A History of Audio DAC and ADC Chip Performance

Introduction

In this day and age, the humble DAC may be among the most overrated components in audiophile circles – chances are, if you bought a decently-performing one in the last decade and change, it'll sound just as good as any other (well, maybe stay clear of 0 dBFS digital levels by 2-3 dB, not all of them have headroom for intersample-overs). Even onboard audio of the fancier kind (ALC1220) is capable of performance that would have been considered rather high-end years ago these days, assuming the board manufacturer didn't screw up the implementation. A dynamic range well exceeding 110 dB(A) (when there is a volume control following, you can easily get away with 90) and THD+N of something like -90 dB aren't a truly major challenge any more. Only preamp replacement use tends to demand over 120 dB of dynamic range.

Now, that being said, there still are some use cases where you need converters with above average performance. The things have gotten so wildly good that a top-grade DAC will match or outclass even a decent all-analog preamp in performance (minus the ability to connect a bazillion different sources, that is), and if you need a fairly basic but still top-performing unit they're not even that expensive (around US$500). On these, you'll see dynamic range approaching or even exceeding 130 dB A-weighted, 1 kHz harmonics 120-130 dB down or even lower, jitter at irrelevant levels, and often a choice of various reconstruction filters as well (not all of which are all that useful).

Besides, I don't know about you, but I consider nerding out about converter specs fun, and I like history. It turns out there is quite a lot of that, and the recording side of things is at least as interesting if not even more so. This topic even grew so large that I eventually decided it deserved a page of its own instead of being stuck on my audio "blogalike" that no search engine seems to like very much. So let's get started with a look at DAC evolution, shall we?

The Playback Side: DACs

So how did we get here? Well, if you start digging through a bunch of datasheets (Datasheet Archive usually has them) and do some Internet Archive sleuthing to pin down release dates it turns out that a lot of the heavy lifting has already been done over 15 years ago, as the following table of A-weighted dynamic range and THD+N specs of converter ICs shall illustrate (grouped by year and sorted by performance, 2-channel DACs with 1-bit or multilevel delta-sigma architecture with integrated digital filter unless indicated):

Date Manuf.
(Brand)
IC Type DR A-wtd. (mono mode) THD+N 0dBFS Notes
32X Speed DACs (up to 1536 kHz, 32 bit, DSD)
2022-09 AKM AK4499EX + AK4191 135 dB 138 dB -124 dB 7-bit DAC + DS modulator / interface; PCM 1536k + DSD 1024; relaunch post factory fire
16X Speed DACs (up to 768 kHz, 32 bit, possibly DSD)
2023-03 ESS Tech ES9039Q2M
ES9069
130 dB 133 dB -120 dB PCM 768k + DSD 1024; ES9069 w/ MQA dec
2022-05 ESS Tech ES9039PRO
ES9039MPRO
132 dB 140 dB -122 dB 8-channel; PCM 768k + DSD 1024; M w/ HW MQA dec
2022-05 ESS Tech ES9027PRO
ES9027MPRO
124 dB 130 dB -114 dB 8-channel; PCM 768k + DSD 1024; M w/ HW MQA dec
2022-06 ESS Tech ES9017 120 dB 126 dB -110 dB 8-channel; PCM 768k + DSD 512; lo-pwr
2021-04 Rohm BD34301EKV 130 dB -115 dB filters fast / slow rolloff (-192k), off (-768k)
2021-04 ESS Tech ES9033Q 122 dB
(w/ DRE)
115 dB
(w/o DRE)
-108 dB 2Vrms unbal line drivers; built-in PLL
2020-12 ESS Tech ES9080Q 120 dB
(w/ DRE)
112 dB
(w/o DRE)
127 dB
(w/ DRE)
-108 dB
(-112 dB, 4ch diff / mono)
8-channel, 2Vrms unbal line drivers, 4ch SE + 4ch diff + mono mode; built-in PLL + SPDIF out
2020-05 AKM AK4462 117 dB -107 dB non-TOTL
2020 Realtek ALC4080
[CODEC]
120 / 110 dB (port D / E) -90 dB (-3dBFS) USB codec, 2-in 10-out, 32 bit, DSD; ~ALC1220 (HDA)
2019-11 AKM AK4468 117 dB -107 dB non-TOTL; 8-channel; AK4458 update
2019-02 AKM AK4499 134 dB 140 dB x1
137 dB x2
-124 dB 4-channel
2019-01 ESS Tech ES9068Q
[CODEC]
126 dB -120 dB mobile codec w/ basic SE stereo line-in + differential mic-in; ES9068AS first seen late 2020
2018 AKM AK4377 118 dB* -109 dB mobile DAC w/ PGA and headphone driver, 32 bit in, DSD (DoP), max output 1.1 Vrms, Class G analog, *) min onoise -127 dBV; tiny BGA; ≠AK4377A!
2017-12 AKM AK4493 123 dB 128 dB -113 dB
(-110 dB, mono)
large amplitude mode + mono mode similar to PCM1792/94
2017-01 AKM AK4492 127 dB -115 dB mobile DAC
2016-07 AKM AK4497 128 dB
131 dB (2.8V)

133 dB (2.8V)
-116 dB
-113 dB (2.8V)
normal (2 V out) and high gain (2.8 V out) modes
2016-01 ESS Tech ES9038PRO 132 dB 140 dB x1
137 dB x2
-122 dB 8-channel; up to 1536 kHz w/ ext. filter
2016-01 ESS Tech ES9028PRO 129 dB 135 dB x1
133 dB x2
-120 dB 8-channel; up to 1536 kHz w/ ext. filter
2016-01 ESS Tech ES9026PRO 124 dB -110 dB 8-channel, no mono mode; up to 1536 kHz w/ ext. filter
2016 ESS Tech ES9038Q2M 128 dB -120 dB mobile DAC, lower power than 9028Q2M
2014-12 AKM AK4452
AK4454
AK4456
AK4458
115 dB -107 dB non-TOTL; 2/4/6/8-channel; 3rd-order mod at 256fs
2014-05 AKM AK4490 120 dB 123 dB -112 dB
2013-12 AKM AK4495S 120 dB
123 dB (7V)
123 dB
126 dB (7V)
-101 dB
-105 dB (7V)
analog supply VDDL/R up to 7 V rather than the standard 5 V; update: AK4495 (2014)
Octuple Speed DACs (up to 384 kHz, 32 bit, possibly DSD)
2020 ESS Tech ES9281A PRO
ES9280A PRO
[CODEC]
122 dB* -112 dB USB codec w/ PGA + headphone driver, max output = 2 Vrms; *) SNR = 130 dB
2017-10 Cirrus Logic
(Crystal)
CS43198 130 dB (44.1k)
128 dB (192k)
-115 dB (44.1k)
-112 dB (192k)
-105 dB (2V)
standard output = 4.9 Vpp (1.7 Vrms), 5.7 Vpp (2.0 Vrms) optional, tiny BGA, Class H analog, 26 mW idle; DAC-only version of CS43131
2017-10 Cirrus Logic
(Crystal)
CS43131 128 dB (192k) -112 dB (192k)
-105 dB (2V)
mobile DAC w/ headphone driver, standard output = 4.9 Vpp (1.7 Vrms), 5.7 Vpp (2.0 Vrms) optional, tiny BGA, Class H analog, upgraded CS43130
2017 AKM AK4376A 116 dB* -107 dB mobile DAC w/ PGA and headphone driver, 32 bit in, max output 1.1 Vrms, Class G analog, *) min onoise -125 dBV; tiny BGA; update: AK4377A (2018)
2016-12 Cirrus Logic
(Crystal)
CS4399 130 dB (44.1k) -108 dB (44.1k) max output = 4.9 Vpp (1.7 Vrms), tiny BGA, Class H analog, 23 mW idle; DAC-only version of CS43130
2016-12 Cirrus Logic
(Crystal)
CS43130 130 dB (44.1k) -108 dB (44.1k) mobile DAC w/ headphone driver, max output = 4.9 Vpp (1.7 Vrms), tiny BGA, Class H analog
2016-08 ESS Tech ES9218 121 dB -114 dB mobile DAC w/ headphone driver, max output = 2.0 Vrms
2016-02 ESS Tech ES9118 120 dB -112 dB mobile DAC w/ headphone driver, max output = 1.1 Vrms
2015 ESS Tech ES9028Q2M 129 dB -120 dB mobile DAC
2014 ESS Tech ES9018K2M 127 dB -120 dB mobile DAC
2014 ESS Tech ES9016K2M 122 dB -110 dB mobile DAC
2014 ESS Tech ES9010K2M 116 dB -106 dB mobile DAC, single supply
2014 TI (Burr-Brown) PCM5242 114 dB -94 dB (-1dBFS) non-TOTL; 4.2 Vrms ground-referenced diff output; PCM5252 similar + miniDSP
2011 TI (Burr-Brown) PCM5102 112 dB -93 dB (-1dBFS) consumer; 2 Vrms ground-referenced SE output; update: PCM510xA (2012) – 1.8 V digital supply option; PCM512x similar + processing; PCM5142 similar + miniDSP
" " PCM5101 106 dB -92 dB (-1dBFS)
" " PCM5100 100 dB -90 dB (-1dBFS)
Quad Speed DACs (up to 192-216 kHz, min. 24 bit, possibly DSD)
2018 Realtek ALC1200
[CODEC]
110 dB (front) -88 dB (-3dBFS) HDA codec, non-TOTL, 2-in 10-out; meas thread
2017 Realtek ALC1220
[CODEC]
120 / 110 dB (port D / E) -90 dB (-3dBFS) HDA codec, 2-in 10-out, 32 bit; meas thread
2017-11 AKM AK4558
[CODEC]
108 dB -100 dB codec, low power; 32 bit in, SE out
2015 AKM AK4375A 110 dB* -99 dB mobile DAC w/ PGA and headphone driver, 32 bit in, max output ~1 Vrms, Class G analog, *) min onoise -119 dBV; tiny BGA
2013 Realtek ALC1150
[CODEC]
115 / 110 dB (front diff / SE)
96 dB (other)
-88 dB (-3dBFS) HDA codec, 2-in 10-out
2012 AKM AK4413 120 dB 123 dB -102 dB 4-channel, 24 bit; update: AK4414 (2013-05) w/ 32 bit
2012 AKM AK4482 111 dB -100 dB non-TOTL, AK4385/82A/81 pinout
2011-01 AKM AK4621
[CODEC]
115 dB -100 dB filters: sharp, slow, short GD sharp; successor of AK4620B
2011 ESS Tech ES9023 112 dB -94 dB 2Vrms unbal line driver, +3.3-3.6V single supply; replaces similar ES9022 (2009)
2011 Realtek ALC892
[CODEC]
95 dB -84 dB (-3dBFS) HDA codec, 2-in 10-out, non-TOTL, typical midrange onboard
2010-02 AKM AK4480 114 dB 117 dB -100 dB non-TOTL, 32 bit
2009 ESS Tech ES9018 129 dB 135 dB -120 dB 8-channel, 32 bit; spec later updated to indicate 384 kHz support; up to 1536 kHz w/ ext. filter
2009 ESS Tech ES9012 133 dB -120 dB 32 bit
2009 ESS Tech ES9016 124 dB -110 dB 8-channel, non-TOTL, 32 bit; spec later updated to indicate 384 kHz support
2009 Wolfson Micro WM8742 123 dB 126 dB -100 dB non-TOTL, cheaper version of WM8741
2009 AKM AK4392 120 dB -103 dB 32 bit; AK4397 w/ filter update?; ~AK4390 (2009-07, smaller package, no DSD)
2009-05 AKM AK4425
AK4426
106 dB -91 dB 2Vrms unbal line driver, +5V single supply
2009-06 AKM AK4440 105 dB -93 dB 8ch, 2Vrms unbal line driver, +5V single supply
2008 ESS Tech ES9008 128 dB 134 dB -118 dB 8-channel; +3.3V analog
2008-12 AKM AK4399 123 dB 126 dB -105 dB 32 bit, short group delay filter option; +5V supplies
2008 ESS Tech ES9006 120 dB -102 dB 8-channel; +3.3V analog; update: ES9006A (2014)
2008-10 TI (Burr-Brown) PCM1789 113 dB -94 dB ~like PCM1791A + PCM1793
2008 Realtek ALC889
[CODEC]
108 dB -90 dB (-3dBFS) HDA codec, 2-in 10-out
2008 Realtek ALC887
[CODEC]
97 dB -92 dB (-3dBFS) HDA codec, 2-in 8-out, non-TOTL
2007 Wolfson Micro WM8741 125 dB 128 dB -100 dB
2007 AKM AK4397 120 dB -103 dB 32 bit, basically AK4396 with lower distortion
2007 Cirrus Logic
(Crystal)
CS4350 109 dB
(101 dB, SE)
-91 dB diff or SE out, PLL + fs autodetect, suspected IIR filter; replaces CS4391(A)
2006 AKM AK4387
AK4388
106 dB -90 dB consumer, SE outputs; AK4384 ~replacements (87 serial, 88 parallel interface)
2006-12 AKM AK4556
[CODEC]
106 dB -90 dB codec, low power; 24 bit in, SE out
2006 Analog Devices AD1988A
AD1988B
[CODEC]
95 dB
101 dB
-83 dB
-84 dB
HDA codec, 6-in 10-out
2005-07
2005-01
 
