by Michael Shelton
[Editor: This article is a compilation of the author’s numerous project postings in the DIY Workshop forum. The projects include portable headphone amps with current-boosted output stages, amps with VU meters, amps in mint box enclosures. Accompanying each project description is a set of links to the original forum threads that introduced the project.
However, these descriptions may include corrections and updates not found in the threads. And there may be other threads not listed here that mention these projects. The reader is encouraged to browse/search the DIY Workshop forum for more information. “Apheared” is the author’s username in the forums.]
Apheared’s 47 Amplifier
[Editor: Apheared’s 47 amplifier has a current-boosted output stage. It developed from his work with the pocket amp by Chu Moy.]
I live in Manhattan on the Lower East Side. Manhattanites are the perfect headphone candidates. It’s all you get to listen to. We don’t HAVE cars. We walk or subway or cab ($$) everywhere. I listen to headphones 4-10 hours everyday. I built the Jasmin Levallois’ version of the pocket amp using the OPA4134 (the quad version of the Burr Brown OPA134), but it got distorted very quickly and had no stomp, even for Grado SR-80s. It just didn’t have enough juice. I fixed that by rebuilding the amp and paralleling the opamps suggested by Skippy and the circuit found in Burr-Brown’s AB-051 application note: Double The Output Current To A Load With The OPA2604 Audio Opamp.
If you use Grados, build this. It is, by far, the best sounding OPA variant I’ve built (I’ve built many, many times many). It is so much better than just a single output stage. It has full, deep, effortless sound and worth that extra chip, believe me. It’s not just for blasting; it changes the sound dramatically for the better at all volume levels. If you want a crossfeed section, add it yourself – after exaustive testing I do not like any of them, at all. It only has a gain of 3. What? Apheared building something with only x3 gain? Yes. It’s intended only for Grados with a line-level source. MAYBE low ohm earbuds, and MAYBE up to 60 ohm, efficient cans. That’s it.
The resistor-divided 9V battery in the original spec it will not power it very well or for very long – maybe 2 hours. Since I am willing to tote around a headphone amp weighing up to a pound, I used two 8-cell AA battery holders (Radio Shack 270-407) to get ±12V (two 8-cell packs will give 200 hours of runtime depending on volume level and headphone load). The PacTec case (Radio Shack 270-214) has room for both battery holders, and with the heavy-duty plastic topped 9V connectors, they fit perfectly. No rattle. I can’t believe how much difference the voltage makes. I think current has something to do with it too, especially using the 9V divided. The battery was hitting it’s limit for getting current to the four opamps, and it was “starving” them.
There is a definate change to the circuit when running the amps parallel. Bass especially. At loud volumes strong bass notes either don’t have the power and kinda go “pluah” and make this muddy fuzz sound, or you just don’t hear them at all. Wired up like this with the volume loud, the same bass note will RATTLE YOUR TOES. However, even when not playing it ludicrously loud, it sounds much better to me anyway. Bass is very tight. Imaging is about the same, but at the volume level that the non-parallel’d amp starts to get that harsh, inside-your-ears ringing brightness (that’s dangerous) the parallel’d one still has full transient punchiness, and beyond.
NOTE: This thing can probably fry your $200 cans easily. It can also seriously hurt you. I’ve measured 2.5V coming outta each side. If I lay the Grados on the couch, and GO OUTSIDE, I can hear them playing through a steel door with ambient outside noise and 20-25′ distance, enough that you can clearly hear the lyrics. With the dual 12V and the parallel opamps, it quintupled the output. At least.
The VU Metered Amplifiers
[Editor: Apheared has built several amps with LED and mechanical VU meters. The National LM3915 driver used in the LED meters has been discontinued and replaced by the LM3915N-1, which is an 18-pin DIP. At the time of this writing, Circuit Specialists still had the older LM3915 in stock for $1.90 each.]
While waiting for more OPAs and other asst. opamps to arrive, I got bored. I got some LM3915 LED VU driver ICs and a few 10-segment low mA LEDs. Wiring them up is pretty simple normally, all LED anodes wire to voltage and all the LED cathodes wire to individual pins on the 3915s. By default the ICs are set for line level audio, just add the resistor of your choice (I used 1.2K ohms) to pick the current for the LEDs and you’re done. I wired them so all 20 LEDs lit plus both ICs takes 18mA.
I will never use perfboard again. Well, at least not the kind without pads. This stuff sucks to solder on, if you make a good electrical connection the component might still be loose because there’s nothing to make it stick to the board. I ended up making it structurally sound by globbing up solder on the leads. Just like those example photos of how NOT to solder.
Luckily I had some 20-pin sockets from another experiment I’m doing, because the channel where the control board would be mounted wouldn’t have made the LEDs even close to flush with the panel. The socket was the perfect height though, the arrays ended up being 1/16″ from the front of the faceplate.
