Balancing Act

[purrin]

When talking about the BA you really need to specify what flavor BA you are talking about.

I've got some vintage Osram PX25s in mine. I've heard other PX25s, PX4s, 300Bs; and I haven't heard a tube that made the BA superior in bass control to something like a Dynahi. Heck, even the Super 7 has tighter more controlled bass than any of these BA configurations. I would have no idea how the 307A's would perform in this regard.

Edited April 14, 2012 by purrin

[morphsci]

The flavors I was referring to are the PX4/300b, 307A and AD1. There are clear differences among those three configurations. The AD1 has remarkable treble extension and the 307A has a tight and clear bass. But I am not a bass head.

[purrin]

LOL, The 307 doesn't count. Aren't there only two in the universe?

[morphsci]

I thought there were 5, but I could be mistaken. I know for sure that there is at least one.

[Voltron]

I bet there are at least 5 of the 307s. I think there are only two AD1s (mine and Dan C's), but I am not sure about that either.

[recstar24]

Is there some secret sauce to optimizing the OPTs to sound good with a wide variety of headphone impedances? I am just wondering since there is only a single tap off the Electraprints in the BA, but some other commercial offerings have multiple output taps off their transformers.

Not an expert, but I believe if the transformer has enough inductance to "swamp" the impedance of whatever it is driving you are typically ok (minus any power transfer issues driving a high impedance can with a low output tap).

[spritzer]

With only one winding then the trick is to pick some compromise which means the tubes never see a load which they don't like. The tubes are also driven well below their limits which helps in this regard. Nothing beats multiple taps though...

Some current assembly pics of the BA:

It's better than the older amps but looks like there is at least one cold solder joint on the tube sockets (top 4 pin with blue wire attached) and he's still using those awful rotary switches. Looks like somebody gave Alpha a call and ordered some 24mm pots...

[pigmode]

The AD1 has remarkable treble extension...

I need to find out about future availability of the AD1 BA.

[grawk]

Call Craig then. Or al.

[dsavitsk]

Is there some secret sauce to optimizing the OPTs to sound good with a wide variety of headphone impedances? I am just wondering since there is only a single tap off the Electraprints in the BA, but some other commercial offerings have multiple output taps off their transformers.

To grossly oversimplify, there are a number of ways to do it. One, as Ryan suggests, is to just have tons of inductance (this is the Lundahl way). Downside is more windings, and possible loss of linearity.

Or, you can put a resistor in parallel with the output. For instance, on the "16 ohm" winding, with a 20R resistor, Grados would appear as 12R, LCD's as 14, Senns as ~18. All of these are likely close enough (like everything, the lams in transformers have a big +/- to their specs and you would generally design for worst case.) The upside is that if the load has impedance peaks, this reduces that effect to an extent (though this is hardly the case with most headphones). The downsides are that, unlike with multiple taps, there is a different power rating into different phones, and like a OTL amp, in order to have the voltage swing for high Z phones, you'll have too much gain for low Z phones. It also often requires a bigger core*, which lowers the quality, and increases the cost of the OPT.

* for a given core size, each wire loop can handle a certain voltage before saturation (audio transformers are rated in voltage swing, not wattage as is commonly used). So, as the voltage needs of the primary increases, the number of windings must increase. For example, to put 1W into 300R phones with a 5K:300 transformer requires a primary voltage swing of 70Vrms. However, to do the same with a 5K:16 transformer, assuming a parallel resistor, the primary voltage swing is ~300Vrms. Thus, not only a bigger transformer with more copper, but also much more stress on the tubes, and forcing them to operate in a less linear range.

Or, you can have multiple taps. The down side to that is it costs more, and makes the winding more complicated, and if there are too many taps, you can start to lose linearity. Where that happens depends on the particulars.

Every designer makes the tradeoffs they think are appropriate, so there is no "right" way to do this.

[deepak]

Thanks guys

[dsavitsk]

The other thing is I noticed is how thin, tightly packed, and precisely cut the laminations are (you can just barely make out the individual laminations in the photo Mike posted.) Would be a serious pain to put together if the E's and I's alternated with each layer, which I believe could be the case.

They are almost surely standard M6 lams with the E's on one side and the I's on the other -- it is a SE amp and the transformers need to be gapped.

[kevin gilmore]

There is a better way to make output transformers. Just like Mcintosh did on their 150 and 300 watt tube amps way back when.

A total of 4 output windings, bifilar wound with the primary. In this case, lets say each winding was designed for 120 ohms.

All 4 windings in parallel, and you have 30 ohms. 2 windings in series in parallel with the other 2 windings and you have 120 ohms.

All 4 in series and you have 480 ohms. Perfect for all the popular headphones out there, no resistor loading needed.

[recstar24]

I believe electraprint may wind their secondaries in a similar fashion - the multiple taps are wired in a way where the whole secondary windings are utilized regardless of which tap you are using. On my old moth s2a3 there was 4 different configurations - 4,8,16,32 (or something like that) - each involved placing different labeled parts of the winding screwed onto various bolts.