[DarthMarth]

I am making my way towards the uber-transparent HD800 and Beta 22 combo. I just ordered the HD800s yesterday and am in the planning stages of the Beta 22. (I have a Crack with Speedball until it's finished) Currently the plan is for a 5-board Beta 22 for balanced or actively grounded single-ended listening. Since this is a few steps above building the Crack, I thought I'd ask some questions for anyone who has built a Beta 22 or is knowledgeable about DIY amps.
 
1. I am unfamiliar with the process of setting up the chassis. I have been looking through the catalogs of the recommended manufacturers (Par Metal appears to be out of commission for now). The logistics of finding a box capable of fitting two Epsilon 22 boards aside, it looks like you're supposed to order the case without any screw holes or other openings for the external parts to come through. Could someone explain the best process for getting the case machined? Are you supposed to be able to do that yourself as well?
2. Does actively grounding the single-ended channel make a perceptible difference?
3. If yes, I'm thinking 5-board instead of 6-board since I don't anticipate any tandem listening sessions. Nonetheless, is there any reason I couldn't wire both single-ended channels to share a single active ground board? What effect would this have?
4. If anyone has a suggestion for a good, highly transparent balanced DAC to eventually go with this setup, state it. Price is not an issue as long as you tell me why it's worth it.

 

[jcx]

active gnd is only useful if you have single polarity supply that needs "splitting"
 
the "3-channel" fad is not technically justified - by engineering theory or measurements
 
the lowest parts count "ultimate" performance B22 configuration for driving the HD800 to >120 dB SPL peaks would be 2 channel, "dual mono" ( 2x isolated +/30 V regulated supplies) and 4-pin "balanced" cable/connector bringing each driver gnd/return to each respective channel amp board
 
4 board "balanced" - really Bridged Output - is not needed for the Vswing with HD800
 
if you want balanced input then drop a few $100 on quality input transformers or even transformer attenuators (which brings it up to $K)
 
 
 
toroid supply xfmr are also faddish - split bobbin EI can have 5-10x less parasitic pri-sec couping C - better isolate from line conducted noise, reduce circulating chassis current in a system
 
use the extra chassis space to keep EI core line transformers away from signal - distance is the cheapest mag field attenuator - add inner perforated iron box enclosing supply for even better mag field shielding

 

[DarthMarth]

Unless you're an engineer in the sound industry, I'm disinclined to take your word over everyone else's.
 
Also, I am in the process of selecting cases. I have heard that aluminum is not as good as steel for blocking EMI from the toroidals. Is this true? I am thinking some Hammond rack-mountable cases, but they are all aluminum.

 

[cfcubed]

Just letting you know I think jcx knows his stuff.   But there is room for personal preference/priorities within the bounds of fact.  For instance I prefer toroidal transformers for their smaller/weaker magnetic field & have come about having 2-box (separating ps/transformer from amp) builds.
 
Think you can find some related discussion here that jcx & myself & others have been involved in that may help your decision process (e.g. do more searching/reading of these & amb forums).  For example my ears, music & opinion has left me with quite happy w/a decent 2-box/2-channel beta22 as best bang for the buck (having built or heard differing 2-, 3- & 4-channel builds).  In my opinion the complexity / size / expense of > 2-channel amps is not cost-effective WRT other equipment in my chain.  Esp. for dynamic cans, think the leap an electrostatic setup may make more sense at that point.
 
But we research, build & form our own opinions about our point on the cost-benefit curve.   Good luck & have fun.

 

[wje]

Quote:

Unless you're an engineer in the sound industry, I'm disinclined to take your word over everyone else's.
 
Also, I am in the process of selecting cases. I have heard that aluminum is not as good as steel for blocking EMI from the toroidals. Is this true? I am thinking some Hammond rack-mountable cases, but they are all aluminum.

 
Why ask for advice, then cut apart the responses that you get?  I'm confused.  

 

[DarthMarth]

I guess I'm confused as to why people in other Beta 22 threads get straight answers and I get answers saying that AMB's design philosophy is mistaken. I plan for this to be the most transparent amp I own in the foreseeable future, and with plenty of time and money but no ability to preview the options for myself, I am inclined to go for the "safer" option that will give the best quality sound. Maybe I fall farther up the cost-benefit curve than you.
 
Also, if building the Beta 22 requires the electrical knowledge that jcx's reply presupposes, I have a lot of learning to do. I thought that the point of "balanced"/bridged audio was to minimize interference, not just to allow a greater VSwing.