AKM AK4620B
AK4620A
[CODEC]
115 dB -97 dB no standalone counterpart, digital filter = AK4394/96; AK4620A short-lived
2005-05 Cirrus Logic
(Crystal)
CS4270
[CODEC]
105 dB* -95 dB SE output; DR generally beats spec; see CS4391 (1999)
2005 Realtek ALC882
[CODEC]
101 dB -90 dB (-3dBFS) HDA codec, 2-in 8-out
2004 Wolfson Micro WM8740 117 dB 120 dB -104 dB
2004-08 AKM AK4396 120 dB -100 dB see AK4394 (1999)
2004-09 AKM AK4589
[CODEC]
114 dB -94 dB 2-in 8-out; so-so filter; SPDIF RX/TX
2004-02 AKM AK4626
AK4628
[CODEC]
106 dB -90 dB consumer 2-in 6/8-out; SE output; so-so filter; AK4588 (2004-06, 2/8) adds SPDIF RX/TX
2004-01 Cirrus Logic
(Crystal)
CS4344
CS4345
CS4346
CS4348
105 dB -90 dB consumer, 16-24 / 16-24 / 24 / 16 bit in, SE out; 10-pin package; CS433x successor
2003-12 TI (Burr-Brown) PCM1792
PCM1794
127 dB 132 dB -108 dB large amplitude mode + mono mode; update: PCM1792A (2004), also DSD1792, DSD1794
2003-12 TI (Burr-Brown) PCM1796
PCM1798
123 dB -106 dB non-TOTL; PCM1798 with PCM1794 pinout; PCM1796 update: PCM1795 (2009) accepting 32-bit data
2003-03 Cirrus Logic
(Crystal)
CS4398 120 dB -107 dB DSD support; mainstream CS43122 replacement
2003 TI (Burr-Brown) PCM4104 118 dB -100 dB 4-channel
2003-03 Cirrus Logic
(Crystal)
CS4272
CS4271
[CODEC]
114 dB -100 dB see CS4392 (2001)
2003-03 TI (Burr-Brown) DSD1791
DSD1793
PCM1791A
PCM1793
113 dB -94 dB non-TOTL; HW/SW cntl, w/ and w/o DSD support; can use ext filter except PCM1793
2003-08 Wolfson Micro WM8776
[CODEC]
108 dB -97 dB consumer codec, SE in/out, vol ctl + mux
2003-04 TI (Burr-Brown) PCM1753
PCM1754
PCM1755
106 dB -94 dB like PCM1742KE but +5V single supply / different filter; various interfaces
2002 Analog Devices AD1955 120 dB 123 dB -110 dB
2002 Wolfson Micro WM8718 111 dB -100 dB ~WM8716 level but uses simpler WM8728 filter
2002-09 AKM AK4381 108 dB -94 dB non-TOTL; compatible AK4382A downgrade; update: AK4385 (2003)
2002-09 AKM AK4384 106 dB -94 dB consumer, SE outputs; ~AK4359 8ch (2004-02)
2002 Cirrus Logic
(Crystal)
CS4340A
CS4341A
101 dB -91 dB consumer; update of 96 kHz CS4340/41, quad speed filter dubious
2001 Wolfson Micro WM8728 106 dB -97 dB released w/ DSD support, dropped?
2001-01 Burr-Brown PCM1738 117 dB -108 dB DSD support; PCM1730 (2001-11) w/o DSD
2001-05 Cirrus Logic
(Crystal)
CS4392 114 dB -100 dB non-TOTL; DSD support; ~CS4271/72 and CS42426/28 DAC part, plus 6/8-ch CS4362(A), CS4382(A), CS4383, CS4365, CS4385(A)
2001 Wolfson Micro WM8716 112 dB -92 dB
(-97 dB, -1dBFS)
PCM1716 compatible, 64-level (6-bit) mod
2001-04
2001-01
AKM AK4383
AK4382A
112 dB -94 dB non-TOTL; AK4383 w/ DSD; ~AK4358 8ch (2003); AK4382 (2000-08) short-lived; AK4381 (2002) compatible w/ worse perf
2001-02 Burr-Brown PCM1608 105 dB -94 dB 8ch; 8-level 4th order delta sigma; PCM1600/01 update
2001-02 AKM AK4355 106 dB -90 dB 6-channel, non-TOTL
2000-12 Cirrus Logic
(Crystal)
CS43122 122 dB -102 dB CS4396 pinout; whitepaper
2000-08 AKM AK4395 120 dB -100 dB see AK4394 (1999)
2000-12 NPC SM5865CM 117 dB -110.5 dB mono 31-level DAC w/ 3rd-order noise shaper for use w/ ext. filter
2000 NPC SM5865BM 114 dB -108 dB mono 23-level DAC w/ 3rd-order noise shaper for use w/ ext. filter
2000 Analog Devices AD1852 114 dB 117 dB -102 dB
(-105 dB mono)
non-TOTL (still very good filters)
2000 Analog Devices AD1854K*
AD1854J*
113 dB
108 dB
-101 dB
-97 dB
non-TOTL; filter stopband very average; double speed: AD1855
2000 NPC SM5866AS 112 dB (PCM)
115 dB (DSD)
-106 dB (PCM)
-109 dB (DSD)
mono 1-bit DSD + multilevel PCM DAC for use w/ ext. filter, SACD application
2000-12 TI (Burr-Brown) PCM1742KE
PCM1742E
106 dB
100 dB
-94 dB
-90 dB
quad speed version of PCM1748
2000-04 Burr-Brown PCM1604
PCM1605
105 dB -95 dB 6ch; 8-level 4th order delta sigma; PCM1600/01 update
1999-11 AKM AK4394 120 dB -100 dB upgrades:
AK4395 (2000-08) – better filters,
AK4396 (2004) – lower Pdiss + DSD but 4394 filters again
1999-09 Cirrus Logic
(Crystal)
CS4396
CS4397
120 dB -100 dB CS4397 w/ DSD + HDCD support
1999 Analog Devices AD1853 116 dB 119 dB -104 dB
(-107 dB mono)
SNR = DR + 1 dB (idle tones!)
1999-09 AKM AK4356 112 dB -94 dB 6-channel; filter ~AK4394?
1999-09 Cirrus Logic
(Crystal)
CS4391 108 dB -94 dB non-TOTL; quad speed filter dubious; minor update: CS4391A (2003); ~CS4270 DAC part, CS4361
1999-11 Burr-Brown PCM1737
PCM1739
106 dB -96 dB PCM1737 w/ programmable settings
Double Speed DACs (up to 96-108 kHz)
2016 Tempo Semi TSCS25x
[CODEC]
102 dB -84 dB codec, mobile; HP amp, compressor etc.; 32 bit in, SE out; data for 1.9 V
2005 Cirrus Logic
(Crystal)
CS42L51
[CODEC]
98/95 dB -86/-88 dB mobile; 24 bit in; capless HP out; data for 2.5/1.8 V, single speed
2000-02 AKM AK4528
[CODEC]
110 dB -94 dB AK4524 upgrade with better ADC
2000 AKM AK4527(B)
[CODEC]
106 dB -90 dB codec, 2-in 6-out; 24 bit in; AK4526 successor; AK4527B 2001-02
2000-12 TI (Burr-Brown) PCM1748KE
PCM1748E
106 dB
100 dB
-94 dB
-90 dB
24 bit in; like PCM1728 but worse digital filter; quad speed: PCM1742
2000 Philips UDA1328T 103 / 106 dB -90 / -95 dB consumer, low-power (3 V), 6ch; 24 bit in; SE out (1 V) / ch1/2 diff out (2 V)
2000-01 Cirrus Logic
(Crystal)
CS4228
[CODEC]
103 dB -90 dB codec, 2-in 6-out, 24 bit in; SE output; update: CS4228A (2001-04, 100 dB, -90 dB)
2000-09 AKM AK4552
[CODEC]
100 dB -88 dB codec, low power (3 V); 24 bit in, SE out
2000-04 AKM AK4380 100 dB -88 dB consumer, SE outputs
2000-09 Cirrus Logic
(Crystal)
CS43L42 93/91 dB (HP)
96/94 dB (line)
-76/-82 dB (HP)
-84/-80 dB (line)
mobile, HP + line-out; data for 3.0/1.8 V, single speed; +CS43L43 w/o line-out
1999-01 AKM AK4524
[CODEC]
110 dB -94 dB 24 bit in
1999-10 Burr-Brown PCM1600
PCM1601
105 dB -95 dB 24 bit in, 6ch; 8-level 4th order delta sigma
1999-07 Burr-Brown PCM1732 104 dB -96 dB 24 bit in; good filter performance; HDCD decoder
1999-01 Cirrus Logic
(Crystal)
CS4340
CS4341
101 dB -91 dB* *) originally -88 dB; consumer, 24 bit in; CS4341 w/ digital volume
1998-11 AKM AK4393 120 dB -100 dB 24 bit in; EOL 2012/13
1998-07 Burr-Brown PCM1704
(-/-J/-K)
110 / 110 / 112 dB
(102 / 106 / 112 dB min)
-92 / -96 / -102 dB
(-90 / -92 / -96 dB max)
1-channel, 24-bit in, up to 16X OS; multibit sign-magnitude converter
1998-06 Burr-Brown PCM1728 106 dB -96 dB 24 bit in; like PCM1716 but some functions removed; still active 2023-08!
1998-04 Cirrus Logic
(Crystal)
CS4334
CS4335
CS4336
CS4337
CS4338
CS4339
96 dB -88 dB consumer; 16-24 / 16-24 / 24 / 20 / 16 / 18 bit in; 8-pin package; CS4334/5/8/9 only EOL'd in mid-2022
1998-01 Burr-Brown PCM1733 95 dB -83 dB consumer, 18 bit in, SE out; 5-level 3rd-order delta sigma; still active 2023-09!
1997-12 Burr-Brown PCM1716 106 dB -96 dB 24 bit in; good filter performance; see WM8716
1997 AKM AK4324 105 dB -94 dB 24 bit in
1997-11 AKM AK4526
[CODEC]
100 dB -90 dB 2-in 6-out; ~AK4321 but 24 bit in
1996-10 AKM AK4321 100 dB -90 dB 20 bit in; ~AK4520A (1997) DAC
1996(?) NPC SM5864AP 98 dB -96 dB mono DAC for use w/ ext. filter at 8fs; 20 bit in; 32fs 4th-order delta-sigma mod
1994(?) Burr-Brown PCM1710U 98 dB -92 dB 20 bit in; 5-level (!) 4th-order delta-sigma
1993 Burr-Brown PCM1702
(-/-J/-K)
106 / 108 / 110 dB
(100 / 102 / 104 dB min)
-92 / -96 / -100 dB
(-88 / -92 / -96 dB max)
1-channel, 20-bit in, 8X OS, 16X OS max; multibit sign-magnitude converter; not adjustable
1993(?) NPC SM5872A 98 dB -94 dB 16 bit in, 8X OS; 32fs (16fs) 4th-order 13-level delta-sigma mod; SM5872B single speed only
1993(?) NPC SM5871A 94 dB -93 dB 16 bit in, 8X OS; 32fs 3rd-order 11-level delta-sigma mod
1992 Burr-Brown PCM67P
PCM69P
(-/-J/-K)
100 / 106 / 106 dB
(94 / 100 / 100 dB min)
-86 / -91 / -95 dB
(-82 / -88 / -92 dB max)
1-channel, 18-bit in, 8X OS, 16X OS max; hybrid multibit R-2R + 1-bit delta-sigma converter
1991 Philips TDA1547 + SAA7350 108 dB* -101 dB 20-bit in, 8X OS, stereo 1-bit converter + 3rd-order modulator, integrated switched-capacitor filter; DR spec is unweighted 20-20k
1990(?) Ultra Analog D20400 112 dB*
(110 dB min)
-102 dB
(-100 dB max)
*) DR unwtd 20-20k; 20-bit in, 8X OS; potted hybrid module multibit converter supposedly using trimmed AD chips
1990(?) Burr-Brown PCM63P
(-/-J/-K)
100 / 104 / 108 dB
(96 / 100 / 104 dB min)
-92 / -96 / -100 dB
(-88 / -92 / -96 dB max)
1-channel, 20-bit in, 8X OS, 16X OS max; multibit sign-magnitude R-2R ("Colinear") converter; data for unadjusted; adjusted DR approaching idle channel SNR of 120 dB(A)
1990? Analog Devices AD1862N
(-/-J)
105 dB
(102 dB min)
-94 / -98 dB
(-92 / -96 dB max)
1-channel, 20-bit in, 16X OS max; multibit R-2R converter; data presumably for unadjusted
1990(?) Burr-Brown PCM1700
(-/-J/-K)
94 / 96 / 98 dB
(88 / 88 / 94 dB min)
-88 / -94 / -98 dB
(-82 / -88 / -92 dB max)
18-bit in, 8X OS, 16X OS max; multibit converter
1990(?) Analog Devices AD1851
AD1861
(-/-J)
101 dB
(88 / 96 dB min)
-88 / -90 dB
(-82 / -88 dB max)
1-channel, 16/18-bit in, 16X OS max; multibit R-2R converter; monotonicity 14/15 bits; data presumably for unadjusted; AD1851 still active 2023-08!
Single Speed DACs (up to 48-54 kHz)
2003-06 Burr-Brown PCM2704
PCM2706
98 dB -84 dB / -78 dB* USB DAC + HPA, 16 bit in, 2706 w/ SPDIF; *) bus-powered
2002-06 Burr-Brown PCM2904
PCM2906
93 dB -86 dB USB codec + HPA, 16 bit in, 2906 w/ SPDIF; active 2023-08
2000-05 Burr-Brown PCM2702 100 dB -94 dB first USB DAC, 16 bit in
1999 Alesis Semi (Wavefront Semi) AL1201 107 dB -90 dB 24 bit in, max 55 kHz; 5th-order 1-bit mod @ 128x OS w/ 2nd-order switched-cap + regular LPF
1999-03 AKM AK4352 94 dB
(96 dB)
-85 dB
(-89 dB)
18 bit in, SE out; low power DAC (Vdd = 1.8-3.6 V, 6 mW @ 2 V), multilevel mod; replaces AK4350; specs for 2V (3V)
1998-09 Cirrus Logic
(Crystal)
CS4223
CS4224
[CODEC]
105 dB -97 dB codec, 24 bit in
1998-12 Philips UDA1341TS
[CODEC]
100 dB -91 dB codec, low-power (3V); 20 bit in, SE out; PGA + AGC, DSP
1997-09 Cirrus Logic
(Crystal)
CS4390 106 dB -98 dB 24 bit in; CS4329 pinout
1997-05 Philips UDA1340
[CODEC]
100 dB -85 dB codec, low-power (3V); 20 bit in, SE out; update: UDA1344TS
1997-07 Cirrus Logic
(Crystal)
CS4222
[CODEC]
99 dB -88 dB codec, 20 bit in
1997 Analog Devices AD1857
AD1858
94 dB -90 dB consumer, 20 bit / 16 bit in; multibit mod w/ switched cap filter; SE voltage out
1997-11 AKM AK4350 92 dB -85 dB 18 bit in, SE out; low power (Va = Vd = 2.0 V) DAC, multilevel mod; update: AK4352
1996-03 AKM AK4323 100 dB -90 dB 20 bit in; integrated PLL clockgen; similar to AK4321 (1996) and AK4520A (1997) DAC but single speed only
1996-09 Cirrus Logic
(Crystal)
CS4226-KQ
CS4226-BQ
[CODEC]
98 dB
96 dB
 
