How were Chord's current DACs and the M Scaler developed, and how long did each take?

Short version

He links each product to earlier work. In December 2017 he said DAVE was the project he spent the most time on; in 2018 he said the Hugo M Scaler uses the same filter as the Blu Mk2 (CanJam London, July 2018), that Hugo 2 reused DAVE modules (Porta-Fi interview in Singapore, published April 2018) and, in May 2018, that Hugo TT2 took its output stage from his power amplifier work; and in 2022 he called Mojo 2 a roughly four-year project in which "the problem of getting better performance" was power dissipation article, Dec 2017Pursuit Perfect System 5:42article, Apr 2018Head-Fi, May 2018Darko.Audio 1:06Audio Force 3:29. He gave explicit durations for DAVE, the M Scaler and Mojo 2, and none for Hugo 2, Hugo TT2 or the original Mojo. Many figures below come from talks (items beginning RWT-), so check the recordings before relying on an exact number Darko.Audio 50:26.

How he says his projects run

In December 2017 he said he typically goes through "3 to 6 batches of prototypes, particularly for new product categories" article, Dec 2017. In August 2020 he wrote that "projects take a minimum of 18 months to go from concept to start of production" Head-Fi, Aug 2020. In March 2019, writing about TT2, he said "I take as long as it takes to get things right" Head-Fi, Mar 2019. In January 2022 he said it is "normally about two to 3 years to develop a product" Darko.Audio 50:26. In an interview published October 2026 (recording date not stated; its transcript places the recording before May 2026) he said Chord pays him only when products are in production: "I never charge them anything until they have a finished product" HiFi Buys 2:01.

Background: Hugo (2012 to 2014)

Hugo matters here because he ties DAVE and the later products to it. In August 2014 he said "Hugo was kind of an accident", that John Franks had identified that a portable DAC would sell, and that "in Autumn 2012 we went ahead with the project"; "Hugo got all the code that had been intended for the reference DAC" article, Aug 2014article, Aug 2014. In April 2014 he said "Some of the intellectual property (IP) for Hugo started 30 years ago, so one could say that it took 30 years to develop" Head-Fi, Apr 2014, and that he had "spent the last 7years on R and D" on DAC performance Head-Fi, Apr 2014. He dated first listening twice without relating the two in these posts: in June 2017 he wrote "until Oct 2012 when I listened to the first Hugo", and in October 2014 he wrote "Hugo is one year old today", describing a first listening test on the fully tested and de-bugged code on Sunday 20th Oct, which on its face places that test in October 2013 Head-Fi, Jun 2017Head-Fi, Oct 2014.

DAVE

Why and on what hardware. In February 2016 he wrote "Dave primarily was about finding what the factor was that I had done with Hugo, then maximize it" Head-Fi, Feb 2016. In the Hifistatement interview (recorded about January 2016, published July 2016) he said "I've got an FPGA that's 10 times larger" than Hugo's, which gave room for bigger WTA filters and noise shapers Hifistatement 4:16. In July 2016 he said "Dave's first prototype PCB was designed over three years ago", when the FPGA was chosen Head-Fi, Jul 2016. In February 2016 he also said he designed DAVE to "my own targets, not something Chord was setting" Head-Fi, Feb 2016.

How long it took. His figures differ across posts. He said "Dave took over 2 man years" (May 2017), "over 2 and a half years" (December 2017), "2 and a half full time years" (December 2018), "it took 2 to 3 years to develop" (June 2019), and "Dave took 2 and a half years to design" (September 2019) Head-Fi, May 2017article, Dec 2017Head-Fi, Dec 2018Head-Fi, Jun 2019Head-Fi, Sep 2019. In August 2020 he wrote "Dave took more than 3 years", in the context of projects taking at least 18 months from concept to start of production, and in July 2022 "Dave took nearly 3 years to develop" Head-Fi, Aug 2020Head-Fi, Jul 2022. In 2023 talks he said "two and a half, 3 years worth of development" and that DAVE "was two and a half years under development" Dusty Audio 1:03:33Soundnews 52:07. This archive does not show him reconciling these figures.

Milestones he gave. In November 2015 he said "Dave had 4 prototype batches (ten units per batch)" Head-Fi, Nov 2015. In July 2016 he said final production PCBs were made in Spring 2015, tested in the Summer, "when a production BOM was finally released", and "Then it took to December for me to finish the code to my satisfaction" Head-Fi, Jul 2016. In November 2015 he reckoned "I have spent 5,000 listening hours on it (first prototype to production)" Head-Fi, Nov 2015.

