Fully balanced OPA861 I/V stage
OPA861 is an operational transconductance amplifier. It has super high slew rate up to 900V/us with wide bandwidth of 80MHz. In current input mode, the input impedance can be low around 10.5 ohm. The performance is pretty much like a super transistor. The ultra-high speed and very low current mode input impedance make it almost the perfect component for I/V converter besides the -57dB 3rd-harmonic.
http://www.ti.com/lit/ds/symlink/opa861.pdf
Audial designed an OPA861 I/V stage for TDA1541A DAC in 2013 with good reputation from audiophiles. That encourages me to try a fully balanced OPA861 I/V stage for ESS9028Q2M and ESS9038Q2M DAC HAT.
Please see the attached schematic for principle details. This I/V stage will also have zero feedback similar to transformer I/V, so nature sound style is expected.
PCB order placed. Hopefully I can start to evaluate this balanced OPA861 I/V stage very soon.
Ian
OPA861 is an operational transconductance amplifier. It has super high slew rate up to 900V/us with wide bandwidth of 80MHz. In current input mode, the input impedance can be low around 10.5 ohm. The performance is pretty much like a super transistor. The ultra-high speed and very low current mode input impedance make it almost the perfect component for I/V converter besides the -57dB 3rd-harmonic.
http://www.ti.com/lit/ds/symlink/opa861.pdf
Audial designed an OPA861 I/V stage for TDA1541A DAC in 2013 with good reputation from audiophiles. That encourages me to try a fully balanced OPA861 I/V stage for ESS9028Q2M and ESS9038Q2M DAC HAT.
Please see the attached schematic for principle details. This I/V stage will also have zero feedback similar to transformer I/V, so nature sound style is expected.
PCB order placed. Hopefully I can start to evaluate this balanced OPA861 I/V stage very soon.
Ian
Attachments
Last edited:
Thank you for providing information as a proof that I said what ESS chip does not need to be designed in mono block and parallel configuration.
The current output of ESS chip is large enough, and it is easy for negative factors to be greater than positive ones when it be designed in mono block and parallel configuration. This is why all of the products you said it have been discontinued, just after 9018 DAC was launched.
Last edited:
Thank you for providing information as a proof that I said what ESS chip does not need to be designed in mono block and parallel configuration.
The current output of ESS chip is large enough, and it is easy for negative factors to be greater than positive ones when it be designed in mono block and parallel configuration. This is why all of the products you said it have been discontinued, just after 9018 DAC was launched.
The pure audio lab once introduced the 9018 and 9038 DACs. Lotus DAC3 uses two 9018s, but in Lotus DAC5 uses one 9038. Of course not because the 9038 chip is more expensive. As early as pure audio launched mono dac, I told him that this is meaningless and just increase the cost. At the time, he was still very unconvinced. As a result, his product could talk what he gave up the design of mono dac.
The current output of ESS chip is large enough, and it is easy for negative factors to be greater than positive ones when it be designed in mono block and parallel configuration. This is why all of the products you said it have been discontinued, just after 9018 DAC was launched.
The pure audio lab once introduced the 9018 and 9038 DACs. Lotus DAC3 uses two 9018s, but in Lotus DAC5 uses one 9038. Of course not because the 9038 chip is more expensive. As early as pure audio launched mono dac, I told him that this is meaningless and just increase the cost. At the time, he was still very unconvinced. As a result, his product could talk what he gave up the design of mono dac.
Hi Ian,
I recently saw an article about a joint collaboration on a new dac named "Kantana". Its between Allo.com and a company named SparkoS Labs Inc.
This SparkoS Labs might be an interesting company for you because they advertise with a product called "discrete opamps";
Sparkos Labs has created a family of discrete op amps in an 8 pin DIP compatible package optimized for high performance audio applications. These devices are drop in replacements for many common, yet inferior audio op amps and are uniquely compensated for trouble-free swap out into virtually any circuit. Having a footprint of a mere 0.33 square inches, these devices are one-third the size of any other discrete op amp on the market. Class A biasing and high output current capability coupled with a proprietary compensation scheme requiring multiple NPO dielectric capacitors make these discrete op amps impossible to fabricate as a monolithic IC. These devices utilize matched transistor pairs encapsulated within a single device package for the input stage and internal current mirrors, which retains the advantage that monolithics have with device matching. Each device is fully specified, and is available as a single or dual discrete op amp.
You can find them at Home - SparkoS Labs Inc.