-88 dB
-86 dB
 
codec, 2+1-in 6-out, 20 bit in; SPDIF receiver, also available w/o as CS4227 (1998)
1996(?) Cirrus Logic
(Crystal)
CS4330
CS4331
CS4333
94 dB -86 dB consumer, 18 bit in, SE out; 8-pin package; 128fs 3rd-order delta sigma w/ switched capacitor filter; single +5 V or (-K grade only) +3V supply
1996 AKM AK4314 92 dB -84 dB consumer, 16 bit in, low-power (3 V), SE out; digital bass boost + volume; 64fs 1-bit mod w/ switched cap + regular LP filter
1996 AKM AK4315 91 dB -84 dB like AK4314, but master clock 256fs/384fs instead of 384fs only
1995-09 Cirrus Logic
(Crystal)
CS4329 106 dB* -97 dB *) by 1998 – original spec 105 dB; 20 bit in; upgrade: CS4390
1995-09 Cirrus Logic
(Crystal)
CS4327 100 dB* -93 dB* *) by 1997, original spec 101 / -90 dB; consumer, 20 bit in, SE outputs; 128fs delta-sigma
1995 AKM AK4320 100 dB -84 dB 20 bit in
1995 AKM AK4319 96 dB -90 dB 18 bit in; 4th-order 1-bit mod w/ switched capacitor filter; less out-of-band noise than AK4318; update: AK4319A (1998, 92 dB, -87 dB)
1994(?) Burr-Brown PCM1715U 98 dB -92 dB 16 bit in; 5-level (!) 4th-order delta-sigma; same silicon as PCM1710U?
1994(?) Burr-Brown PCM1712U 94 dB -87 dB 16 bit in; 5-level (!) 3rd-order delta-sigma
1994(?) AKM AK4313 92 dB -88 dB 18 bit in, low-power (3 V); 64fs 4th-order 1-bit mod w/ switched capacitor filter
1993 Cirrus Logic
(Crystal)
CS4328-K
CS4328-B
97 dB
95 dB
-93 dB
-88 dB
18 bit in; first with all-switched-capacitor analog filtering; see The Audio Critic, Issue 21; also see patent US5220483A
1993(?) AKM AK4318 97 dB -92 dB 18 bit in; 64fs 4th-order 1-bit mod w/ switched capacitor filter; update: AK4319 (1995, less out-of-band noise)
1992? Cirrus Logic
(Crystal)
CS4303 107 dB -100 dB 18 bit in; 5th-order 1-bit mod; = AK4303
1992(?) AKM AK4316 92 dB -80 dB 16 bit in; 64fs 4th-order 1-bit mod w/ switched capacitor filter
1991(?) Analog Devices AD1866 94 dB
(90 dB?)
-86 dB
(-80 dB max)
2-channel, 16-bit in, 8X OS max; multibit R-2R converter, no adj; still active 2023-08!
1990(?) Sony CXD2552Q + CXD1244 >96 dB <0.003% THD stereo 1-bit converter (3rd-order) + 4/8X OS digital filter for CD player (CXD1125Q etc.)
1989 Burr-Brown PCM61P
(-/-J/-K)
94 / 96 / 100 dB
(88 / 94 / 94 dB min)
-88 / -94 / -98 dB
(-82 / -88 / -92 dB max)
1-channel, 18-bit in, 4X OS, 8X OS max; multibit R-2R converter; data for unadjusted
1988 Burr-Brown PCM58P
(-/-J/-K)
100 / 100 / 102 dB
(94 / 94 / 100 dB min)
-94 / -96 / -100 dB
(-92 / -94 / -96 dB max)
1-channel, 18-bit in, 4X OS; multibit R-2R converter; data for unadjusted
1988(?) Analog Devices AD1856
(-/-J/-K)
95 dB
(88 / 88 / 94 dB min)
-94 dB THD
(-82 / -86 / -92 dB max)
1-channel, 16-bit in, 8X OS max; multibit R-2R converter; monotonicity 15 bits; data presumably for unadjusted; still active 2023-08!
1987? Philips TDA1541A
( - / R1 / S1 / S2)
102 / 103 / 107 / 107 dB -95 / -95 / -95 / -97 dB 1-channel, 16-bit in, 4X OS, max 8X OS; multibit R-2R converter, selected for differential nonlin.
1986? Burr-Brown PCM56P
(-L/-/J/K)
95 dB
(80 / 88 / 88 / 94 dB min)
-94 dB THD
(-80 / -82 / -88 / -92 dB max)
1-channel, 16-bit in, 4X OS; multibit R-2R converter
1985? Burr-Brown PCM54*P
PCM55*P
(H/J/K)
94 / 94 / 100 dB
(88 / 88 / 94 dB min)
-94 dB THD
(-82 / -88 / -92 dB max)
1-channel, 16-bit in, multibit R-2R converter; PCM54 ±12V, PCM55 ±5V
1984? Sony CX20152 94 dB -90 dB 1-channel, 16-bit in, 2X OS (L/R alternating); multibit converter
1983? Burr-Brown PCM53*P (J/K) 94 dB (88 / 94 dB min) -94 dB THD
(-88 / -92 dB max)
1-channel, 16-bit in, 4X OS; multibit R-2R converter; backstory and references

There is obviously more to DAC performance than just these numbers, notably digital filter performance (many more recent DACs are not exactly pretty in terms of periodic ripple, and lots of DACs capable of 384 kHz and up essentially "cheat" by bypassing most of their digital filtering beyond 192 kHz), handling of clock jitter (some like ESS essentially have built-in ASRC), 0dBFS+ handling (anything with built-in ASRC like ESS or TI PCM179x tends to hard-clip at 0 dBFS, more traditional parts generally have about 2 dB and change of headroom), power consumption and features, but as an illustration of progress I figured this table should do.

You can see how in just 6 years, delta-sigma converter IC dynamic range increased by 20 dB, and 30 dB over a decade, absolutely massive improvements to the point where analog noise became a dominating factor. In the same decade, distortion went down by 15-17 dB as well, which if taken together with the increased demands brought about by reduced noise and inevitably limited maximum power draw must mean substantially more complex analog circuitry as well.

If you were an equipment manufacturer in the late '90s, on the one hand you'd be lucky in that you could probably present better performance figures than devices just a few years old, but on the other hand the performance of your brand spanking new model would soon be considered obsolete as well; if you're lucky, you would be able to adopt a new better replacement part. Things have not been nearly as turbulent again since the mid-2000s, although with the advent of DAC + headphone amp ICs like the CS43130/131 (DAC-only cousin: CS43198) in recent years, near-130 dB dynamic range has never been more pocket-sized or affordable.

You may have noticed that Crystal Semiconductor did not have a TOTL double speed DAC, while AKM did. We'll find sort of the reverse in the ADC world, with AKM coming out with their last TOTL single-speed ADC a few weeks after Crystal brought out their first double-speed model.

AKM in recent years seems to have been going with a "throw shit at the wall and see what sticks" approach, somewhat reminiscent of Burr-Brown in the early '90s. Clearly, there isn't much left for DAC performance to go at this point – dynamic range is pretty much maxed out, just distortion could still go down some more.

Note: I'm mostly listing delta-sigma DACs with built-in digital filters here, which in the beginning weren't nearly as high-end. If you wanted best dynamic range in the '90s, you'd generally come up with some contraption involving traditional multibit DACs (I have added some now) or even hybrids using a separate digital filter as it was common in the olden days. The Pacific Microsonics Model One ADC + DAC was reputed to be able to pull off <-120 dBFS of noise in loopback (D/A --> A/D), way back in the mid-late '90s. Mind you, these units sold at cost (!) for around 15 grand or something, very very advanced stuff for the day.

One thing that is very much different in traditional multibit vs. delta-sigma DACs is that dynamic range is limited by low-level nonlinearity rather than noise. (I am listing DR as THD+N at -60 dB relative to full-scale, i.e. -42 dB becomes -102 dB etc.) Multibit DAC ICs were therefore commonly selected and/or provided with external trimming options for at least the most significant bit, sometimes even more. The issue potentially comes back with multilevel delta-sigma designs but those are generally using several converters of a few bits that are being picked at random, which randomizes their quantization noise as well which can then be noise-shaped out of the audible frequency range.