 

[tomb]

Quote:

I guess I'm confused as to why people in other Beta 22 threads get straight answers and I get answers saying that AMB's design philosophy is mistaken. I plan for this to be the most transparent amp I own in the foreseeable future, and with plenty of time and money but no ability to preview the options for myself, I am inclined to go for the "safer" option that will give the best quality sound. Maybe I fall farther up the cost-benefit curve than you.
 
Also, if building the Beta 22 requires the electrical knowledge that jcx's reply presupposes, I have a lot of learning to do. I thought that the point of "balanced"/bridged audio was to minimize interference, not just to allow a greater VSwing.

AFAIK, jcx is IS an Engineer in the sound industry.  At the very least, I'm pretty sure his profession is in designing opamps and/or opamp circuits.
 
That said, he has not believed in the ppl (Lisa III)/Morsel (PPA, M3)/Tangent (PPA, PIMETA)/AMB method of active ground channels.  Further, he and AMB have had a bit less of a confrontational position than AMB and NWAVGUY, but not by much.  Regardless - for years in the headphone community, the active ground channel was the ultimate technology for headphone amps and I choose to believe that those people know a few things.
 
From reading AMB's posts for years and others - including on Headwize, if you are going to build a 2 channel B22 vs 3 channel B22, then the 3 channel B22 is definitely called for.  It will sound noticeably better than a 2 channel B22.  You can make a simple test with this using a PIMETA.  Ground the output connector to the common ground and it will short out the active ground circuit.  Insulate the output connector from the common ground and the active channel will be "active."  There's no illusion/brain-washing taking place as to which one sounds better - the active ground channel one.
 
However, if you are contemplating building a 4-channel, balanced B-22, then the benefit is double voltage swing, double slew rate, etc.  Yes, the balanced configuration also tends to cancel out distortion.  But the consensus is that the use of an active ground channel in a balanced configuration may have questionable benefits.  Most people stop at 4-channels, iow.  If I'm not mistaken, attempting to setup the active ground in a balanced configuration may actually be detrimental.  Note that 5-channels is not even a configuration that AMB recommends in his B-22 documentation.
 
I can't give you anymore than that - just the anectodal experience I have in reading years of posts, building many amps (not the B-22, but definitely the PIMETA), and the opinions of others that I trust.

 

[DarthMarth]

Thanks, that backstory would have been helpful. I didn't realize there was such a debate going on. Could anyone explain why actively grounding the balanced channel might be detrimental? What if I only connected the active ground board to a single-ended channel and not to the balanced channel? (It would be kind of cool to build an amp with a bunch of internal switches to allow these different configurations to be easily AB tested)

 

[tomb]

Quote:

Thanks, that backstory would have been helpful. I didn't realize there was such a debate going on. Could anyone explain why actively grounding the balanced channel might be detrimental? What if I only connected the active ground board to a single-ended channel and not to the balanced channel? (It would be kind of cool to build an amp with a bunch of internal switches to allow these different configurations to be easily AB tested)

The simplest answer to your question is to PM AMB - get on his forum and PM him, or send him an e-mail.  There's no one better to answer your questions about how to configure a B22.

 

[n_maher]

Having built, owned and worked on 2, 3 and 4ch versions of the beta22 and having dealt with the issues that all of the different types of builds present I'd be pretty content to build a 2ch version.  But don't bother building any of them unless you're going to do the proper planning to get one built right which is no simple task.  I've heard multiple 2, 3 and 4ch versions that had noise floor issues because of compromised layouts because someone didn't want to address the possibility or rushed through the final wiring.  IMO there is still an issue with the output stage being more susceptible to burning up if not treated with caution.  I'm lazy enough at this point to want an amp that I can swap headphones on without worrying about it releasing magic smoke if I don't turn it off first.

 

[KimLaroux]

I think 5 boards may be overkill, especially if you're gonna use the 5th one _only_ in single-ended mode.
 
If I understand correctly, a 3 channel design is kind of a wanna-be balanced design where you replace both "negative" channels using a single one which "amplifies" the 0V reference voltage. It's just a compromise between 4 channels bridged and 2 channels referred to 0V. If you're already gonna build a 4 channel amplifier, you don't need to add a board to get a 3 channel topology for single ended use. I'm probably just as ignorant as you are, but something tells me you could add something between the two "negative" boards to simulate a 3 channels design. I'm thinking something like a resistor adder.
 
Am I wrong? Anyone ever did that?