Pieces of the work. He described redesigning the 164,000-tap WTA filter because he could not get timing closure, and said "it took nine months to get code to work" (2016) and that it took nine months "to design three different filters" (2018) Hifistatement 4:02MajorHiFi 15:26; in March 2016 he wrote "Dave took 9 months and that's only 100,000", about lines of code Head-Fi, Mar 2016. The noise shaper work took "over a 90-day period" in the 2016 interview, and in a podcast recorded in October 2025 (published March 2026) he put the roughly three months in the summer of 2014, "Well, maybe twenty fifteen" Hifistatement 3:05ecoustics 37:55. "About three quarters of the way through Dave's development" he found it sounded too soft, warm and dark, then found the technical reason Hugo had good timing and designed a 256 FS WTA filter Head-Fi, Jun 2016. On distortion, he said the first prototype measured minus 120 dB and a later change cut it by 20 dB, and in 2024 he wrote that the distortion target was a measurement goal and that "there never was a design goal to beat the sound quality" of the dCS Vivaldi RMAF 13:41Head-Fi, Jan 2024.

After launch. Discussing why DAVE gets no code updates in February 2023, he said "Once that project was finished, I'm done with that" Dusty Audio 1:03:33. In September 2019 he wrote "Dave 2 isn't on my design horizon at all", in August 2020 that design work on Dave 2 was not on his horizon, and in February 2022 "I haven't started work on a replacement to Dave at all" Head-Fi, Sep 2019Head-Fi, Aug 2020Head-Fi, Feb 2022.

Hugo M Scaler (and the Blu Mk2 filter it came from)

The wish. In a January 2020 post about January 2015, when Mojo was being designed, he wrote that "the M scaler hadn't started, it was a wish list product" Head-Fi, Jan 2020. He wrote in December 2017 that "Ten years ago I knew that an M scaler (million tap WTA) would be on the horizon" and that five years earlier he had been finishing 1M-tap filter designs to scope out future FPGA needs Head-Fi, Dec 2017. He said he "envisioned needing 1,000,000 taps in the 1980's" article, Dec 2017, and in November 2019 "it did take me 30 years to create the M scaler" Head-Fi, Nov 2019.

The FPGA and the half-million-tap step. In January 2017 he wrote that half a million taps "became possible in spring of 2016", that going from 26,000 to 164,000 taps in DAVE had left him "disappointed" after nine months of design, and that he expected a "worthwhile" but not transformational gain from 0.5M taps Head-Fi, Jan 2017. In 2017 talks he said the Xilinx part had been announced years earlier but was only buyable about 18 months before, that he used it first for the Davina ADC work and then dropped the code into the Blu upgrade, and that he chose half a million taps because he knew he would get timing closure and have no memory problems Hifistatement 0miceblue425 23:55miceblue425 29:37. He said the half-million design "took A couple of months" in a clip from a Pump House slide talk published by Hifistatement in October 2017 (recording date not stated) and "took about three months" in a February 2018 seminar Hifistatement 3:36MajorHiFi 15:26.

The million-tap step. In his January 2017 slides post he wrote that after "a few months over the summer" he heard the half-million filter, then chose to delay the Blu Mk2, redesign the filter architecture, and "December I finally got it finished" Head-Fi, Jan 2017. In another January 2017 post he wrote "I first heard the 512,000 tap WTA filter in late July", listened to the full M scaler "in late November", and had the final version "last Christmas eve", adding that “it took 3 months to code verify and test the M scaler” when going from 512,000 to the full 1M taps Head-Fi, Jan 2017. In December 2017 he dated the same tests differently: "my own listening tests in Aug 2016 (half M taps) and Dec 2016 (full 1M)" Head-Fi, Dec 2017. In a February 2018 seminar he said he had promised the delay would take only three months and "it did take longer than three months", and that the million-tap filter is half a million lines of code; he thought he had it working on Christmas Eve MajorHiFi 16:55MajorHiFi 22:30. He added "Coding this is half a million lines of code" in another clip of that talk Hifistatement 45. At CanJam RMAF in October 2017 (video published December 2017) he said John Franks did not want him to do a Blu 2, and that in this case "it was six months" of work RMAF 58:11. In November 2016 he apologized for the Blu Mk2 delay: the prototype was working in August, "It cost me 2 or 3 months" to improve it further Head-Fi, Nov 2016.

Hugo M Scaler as a product. At CanJam London in July 2018 (interview published August 2018) he said "the filter that's in the Blu is the exact same filter that's inside this device", on the same FPGA Pursuit Perfect System 5:42. In September 2018 he wrote that "The first M scaler was built over 15 months ago", with four hardware versions and "at least 3 or 4 units per design" Head-Fi, Sep 2018. In a July 2018 post presenting the Hugo M Scaler, he wrote that the M scaler needed a new filter architecture of "528 DSP's operating together, with half a million lines of code", and that it "took some months to test and de-bug" Head-Fi, Jul 2018. In a May 2017 post that does not name which M scaler unit he meant, he wrote that the first prototype had no galvanic isolation, the second did and "that made quite a difference to the sound quality", and the production unit was better again Head-Fi, May 2017. He listed tap counts by product: "Blu MKII (2017) 1,015,808 taps" article, Apr 2018.

