Curiously Strong CMOY Headphone Amp Review

[solderdude]

102dB peak level for DT880-600 means about 85dB average on well recorded music.
Decently loud but not very loud for sure.
Loud enough for 'normal' listening.
For the HD600 2V = 110dB peak which is loud enough but as soon as people EQ in some bass it drops quickly.
One would like to stay away from any clipping levels with opamps.

2V is possible with a single 9V battery. With 2x9V on high Z headphones (>150 Ohm) 5V is possible and should be enough for very loud and distortion free HD600/HD650/HD58X/HD660S listening.
The HD560S can get by on a single 9V.

My rule of thumb:
When you want to play uncomfortable loud (for 30 secs. or so) and not run into clipping issues the headphone should be able to reach 120dB SPL
When you want to play comfortable loud (duration of a complete song) without issues then 105-110dB SPL is enough.
When you want to play at 'normal' to moderately loud listening levels 90-95dB SPL should be reached.

Note that dBSPL is not the same as Phon. You may need more dBSPL in the bass than one might think. Especially when a few dB extra bass is used.
People often say 90dB is loud... yes a 90dB 1kHz tone is very loud. Much louder than the average 90dB sounds you often see in those SPL tables.
85dB bass notes and 70dB mids superimposed sounds 'normal' but you need to have an amp that can put out enough voltage.

This is why I made this list so headphone owners can see what they need for their on- and over-ear headphones (IEMs is never a power problem).
Removes the guessing, assumptions and calculations per headphone.

 

[AnalogSteph]

It's interesting to note you had oscillation issues with some opamps.

In my standard test jig, all the opamps I have work just fine with one exception - the LM4562 which oscillates at about 380kHz when connected to any cable longer than about 40cm. Even series resistors in the output don't help.

This is at a low gain of around 2x as well - even a cap across the feedback resistor doesn't help. So although it may be a good opamp, it seems to be difficult to stabilise.

I guess you should have tried higher gain as a last resort, that should eventually do it. What were your feedback network resistor values like? Since you ruled out so many things I'd suspect an issue of output to input coupling, either capacitively or via ground, and a nice resonance peak in FR.

Here's Sijosae's old cMoy output level test results for completeness' sake (pretty standard circuit if memory serves, 4k7/1k):

TDA1308 was actually run out of spec, it's like 8 V max. You can see that NJM4556A is hard to beat for an opamp as far as sheer grunt into 33 ohms goes. Which I guess also is why it was used in the infamous Grado RA-1, though I can't say I would necessarily recommend the circuit as-is.

A while back I ran some load tests (with headphone usage in mind) on a group of opamps and the OPA2134 didn't do very well.

All tests were run with a 15-0-15 psu.

Surprisingly, the NJM2068 was second to the LM4562 which came out on top.

That's odd, an OPA213x should have driving abilities roughly in the 5532 ballpark. Do you remember the circuit values used? All I can imagine is either a big impedance mismatch between the inverting and noninverting inputs (the datasheet recommends <2 kOhms of difference due to the usual input capacitance modulation issues typical for JFET parts), or possibly a fake part that actually was a rebadged TL072 or 4558 (a relative measurement of input noise and input bias current should clear that up - the real deal should be about 3 dB noisier than a 5532 with negligible input bias current, a TL072 would be over 6 dB noisier).

 

[audio_tony]

That's odd, an OPA213x should have driving abilities roughly in the 5532 ballpark. Do you remember the circuit values used?

Douglas Self also found the OPA2134 to lack drive ability. To quote from his book, mentioned below.
"The OPA2134 does not show phase reversal anywhere in the common-mode range, which
immediately marks it as superior to the TL072.
The two THD plots in Figures 4.36 and 4.37 show the device working at a gain of 3? in both
shunt and series-feedback modes. It is obvious that a problem emerges in the series plot, where
the THD is higher by about three times at 5 Vrms and 10 kHz. This distortion increases with
level, which immediately suggests common-mode distortion in the input stage. Distortion
increases with even moderate loading, see Figure 4.38.
This is a relatively modern and sophisticated op-amp. When you need JFET inputs (usually
because significant input bias currents would be a problem) this definitely beats the TL072; it
is, however, four to five times more expensive."