 

[cfcubed]

Quote:

<snip> if you are going to build a 2 channel B22 vs 3 channel B22, then the 3 channel B22 is definitely called for.  It will sound noticeably better than a 2 channel B22.  <snip of attempt to correlate this through hack/test in PIMETA design>

 
Unless one has done some sort of A/Bing or can link to reports of some, specific to comparable beta22 builds, that's a pretty strong "statement" for its preconditions.  I note in my related posts the units I've built & heard, most side-by-side in my home (as Nate likely has as well) to form my posted opinions .  E.g. I've always qualified my opinion of any audible difference amongst them with things like "with my ears & my music & my sources" & the conditions of my comparison.  It's anarchy / the wild west around here, esp the last year or two, but think one should still give any qualifiers/disqualifiers WRT apparent/attempted statements of fact.
 
Quote:

  Maybe I fall farther up the cost-benefit curve than you.

 
True & maybe your ears are "better" than my older, somewhat abused ones   E.g. IIRC the person I sold a 4-ch beta22 to had "golden ears", checked annually, and said he could reliably detect the difference between 2-ch & 4-ch Beta22s and the improvement of balanced was worth it to him.  Note there are factors to balanced/unbalanced A/Bing like how the common source's unbalanced output is derived from its balanced, the cabling & some internals of the amps being compared.
In the end I found more impact & potential enjoyment through changing out cans vs. source/amp topology.  E.g. having a good 2-ch Beta22 reference amp allowed me to tour T1s, LCD-2s & (settling on) HE-500s - not having to go balanced with them helped in that process.  BTW you don't need jcx's level of expertise to build these amps, just careful, methodical process & double-checking things should do it.
 
P.S. WRT Nate's magic smoke reference - Haven't had that across my beta22 builds but know that's a far smaller number than yours.   Note though that I've gone to 1R R34/R35, and retrofitted the builds with same, to reduce the possibility.  As Ti says here: "The default value for these resistors is 0.47Ω. If you have TRS headphone output jacks, the value may be increased to 1Ω to 2.2Ω for added protection. This is because TRS plugs create momentary short circuits while being inserted or removed, and could potentially damage the output MOSFETs."

 

[tomb]

Quote:

 
Unless one has done some sort of A/Bing or can link to reports of some, specific to comparable beta22 builds, that's a pretty strong "statement" for its preconditions.  I note in my related posts the units I've built & heard, most side-by-side in my home (as Nate likely has as well) to form my posted opinions .  E.g. I've always qualified my opinion of any audible difference amongst them with things like "with my ears & my music & my sources" & the conditions of my comparison.  It's anarchy / the wild west around here, esp the last year or two, but think one should still give any qualifiers/disqualifiers WRT apparent/attempted statements of fact.
 

I think I qualified that statement with the PIMETA example - that's definitely a first-hand experience that has no comparison in my mind (ears) with 2 channels.  Granted, I haven't listened to that many B22's in a direct comparison and didn't say so.  However, given the experience I've had with other 3-channel designs and AMB's own descriptions of options, it seems to me a reasonable conclusion. The fact remains that AMB steadfastly supports 3-channel designs (along with the other pioneers I mentioned) 
 
If someone wants to build something superior than simply 2-channels (according to the opinion of the designer), then 3-channels is a beneficial upgrade.  That didn't used to be a controversial statement.  Further, if they're willing to go balanced with all the extra requirements that entails (specially wired headphones, balanced sources, special connectors, double cost, etc., etc.), then 4-channels too, is a valid option with tangible benefits.  If they want both, then a 5-channel configuration might be a cost-savings over a fully-duplicated, 6-channel amp.  2-channel, 3-channel, balanced, 6 channels - several tiers of options described and supported by the designer.  Granted, it may not be yours or someone else's cup of tea, but is that really controversial?
 
BTW, I disagree with your "anarchy/wild west around here," comment. (toungue-in-cheek, though) I would tend to classify it as more akin to the Pied Piper, rather than anarchy. Someone blew the magic pipe of extreme measurement and they followed in droves.
(heavily edited) 
 
P.S. Nate will listen to anything.  Look what he has in his signature. 

 

[wakibaki]

I think tomb's characterization of the behaviour as resembling the Pied Piper and his followers contains an element of truth, but it is worth recognizing that there was a different piper before that and probably a different piper before that. At least one of those pipers though that increasing complication (such as introducing a ground channel) would automatically result in an increase in sound quality, but this is certainly not borne out by instrumented test. Of course not everybody believes that instrumented test shows everything about sound quality, but it is suspicious that 3rd. channel amplifiers have markedly worse performance in these situations.
 
