"Life is what happens beween headphone sessions" unknown
"My profession is digital, my passion is analog" me
My day job finally allowed some time to run some opamp comparisons
For your amusement, below is a pic of an NJM4556 that I blew up when I shorted a voltage probe between +12 volts and an output.
Here is a reminder of the power output with LM4562 opamps.
Let's begin with some facts regarding the outputs of the Sunrise 2 (FETs) and Starlight:
Project Starlight OpAmp Comparison
NOTE: The assembly manual for the Starlight can be downloaded from www.garage1217.com.
NOTE: The article below is from an email question I sent to Frans, the engineer who designed the amps made by Garage1217. The question was “Why the outputs on the SUNRISE 2 (with no music playing, or zero input) run hot while the outputs on the STARLIGHT are slightly warm.”
The Sunrise is running in class-A which is needed otherwise quite audible distortion would be present.
The Starlight is running in AB which is possible due to the feedback in the opamp circuit.
The Starlight does not but basically quadruples the output current of the opamp sections by having 4 opamp section sharing the same load (headphone) so 4x 40mA yields 160mA of output current.
When you are running only 2 NJM4556 (so 1 DIP-8 per channel) you will have the same output current as when 4 LM4562 are used.
When you are running 4 NJM4556 the maximum output current will be 4x 70mA = 280mA, that's about the same as a BUF634.
The opamps run cooler because the idle current is low. Per DIP-8 this is only about 10mA for most of the better opamps.
With 24V over them they dissipate only 0.24W when doing nothing and will get hotter when current is drawn.
For class-A there is always a high idle current that needs to be higher than the highest current that can be drawn by the load.
In the SR-II this is 210mA (thus 21 x as much as the opamp idle current). With the same voltage (24V) this means the dissipated power (= heat) is also 21x times higher thus 5W.
Therefore the class-A stage (SR-II) gets very hot (5W) and the opamps (Starlight, C'Moy) do not (0.24W)
How warm the opamps become is determined by the idle current (also called quiescent current, see datasheet of the opamps) and how much current it must deliver in headphones, which in turn depends on the impedance of the headphone and output resistance setting of the amplifier. The idle (quiescent) current is not the same as the maximum output current.
END OF ARTICLE
Now, on to the comparison stuff. I have listed the headphones and opamps used first.
Opamp Manufacturer Output Current Frequency Response Cost
NJM4556 JRC 70ma 8 MHZ $.57
AD8620 Analog Devices 30ma 25MHZ $25.00
LM4562 Texas Instruments 26ma 10 MHZ $2.67
OPA2132 Burr Brown 40ma 8MHZ $6.21
OPA2134 Burr Brown 40ma 8MHZ $3.05
OPA2227 Burr Brown 45ma 8MHZ $5.00
**The price for one AD8620 at Digi Key runs from $15.35 to $31.01 with a minimum order of 1. These are surface mount only. Cimarron Technologies sells a Brown Dog adapter for 3.00 or an adapter with an AD8620 mounted for 25.00. Soldering an AD8620 (surface mount) to the adapter is not for the faint-of-heart!
The specs are taken directly from the DATA SHEETS of the amps. It isquite obvious that the frequency response range is WAY wider that the audio range (20 Hz to 20,000 Hz)
Amp rolling can be done with just one amp on each side. The output current listed above is DOUBLED; as two amps on each side (stock configuration) quadruples the output. This cuts down $$ while experimenting. YOU CAN EVEN RUN ONE TYPE ON ONE SIDE AND A DIFFERENT ON THE OTHER SIDE FOR A/B COMPARISONS!!
Most opamp reviews are done with a single pair of phones. I used three types:
Allessandro MS-1 (Grado basically); easy to drive.
AKG K702 (56 ohm-known to be “amp finicky”).
AKG Sextett (600 ohm--known to be quite “power hungry”). These actually have seven drivers—the main driver and six “passive radiators” which are energized by moving air from the main driver against the back of the driver—thus more power required. The Starlight has two output resistance settings—2.5 ohms for lower impedance phones like Grado and high impedance like 300-600 ohm (Sextetts) and German models. The AKGs and Grado were run on the LOW setting and the Sextett was run on LOW and HIGH setting. On these phones the volume was the same and the jury is still out on the sound.
These observations are only based on listening tests only. I have digital and analog oscilloscopes and signal generators; my intent here is a “high altitude” observation with a focus on how the opamps perform driving three different headphones volume wise; listening for volume and terms like volume knob at 9’clock, etc.
