[Malvin]

 
Speaking of those observing much worse RMAA results/crosstalk ..
 
These are all DIY builds, and there is no quality control in a normal sense. Parts could be different, build quality can suffer, test setup itself could create additional crosstalk.. I'm not saying this was the case, but it is not impossible..
 
Even RMAA itself produces different results depending on the settings selected, drivers etc.
 
btw: i measured this unit before, and its crosstalk numbers were worse than those posted above (around -70db for 33 ohm load).
 
CKK uses star ground, and the suggested wiring scheme is to keep input jacks isolated from chassis. This was not the case in my unit initially. And even though i didn't hear any noise of reduced performance, i decided to do it right. Crosstalk is better now, and isolating the input jacks was the only change I made, so i cannot think of any other reason for improvement.. so build quality matters..
 

 

[deadlylover]

Quote:

Are you just rolling in the NE5332, or are we going all out? Replacing the 4556, as already established, would invariably be a very bad idea.
If you are going to roll, I'll brave the slap and say at least do the test with the specified opamps a few times, in the same way you wouldn't stamp up and down on the Beta22 a few times before doing the listening comparison

 
Also, it struck me that nobody has ever verified the Beta22 actually measures far beyond audible limits. Based on how the dangerously (Capacitors running 100V out of spec/power supply built entirely by cutting up wallwarts and wiring the bits of about eight of them together) badly designed Singlepower amplifiers enjoyed prominence around here a while back as superb high-end amplifiers, an amplifier that doesn't actually perform that well being described as one of the best in its class would be comparatively unsurprising.

 
Aahahah sorry I wasn't being clear, I meant that I'd slap anyone who said the blind test wouldn't be fair because I'd be using my prototype Obj2 for the comparison, rather than a proper pcb build. Audiophile opamps are too expensive, I ain't going there.

 
Re: Singlepower, the Stax Mafia always knew that they were a steaming pile of crap. The circuit was based on the one that Stax gave away for DIYers, and that was okay, but it was the atrocious build quality that earned SP their reputation.
 
I do think some of the reservations you have with AMB are unfair, but hey, the world would be too boring if people didn't have conflicting views. The very nature of DIY designs is kinda "YMMV", we don't pay a cent to use their designs, so really, they owe us nothing. It's far from false advertising from a big corporation or anything like that. The beta22 was partly designed under simulations, and I agree, it's a shame that nobody has thrown it under a proper audio analyser yet. They're just so bloody expensive.
 
And even if the b22 performs worse on a scope compared to the Obj2, you also have to consider that the b22 is capable of much higher power output, so it's hard to say if it really "loses" or not.

 

[mikeaj]

Maybe the crosstalk naysayers are just not well acquainted with the right (superconducting? 

) 3.5mm jacks?  The plot thickens...
 
Most 3.5mm jacks are rated at 50 mΩ max contact resistance or something similar, unless I'm looking at the wrong ones.  How close are actual values to max (in practice for many things they are not close to max specs, but for this I don't know)?  Ignoring wire and any other impedance, with 33 ohms load, 20*log10(33 / 0.05) = 56.4 dB.  To get like 91.3 dB, you need all relevant impedances to sum to no more than 0.89 mΩ, right?  This is assuming that nothing in the amp itself actually contributing crosstalk higher than that amount, and that the limiting factor really is the jack.  Hopefully I didn't overlook something stupid.
 
I like how with 330 ohms load, RMAA claims less crosstalk than with no load.  Maybe RMAA is up to tricks again, not the same output level set, or something of that nature? 
 
 
With regards to 3 channel, you have to consider output impedance of the active ground too, which is why I'd be great to have that setup detailed once and for all, to eliminate any doubts.
 
 
But yeah I agree with the sentiments about DIY: parts, variances, and nice free designs to use.

 

[bcg27]

^^ I think most people probably use quarter inch jacks if they are going single ended for a desktop amp

Edit: Also can you explain your calculation of the crosstalk? What does the ratio of the load impedance to the output impedance give you? Perhaps also a reference where this formula is detailed?

 

[Malvin]

This test was done using 1/4 jack, which i think is more like 15 mOhm (I can try to locate its datasheet).
 
btw this unit also has 1/8 jack wired in parallel, I can try to rerun these tests in a different configuration..
 
As for the crosstalk being better for 330 Ohm vs no load at all.. I knew somebody would bring this up the difference is really small, and based on my experiments +/- 1db is nothing to talk about - you can easily get this difference by just running two subsequent tests for exactly the same config..