As you can see, by the early '90s multibit DACs still had the upper hand, however there was little in terms of higher performance to be had by 1993. Even the PCM1704, often considered the best multibit audio DAC ever, only had a few dB left to gain over its predecessor while the much cheaper delta-sigma competition was inching closer and closer. (Multibit DACs like these ran at 10-15 bucks a pop, and you needed at least two plus a digital filter / processor IC. High-performance implementations often employed multiple DACs per channel, driving cost up even more.) Once multilevel delta-sigma converters like CS4396 and AK4393 were out and leaping ahead in dynamic range by 8+ dB, the race was pretty much over.

Discrete multibit designs can do better: A Denafrips ARES II gets to DR = 120 dB(A) and around -100 dB of THD+N, the latter rather in line with good multibit DACs from the olden days (matching between channels is also good, and going by SMPTE over level I suspect this may have a sign-magnitude architecture similar to PCM1702/04). About 18 effective bits available. Even back in that day, it appears some trimmed PCM63s could do very well if Krell KPS-20i and Krell KPS-20iL measurements are any indication – the former appears to be truncating internally(!) at 19 bits (THD+N -113ish dB?), the latter seems to make it pretty much all the way to 20 bits, with a THD+N that could be as low as -118 dB. (The PCM1702 ditching external adjustments was not necessarily a good thing, as evidenced by measurements of two Linn Numerik versions equipped with either type, and yes, -K grade in both cases.)
About the best I've seen has to be the Holo Audio May with DR = 130 dB(A) and THD+N at -118 dB, though worst-case gremlins in the SMPTE test came up to about -105 dBFS. Good modern delta-sigma designs still have that distortion beat by about 10 dB, even if that's just measurable rather than audible at this point.
One thing these DACs are extremely good for is out-of-band noise, as you might expect from a multibit. Makes you wonder what a discrete multilevel sigma-delta could do, with perhaps a couple of 6-8 bit DACs randomized and noise shaped to oblivion (you can't run discrete DACs as fast as integrated ones, so you definitely have to start out with several bits more). I wouldn't be surprised if that had already been done, actually.

Here is a neat presentation from back in the mid-'90s comparing different Burr-Brown DAC types at a time when multibit still reigned supreme in high-performance applications.

Go back to table of contents

The Recording Side: ADCs

Recording applications have to deal with potentially quite large amounts of dynamic range, and the music industry was not exactly short of cash in the '80s and '90s, fueling the development of high-performance delta-sigma ADCs which had supplanted multibit ones by the early '90s already. You'll see that in the following table (grouped by year and sorted by performance, differential analog inputs unless indicated):