Kind Regards,
Frank
Sparkos makes some very interesting devices. I have some of their discrete opamps here to try, but haven't done that as-yet. I HAVE used their discrete regulators in a few places and been very impressed with the results!
A VERY good set of comments contrasting the requirements for an I/V opamp (including the Sparkos devices) for the ES9038 are posted by b0bb in this long thread about the LKS ESS-based DAC:
LKS Audio MH-DA003 | Headphone Reviews and Discussion - Head-Fi.org
bobb also posts some great tips on alternative clocks and capacitors in this thread.
I'm following the allo.com Katana thread very closely. Allo has been working hard to transform the RPi audio space at reasonable costs, just as Ian has. Based on what I'm hearing from the prototype of Ian's ES9028Q2M DAC, I'm VERY interested in hearing Allo's Katana... based on their recent upgrade to their top-of-the-middle-end Boss 1.2, they are learning fast and applying their lessons very effectively.
Greg in Mississippi
Last edited:
VERY INTERESTING tangent Ian!
I've been curious about the use of devices such as these for I/V. Joe Rasmussen is also a fan of them and included a set of high-level schematics for their use with and without his alternative filtering configuration:
Practical Implementations of Alternative Post-DAC Filtering
I'm a fan of no and low-feedback setups, based on my few experiments and comments from others. This should be a worthy alternative to the transformer output stage.
Greg in Mississippi
@iancanada
Ian, Why not to implement Buffalo III analog output footprint on your board and use any of existing twistedpearaudio I/V stages available today ... and develop new ones of course.
One of my favourite ones is Legato v3.1. I use it with my es9038q2m.
Last edited:
Hi Ian, just wondering if your controller is available for purchase yet? I would like to try and resurrect my current ES9018 board - the main reason I put it away was because I could not control the ESS volume control from my RPi. Is the volume control driver for the controller available?
Thanks.
Thanks.
Hi TioFrancotirador
That's a very good idea. I like it and will take it into consideration.
Thanks,
Ian
Sorry, I haven't and likely won't soon. Ian got a pair of 1544As, I have several transformer options I'm trying and have to try... in the Lundahl line I have 1570XL & 1684s (both a stock pair and a pair that has been 'nuded').
At some point, Ian & I should work out trading some of the transformer options back and forth.
OTOH, my next planned trials on Ian's ES9028Q2M DAC board is to upgrade it with additional LT3042 regulator board. The devices Ian uses on his boards are on backorder EVERYWHERE, even Linear Technology. BUT Digikey was suggesting a ship date early next week. Once I get them, I'll build up a couple more of his regulator boards and use them in the currently un-filled regulator board spots on the prototype board I have from Ian.
Greg in Mississippi
Hi Frank,
Sparkos has some very interesting products. If I have chance to evaluate them, I'll get them compared with other popular op amplifiers, such as OPA1612, OPA1622, OPA2604 and so no. If the result is promising, I might be considering a standard I/V stage with sockets to adapt with them.
Thanks,
Ian
TI samples. OPA861 and OPA1632.
I'm quite excited about the so many things I'm going to try.
http://www.diyaudio.com/forums/pc-b...i2s-dac-hats-raspberry-pi-37.html#post5454403
OPA861OPA1632samples by Ian, on Flickr
Ian
I'm quite excited about the so many things I'm going to try.
http://www.diyaudio.com/forums/pc-b...i2s-dac-hats-raspberry-pi-37.html#post5454403
OPA861OPA1632samples by Ian, on Flickr
Ian
Thanks Greg for the link. Didn't notice the thread before. It seems Rasmussen has same interest with me. I'll look into it.
You are right. No-feedback and low feedback I/V stage sound fantastic, less flavor with more music.
Clock of a new DAC can go up to 100MHz, I/V stage has to be very high speed to handle the high frequency. Light feedback or no feedback are preferred.
Regards,
Ian
Although I'm not using a DAC which requires such a high clock frequency I do find my I/V stage sounds better when fairly steep filtering is applied between DAC chip and I/V. This way the I/V stage doesn't have to deal with an ultra wide bandwidth, just audio frequencies. I still use zero GNFB I/V though
Although I'm not using a DAC which requires such a high clock frequency I do find my I/V stage sounds better when fairly steep filtering is applied between DAC chip and I/V. This way the I/V stage doesn't have to deal with an ultra wide bandwidth, just audio frequencies. I still use zero GNFB I/V though
Hi abraxalito,
For ESS DACs, we need 100MHz MCLK to make it capable for 384KHz at ASYNC mode, or 90/98 MHz for 768KHz at SYNC mode. And I found they sound better for higher frequency than lower. I/V converters in this case are very sensitive to the bandwidth. Lower bandwidth I/V converters always tend to degrade the sound quality.