There are many reviews via GOOGLE on opamp sound comparisons using terms like sound stage, airy, detail and such. That is not the intent here. The sound on the output side of the opamps is the same as in most reviews with the exception that the output resistance can be changed from 2.5 ohms to 70 ohms.
Detailed sound comparisons with tube and opamp rolling will be addressed later.
The amp has a Sylvania 6922 tube—highly regarded good overall sound.
Two sources used.
Desktop PC running Roxio, which I use to make music CDs driving a MusicStreamer II. This in turn drives an Audio Control C-101 10 band equalizer. Yeah, I know, EQs are cheating and color the sound. OF COURSE THEY DO! That’s the idea. They are great for old recordings. The idea here is how the amp handles what goes into the inputs and comes out of the outputs, regardless of the source sound.
Cambridge Audio Azur 340 CD player going into an Applied Research and Technology (ART) professional 31 band per channel EQ.
(1) Metallica “Enter Sandman”—heavy drums.
(2) Mannheim Steamroller “Catching Snowflakes With Your Tongue”—deep sustained synthesizer BASS and crispy cymbals.
(3) Eagles “Hell Freezes Over”—live acoustic “Hotel California” with three acoustic guitars and a bold kick drum. I heard this at the first Rocky Mountain Audio Fest through a speaker system at about $60,000 cost. The phones aren’t as dramatic as the speakers but each instrument stood out about as well.
(4) Steely Dan “Deacon Blues” needs no introduction. It has many different instruments that are added as it progresses—excellent test for details.
(4) London Philharmonic Orchestra “Eine Kleine Nachmusic 1st Movement”—lots of strings.
The only reason this one is even included is that I happen to some on hand that I borrowed from my Pimeta amp (Tangentsoft.com) This amp is solid state with Burr Brown (Texas Instruments) BUF632 output buffers driven by the AD8620.
The price breakdowns can be seen in a previous paragraph.
The power output is 60 ma combined.
The Burr Brown amps are considered to have a “Laid Back” sound. Let’s call that “wood like”. If they are wood like then the AD8620 has a “metal like” sound. Considering that two are needed for about $50, I don’t recommend it unless it is affordable.
The sound? EXCELLENT! Bass, mids, and highs are clear and balanced. The combined 60 ma current out put drives the Sextetts comfortably. The K702s seem to have more BASS compared to the Sextetts. The Allessandros “almost” require switching out the EQ; boomy—I turned down the volume with EQ in. This brings to mind another good effect of the EQ. You can get more BASS and TREBLE with less volume.
Ahh, the infamous 57 cent Grado RA1 opamp! The RA1 is noted for driving Grados with a good sound to many, but not all. With a 70 ma output times two for a single dip, then it can drive the Sextetts with authority and they barely got warm at high volumes. With four per channel, it can drive insane volume levels with little effort. The K702 sounded solid state like. The Grados sounded great at low levels--high volume brought out the “Grado colored sound”.
OPA2132 / 2134
I grouped these two together as they sound the same to me. At 40 ma each, four will output 160 ma—power is not an issue. These have the Burr Brown “laid back” sound, almost tube like on their own. The Sextetts really stood out on the MIDRANGE and sound stage was like sitting on the stage, especially classical with an orchestra. The volume punch and bass is less that 8620 and 4556. The K702s did not fare as well, OK but not outstanding. The Grados sounded good but the bottom end seemed a little “loose”.
A departure from the 2132/ 2134 amps. It has 5 more ma for a total of 180 ma with two dips. Again, the AKGS and Grado sounded the same in HIGHS and MIDS, but the bass was tighter.
Last, but certainly not least. This is the one that comes with the kit and assembled. Garage1217 feels that this one tested best at their lab. It is a Texas Instrument amp, not Burr Brown, hence the LM instead of the OPA.
Power is 26 ma, which delivers 104 ma total with two on each side.
I see (hear) this one as closer to the AD8620, no Burr Brown “laid back” here. If you don’t care to roll amps, you can rest assured that this one covers most of the bases out there, to me that is. The Sextett was “livelier” as was the K702. The Grado could do anything I wanted with the EQ, but I wound up lowering the lower bass (20 / 40 Hz).
Bear in mind that this is a low tech comparison. Hopefully, more detailed opamp and tube reviews (not comparisons) are forthcoming.
Edited by rmouser - 5/28/13 at 3:51pm