Date Manuf.
(Brand)
IC Type DR A-wtd. (mono mode) VA / V THD+N -1dBFS Filter Ripple, ± Filter Stopband Notes
16X Speed ADCs (up to 768 kHz, 32 bit, possibly DSD)
2023-06 Cirrus Logic
(Crystal)
CS5308P 123 dB ? 3.3 -110 dB ? ? 8ch, hybrid gain ctl
" Cirrus Logic
(Crystal)
CS5304P 123 dB ? 3.3 -110 dB ? ? 4ch, hybrid gain ctl
" Cirrus Logic
(Crystal)
CS5302P 126 dB ? 3.3 -110 dB ? ? hybrid gain ctl
2023-04 ESS Tech ES9826 115 dB (123 dB, ARE) 3.3 -105 dB 8 selectable filters, PGA, opt. DRE (ARE), mic bias, PLL
2022-09 ESS Tech ES9821 120 dB 3.3 -112 dB 8 selectable filters, opt. PLL, smaller package
2020-11 ESS Tech ES9842PRO 122 dB 128 dB 4.5 -116 dB (-118 dB, mono) 4ch, 8 selectable filters + custom
" ESS Tech ES9822PRO 124 dB 127 dB 4.5 -117 dB (-118 dB, mono) 8 selectable filters + custom
2020-11 ESS Tech ES9840 116 dB 118 dB 3.3 -108 dB (-110 dB, mono) non-TOTL, 4ch, 8 selectable filters + custom, ASP
" ESS Tech ES9820 116 dB 117 dB 3.3 -108 dB (-109 dB, mono) non-TOTL, 8 selectable filters + custom, ASP
2020-12 TI (Burr-Brown) TLV320ADC6120 113 dB
(123 dB, DRE)
116 dB 3.3 -98 dB 0.05 dB
0.015 dB (low GD)
-73.8 dB
-86.4 dB (low GD)
fancier PCM1820, =PCM6120 (2022)
" " TLV320ADC5120 108 dB
(120 dB, DRE)
111 dB 3.3 -95 dB " " =PCM5120 (2022)
" " TLV320ADC3120 106 dB 109 dB 3.3 -95 dB " " fancier PCM1821, =PCM3120 (2022)
2016-03 AKM AK5558 115 dB 124 dB 5 -106 dB 0.03 dB (48-96k)
0.02 dB (192k)
0 dB (384k, 768k)
-85 dB non-TOTL, like AK557x, Pdiss ~half
" " AK5556 " 122 dB
" " AK5554 " 121 dB
" " AK5552 " 118 dB
2016-03 AKM AK5538 111 dB 120 dB 3.3 -103 dB 0.03 dB (48-96k)
0.02 dB (192k)
0 dB (384k, 768k)
-85 dB 3.3 V like AK555x
" " AK5536 " 118 dB
" " AK5534 " 117 dB
2015-12 AKM AK5578 121 dB 130 dB 5 -112 dB 0.03 dB (48-96k)
0.02 dB (192k)
0 dB (384k, 768k)
-85 dB 8/6/4/2-channel, selectable filters, short delay filters are IIR
" " AK5576 " 128 dB
" " AK5574 " 127 dB
" " AK5572 " 124 dB
2014-11 AKM AK5397EQ 127 dB 130 dB 5 -108 dB 0.00013 dB -100 dB selectable filters
Quad Speed ADCs (up to 192-216 kHz, 24 bit, possibly DSD)
2020-12 TI (Burr-Brown) PCM1820 113 dB
(123 dB, DRE)
3.3 -98 dB 0.05 dB
0.015 dB (low GD)
-73.8 dB
-86.4 dB (low GD)
2ch like PCM1840
" " PCM1822 111 dB
(117 dB, DRE)
3.3 " " " like PCM1820 but 1 Vrms input
2020 Realtek ALC4080
[CODEC]
110 dB (main)
92 dB (sub)
5 -87 dB (-3dBFS) -0.405/+0.015 dB* -90 dB USB codec, 2-in 10-out; ~ALC1220 (HDA); *) digital + analog filter, 48k
2019 TI (Burr-Brown) PCM1840 113 dB
(123 dB, DRE)
3.3 -98 dB 0.05 dB
0.015 dB (low GD)
-73.8 dB
-86.4 dB (low GD)
4ch, 32 bit, low group delay mode (better than FIR?!), DRE (digitally controlled PGA, max 48 kHz), mic bias
2019 AKM AK5704 105 dB 3.3 or 1.8 -90 dB 0.16 dB -72 dB low power, LN PGA + mic bias, mixer, filters, ALC, activity det
2018 Realtek ALC1200
[CODEC]
102 dB* 5 -80 dB (-3dBFS)* 0.03 dB -80 dB HDA codec, non-TOTL, 2-in 10-out; DR / THD better than spec IRL; meas thread
2017-11 AKM AK4558
[CODEC]
108 dB 3.3 -92 dB 0.03 dB -85 dB codec, low power; SE inputs, 32 bit out
2017 Realtek ALC1220
[CODEC]
110 dB (main)* 5 -85 dB (-3dBFS)* 0.03 dB* -80 dB HDA codec, 2-in 10-out; used on Asus Xonar SE; DR / THD / ripple (much) better than spec IRL; meas thread
2014-03 TI (Burr-Brown) PCM1861
PCM1863
PCM1865
110 dB (106 dB, SE) 3.3 -93 dB 0.05 dB
0.02 dB (IIR)
-65 dB
-75 dB (IIR)
2/2/4ch ADC, PGA, diff/SE input, IIR filter better than FIR?
" " PCM1860
PCM1862
PCM1864
103 dB 3.3 " " "
2013 Analog Devices ADAU1979 109 dB
(103 dB min)
3.3 -95 dB
(-87 dB max)
0.015 dB -79 dB 4ch, input 4.5 Vrms diff, PLL
2013 Analog Devices ADAU1978 109 dB
(103 dB min)
3.3 -95 dB
(-88 dB max)
0.015 dB -79 dB 4ch, input 2 Vrms diff, PLL
2013 Analog Devices ADAU1977 109 dB
(103 dB min)
3.3 -95 dB
(-89 dB max)
0.015 dB -79 dB 4ch, input 10 Vrms diff DC-coupled, PLL, mic bias 5-9 V, diagnostics; weird thing
2013 Realtek ALC1150
[CODEC]
104 / 93 dB 5 -80 dB (-3dBFS) 0.03 dB -80 dB HDA codec, 2-in 10-out
2012 Realtek ALC898
[CODEC]
104 dB 5 -86 dB (-3dBFS) 0.03 dB -80 dB HDA codec, 2-in 10-out; used on Creative Audigy FX
2011-01 AKM AK4621
[CODEC]
115 dB 5 -102 dB 0.005 dB
(0.01 dB, short GD)
-100 dB
(-80 dB, short GD)
successor of AK4620B; short group delay mode (14/fs vs. 39/fs, part IIR?)
2011 Realtek ALC892
[CODEC]
90 dB 5 -84 dB / -85 dB (-3dBFS) 0.03 dB -80 dB HDA codec, 2-in 8-out, non-TOTL, typical midrange onboard
2009 AKM AK5388 120 dB 123 dB 5 -110 dB 0.01 dB -80 dB 4ch; update: AK5388A (2014)
2008 Realtek ALC889
[CODEC]
104 dB 5 -90 dB / -85 dB (-3dBFS) 0.03 dB* -80 dB HDA codec, 2-in 10-out, actual filter ripple way lower than spec
2008 Realtek ALC887
[CODEC]
90 dB 5 -85 dB (-3dBFS) 0.0005 dB -90 dB HDA codec, 2-in 8-out, non-TOTL, typical budget onboard
2007-01 AKM AK5386 110 dB 5 -96 dB 0.005 dB -80 dB SE inputs; upgrade for AK5380/81
2007 Analog Devices AD1974 105 dB
(100 dB min)
3.3 -96 dB
(-87 dB max)
0.015 dB -79 dB 4ch, PLL
2006 TI (Burr-Brown) PCM4220 123 dB 4.0 -108 dB 0.00015 dB
0.001 dB (low GD)
-100 dB
-90 dB (low GD)
low group delay mode, PCM4222 ~same w/ extra features
2006 Cirrus Logic
(Crystal)
CS5368
CS5366
CS5364
114 dB 5 -105 dB 0.017 to 0.02 dB -92 to -97 dB 8/6/4-ch versions of CS5361 (2003); TDM interface; first multich ADC?
2006-12 AKM AK4556
[CODEC]
103 dB 3.0 -91 dB 0.04 dB -68 dB codec, low power; SE inputs, 24 bit out
2006 Analog Devices AD1988A
AD1988B
[CODEC]
90 dB
92 dB
3.3 -81 dB
-82 dB
0.005 dB -100 dB HDA codec, 6-in 10-out
2005-07
2005-01
AKM AK4620B
AK4620A
[CODEC]
113 dB
(110 dB, SE)
5 -100 dB
(-90 dB, SE)
0.005 dB -100 dB ~AK5385B (2005); AK4620A short-lived
2005-05 Cirrus Logic
(Crystal)
CS4270
[CODEC]
105 / 102 dB 5 or 3.3 -95 / -92 dB 0.035 dB -70 dB SE inputs; short group delay filter, IIR?, quad speed filter dubious; ~CS5341 (2003), CS5345 + CS5346 (+PGA, Mux, Mic); data for 5/3.3 V
2005-08 AKM AK5359 102 dB 5 -94 dB 0.04 dB
(0.02 dB, 192k)
-68 dB
(-70 dB, 192k)
consumer, SE inputs; 96 kHz companion: AK5358 (2005-11); update of previous AK5357 (2004)
2005 Realtek ALC882
[CODEC]
90 dB 5 or 3.3 -82 dB (-3dBFS) 0.2 dB -76 dB HDA codec, 2-in 8-out; data for 5 V
2004-11 Wolfson Micro WM8785
WM8786
111 dB 5 -102 dB (-0.1dBFS) 0.005 dB -85 dB
2003 TI (Burr-Brown) PCM4202 118 dB
(115 dB, DSD)
5 -105 dB
(-103 dB, DSD)
0.005 dB -100 dB DSD support; 4ch: PCM4204 (2004)
2003-10 AKM AK5385A 114 dB 5 -103 dB 0.005 dB -100 dB non-TOTL; update: AK5385B (2005, EOL early 2017); ~AK4620A/B (2005) ADC
2003-03 Cirrus Logic
(Crystal)
CS4272
[CODEC]
114 dB 5 -100 dB 0.017 to 0.02 dB -92 to -97 dB short group delay filter, IIR?; ~CS5361 (2002), CS5364, CS5366, CS5368 (4/6/8ch, 2006); matching paper?
" " CS4271
[CODEC]
108 dB " -98 dB " " as above but SE input only; ~CS5351 (2002)
2003 Cirrus Logic
(Crystal)
CS5341
CS5342