 

[cheapskateaudio]

Holy moly, the WAU16-400 PSU really does make a difference for HD650's, much better overall sound signature, more cohesive mids, I did a quick SBT since both wall-warts are under my desk and was able to pick out the better sounding PSU with ease. Still room for improvement though... Damn these wima caps are huge for being 1uf, lol. 
 
edit: Dropped the OPA2134's in the output stage... I think they sound......... a bit better on first listen. After a bit I think they sound congested relative to the 4556's, probably because those are a bit thin sounding, bass is more prominent overall, but relative to the mids may seem lesser, the mids are more there for sure. I'll keep messing around with it later today.
 
edit: I think it's definitive, OPA234's sound better in the output stage. There is less smearing of the sound, midrange clarity is improved, bass is hefty and womps you upside the head.
 

 

[mikeaj]

Quote:

This test was done using 1/4 jack, which i think is more like 15 mOhm (I can try to locate its datasheet).
 
btw this unit also has 1/8 jack wired in parallel, I can try to rerun these tests in a different configuration..
 
As for the crosstalk being better for 330 Ohm vs no load at all.. I knew somebody would bring this up the difference is really small, and based on my experiments +/- 1db is nothing to talk about - you can easily get this difference by just running two subsequent tests for exactly the same config..

Yeah, I know, just couldn't resist going for the easy one. 

  It's just another quirk of RMAA?  You'd think that the tests would be more repeatable than that, but they aren't.
 
 
Quote:

^^ I think most people probably use quarter inch jacks if they are going single ended for a desktop amp

Edit: Also can you explain your calculation of the crosstalk? What does the ratio of the load impedance to the output impedance give you? Perhaps also a reference where this formula is detailed?

As for my calculation, it was just a quickie back-of-the-envelope deal (and disclaimer: I normally stay far away from electronics and instrumentation, so I wouldn't be surprised at an error), making the approximation that the load impedance is way higher than the ground impedance.  Hopefully I didn't misremember something or otherwise make several gaffes, so it's good for you to check up on this.  See below diagram, where the left channel is outputting 1V while the right channel is muted at 0V.  In reality some of the resistors below should be impedances and not purely real.
 

 
Consider the impedance of things that both the left and right channel loops share.  This includes the impedance of the ground jack connection  (R_jack_ground, which may be very different than 0.050 ohms).  So include PCB traces and other things in that R_jack_ground as well--it's not just the jack.  There's more or less 1 V across R_L (I did say back-of-the-envelope), resulting in how much signal across R_R?  For 90 dB crosstalk, the voltage across R_R needs to be like 32 uV if the voltage across R_L is 1 V.

 

[bcg27]

Ah yes thanks for the explanation for some reason I was picturing things incorrectly. Anyway here is a picture of how I see the active ground topology: 
[size=10pt][/size]


 

 
 
So a one volt output on one channel with the other channel's inputs grounded sees the output impedance of it's channel in series with the jack impedance and the load impedance. From that point it then sees a parallel impedance network with one branch consisting of the jack impedance and the output impedance of the ground channel, and the other branch consists of the other load impedance, the jack impedance, and the other channels output impedance. 
 
The total impedance seen at the output of the right channel will be:
 

 
 
So I am going to use a quarter inch jack impedance of .015 ohms and the beta 22 quoted output impedance which is less than .01 ohms (so I’ll use .01 ohms), and the specified load of 33 ohms.
 

 
 
So since I=V/R, I=.0303 mA. Just past Z_R we have dropped V=.0303*(.01+.015+33)=.9992 volts. At this point the current will split between the two parallel branches and drop .0008 volts across both. If we take 20*log(1/.0008) we get 61.9 dB.
 
So I think we are more or less in agreement mikeaj. That being said I cannot explain all the RMAA results saying otherwise. Either the calculation is being done incorrectly or something in RMAA is not operating correctly.
 

 

[cheapskateaudio]

I'm feeling good about this amp with the latest upgrades, upgraded PSU (WAU16-400), upgraded caps (Panasonic Polypropylene 2.2uf coupling caps), different op-amps (NJM4562, so smooth), it's getting better with each step.
 
Now I will try using better panasonic electrolytic caps in the power supply and see if that helps things a bit more. Really like the sound right now though, I could live with it at this point, previously, not so much, nice amp, had potential, not quite *there* to my ears.
 
edit: added the panasonic electrolytics in the power supply, same 470uf, but higher end I suppose and boOoOOom, goes the bass... Rhythm went up another notch, coherency up another notch, good upgrade that. Increasing the uf, while cautioned against by NwAvGuy, seems like it might do something positive. Heck if I know why he says not to do that, maybe someone with more circuit knowledge can elucidate. But with how the bass is sounding right now, I WANT MOARRR.
 
