontinued jfet for the Zen IV - the op amp multiple feedback diff to SE converter circuits should work fine
at the pcm1794 current level you may need to parallel the fets, increase bias, add offset canceling input current
having just reread Nelson's Zen IV article its amusing to see he used '70s era op amp examples in showing their "poor" performance for DAC I/V - and admitted he couldn't even compensate them properly
op amps have gotten lots better - especially in the last decade or so with telcom xDSL, medical imaging, other industrial/scientific applications outstripping audio performance demands
its not obvious what's wrong with the datasheet app/measurement circuits - particularly if you assuage your high frequency IMD suspicions with order of magnitude faster, more linear input op amps
with mA offset the I/V op amps are Class A biased so even the shown 5534 will be very good - thats what they claim the numbers were taken with
there are technically better op amps - Analog's "highly linear input" op amps appear to be hiding real innovation applicable to I/V with their rather shy advertising
there are also fet input op amp that should be very good - several recent op amps are targeted at OPA627 but faster, built on modern processes
even if you can source, match the discontinued jfet for the Zen IV - the op amp multiple feedback diff to SE converter circuits should work fine
at the pcm1794 current level you may need to parallel the fets, increase bias, add offset canceling input current
Last edited:
I think you can seriously consider utilizing only one of the differential output phases. The distortion will be beneath that of the Zen I/V without having to worry about differential analog signal processing after the PCM1794A. See the below linked thread for an excellent series of PCM1794A distortion measurements made by Smms73.
http://www.diyaudio.com/forums/digital-source/221743-testing-pcm1794.html
http://www.diyaudio.com/forums/digital-source/221743-testing-pcm1794.html
you could use a transfomer for example?
have read at my Website (in signature), sure this will give you some ideas how it all can be done ....
jcx, I have an op-amp stage (like the datasheet) I am going to evaluate, but there are surely many that rate the minimalist IV approach, so I wanted to evaluate that too. The Zen i think might end up slightly cheaper too, (well, depends.. powering it may be trickier too) Sourcing the jfet is admittedly a problem, though. What would you recommend for the IV and Diff/filter stage op-amps.. something affordable (ie not OPA672 territory)?.. i'm only really (somewhat) familiar with TI's offerings and have been looking at something like the OPA2134s.
Ken, thanks for the input.. interesting.. I may try an IV stage like yours, what did you use for your linestage?
dddac, from what I have seen any decent trasformers are very expensive, so that rules it out for me.
Ken, thanks for the input.. interesting.. I may try an IV stage like yours, what did you use for your linestage?
dddac, from what I have seen any decent trasformers are very expensive, so that rules it out for me.
I had been using a Forte 44 discrete JFET linestage. Then, I tried directly connecting my home brew PCM1794A DAC directly (A.C. coupled) to my power amp. The digital fullscale analog output amplitude of my DAC is 200mVRMS (generated by passive resistor I/V, with no active gain stage post the PCM1794A). The improvement over my Forte 44 was so great that I've left it that way, forgoing the convienence of a volume control. Evetually, I plan to access the digital volume control feature built in to the 1794A, but, for the time being; 
I had been using a Forte 44 discrete JFET linestage. Then, I tried directly connecting my home brew PCM1794A DAC directly (A.C. coupled) to my power amp. The digital fullscale analog output amplitude of my DAC is 200mVRMS (generated by passive resistor I/V, with no active gain stage post the PCM1794A). The improvement over my Forte 44 was so great that I've left it that way, forgoing the convienence of a volume control. Evetually, I plan to access the digital volume control feature built in to the 1794A, but, for the time being;
So, you don't have any low pass filtering between the DAC and your power amp?
There are two issues here, not one. First, you need to figure out how you want to perform the current to voltage conversion. Second, you have the stated problem of converting the signal from differential to single ended.
You can use one leg of the DAC output as a single ended source, as Ken suggested. It apparently works ok.
You can get a different amplifier that has differential inputs, or you can get two amps per channel and bridge them.
You can use the opamp circuit shown in the PCM179X datasheets, or some version thereof.
You can use a discrete IV converter that also does the differential to single ended conversion. There are several threads on this type already.