105 / 102 dB 5 or 3.3 -98 / -95 dB 0.035 dB -70 dB SE input, short group delay filter, IIR?, quad speed filter dubious; ~CS4270 (2005) ADC part, CS5345 + CS5346 (+PGA, Mux, Mic); data for 5/3.3 V
2003 Cirrus Logic
(Crystal)
CS5340 101 / 98 dB 5 or 3.3 -94 / -91 dB 0.035 dB -70 dB low-cost version of CS5341
2002-01 AKM AK5394A 123 dB 5 -110 dB 0.001 dB -120 dB SOTA for over a decade, EOL ~late 2018; replaced short-lived AK5394
2002-12 Cirrus Logic
(Crystal)
CS5381 120 dB 5 -110 dB 0.017 to 0.02 dB -92 to -97 dB short group delay filter, IIR?
2002 Cirrus Logic
(Crystal)
CS5361 114 dB 5 -105 dB 0.017 to 0.02 dB -92 to -97 dB non-TOTL, short group delay filter, IIR?; 5th-order multibit mod; ~CS4272 (2003) ADC part, CS5364, CS5366, CS5368 (4/6/8ch, 2006); matching paper?
2002 Cirrus Logic
(Crystal)
CS5351 108 dB 5 -98 dB 0.017 to 0.02 dB -92 to -97 dB SE input, otherwise like CS5361; ~CS4271 (2003) ADC part
2001-11 TI (Burr-Brown) PCM1804 112 dB 5 -102 dB 0.005 dB -100 dB
2001-01 AKM AK5394 123 dB 5 -105 dB 0.001 dB -120 dB quickly superseded by AK5394A (2002)
Double Speed ADCs (up to 96-108 kHz, 24 bit)
2016 Tempo Semi TSCS25x
[CODEC]
90 dB 1.9 -80 dB ? ? codec, mobile; mux, PGA mic bias; SE/diff inputs, 32 bit out (17 bits of noise…)
2014 AKM AK5720 102 / 98 dB 5 to 3.0 -94 dB 0.16 dB -71/-72 dB low power, SE input, preamp +15 dB; data for 5/3 V, 0 dB
2011-03 TI (Burr-Brown) PCM1808 99 dB 5 -93 dB 0.05 dB -65 dB non-TOTL, SE input
2008-11 TI TLV320ADC3001
ADC3101
ADC3100
92 dB 3.3 to 3.0 -90 dB (THD) 0.1 dB -73 dB low power, PGA, LN mic bias, miniDSP; 3101 w/ dig mic in, 3100 w/ 4 inputs instead of 6
2005-09 TI (Burr-Brown) PCM1807 99 dB 5 -93 dB 0.05 dB -65 dB non-TOTL, SE input
2005-06 TI (Burr-Brown) PCM1803A 103 dB 5 -95 dB 0.05 dB -65 dB non-TOTL, SE input; PCM1803 update
2005-11 AKM AK5358 102 dB 5 -92 dB 0.04 dB -68 dB consumer, SE input, ~AK4589 / AK4588 / AK4628 ADC; 96k companion to AK5359 (2005)
2005 Cirrus Logic
(Crystal)
CS53L21 98 / 95 dB 2.5 or 1.8 -88 / -86 dB -0.09/+0.17 dB -33 dB low power, LN mic bias + preamp, PGA + Mux; data at 2.5 / 1.8 V; see CS42L51
2005 Cirrus Logic
(Crystal)
CS42L51
[CODEC]
98 / 95 dB 2.5 or 1.8 -88 / -86 dB -0.09/+0.17 dB -33 dB codec, like CS53L21
2004-02 AKM AK4626
AK4628
[CODEC]
102 dB 5 -92 dB 0.04 dB -68 dB consumer codec, 2-in 6/8-out, SE input, ~AK5358; AK4588/89 (2004-06/09) adds SPDIF RX/TX
2004-11 TI (Burr-Brown) PCM1803 103 dB 5 -95 dB 0.05 dB -65 dB non-TOTL, SE input; update: PCM1803A (2005)
2004 AKM AK5357 102 / 100 dB 5 to 3.0 -88 / -86 dB 0.04 dB -68 dB consumer, SE input; data for 5/3 V
2003-08 Wolfson Micro WM8775 102 dB 5 to 3.0 -90 dB (0dBFS)
-95 dB (-3dBFS)
0.01 dB -65 dB consumer, SE input, PGA + mux; ~WM8770 + WM8776 ADC; data for 5 V
2003-05 AKM AK5384 107 dB 5 -100 dB 0.005 dB -80 dB 4ch
2002 Analog Devices AD1871 105 dB
(103 dB unwtd.)
5 -85 dB
(-0.5dBFS)
0.01 dB -120 dB diff/SE input, PGA + mux, modulator out
2001-04 AKM AK5380 106 dB 5 -96 dB 0.005 dB -80 dB non-TOTL, SE inputs; update: AK5381 (2002), compatible but lower perf: AK5357 (2004), AK5358 (2005)
2001-11 TI (Burr-Brown) PCM1802 105 dB 5 -96 dB 0.05 dB -65 dB non-TOTL, SE input
2001 Philips UDA1361TS 100 dB 3 -88 dB 0.01 dB -70 dB low-power, SE inputs
2001-05 Cirrus Logic
(Crystal)
CS53L32A 96 / 93 dB
(98 / 94 dB, 96 kHz)
3.3 to 1.8 -88 dB
(-85/-88 dB, 96 kHz)
-0.09/+0.17 dB
(-0.09/+0 dB, 96 kHz)
-60 dB
(-48 dB, 96 kHz)
low power, SE inputs, PGA + mux; data for 3.0/1.8 V
2001-04 Cirrus Logic
(Crystal)
CS5333 96 / 91 dB
(98 / 94 dB, 96 kHz)
3.3 to 1.8 -88 dB
(-85/-88 dB, 96 kHz)
-0.09/+0.17 dB
(-0.09/+0 dB, 96 kHz)
-60 dB
(-48 dB, 96 kHz)
low power, SE inputs; data for 3.0/1.8 V
2000-02 AKM AK4528
[CODEC]
108 dB 5 -94 dB 0.005 dB -80 dB AK4524 upgrade w/ diff input
2000 AKM AK4527(B)
[CODEC]
102 dB 5 -92 dB 0.005 dB -80 dB codec; AK4526 sucessor; AK4527B 2001-02
2000-01 Cirrus Logic
(Crystal)
CS4228
[CODEC]
102 dB 5 -90 dB 0.01 dB* -80 dB* *) single speed mode; codec, 2-in 6-out; update: CS4228A (2001-04; 97 dB, -95 dB)
2000-09 AKM AK4552
[CODEC]
97 dB
(100 dB, 96 kHz)
3.0 (4.0 to 2.4) -89 dB 0.1 dB -65 dB codec, low power; SE inputs
1999-01 AKM AK5393 117 dB 5 -105 dB 0.001 dB -110 dB 4th-order cascaded multilevel mod; Paper by Fujimori; EOL ~late 2018
1999-01 AKM AK5383 110 dB 5 -103 dB 0.001 dB -110 dB non-TOTL, AK5393 compatible; EOL 2012/13
1999-01 AKM AK4524
[CODEC]
100 dB 5 -90 dB 0.005 dB -80 dB SE input
1999-06 AKM AK5353 96 / 92 dB 5 to 3.0 -84 dB ? -65 dB consumer, SE inputs; data for 5/3 V
1997-07 Cirrus Logic
(Crystal)
CS5396
CS5397
120 dB 5 -105 dB 0.005 dB
(0.008 dB, low GD)
-117 dB
(-86 dB, low GD)
CS5396 for audio, CS5397 w/ non-alias filter for meas., early multilevel (3) w/ 7th-order mod, opt. shaped dither for 20/18/16 bit; papers #1, #2, #3
1997-01 AKM AK5352 104 dB 5 -97 dB 0.005 dB -80 dB non-TOTL, 20 bit out, early multilevel mod; 96 kHz companion to AK5351
1996(?) AKM AK5340A 96 dB 5 -92 dB 0.01 dB -88 dB 18 bit out; 64fs 5th-order delta sigma mod; double speed version of AK5340
Single Speed ADCs (up to 48-54 kHz)
2013-07 Cirrus Logic
(Crystal)
CS53L30 91 dB 1.8 -84 dB 0.14 dB -70 dB low power, 4ch, 24 bit out; LN mic bias + preamp, SRC
2010-08 AKM AK5373
AK5374
91 dB 3.3 -85 dB -0.07/+0.16 dB -68 dB USB, diff in, 24 bit out; mic preamp + PGA
2008-08 TI (Burr-Brown) PCM1870A 90 dB 3.3 -81 dB 0.02 dB -60 dB low power, 16 bit out, mic bias + preamp
2007-03 Burr-Brown PCM2900
PCM2902
89 dB 5 -80 dB 0.05 dB -65 dB USB codec + HPA, 16 bit out, 2902 w/ SPDIF; rev. C (2011) active 2023-08
2002-06 Burr-Brown PCM2904
PCM2906
89 dB 5 -80 dB 0.05 dB -65 dB USB codec + HPA, 16 bit out, 2906 w/ SPDIF; active 2023-08
2001-01 AKM AK5354 89 dB 2.5 (3.3 to 2.1) -84 dB 0.1 dB -65 dB low power, 20 bit out
1999 Alesis Semi (Wavefront Semi) AL1101 107 dB 5 -95 dB 0.025 dB -76 dB 24 bit out, max 55 kHz, 5th-order 1-bit mod @ 64X OS
1999 Burr-Brown PCM1801 93 dB 5 -88 dB 0.05 dB -65 dB consumer, 16 bit out, SE input; still active 2023-08!
1998-09 Cirrus Logic
(Crystal)
CS4223
CS4224
[CODEC]
105 dB 5 -97 dB 0.01 dB -80 dB codec, 24 bit out
" " CS4220
CS4221
[CODEC]
100 dB 5 -92 dB
1998 AKM AK4523
AK4522
[CODEC]
100 dB 5 -92 dB 0.005 dB -80 dB codec, 20 bit out; evolution of AK4520(A) (1996/97)
1998-12 Philips UDA1341TS
[CODEC]
97 / 100 dB 3 -85 / -90 dB 0.05 dB -60 dB codec, low-power; SE/diff inputs, 20 bit out; PGA + AGC, DSP
1997-11 AKM AK5392 116 dB 5 -105 dB 0.001 dB -110 dB 24 bit out, early multilevel mod; EOL ~2012
1997 Cirrus Logic
(Crystal)
CS5394 117 dB 5 -103 dB 0.005 dB -117 dB 24 bit out; 3-level w/ 7th-order mod; cut-down CS5396, slightly earlier?
1997-09 Cirrus Logic
(Crystal)
CS5360 105 dB 5 -95 dB 0.0025 dB -85 dB non-TOTL, 24 bit out, CS5335 update
1997-02 AKM AK5351 103 dB 5 -97 dB 0.005 dB -80 dB non-TOTL, 20 bit out; 48 kHz companion to AK5352, update of AK5350 (1996) in smaller package
1997-11
 