By the way, this amp drives the ever loving daylights out of the HD448 (and probably other low impedance headphones), a highly flawed headphone yes, but really quite musical and fun once you get used to it's wacked FR curve.
 
Picture of all the updates so far:

 
Op amps:

 

[deadlylover]

Quote:

I'm feeling good about this amp with the latest upgrades, upgraded PSU (WAU16-400), upgraded caps (Panasonic Polypropylene 2.2uf coupling caps), different op-amps (NJM4562, so smooth), it's getting better with each step.
 
Now I will try using better panasonic electrolytic caps in the power supply and see if that helps things a bit more. Really like the sound right now though, I could live with it at this point, previously, not so much, nice amp, had potential, not quite *there* to my ears.

Aahahah nice to see you having fun there with the modifications, that's true DIY spirit there.
 
I hear V-Caps are all the rage these days?

 

[cheapskateaudio]

 
Quote:

Aahahah nice to see you having fun there with the modifications, that's true DIY spirit there.
 
I hear V-Caps are all the rage these days?

Thats just wrong. But I could drill holes where the 9V go and strap those in with some zip ties and run hookup wire to the board... 

 

[Vladco]

Quote:

I'm feeling good about this amp with the latest upgrades, upgraded PSU (WAU16-400), upgraded caps (Panasonic Polypropylene 2.2uf coupling caps), different op-amps (NJM4562, so smooth), it's getting better with each step.
 
Now I will try using better panasonic electrolytic caps in the power supply and see if that helps things a bit more. Really like the sound right now though, I could live with it at this point, previously, not so much, nice amp, had potential, not quite *there* to my ears.
 
edit: added the panasonic electrolytics in the power supply, same 470uf, but higher end I suppose and boOoOOom, goes the bass... Rhythm went up another notch, coherency up another notch, good upgrade that. Increasing the uf, while cautioned against by NwAvGuy, seems like it might do something positive. Heck if I know why he says not to do that, maybe someone with more circuit knowledge can elucidate. But with how the bass is sounding right now, I WANT MOARRR.
 
By the way, this amp drives the ever loving daylights out of the HD448 (and probably other low impedance headphones), a highly flawed headphone yes, but really quite musical and fun once you get used to it's wacked FR curve.
 
Picture of all the updates so far:

 
Op amps:

Cmon opamp is opamp, power supply is power supply, caps are caps, interconnects are interconnects unless you prove to yourself other wise.
Also you could install small decoupling cap under the board the board on + and - power supply legs of opamps. 10 to 100 nF  It may or may not help.
Vlad

 

[Willakan]

I really, really would avoid touching the output stage. The NE4556 is uniquely suited to its role in that it can sink a helluvalot of current with very little distortion. You are definitely not getting improved performance by replacing them with other DIP-8 opamps - check the datasheets, nothing else comes close.
 
The difficulty with DIY modifications is that it is very difficult to compare before/after due to the time required to switch the parts, making any such impressions incredibly bias-prone.
 
@bcg27:
 
RMAA vs maths. I know which I'd put my money on!

 

[Vladco]

Sorry no one size fit all
Vlad

 

[cheapskateaudio]

Quote:

I really, really would avoid touching the output stage. The NE4556 is uniquely suited to its role in that it can sink a helluvalot of current with very little distortion. You are definitely not getting improved performance by replacing them with other DIP-8 opamps - check the datasheets, nothing else comes close.
 
The difficulty with DIY modifications is that it is very difficult to compare before/after due to the time required to switch the parts, making any such impressions incredibly bias-prone.
 
@bcg27:
 
RMAA vs maths. I know which I'd put my money on!

Listening today the sound is MUCH fuller. I admit, I piled on the mods in short order, so, what caused what is sort of up for grabs, but I noticed a change in the sound with each upgrade. Generally the sound progressively became more filled out with each addition. The opamps were a little weird, the 2134 sounds better and worse, the 2227 sounded flat, the 4562's were my favorite of those 3 by far.
 
So I switched back to the 4556's and the clarity increased but without the harshness now. Sounds damn good. The 4556 is the definite winner with all the updates added, this amp is a beast.  Bass is titanic, freaking amazing bass... 
 
The other op amps are definitely doing some weird things to the sound in this circuit, not all bad though I kind of like the smoothness of the 4562.