You can do a lot of different things to get what you want, obviously. Only you can decide which is best, and that's what everyone seems to be doing here.
Hi, Dirk,
Not in the DAC itself. My power amp inputs feature a D.C. blocking cap. (Black Gate NX bipolar), and a 60KHz low pass filter.
dirk; the question began as a solution for getting diff-> SE for a Zen.. it would still be nice for a solution to that; using op-amps or half the chip aren't ideal solutions with the zen. But i have somewhat gone off the Zen (parts availability, +-30v power needed). What are the popular discrete IV converters that also do diff -> SE here? Most that i come across here are differential output.
Since it is so simple, I will also try a passive resistor output to compare against the op-amp circuit. My power amp has an input capacitor but not a low pass filter.. what should i do for this??
Since it is so simple, I will also try a passive resistor output to compare against the op-amp circuit. My power amp has an input capacitor but not a low pass filter.. what should i do for this??
This sounds exactly like the scheme that Raleigh Audio utilizes for their passive RAKK DAC output module. My concern with this solution is that the real source impedance driving the transformer primary is equal to that of the DAC chip's current outputs. This maybe fine for older DAC chips, such as the PCM1794 and AD1865, which feature approximately 1k ohm source impedances. However, newer DAC chips, such as PCM1794A, feature output impedances much higher that this, possibly above 1 Megohm. High source impedance would seem to present significantly adverse affects if used to drive an audio transformer primary.
Have you measured the THD, and frequency response at the transformer secondary? If so, what can you report?
I tried it with another current output DAC (TDA1541A), but I believe it can be easily adopted to PCM1794. The THD and frequency response depends on the actual transformer used, here are my results:
http://www.diyaudio.com/forums/digi...s-valve-output-stage-lundahl-transformer.html
This is just a proof-of-concept, the S/N is not the best, but I am working on it...
Below, is a link to that Raleigh audio transformer based PCM1794A interface to which I referred. There are no termination resistors connecting the DAC current outputs to ground. The DAC's output bias current is counter conducted through split primary windings to ground, producing a mutually cancelled flux, just as if there were no bias current present. Differential signal current produces an proportional imbalance in the canceled flux, thereby, inducing signal in the secondary.
This all is well and good, except that without resistors terminating the DAC current output pins, the real source impedance is fully determined by the DAC chip's output impedance. High ohmic source impedance is normally quite detrimental to transformer performance. Perhaps, I'm missing something here.
http://www.raleighaudio.com/passive_output.htm
Last edited:
The source resistance of the RAKK solution is the 3k I/V resistor divided by the turns ratio squared, that is 187.5R. This resistance together with the inductance of the transformer (seen at the primary) determines the low-frequency cutoff. It also contributes to damping the high-frequency ringing that is due to stray capacitance and inductance. Just like in the case of a pentode PP output transformer.
I read your comment as saying, that the reflected impedance from the secondary not only forms the load presented by the primary (agreed), but also forms it's own source impedance to drive that primary. I do not believe this second part is correct.
Well, lets draw the equivalent circuit (in head):
Current source (I) - transformer (1:n) - load (R) parallel connected
This can be redrawn to:
Current source (I) - reflected load parallel connected (R') - transformer
Which is eqivalent to:
Voltage source (U) - reflected load serially connected - transformer
Where U = I * R' and R' = R / n^2
Current source (I) - transformer (1:n) - load (R) parallel connected
This can be redrawn to:
Current source (I) - reflected load parallel connected (R') - transformer
Which is eqivalent to:
Voltage source (U) - reflected load serially connected - transformer
Where U = I * R' and R' = R / n^2
Except that the reflected load is due to inductive coupling, I believe that is the key distinction. This does seem a bit confusiing. Every transformer datasheet I've seen which graphs THD versus driver source resistance (see, Jensen's, for example) depicts THD substantially degrading as the driver source resistance increases. The related test circuits show the driver circuit to have the specified source resistance, they do not show that driver source resistance as being self-established by the reflected scondary load. Most datasheets specify that the driver source impedance should be a low as possible, right down to zero ohms.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.