Hugo 2

In April 2018 he said "90% of the technologies implemented on Hugo 2 was referenced or adopted from DAVE", because Verilog modules that pass listening and simulation tests go into his library and are reused article, Apr 2018. In June 2017 he wrote that Hugo 2 had an FPGA "about 5 times more powerful" together with the knowledge gained from DAVE Head-Fi, Jun 2017, and called Hugo 1 to Hugo 2 "frankly a complete re-design" Head-Fi, Jun 2017. In March 2017 he said "I went through 4 prototypes batches" Head-Fi, Mar 2017, and in January 2018 that "Hugo 2 took 5 new PCB's" to reach the performance he requires Head-Fi, Jan 2018. In September 2017 he wrote that he first used the prototype on a flight in the previous December Head-Fi, Sep 2017. In February 2017 he said eight weeks of fine tuning had left the FPGA at "100% of the memory, 100% of the DSP cores running at the fastest rate possible, and 98% of the logic", after which he would release production code Head-Fi, Feb 2017. He tied that capacity to the larger battery and case, which let him use "100% of the capacity of the FPGA" Head-Fi, Sep 2017. He gave no total start-to-finish duration for Hugo 2.

Hugo TT2

The original Hugo TT project, he said in February 2015, "started in Feb 2014" and used batteries as the low-risk choice Head-Fi, Feb 2015. For TT2, in May 2018 he said "Initially TT2 had a beefed up Hugo 2 OP stage", that the power amplifier project's more complex output stage was redesigned, and that he then put the new stage into the next TT2 prototype Head-Fi, May 2018. He also wrote that "the TT2 OP stage comes from the power pulse array design" Head-Fi, May 2018. In the same May 2018 post (the archive dates it from the post ID) he said he had been "listening to many TT2 prototypes now since last September" Head-Fi, May 2018. On why TT2 exists, he wrote in January 2020 that "I got fed up with complaints that Hugo 2 wasn't capable of driving some easy to drive headphones - so with TT2 I decided to use a sledgehammer to crack a nut" Head-Fi, Jan 2020. On results, in March 2019 he said "I was expecting TT2 to be a lot closer to Dave than it is" Head-Fi, Mar 2019, and in August 2022 that final listening tests showed DAVE "a lot more transparent than TT2", and that "there is no cost saving problem with TT2" Head-Fi, Aug 2022. He also said the bill of materials ended up "greater than expected" Head-Fi, Mar 2019. He gave no total duration for TT2.

Mojo and Mojo 2

Mojo. At CES in January 2016 he said "this Mojo project started with Hugo and the success of Hugo" and "You could argue it's actually taken 30 years' worth" of development What Hi-Fi? 7:06What Hi-Fi? 2:43. In a January 2020 post he placed its design in January 2015 Head-Fi, Jan 2020, and said "Mojo production was planned a whole year in advance, and that was just from obtaining the parts POV" Head-Fi, Jan 2017. At CanJam NYC (late February 2022, date unconfirmed; published March 2022) he described the power limit that shaped it: on the original Mojo he used about a third of the FPGA "because I was limited on power" MFG Audio Reviews 24:26. He gave no overall duration for the first Mojo.

Mojo 2. In January 2022, asked whether COVID had accelerated it, he said "No, it actually did the opposite because of problems in getting hold of parts", that it "started as a design project in twenty eighteen", and that there were "five different versions of the design" on the PCB side Darko.Audio 1:06Darko.Audio 1:25. He called four years "a little bit untypical" against a normal two to three years Darko.Audio 50:26. His own post that day also said the project started in 2018 "with many prototypes" Head-Fi, Jan 2022. In September 2022 he said he and John Franks talked about Mojo 2 "about four or 5 years ago, so the project started 5 years ago", a count that does not match 2018 on its face The Source AV 33:15. The design brief required Poly compatibility Darko.Audio 2:22. In January 2022 he said he had to save "two hundred milliwatts" elsewhere so the FPGA could run at full capacity; in April 2022 he put the saving at "some 300 or 400 milliwatts", and at CanJam NYC (late February 2022) and in April 2022 he said it used 98% of the FPGA, up from about a third Darko.Audio 38:02Audio Force 4:16MFG Audio Reviews 25:54. The DSP core "took a year or two to put into place" in a September 2022 talk, and in 2023 he mentioned "the three or four months that went involved into DSP core and evaluating it" The Source AV 34:07The Source AV 37:34. In the HiFi Buys interview published October 2026 (recording date not stated) he said that, on a scale with the original Mojo at 0% and Hugo 2 at 100%, his target was to get Mojo 2 "within two-thirds of the sound quality", that "I hit the two-thirds on the first prototype", and that the boards still went through "quite a lot of iterations" for reliability and fit HiFi Buys 2:29:40HiFi Buys 2:31:58.