AKM AK4526
[CODEC]
100 dB 5 -90 dB 0.005 dB -80 dB codec, 2-in 6-out-24/96, ~AK5350 (1996) / AK4520A (1997)
1997-02
 
AKM AK4520A
[CODEC]
100 dB 5 or 3.0 -90 dB 0.005 dB -80 dB codec, 20 bit out, early multilevel mod; ~AK5350 (1996); update of AK4520 (1996-12, 98 dB, -90 dB)
1997-07 Cirrus Logic
(Crystal)
CS4222
[CODEC]
99 dB 5 -90 dB 0.01 dB -80 dB codec, 20 bit out
1997-05 Philips UDA1340
[CODEC]
95 dB 3 -85 dB 0.05 dB -60 dB codec, low-power, SE inputs, 20 bit out; update: UDA1344TS
1996-11 AKM AK5391 113 dB 5 -97 dB 0.001 dB -110 dB 24 bit out, early multilevel mod
1996-11 Cirrus Logic
(Crystal)
CS5335 105 dB 5 -95 dB 0.0025 dB -85 dB non-TOTL, 20 bit out, 4th-order mod @ 128X OS
" " CS5334 100 dB " -90 dB
1996-09 AKM AK5350 100 dB 5 -94 dB 0.005 dB -80 dB non-TOTL, 20 bit out, early multilevel mod; update, smaller package: AK5351 (1997)
1996? AKM AK5340B 100 dB 5 -95 dB 0.01 dB -88 dB non-TOTL, 18 bit out; 64fs 5th-order delta sigma mod; updates AK5340/A (96 dB, -92 dB)
1996-09 Cirrus Logic
(Crystal)
CS4226-KQ
CS4226-BQ
[CODEC]
95 dB
93 dB
 
