dsavitsk

For this amp, unless someone has a wildly different take on it than me, I think we are going to do away with the WE option. I think it sounds (and measures) worse. But, as for the larger comment, I was leaving biasing out of the article as it was only meant to deal with parafeed specific issues. Maybe I'll do a separate short article on biasing issues, though there seems to be plenty out there on the topic.

up is 300, down is 32. Article updated

This is just symantecs -- I consider the flow of electrons to be -= to the flow of current. In solid state world, it is, as you say, the opposite. Bah -- this is what proofreaders are for -- I said it exactly backwards You can indeed think of it as an inherent resistance in the tube. In phono stages where you need a certain rp for filter purposes, it is common to augment the rp with an additional resistor. That, I'm afraid, it way too big of a topic for here, and something I can't help much with -- you''ll need to read up on transformer theory.

A couple of the questions here suggested that a more thorough explanation of this amp might be in order. So, to that end I wrote up a little primer on parafeed amps for headphones. This is still a bit of a rough draft, but I posted it in case it is helpful to anyone. Comments on where I might clarify things are appreciated. http://www.ecpaudio.com/pdf/parafeed_basics.pdf

Hard to know whether the imbalance will matter. Yours do seem a little more off than most I have used. Ideally, you can find a pair of tubes that are matched for both mu and gm. But, the way to do this is to put them in the circuit, run a test tone through it, and measure the output. A 60Hz tone and a basic multimeter measuring AC is the easiest way to do this.

Regardless of the difficulty of sourcing the transformers, building that amp is difficult. There are lots of "real" obstacles to having it in one's living room. There's a reason there is only one Frank around ...

Simple in terms of parts count, perhaps, but, getting that thing quiet and stable is real work.

Guys -- there is a fair amount of play front to back in the IEC -- the holes in the board are round, but the pins are wider than they are deep. If you are having issues lining things up, you might try desoldering the IEC, then use the cutout back to line things up, then resolder it. As a general rule, if you have multiple board mounted parts to line up on a panel, it is always easier to mount them to the panel first, then solder.

Not surprising, though if they were matched then you might want to check the CCSes or the bias. Jumpers set the bias -- CCS is always on. Gain is what it is. It is slightly adjustable depending on the impedance of the bias and the amount of current, but not enough here that you'd ever notice. I have my doubts that this is appropriate for orthos, though there is no reason to not turn the dial up to 70 or 100%. [ATTACH=CONFIG]3386[/ATTACH]

15 * 24 = 360 You have 110 hours to spare.

Sure. If there is a big signal swing on the grid and a lot of the current is being delivered to the load, the LEDs current should change with signal. I'd think it would need to be really low frequency to see.

Yes, when you put the legs through the holes in the circuit board, melt a little solder on the juncture to hold them in place.

My wife refers to me as the safety inspector ... Hard to imagine that that's heater noise. Worth a try I suppose. Once in the case and with standoffs in all the holes in the center, it is definitely more stable. But, particularly if it grows a little in length, a slightly thicker board might be prudent.

And a ridiculously long amp -- what kind of a jerk would make an amp so long and skinny?

Here are some distortion + noise plots for normal listening levels. The first is from the 32 ohm tap, the second is from the 300 (both loaded with resistors). THD of the first is ~0.144%. THD of the second is about 0.047%. As you push the volume, THD will go up into the 1 to 2% range -- seems to be the limit of the tube. When I get a chance, I'll post distortion at particular power levels. Also, we aren't sure if the 60Hz and other LF noise is magnetic coupling between the PT and OPTs, or if it is PS noise. Probably the former which means probably adding an inch of length to the thing. Oh, most of the the higher (4th and 5th, etc.) harmonic components are an artifact of the sound card.

Even if that part is insulated, there is plenty of stuff around it that isn't. Didn't you just get a nasty shock recently?

Please don't touch high voltage parts.

A quick note, it looks like distortion drops a ton with higher current. I'll post more details later, but upping CCS current to ~18mA (replacing R4R and R4L, the 75R resistor, with a 56R one) seems to drop it by an order of magnitude.

I think we are going to switch to a split case design so that the top just lifts off (rather than slides as with the Hammond) to allow easier tube access. As you note, circuit ground and safety earth are connected together. I kind of like it that way for simplicity. But, we'll have to see if anyone has hum issues. I'd rather not add a toggle for both cost and space reasons if we can avoid it. And, since headphones are on people's heads, in a HV setup, I do feel better with everything connected to earth.

Hammond made these (about) 13 inches long. The original two were too short. This is one of the things that will get worked out in the finals

One thing that I am not sure has been said yet -- the mounting holes are all ground -- make sure that at least one of these is making good contact with the case (with a screw and metal standoff and lock washers) such that the case is well connected to safety ground.

I sure hope not. If that's the case, you got one on the wrong bobbin and it should not fit the board at all. Pin 1 is offset. Yes -- the PCB was designed before we knew which pin would be the shield. So, solder it without connecting it and check for hum. If there is a little, connect it and it should go away (on breadboard, the spacing of the transformers was just enough to eliminate it, but even a 1/2" closer made it audible) -- but, if we can get away without the shield, it saves everyone $6 per kit on the finals All cathodes go toward the middle of the PCB. What, you can't read my mind? The order of B-C-E seems consistent across devices. But, on the SOT-32s, pin one is B while on the TO-225 and TO-126 pin 1 is E. I find figuring out which pin is 1 on these things tough enough, so this extra wrinkle is a real pain. Most likely, we'll supply devices that are labeled. -d

Jack and I are still finalizing designs, so I don't actually have any to send you. And, if it helps relieve the compulsion, a pair will cost more than the rest of the kit -- way more that the rest of the partial kit. Mouser lists 4 different MJE350's -- any of them will work, but check the pinout carefully. There's SOT-32, TO-225, and TO-126 and while the cases are the same, the pinouts are different. Literally anything from 0.1 to 1000u is probably fine. Just find something that fits. We don't care about performance as this is just for the heater supply. These will see full unloaded B+ at startup until the tubes start conducting, so 300V is minimum -- use the spec'd parts here. One thing -- if you have everything else except these, don't sweat it -- it will probably run fine without them.

Taking the signal from the drain of the other jfet would be inverting. This is basically a long tailed pair phase splitter with the inverted output ignored -- which is what the Melos is, too.