The M Scaler's successor, Quartet (not one of the listed products)

He said Quartet "started in 2018" and "it's taken 8 years" (interview published March 2026, recording date not stated), that the research "started in 2018, when the M Scaler was finished" (interview recorded in autumn 2026, published October 2026), and that finishing it "took a lot longer than I expected": mathematical tools developed by his son Daniel exposed other issues, "And that took an extra 2 years" (October 2026) Audio Force 9:36article, Oct 2026Distorsiuni Interioare 8:25Distorsiuni Interioare 9:38. In August 2026 he called it “8 years of development, by far the most complicated product I've ever worked on” The Source AV 2:17. In an interview given at a Munich show (most likely May 2024, date unconfirmed; published February 2025) he said "This is a project that I started 6 years ago", in line with his other 2024 counts: "working on for 6 years" at Munich in May 2024, and "since 2018. So 6 years" in September 2024 audio-2G 4:00ConvinceMeAudio 42The Source AV 14:19. In a podcast published October 2026 (recording date not stated) he said of the research stage that "that research took a good 6 years" Part-Time Audiophile 52:36.

What he has not said

He gives no start-to-finish duration for Hugo 2, Hugo TT2 or the first Mojo; the nearest statements are about prototype batches, a prototype used in December 2016, prototypes heard since the previous September, or when Mojo was being designed Head-Fi, Mar 2017Head-Fi, Sep 2017Head-Fi, May 2018Head-Fi, Jan 2020. He did not give one agreed figure for DAVE, and the numbers he gave sit between over two man years and more than three years, with the 2020 figure framed as concept to start of production Head-Fi, May 2017Head-Fi, Dec 2018Head-Fi, Aug 2020. For Mojo 2 his start dates differ by source (2018 against "5 years ago" in September 2022), and this archive does not reconcile them Darko.Audio 1:06The Source AV 33:15. For Quartet, his 2024 statements of 6 years, including an interview published February 2025 but given at a Munich show (most likely May 2024), agree with the 2018 start he gave in September 2024 and March 2026 audio-2G 4:00The Source AV 14:19Audio Force 9:36.

Quotes

“just calculated I have spent 5,000 listening hours on it (first prototype to production)”2015-11-22 Head-Fi, Nov 2015

“projects take a minimum of 18 months to go from concept to start of production; Dave took more than 3 years.”2020-08-22 Head-Fi, Aug 2020

“it did take me 30 years to create the M scaler”2019-11-17 Head-Fi, Nov 2019

“90% of the technologies implemented on Hugo 2 was referenced or adopted from DAVE.”2018-04-04 article, Apr 2018

Dated positions

Caveats and open points

• His DAVE duration figures differ by post (over 2 man years, 2.5 full-time years, 2 to 3 years, nearly 3 years, more than 3 years); this archive does not show him explaining the differences, so no change of view is claimed.
• Mojo 2 start dates differ between sources (2018 against 5 years before September 2022); the items do not reconcile them. For Quartet, the audio-2G interview published February 2025 says "6 years ago", but it was most likely given at a Munich show in May 2024 (date unconfirmed); with that date the 6 years and the 2018 start agree.
• Head-Fi, May 2018 is dated by the archive from its post ID, so 'since last September' in a May 2018 post is placed at September 2017 by inference.
• The first Hugo listening is dated October 2012 in a 2017 post and by a 2014 'one year old today' post; he does not relate the two in those posts.
• No total start-to-finish duration for Hugo 2, Hugo TT2 or the original Mojo appears.
• Quartet is included because he describes it as starting when the Hugo M Scaler was finished.
• Mojo 2 power saving: he said 200 milliwatts in January 2022 and 300 or 400 milliwatts in April 2022; the items do not reconcile them.
• The million-tap step: in January 2017 he wrote it took 3 months to code, verify and test; in a February 2018 seminar he said it took longer than the three months he had promised. Both are reported; the items do not reconcile them.
• Talk dates use the recording or event date where known: CanJam RMAF October 2017 (video published December 2017), CanJam London July 2018 (published August 2018), the ecoustics podcast recorded October 2025 (published March 2026) and the SoundNews interview recorded in autumn 2026 (published October 2026).
• The Hifistatement Blu MkII clips (published October 2017) come from a slide talk at Chord's Pump House; no recording date is given. The HiFi Buys interview (published October 2026) and the Part-Time Audiophile podcast (published October 2026) have no stated recording date; the HiFi Buys transcript places its recording before May 2026.

Topics: Development history of released products