5 -88 dB
-86 dB
 
0.01 dB -80 dB codec, 2+1-in 6-out, 20 bit out; SPDIF receiver, also available w/o as CS4227 (1998)
1996(?) Cirrus Logic
(Crystal)
CS5330A
CS5331A
94 dB 5 -84 dB 0.05 dB -80 dB consumer, SE input, 18 bit out; 128fs delta-sigma; 8-pin package
1996 AKM AK5330 90 dB 5 -84 dB 0.1 dB -70 dB consumer, SE in, 18 bit out; 64fs 4th-order delta-sigma
1995 TI TLC320​AD58C 95 dB 5 -93 dB 0.01 dB -80 dB 18 bit out
1994? AKM AK5340 96 dB 5 -92 dB 0.01 dB -88 dB non-TOTL, 18 bit out; 64fs 5th-order mod; updates: AK5340A/B
1994(?) Analog Devices AD1877 92 dB 5 -90 dB 0.006 dB -90 dB non-TOTL, 16 bit out, 4th-order mod; still active 2023-08!
1993(?) AKM AK5390 112 dB ±5 -103 dB 0.005 dB -100 dB 20 bit out, = CS5390; SNR=110dBA, SNDR=104dB; AES paper
1993-10 Cirrus Logic
(Crystal)
CS5390 110 dB ±5 -100 dB 0.005 dB -100 dB 20 bit out, 5th-order, =AK5390
1993(?) Burr-Brown PCM1760 + DF1760 108 dB ±5 -92 dB 0.001 dB -100 dB 20 bit out, 4th-order 4-bit delta sigma ADC + digital filter
1993-03 (?) Cirrus Logic
(Crystal)
CS5389 107 dB ±5 -100 dB 0.01 dB -80 dB 18 bit out, 5th-order, =AK5389 (SNR=106dBA, SNDR=102dB); AES paper
1993(?) Philips SAA7360 97 dB 5 -90 dB 0.001 dB -93 dB 18/16 bit out, 3rd-order mod
1993(?) Cirrus Logic
(Crystal)
CS4225
[CODEC]
85 dB 5 -85 dB 0.1 dB -75 dB codec, 2+1-in 4-out, 2x 16 bit + 1x 12 bit out
1992(?) Cirrus Logic
(Crystal)
CS5336 (-K/-B/-T) 95.7 / 93.5 / 92 dB ±5 -92.7 / 89 / 86 dB 0.01 dB -80 dB non-TOTL, 16 bit out, non-alias filter
" " CS5338
CS5339
95.7 dB " -92.7 dB 0.01 dB -80 dB non-TOTL, 16 bit out
1992? AKM AK5339 94 dB ±5 -90 dB 0.01 dB -80 dB non-TOTL, 16 bit out; 1-bit, 4th-order mod; =CS5338/9?
1992(?) Cirrus Logic
(Crystal)
CS5349 90 dB 5 -87 dB 0.01 dB -80 dB non-TOTL, 16 bit out
1992? AKM AK5345 91 dB 5 -86 dB ? ? non-TOTL, 16 bit out; newer similar 3.3V model: AK5343
1991(?) Analog Devices AD1879 105 dB
(103 dB unwtd.)
106 dB unwtd. ±5 -96 dB
(-98 dB, trimmed)
0.0005 dB -115 dB 18 bit out, 5th-order; 16 bit out: AD1878; filter design paper (errata), related patent
1990 Ultra Analog ADC 20048 108 dB unwtd. ±15 -96 dB
(-2dBFS)
0.05 dB -80 dB* *) up to fs + 20 kHz, max -50 dB above (up to 1.5 MHz); 20 bit out, 4th-order, multilevel delta-sigma using 4-bit flash ADC in 2-chip hybrid module; ±15 V analog and +5 V digital supply; datasheet, AES paper, article with backstory in Studio Sound 05-1990, pp. 32-36, modulator patent
1990(?) Crystal CS5328
CS5329
97.1 dB 100.1 dB ±5 -94.5 dB
(-97 dB, mono)
0.0005 dB -86 dB 18 bit out, 4th-order, = AK5328/9; 5329 w/ non-alias filter; AES paper
1990(?) Burr-Brown PCM1750 90 dB ±5 -90 dB (ext) (ext) 18 bit out, 2ch SAR ADC for use at 4X OS, ext analog + digital filters
1989 Crystal CS5326
CS5327
95.7 dB ±5 -92.7 dB 0.0005 dB -86 dB 16 bit out, 4th-order, = AK5326/7; 5327 w/ non-alias filter; SNR=96dBA, SNDR=93dB; AES paper
1989(?) Burr-Brown PCM78P 94 dB ±12 -88 dB (ext) (ext) 18 bit out, single SAR ADC for use at 4X OS, ext analog + digital filter
1988(?) Crystal CSZ5126
(CS5126*)
92 dB unwtd. 95 dB unwtd., 2X OS ±5 -92 dB (ext) (ext) *) renamed by 1995; 16 bit out, 1 charge redistribution SAR ADC w/ multiplexed stereo input (or mono at 2X OS), ext analog + digital filter; audio version of CS5101A
1984? Sony CX20018 >86 dB (>90 dB, PCM- 501ES) ±5 -86 dB (ext) (ext) 14/16 bit out, 1 multibit ADC w/ multiplexed 2-ch input, ext analog filter; update CXA1144S

By 1989/1990, converter performance had already reached / exceeded the limits of 16-bit samples, and hybrids were scratching at the limits of 18 bits. At that time, a split analog supply was still necessary. It would take until 1993 for ICs with a single +5V analog supply to jump the same hurdle, with their split supply cousins still being 11-13 dB ahead. Those would only become obsolete with new 24-bit single-supply models forging ahead in 1996/1997. While in these 7 years the absolute limits of dynamic range may only have moved by about 10 dB if we count the ADC20048 hybrid, single-supply converters were pushed from a modest 90 dB(A) by 1992 to an unprecedented 116-117-120 dB(A) by 1997. That's impressive!
By contrast, things have been moving a lot more slowly since then. By 2001/2 we had arrived at 123 dB(A), and by 2006 the same was feasible on only 4.0 V. Only in 2014 did we get 127 dB(A) on 5 V. Since then new designs around 4.5 V have mainly concentrated on pushing THD+N instead of further increasing DR. By the mid-2010s 3.3 V ADCs had moved into midrange recording ADC terrain with DR slowly creeping up and the addition of differential inputs, and by 2019/2020 it became clear that the analog side would support more if you insisted, culminating in the first straight 123 dB(A) parts being released in 2023.

The first AKM and Crystal ADCs were jointly developed (and produced by AKM), which however seems to have stopped in the mid-'90s, some time after Crystal was taken over by Cirrus Logic in 1991.

When Crystal brought out their new double-speed CS5396 just weeks before the single-speed AK5392 came out, someone at AKM must have been cursing! By 1997, the writing was definitely on the wall – 96 kHz sampling had been making its way into studio ADCs in the mid-'90s. Funnily enough, AKM did actually have the first 96 kHz capable ADC out, the midrange AK5352, but things must have been moving more slowly for the TOTL model.

1990s High-End Pro Audio ADC Solutions

ADC engineering for professional applications was at a very high level back in the '90s. While researching high-end ADC performance I came across a news article in the March 1995 edition of Studio Sound magazine which read:

4D goes to third generation

The Deutsche Grammophon Recording Centre has developed a third generation upgrade of the Stage Box system central to the 4D recording chain. All recordings made by the Recording Centre since October 1994 have used the new DG AD III technology, whose convertors feature the new Crystal CS5390 delta-sigma 20-bit A-D convertor ICs to provide 23-bit digital-floating delta-sigma A-D conversion.
The process employs two 20-bit convertors, one handling the input signal at unity gain and the other operated with 18dB gain. A sophisticated DSP algorithm regulates the crossfade between the two convertors, producing three bits of supplementary resolution. The DSP program was modified to allow the DSP chip to handle 20-bit convertors at its inputs and a 24-bit wordlength at its outputs.
Quoted specifications include THD+n of -121dBFs with an input of 997Hz at -30dBFs and linearity errors within 1dB down to -135dBFs, together with a largely flat noise-spectrum.
A further improvement is the development of the Authentic Clock Recovery system, permitting superior reconstruction of the master clock signal under real world operating conditions such as long cable runs and numerous interconnected PLLs, where phase modulation of the clock, jitter, becomes a limiting factor on overall system performance. Because Authentic Clock Recovery uses crystal PLLs driven at 512Fs, as opposed to the current 256Fs standard, A-D conversion at up to 96kHz is possible, with full oversampling capability.

Deutsche Grammophon, Germany. Tel: +49 4044 181115.

They were using a composite ADC topology all the way back in 1994! (You may be familiar with the concept from the Sound Devices MixPre II series recorders, or more recently, the Zoom F6 recorder and UAC-232 audio interface or the RØDE NT-1 5th generation XLR/USB microphone, all with float32 output.) I am not sure what exactly they were doing with the 512Fs clock – either the ADC was being overclocked by 100% or they were using two ADCs interleaved so that effective sample rate would be doubled, the difference being that anti-alias filter bandwidth of the former would be twice that of the latter. All times two for the composite ADC action, of course.

I can only assume that they needed such a high level of A/D performance to cover the entire dynamic range of a large-diaphragm condenser microphone, or at least enough of it to get by with a constant mic gain in practice. Some modern LDCs could make even the highest-performance ADCs break a sweat.

By 1998, studio ADC dynamic range had increased by another 10 dB albeit at high input levels, with e.g. the Prism Sound Dream AD-2 posting specs of DR > 131 dB(A) at +28 dBu in, -1 dBFS THD+N < -108 dB and any spurious anharmonics below -112 dBFS (-1 dBFS level) to -140 dBFS (-60 dBFS level). That's the entire unit, of course! Since 0 dBFS level can be turned down to +5 dBu and dynamic range at +18 dBu remains >129 dB(A), total non-instantaneous dynamic range covered must equal at least 139 dB(A), only 3 dB short of a modern RME ADI-2 Pro FS. I've done some math using the RMS unsumming calculator, and effective input noise comes out to about -115 dBu(A), so it probably is more like 141 dB(A) total and as such just 1 dB short.

Go back to table of contents

Go To…

Quick Navigation


© Stephan Großklaß 2023. Commercial duplication of this content, including eBay descriptions and similar, with prior permission only.

Contact me

Page created: 2021-07-09
Entry first created: 2021-02-12
Last modified: 2023-09-22