Why do we need I/V stage for AD1865

I have designed and built AD1865 based DACs in Vout mode, and in passive I/V curret output mode. My experience is that while much depends on the implementation of the modes, I obtained MUCH better sound from the current outputs via passive I/V. I've used passive I/V resistor values as large as 330R without aubible distortion, but find that lower value resistors give a subjectively more focused and organized kind of sound that's particularly obvious in the bass. I currently use 75R, but want to try 50R in the near future. Follow this hyperlink for further detail. http://www.diyaudio.com/forums/digi...-6dj8-srpp-mod-suggestions-4.html#post3432652

Another intriguing option, which I haven't yet had oportunity to try myself, is proposed by 'Vandenberg' on his homepage. It intrigues me because it follows a basic topology a number of us have found to enable outstanding sound with current output DACs. The basic topology is: low value passive I/V resistor, followed by a passive low pass filter, followed by a medium gain active voltage amplification stage. Vandenberg implements that topology with the AD1865 by re-purposing the AD1865's internal op-amp as a voltage amplifier (instead of transimpedance amplifier, as intended). This is easily done by inserting two external resistors in the op-amps's feedback loop via it's connection pins.

http://www.diyaudio.com/forums/atta...d1362304514-ad1865-dac-singing-ad1865-diy.doc

http://www.diyaudio.com/forums/digital-line-level/231240-ad1865-dac-singing.html
 
Last edited:
...In fact, do you think it is possible to take the output from pins 8 and 17 of the dac and get i-out?

I-out is directly available via pins 4 and 21. The AD1865 may be configured to operate in either V-out or I-out mode. I don't find the V-out mode to sound particularly pleasing, but that's only my opinion.

Also, in my experience, there is no perceptible benefit of correcting the 11uS linear phase interchannel delay, at least when listening to loudspeakers. All of those shift register chips may increase jitter though.
 

lordearl

Member
2007-06-14 1:42 pm
Ken - this is just an unbalanced dac which I'll be running into a headphone amp. The headphone amp has enough gain, so I could use i-out on the dac if you think it sounds better, just not sure how to implement i-out given the pcb L and R out are taken from the v-outs of the dac chip
 
Ken - this is just an unbalanced dac which I'll be running into a headphone amp. The headphone amp has enough gain, so I could use i-out on the dac if you think it sounds better, just not sure how to implement i-out given the pcb L and R out are taken from the v-outs of the dac chip

Okay, I think I understand. Yes, perhaps the least PCB destructive and most direct way would be to CAREFULLY and quickly desolder then lift pins 4 and 21. I would only try this if my AD1865 is in the robust DIP package, but not in the more fragile SOIC package. Then, carefully solder the I/V resistors from each IOUT pin to analog ground. I also solder an 500pF NP0 type ceramic bypass capacitor in parallel with the I/V resistor for some minimal ultrasonic and RF filtering.

As for the value of the I/V resistor, I've used values as high as 330R with the AD1865 without audible distress. However, lower values do sound better to my ears - no, not overtly sounding lower in distortion, but sounding more focused and organized, particularly in the bass. The lowest value resistor you can use will depend on the gain of your headphone amp, and the maximum drive voltage required by your headphones at your loudest listening level. You don't want to determine the I/V resistor value empirically, by soldering and desoldering resistors as that can too easily lead to irreparable damage. Rather, compute the minimum useable resistor value, and try to happily live with the resulting volume range.

For an example, should your maximum comfortable headphone drive voltage be 5 volts peak, and the voltage gain of your headphone amp be x20 (+26dB), then the fullscale peak signal voltage across the I/V resistor should be set to 5 volts / 20 = 250mV. The fullscale signal current from either IOUT pin of the Ad1865 is 1mA, so, you simply apply Ohm's Law to find the required I/V resistor value. Which, in this example, is 250 ohms. You may discover, however, that your headphone amp has insufficient gain to make this work, meaning, resulting in an I/V resistor value much above 330 ohms. In which case, you may still want to try this passive resistor I/V modification to subjectively ascertain how promising it would be if implemented in a project specifically designed to utilize it.
 
Last edited:

lordearl

Member
2007-06-14 1:42 pm
Cheers Ken,
I have ordered two pcbs from PCBFAST so will build them with identical components - one utilising v-out, the other i-out and we can compare the differences. It is a pity AD never gave some more informative documentation about the onboard opamp they designed for the AD1865 - thought I do notice that the datasheet recommends using NE5532 as a buffer after the v-output, which is interesting in itself - perhaps BUF634 or even LME49720 might be better choices?
 
They do provide the following bit of information about the internal opamp:

"The output amplifiers use both MOS and bipolar devices and incorporate an all NPN output stage. This design technique produces higher slew rate and lower distortion than previous techniques. Frequency response is also improved. When combined with the appropriate on-chip feedback resistor, the output op amps convert the output current to output voltages."

Sorry, but I can't offer knowledgable advice regarding external opamp selection. I gave up utilizing monolithic opamps in my audio designs some time ago.
 
not only the opamp

it is not only the opamp which is very very good the way i use it.
i use 2 ad1865 chips per channel and 1 opamp ,the resistors are mounted on the back side of the chips , no trouble with powersupply of the opamp and a very good signalreturn path.
so a very short signal path .
futhermore a have a very low jitter of the clock (radiochip is in slave mode).
the power supply is also very good ( no ground loops).
 

lordearl

Member
2007-06-14 1:42 pm