- 200812 ( 1)
- 200811 ( 2)
- 200809 ( 1)
- 200808 ( 1)
- 200806 ( 2)
- 200805 ( 2)
- 200804 ( 2)
- 200803 ( 5)
- 200802 ( 4)
- 200801 ( 1)
- 200712 ( 5)
- 200711 ( 3)
- 200709 ( 1)
- 200706 ( 1)
- 200704 ( 4)
- 200703 ( 1)
- 200702 ( 3)
- 200701 ( 5)
- 200612 ( 3)
- 200610 ( 2)
- 200609 ( 2)
- 200608 ( 3)
- 200607 (10)
- 200606 (10)
- 200605 ( 9)
- 200604 ( 7)
- 200603 ( 9)
- 200602 (11)
- 200601 (26)
- 200512 (20)
- 200511 (22)
- 200510 (19)
- 200509 (13)
- 200508 ( 2)
- 200507 ( 3)
- 200506 ( 7)
- 200505 (26)
- 200504 ( 9)
- 200503 (15)
- 200502 (20)
- 200501 (28)
- 200412 (16)
- 200411 ( 5)
- 200410 (18)
- 200409 ( 7)
- all
blogroll
Mon, 15 May 2006
business is good. business is booming
I've been doing my best impression of a shiftless layabout for the past week—they comp'ed us a week of vacation after tapeout to thank us for all the overtime we put in, so I've been catching up on life—but I did manage to do a couple interesting things.
First, embarassingly enough, Mike convinced me to start playing City of Villains (remember last summer and City of Heroes? I sure don't), but I've only played a small bit of it thus far. It does seem damn cool, and I'm intrigued by the way they've subtly changed the hero archetypes to make altogether different-feeling characters that are still very familiar in their powers. May, you gonna start playing with me now?
Second, I decided that I'm done "missing out" on that "magic vacuum tube sound," so I've designed and am now in the process of building the amplifier you see to your right. whamp (Wahby's headphone amplifier) is the best name I've come up with thus far, but please feel free to suggest others. Note that a couple years designing circuits for ICs has rendered me completely incapable of anything approaching simplicity. Also, the extent to which an op-amp with a non-inverting second gain stage is a pain in the ass to compensate is left as an exercise to the reader (and then note my solution, a pretty cute one if I do say so myself).
For those of you about to protest, "but that's mostly solid-state!" you're absolutely right, but consider that the majority of the amplification is coming from the vacuum tubes, and thus the majority of the audible distortion (which is what really creates the "magic"). Put another way: input refer all the distortion, and all but the distortion from the 12AX7s is attenuated by the gain of the input diff pair. Thus, the "sound" of the amplifier will be mostly dictated by the input pair. This claim is supported by my simulation results, which show that most of the harmonic distortion is in the even-order overtones, the characteristic distortion profile of a vacuum tube amplifier. By the way, I expect that at reasonable volumes I'll get less than .002% THD into my 250 Ohm beyerdynamic DT-880s, which may actually mean that I've designed it so well that it doesn't sound like a glass amp after all. At least it'll look cool...
Edit: updated the schematic to the latest version, which now uses a slightly different compensation scheme and some snazzier output transistors.
[ permalink | 5 comments ]
writebacks
Gautham wrote
How much output power are sourcing with that beast? Four metal can transistors in your output stage? Damn. I guess it means that you don't have to worry about output short protection much.
Speaking of which, is there any output short protection in that circuit? Can't find any...
Gautham wrote
Oh, duh, those aren't metal-can transistors...I need more sleep...
rsw wrote
Actually, I was going to the use the SOT223 versions of the 2222 and 2907. They dissipate power better than the metal can versions of same (which are available, in fact).
I'm not sourcing much power, really. I was thinking of maybe going with bigger output transistors, but since I'm only putting out a few hundred milliwatts, I figured that I'd rather have smaller, cheaper, faster transistors in the output stage.
The output short protection isn't depicted, but basically the 24 Ohm resistors will really be 18 Ohm resistors and they'll be in series with a shared 2.7 Ohm resistor. The short protection will be a BJT connected from the gate of the output device to the output node with the base on the other side of the 2.7 Ohm resistor.
May wrote
If you don't mind granting me amnesty from that promise, I think I'll stay away from COV for now. My wrists aren't getting any better...maybe because I always neglect to take stretch breaks. If my body weren't breaking down with no hope of me not spending <6 hrs a day on a computer...you'd have a done deal.
rsw wrote
dude
Make intel buy you a kinesis. If you don't, I'm gonna send you one of mine.