You will not need to look far to find another piper who will insist that 'minimalism' is the route to ultimate sound quality.
 
Much of high-end audio is fashion. Many claim to hear improvements, such as balanced over single ended, but there is considerable evidence to suggest that these improvements are more in the mind of the listener than a reflection of the reality of the situation.
 
I like to build equipment that meets a very high technical standard, but this is more for the satisfaction of knowing that it has a very high technical performance than because I think that there is any audible improvement to be obtained.
 
If, like me, you are in the fortunate position that you can afford the cost of pursuing technical excellence, then why not build whatever you want? I would be careful, however, about imagining that the sound will be better than equipment costing much less. A Sansa Clip+ will drive most headphones to quite adequate levels with a sound quality affording no less pleasure than equipment costing 100 times more, if your interest is MUSIC. If your interest is electronic jewellry, one-upmanship or the pursuit of specifications, then by all means pursue your interests, but beware that you do not disguise your activity to yourself by allowing yourself to be seduced by the audiophile mindset. Just be honest. There's nothing wrong with being an equipment buff, but an equipment buff who insists that his behaviour is driven by some esthetic pursuit of ultimate 'sound quality' is a bore and one whose self-deception the very large majority of people see right through.
 
Headphones and speakers are the weakest link in any system, and the ones that show the greatest variation in SQ, but nobody needs to spend a fortune to get something that will equal or outperform the equipment enjoyed by music lovers as little as 10 years ago.
 
If you truly need an amplifier to drive inefficient phones you have bought or are thinking of buying, you could do worse than asking for recommendations as to what will do the job at minimum cost.
 
w

 

[jcx]

I've spent decades designing electronics for Industrial/Scientific Instrumentation professionally - lots of odd transducers, some with cables > 100', some getting to within a few dB of the fundamental physical limits
 
at Foxboro my designs were reviewed by 1/2 dozen other engineers with combined >100 man years experience in analog instrumentation design - my hardware's performance verified by separately managed test group with $ Million labs
 
40 MHz processors, DSP within inches of Av 4000 amplifiers, 16 bit ADC, had to pass independent lab tests for CE EMI susceptibility
 
 
and I pursued audio electronics designing, building my own phono preamp, power amp (with cascoded output Q), built speakers - before graduation
 
probably have read >90% of JAES amplifier electronic design articles from the 70's forward
 
I irregularly attend Boston Audio Society meetings, demos, site tours
 
 
I actually contributed a tiny bit to the B22 design in the headwize thread - suggested bootstrapping the output MOSFET cascodes, unequal gate stoppers for N,P MOSFET, output takeoff & gnd return co-location
 
 
I have given the technical reasoning, some estimates of numbers behind my claim that "3-channel" doesn't do several of the "good" things the proponents claim - and pointed to problems they ignored - and the problems were verified by measurements
 
both the much larger common impedance the TRS connector gnd contact and the cross IMD of some "3-channel" amps have been measured
 
 
balanced signal transfer in audio is good - dual identical amps (4+ board B22) isn't the best way to do it - common mode noise is amplified equally by both amps - it can still interact with the diff mode signal to create IMD
 
for extreme EMI, bad gnd loop environments of touring shows, large studios, pro equipment uses audio line transformers to implement balanced signal I/O - but inside most equipment signals are single ended
 
by using a pro signal level line input transformer to receive, convert to single ended, a consumer level audio signal you get inaudible distortion, "true balanced" input, high EMI and gnd loop immunity
 
a single quality audio line input transformer will cost as much as a single B22 board kit - but you can save the extra cost of a 4 gang input attenuator
 
 
 
I did just review the B22 schematic, ran a quick sim of the output stage with Cordell's hand tweaked LTspice VDMOS models - at the level that would give 100 dB SPL out of the HD800 the output stage distortion was -130 dB @ 1 kHz
distortion is still below -90 dB @ 10 kHz, 120 dB SPL level, 2nd and 3rd nearly same level
I think that suggests a 4 board build's potential even harmonic cancellation can't be audible - since the distortion doesn't clear the noise floor at all day safe listening levels, is thoroughly masked at earsplitting dynamic peak level - your ears will be distorting much more
 
In fact the diodes in the offset adjust circuit contribute more distortion (after allowing for loop gain reducing output distortion)