Here is my test circuit.

THD for OPA2134 from Self's (highly recommended) book "Small Signal Audio Design"

 

[kn0ppers]

If you only want to use one dual OpAmp in your CMOY, OPA1688 is probably among the best options.

Also, for comprehensive OpAmp measurements:

http://www.nanovolt.ch/resources/ic_opamps/index.html

Says about OPA2132/34:

A dual amplifier with JFET inputs, topology is unknown. Voltage noise is modestly good only, input offset is rather good for a JFET amplifier though. Single and quad version are available (OPA132 and OPA4132); OPA134/OPA2134/OPA4134 appear to be versions with relaxed DC precision specification.
The basic transfer linearity is pretty good, although it degrades somewhat at higher frequencies. Common-mode effects are relatively well controlled for a JFET input devices, though resulting distortion is very significant at higher frequencies nonetheless. Heavy output loading causes serious distortion, with thermal effects and harmonics up to high frequencies visible.
Good overall cost-performance ratio for a JFET device. For lowest distortion attention to common-mode and especially output loading effects
needed. Suitable upgrade for TL072 amplifiers where the higher quiescent current is no concern.

 

[AnalogSteph]

If we're doing some extrapolation from Douglas Self's data, I would expect maybe 0.002% for the OPA2134 driving 2.7 V into 270 ohms at 1 kHz - so a few dB worse than the NE5532 for sure, but not more than an order of magnitude (!). My bets would be on a fake chip still.

 

[Trouble Maker]

Gotta slow down on trying to eat all of your harvest, no wonder your septic tank is full, that's a lot of fiber. Don't you have neighbors?? lol

He could barter crops for weeding or lifting heavy objects (speakers).

 

[audio_tony]

If we're doing some extrapolation from Douglas Self's data, I would expect maybe 0.002% for the OPA2134 driving 2.7 V into 270 ohms at 1 kHz - so a few dB worse than the NE5532 for sure, but not more than an order of magnitude (!). My bets would be on a fake chip still.

Bear in mind those figures I posted were into a 270ohm load - Self's tests were into a 2.7k load.

Here's the same opamp, same circuit etc. into a 2.7k load.

But compare that to an NJM2043DD - less harmonics and THD a factor of 5 lower.

 

[Helicopter]

I might be wrong, but is this the first ever decapitated recommendation? Cool!

You beat me to it. Though I would probably do an O2 if I were ready for another science project.

 

[jasonhanjk]

You beat me to it. Though I would probably do an O2 if I were ready for another science project.

My next project is actually an O2 redesign. My design's noise will be better than the current O2 design. However I'm still not satisfied it's 32 ohm driving capability.

 

[preload]

This is why I made this list so headphone owners can see what they need for their on- and over-ear headphones (IEMs is never a power problem).
Removes the guessing, assumptions and calculations per headphone.

That is such a cool list. Is everyone in the Netherlands as awesome as you?

 

[preload]

Anyway, my subjective remarks are from my overall experience and feeling about a product. It is an additional data point above and beyond what the measurements say. If all you want to go by is measurements as you ask, then you should just ignore my opinion/recommendation and do just that.

Okay, that makes sense then. I was trying really hard to reconcile the headless panther rating with the measurements, particularly since it seemed like the CMoy was being "penalized" because it did not have high power output and it was only intended for high impedance cans (as Solderdude confirmed).

Well, since the final rating is "subjective," I'll add "mine" to the broader discussion. To me, the CMoy should have the highest subjective rating. The CMoy, as many of us know/remember, was THE amp that ignited the hi-fi headphone amplifier movement. Before the CMoy, everyone pretty much used the headphone jack on their receiver or the 1/8" output of, say, a discman. Maybe there were a few esoteric high end amps, but headphone amps at the time were definitely not mainstream. The Headwize website, created by Chu Moy, was a wonderful little forum with a real DIY spirit. And for many, the CMoy amp was their first real DIY electronics project. And how cool was it to create a miniature headphone amplifier using an Altoids can? It really embodied the "maker" culture, and I bet Chu Moy inspired quite a few people to pursue a career in engineering.

 

[JohnYang1997]

Okay, that makes sense then. I was trying really hard to reconcile the headless panther rating with the measurements, particularly since it seemed like the CMoy was being "penalized" because it did not have high power output and it was only intended for high impedance cans (as Solderdude confirmed).

Well, since the final rating is "subjective," I'll add "mine" to the broader discussion. To me, the CMoy should have the highest subjective rating. The CMoy, as many of us know/remember, was THE amp that ignited the hi-fi headphone amplifier movement. Before the CMoy, everyone pretty much used the headphone jack on their receiver or the 1/8" output of, say, a discman. Maybe there were a few esoteric high end amps, but headphone amps at the time were definitely not mainstream. The Headwize website, created by Chu Moy, was a wonderful little forum with a real DIY spirit. And for many, the CMoy amp was their first real DIY electronics project. And how cool was it to create a miniature headphone amplifier using an Altoids can? It really embodied the "maker" culture, and I bet Chu Moy inspired quite a few people to pursue a career in engineering.

The issue is not that the theoretical CMOY design is bad. It's the final result seeing here of this particular CMOY. Like Amir said, he recommended it for DIY.
1, Simply put an opa2156 in there would be much better, being rail-rail both in and out with 100mA output current.
2, Build a pcb for this with excellent layout and grounding topology. With the same components, distortion won't be as high. In fact can go under -120dB.
3, Use two 9Vs than one. Dramatically increase the output swing. You get 80mW+ for 300ohm.

 

[tvrgeek]

Line many, I built one of these years ago. Used it mostly to do subjective comparisons on op-amps. It drove my Grados OK, but a bit weak on my old Yamahamers. They were low-ish impedance. I did the 9V with a splitter, then went to two 9Vs. Better but did not fit in the can.

Ear buds had not been invented when the CMOY was published. I think that is now where the market has gone. They were to get better sound from your diskman.

 

[Nango]

102dB peak level for DT880-600 means about 85dB average on well recorded music.
Decently loud but not very loud for sure.
Loud enough for 'normal' listening.
For the HD600 2V = 110dB peak which is loud enough but as soon as people EQ in some bass it drops quickly.
One would like to stay away from any clipping levels with opamps.

2V is possible with a single 9V battery. With 2x9V on high Z headphones (>150 Ohm) 5V is possible and should be enough for very loud and distortion free HD600/HD650/HD58X/HD660S listening.
The HD560S can get by on a single 9V.

My rule of thumb:
When you want to play uncomfortable loud (for 30 secs. or so) and not run into clipping issues the headphone should be able to reach 120dB SPL
When you want to play comfortable loud (duration of a complete song) without issues then 105-110dB SPL is enough.
When you want to play at 'normal' to moderately loud listening levels 90-95dB SPL should be reached.

Note that dBSPL is not the same as Phon. You may need more dBSPL in the bass than one might think. Especially when a few dB extra bass is used.
People often say 90dB is loud... yes a 90dB 1kHz tone is very loud. Much louder than the average 90dB sounds you often see in those SPL tables.
85dB bass notes and 70dB mids superimposed sounds 'normal' but you need to have an amp that can put out enough voltage.

This is why I made this list so headphone owners can see what they need for their on- and over-ear headphones (IEMs is never a power problem).
Removes the guessing, assumptions and calculations per headphone.

Re/ your list: stated power for SPL 90dB and 120dB is always the same?? Pls check.

 

[solderdude]

90dB column is in mW the 120dB column in Watt.

 

[Nango]

Ok, I see.