A filter before I/V is really a good idea. But how to do it at true current mode with very low input impedance?
Regards,
Ian
Hi Ian
I've not played around with ESS DACs, simply because I don't wish to sign the NDA and then be prevented from sharing what I learn.
With current out DACs having high impedance (current source) outputs the following impedance of the filter isn't too critical, it just needs to be lower than (say) 1% of the output impedance. Which is fairly easy when a CS output DAC has 10's of kiloohm Zout. LC filter tables in Zverev (the bible of passive filter designers) have a column for where the driving impedance is infinite.
With your ESS DACs the Zout I doubt is constant and likely to be in the 100's of ohms. So you'd likely want a filter impedance of the order of 1 or 2ohms. This is quite possible though doing it with 'audiophile approved' capacitors would be really hard. I've had good results in my filters using X7R capacitors which do have low enough losses to be practical in such low impedance networks but I've not yet gone down to 1ohm terminating impedance.
If you'd like to see what a low impedance filter might look like in practice I put one idea up on my blog a few years back, but didn't get around to building it - http://www.diyaudio.com/archive/blogs/abraxalito/1141-just-fun-ultra-low-impedance-aif.html
You'll see from the schematic it calls up very high Q inductors with just 3mohm ESRs, I envisaged using ferrite beads but I've since found they're not really very stable to use in filters. Lower Q would probably still work fine in practice, with today's popularity of very high current high frequency switching supplies, suitable Q inductors shouldn't be hard to find. Audiophiles (of both subjectivist and objectivist persuasion) though will not like the presence of so many X7R capacitors
I've not played around with ESS DACs, simply because I don't wish to sign the NDA and then be prevented from sharing what I learn.
With current out DACs having high impedance (current source) outputs the following impedance of the filter isn't too critical, it just needs to be lower than (say) 1% of the output impedance. Which is fairly easy when a CS output DAC has 10's of kiloohm Zout. LC filter tables in Zverev (the bible of passive filter designers) have a column for where the driving impedance is infinite.
With your ESS DACs the Zout I doubt is constant and likely to be in the 100's of ohms. So you'd likely want a filter impedance of the order of 1 or 2ohms. This is quite possible though doing it with 'audiophile approved' capacitors would be really hard. I've had good results in my filters using X7R capacitors which do have low enough losses to be practical in such low impedance networks but I've not yet gone down to 1ohm terminating impedance.
If you'd like to see what a low impedance filter might look like in practice I put one idea up on my blog a few years back, but didn't get around to building it - http://www.diyaudio.com/archive/blogs/abraxalito/1141-just-fun-ultra-low-impedance-aif.html
You'll see from the schematic it calls up very high Q inductors with just 3mohm ESRs, I envisaged using ferrite beads but I've since found they're not really very stable to use in filters. Lower Q would probably still work fine in practice, with today's popularity of very high current high frequency switching supplies, suitable Q inductors shouldn't be hard to find. Audiophiles (of both subjectivist and objectivist persuasion) though will not like the presence of so many X7R capacitors
Maxwell 350F ultracapacitor
Bought those Maxwell 350F ultracapacitors from Mouser,
https://www.mouser.ca/ProductDetail/723-BCAP0350E270T11
ESR only 3.2 mOhms, terrible peak output current up to 170A. Noise floor should be undetectable.
Any possible to challenge my LifePo4 battery as a super power supply?
Maxwell350FultraCapacitor by Ian, on Flickr
Ian
Bought those Maxwell 350F ultracapacitors from Mouser,
https://www.mouser.ca/ProductDetail/723-BCAP0350E270T11
ESR only 3.2 mOhms, terrible peak output current up to 170A. Noise floor should be undetectable.
Any possible to challenge my LifePo4 battery as a super power supply?
Maxwell350FultraCapacitor by Ian, on Flickr
Ian
You are right. Have to use 2 cell in serial for 3.3V or 5V. The ESR will become 6.4mOHMs in this case.
Found this one, 325F/2.7V, ESR is only 1.7 mOHMs for one cell. Becomes 3.4 mOHMs when 2 in serial. From Maxwell / Nesscap. Maybe this one is even better.
BCAP0325 P270 S17 Maxwell / Nesscap | Mouser Canada
Regards,
Ian
Last edited: