Audio-gd discrete op-amps reviewed: OPA-Earth, OPA-Moon, OPA-Sun v.2

[Currawong]

Andrea, to your credit, the LME49710 is very good, about the same as the Earth, if marginally less detailed. To your discredit, being right about something doesn't change that you've been banned and are about to be again.

 

[majkel]

I didn't own the aforementioned LME op-amps in the time of the review but added some remarks later in the thread. The LME49860 is great, something between the thickness of the LM4562 and the cold neutrality of the LME49720. However a pair of the AD797BRZ is slightly better, just slightly. On top of this I would place the decompensated op-amps, when you fine tune their feedback and/or custom compensation loops, they outperform even the AD797BRZ pair. These are OPA228, OPA2228 and AD8021 - no joke. Especially fine-tuning the AD8021 is kind of surprising. I had best results with no compensation capacitance between pins 4 and 5 and about 200 ohms Rf.

Regarding the AD744 - it's cold when supplied with low voltages, thicker sounding when supplied closer to the maximum voltage rating, i.e. +-15V.

 

[majkel]

OPA2227 is terrible, worse than the OPA2134. I don't understand using the Panasonic FC series for audio, either. For digital the Sanyo WG seems the best, for analog - here you have quite a big choice - BG, Elna Tonerex, Elna Silmic II, Panasonic PX. When using the OPA2228 the compensation must be added for gains lower than 5. In the DAC I'm using it's just plug and play, no oscillations. The AD8021 won't work in it without circuit ingerence but I tried it in a headphone amp in the VGND section after the voltage divider, and it worked best, better than the AD797BRZ or some other high grade single op-amps.

 

[jamato8]

I have used the FC series and have heard it as the only cap in a portable and frankly I found the cap to perform well.

 

[majkel]

The LT1364 plays simplified and a bit colored music. In unresolving setups it's helpful, starting from some level of quality, it is annoying and killing nuances. This op-amp sounds nothing more than a bit defective LT1028 pair.

Regarding the Panasonic caps - the FG series is more pleasant sounding than the FC, the same applies to the FA and the PX series, where the latter is truly audio dedicated, AKA Pureism.

 

[majkel]

Quote:

Originally Posted by Horse /img/forum/go_quote.gif You keep showing me that you only or mostly listen to detail and the "formal" side of sound.

You are wrong, period. I care for real presentation, full of dynamics, timbre, touch, air, intimacy, just everything.
Quote:

For me tonality is vital, and there lies much of the emotion. And BTW the LT1364 does NOT sound like a less resolving LT1028, since it doesn't have the dryish and slightly lightweight mids, and slightly skeletal bass, of the LT1028. The LT1028 on the other hand doesn't have what I call the "warm luminosity" of the LT1364.

Leave the CN8 series and try the LT1028ACN8, it's more lush than the LT1364 which has colored trebles and a kind of imaging instability causing bluriness. Try the AD8599 if you haven't so far. Up to your preferences IMHO.

 

[majkel]

I don't agree, the statements about the LM6172 mean the guy has got not good enough equipment. The LM4562 is audibly, obviously better for audio than the LM6172. However, it might be true both your impressions for the Xenos ans his impressions for the DIYEDEN. If you tweak the DAC totally (caps, rectifiers, transformer?, Toslink supply?, digital parts decoupling, power supply section rebuild) you can compare the op-amps again - on another quality level this time. The same applies to Mr Jim. Have fun!

 

[majkel]

Quote:

Originally Posted by Horse /img/forum/go_quote.gif P.S. The guy also says that the OPA2604 is "dirty sounding" and prefers the OPA2134

OPA2134 has got pumped up bass and nasal coloration. I still prefer the OPA2604 while both aren't interesting for me.

I compare, also side by side, to more sophisticated DACs than you've mentioned - more like AudioNemesis DC-1, CEC TL51, NorthStar M192, Theta DSP Pro Va, Accuphase DC801. So, save me that drama about a $50 DAC. Regarding the buffers - I won't tell you what buffers I use, they just keep me laughing at the BUF634, but it's another story.

One more word about the dryness present everywhere - try good, audio grade capacitors. Leave the Pana FC. You have a choice of black gate's polar and non-polar, different flavors of Elna, Nippon Chemicon (LXG or LXY maybe), Rubycon (YXF?), some plyester film caps for decoupling the 'lytics, some bipolar to decouple bigger electrolytics, etc. Try it. See for yourself why controlled avalanche diodes are sonically different than soft recovery diodes, why both are better than standard rectifier diodes. See how the DAC chip reacts to more stabilized analog voltage. When you have it verified and chosen, start listening. The end of it will be - compensated op-amps are the bottleneck, no matter which one you choose. Uncompensated op-amps properly used are in a different league. Make it audible. I made several turns around the OPA2228 until my circuit got so transparent that all the rest fell off. In it's sound is just more freedom.

 

[majkel]

My DAC went through many more upgrades since the review time. I have double stabilized 5V, one regulator after another. No Silmic caps, Rubycon YXF + film cap, then Elna + film cap before another regulator, then Rubycon BG STD on the supply, and lots of different caps near the DAC chip. Silmic II is there as well.

I have no clock onboard as I have precise clocking in my digital transport, based on a 5ppm gold pin quartz, so the DAC works synchronously (which I find the best way to design digital electronics - keep it synchronous where possible is what I was taught during the studies).

My modified Technics SL-PS840 sounds 1-2 levels inferior to my DAC while being comparable to the DAC devices I mentioned previously. I prefered my DAC to the Northstar before many later upgrades, with the standard EI transformer, normal diodes, poorer supply decoupling, don't remember what else.

I ordered the LT1358 just for fun and will let you know what I think about it.

What I think about coaxial cables is that they suck when you have no isolation transformer at least on one side of the path as you create a ground loop between devices. The Toslink keeps them electrically separated. For higher transfer and powerful, isolated transports like the gorgeous Accuphase DP800, I'd go with a generic 75 ohm digital TV cable which is all this transport needs. It's powerful enough to squeeze the bits thru a normal cable - confirmed.

 

[majkel]

Quote:

Originally Posted by Horse /img/forum/go_quote.gif No with the coaxial you don't create a feedback loop (if the transport is just that), at most you create a ground chain and that's no bad thing.

Should have been a ground loop, just a typo. It's a bad thing. One ground connection sometimes is a good thing. I tried several optical cables. I agree that some are crap. On the other hand, a good transport will work well with a generic coaxial cable, and I could live with that but ATM I am using a transport which I like very much due to magnetic laser mechanism (no tooth wheels) and there was just the Toslink as the digital output. I am to lazy, and see no reason, to install the coaxial output. My another transport will be something reading denser formats than the RBCD but no sooner than when my current CDP fails.
The matter with the Toslink is that both transmitter and receiver need supply tweaking just like other parts of the CD player or the DAC when you want to improve it. After that you can see what a decent optical cable is capable of. The simplest trick is to add a roughly 100uF tantalum capacitor to the Toslink transmitter and receiver supply, between the GND and +5V pins respectively.

 

[Pluto2]

>To sum it up - even if you were able to identify an individual "best opamp", cascading a few of them will always produce flawed and worse sound than more imperfect individual opamps sensibly and competently combined both for sound and specific electrical virtues.

Hi Horse, this combination (LME49860 + LT1358) sounds interesting, have you ever tried it?

I am thinking of upgrading my 2-way active crossover that contains six LM4562s per channel with those perhaps....instead of the discretes which I had in mind....suggestions?

 

[Pluto2]

Thanks Horse, will be interested in hearing the report from majkel on the LT1358 as well

 

[majkel]

The THS4032 is a bit dry and colorless. Regarding the LME49860 - in my setup it's a sweet spot between the LM4562 and the LME49720 but I still prefer the AD797BRZ pair. On fast, acoustic music with lots of classic guitars playing at the same time the AD maintains the resolution and instrument separation better. On the other hand, there is still one music dimension missing due to overcompensation. When you compare the AD797BRZ to the AD8021 with matched feedback loop, the latter is better. Possibly, do it without caps, just the Rf matching (see datasheet). However, when compensated op-amps are in the further signal path, it's obviously impossible to hear what I'm talking about.

 

[Pluto2]

Excellent gentlemen. Soon will order some LME49710s, LME49860s (mainly), and LT1358s, together with several discretes I already have for mix and match fun : )

 

[majkel]

Quote:

Originally Posted by Horse /img/forum/go_quote.gif The THS4031 is luminous sounding, more so than the blueish and a bit sad LME49xxx & LM4562. It is a bit colorless if not properly matched with the rest of the active parts (DACs, opamps...). Anyway my complaints with the LME's don't lie in the dimensions of the stage (pretty secondary aspect for me). BTW for tonal quality my favorite in the series seems to be the LME49710. BTW there's the THS4051 that's warmer (due to the "5" in its name ) than the THS4031. I prefer the THS4051. It's more stable as well. Then again, the LT1358 has both the 3 and the 5 in its sonics so for any lone opamp application it's my favorite of the bunch. Haven't tried the AD8021, anyway its numbers anticipate that it's close to the LT1028 for tonal quality, so realistically not quite a favorite for me. While the AD8022 had too much "2" (greenish sound), so it's even worse.

Cool! Op amp numerology.

Could you provide us some rules how to resolve the sound from the numbers? Hearing is so subjective, numbers are so scientific...

OK, I acquired the LT1358. This one is better than the LME flavors, indeed but in my current diode/cap/transformer/other stuff a pair of the LT1028ACN8 sounds a bit more liquid and natural. The LT1358 sound is a bit sweeter but also a tad sticky and confined. Actually there is something common between the LT1358 ant the LT1364 in low treble region which is overemphasised and smoothed at the same time. The OPA2228 is still another league with its independent source imaging and acoustics. So, there was no surprise when I tried the OPA228P which I received as well, that I liked them even more due to bigger soudnstage and more stable imaging.