After cutting out a 1×1″ hole in the faceplate carefully (measure twice cut once) I got some smoked gray transparent sticky film from an arts & crafts store. 1 yard of 3′ wide was $1.29. It looks just like film for tinting car windows. It sticks once and that’s it so position carefully! I got it right on the 2nd try and trimmed the edges. Damn, that looks sweet! Total cost, $12.00
I only ran into one problem wiring it up. The volume pot makes the signal drop even though it’s spliced into the line in before the pot. Ugh. Knowing that a diode would fix it, I tried a few different ones and stuck with 4002s. But THEN the diode is dropping voltage and the LEDs are lighting lower. I know I could have solved it mathmatically but ole newbie Apheared did the brute force method: I stuck in my test tone CD, hooked up the voltage reference/bias pin via a pot, and “calibrated” it. Then measured the pot and stuck in the nearest value resistor (1.2K ohms). It’s probably not accurate at all, but hey, it’s just for “ooh aah” factor anyways. I just adjusted the pot until I liked the brightness, and it just ended up being 18mA. I also had to add bypass capacitors the +12V supply. The bottom 2 LEDs kept staying lit. Noise. Just bypassing the power supply got rid of it (or at least lowered it so the LEDs stay off with no signal)
This amp is a current-doubled OPA132. The Middle Atlantic 2U steel box is probably the same one Tim Harrison used for his Szekeres/preamp. What a pain in the ass. I hate steel. Too hard to work with, without having a workshop. And cutting out the hole for the IEC power jack took too much time and blood. I will never, NEVER work with a steel case again.
Throw in a surplus $10 ±15V, 400mA Power One supply. And for kicks, let’s add a few LM3915s (calibrated this time!) and 20 leds on the front panel. Add some nice rack handles and ferrules, an indicator-on power switch, and a pretty knob.
The first LED lights at 50mV and the 10th lights at 1.25V, with the rest 3dB apart. The input is not rectified, just the straight line-level signal, so they flicker with the sine (can only light on positive swing). The LEDs are wired for 30mA each, and will light up a dark room with a beating green glow.
Thanks in part to a certain person’s obsession with locating this case, I have tracked down a fairly easy source for those wanting this Middle Atlantic 2U Steel Chassis – Sweetwater Sound Inc. And the price is good, $34. These are hard to find/get in general. I wish I had known this when I was after one.
They also have the 10″ deep one with an aluminum faceplate, MUCH easier to work with!! I’m building a non-headphone amplifier into one. (but almost double the price @ $64) Trust me, it’s worth it. Less bleeding. I paid $75 for mine, so also good price.
[Editor: The next amplifier with VU meters is based on the class A MOSFET headphone driver by Greg Szekeres.
A DC-coupled Szekeres with an OPA134 gain stage. This is easily my best version too. This one, I think a tear rolled down my cheek… so clean, so gorgeous, so bubbly-hot-tub warm. Like everything, it’s only really time-consuming at the beginning when you’re lost. Once you get comfortable doing it, it can be really fast to build one. An hour or two.
1) Purists, go home. All those that think adding a gain stage using an OPA is degrading, all I can say is “TRY IT.” I have two szekeres here side by side, one gained and one not. Yes, you can hear the OPA, but I bet you’ll like it. OPAs run just fine biased off a single supply like that. There’s still more to go, but it beats a stock non-gain szekeres by 1000%
2) Newbies afraid of the power supply part, take heart! While alot of people are scrounging ebay for Lambdas and other power supplies, surplus ones can work as good as anything you could build. Case in point – this one is powered by an open-frame Power One International Series HC-15-3A, cost me a whopping $18. (even brand new they’re under $50 and easily available) This was new surplus, even had the inspection and testing sheet inside – V: 14.992V, ripple: 0.000V (immeasurable). Beats that $30, 13.8V supply from Radio Shack with a baseball bat. I’m sticking a 2nd one in a box with binding posts as a bench supply.
Ah, what a beast. Without R2/R3 in the DC-coupled gainstage, I’m losing some control here. In the AC coupled designs, adjusting R3 (bias) in the R2/R3 divider lets you set how much DC “piggyback” you want your audio to ride on, which controls how hot the MOSFETs get. Without them, I’m getting 6VDC at the output before capacitors. This same 6V is what’s on top of R4 (quiescent), 6.19V to be exact. So I guess the quiescent resistor is the only way to control dissipated heat in this configuration? I dunno, I don’t have a variety of 15W wirewounds to experiment – I only bought specific values.
The MOSFETs are a bit hotter. I had to swap heatsinks for a solid bar aluminum ones, and it’s all good now. Stabilizes at 64 degrees C/100 degrees F on the chips themselves after warming up 30 minutes. I really should grasp thermic transfer a bit better; I love how much hotter the heatsinks are than the chips/resistors. 88 degrees C/197 degrees F! That’s nifty, ain’t it?
Well, you want it to run hot anyway, just not self-destructive hot. It’s still hotter than a stock value non-gained one. Anyway the heat is within limits and it hasn’t blown a gate in about 40 hours of running and it sounds simply incredible. In a non-vented chassis, it stays at approximately 100 degrees C, which was exactly what I was shooting for. Non-vented enclosures compound the heat issue bigtime though, DON’T use one without knowing exactly how hot it gets in there.
I went a bit nuts on capacitors. In this design, every cap makes an RC filter and I carried it to the Nth degree of overkill. The input is 1uF/.1uF parallel and the output is 2200uF/3.9uF parallel. The only one I didn’t change is the 1uF bias stabilizer for the opamps. And they are all esoterics: FastCaps, WonderCaps, TrueCaps, Muse, Fine Golds. (the pot is the Noble 204Y 25mm.) The coolest thing about capacitors in audio is it’s “tunable” without doing filter circuits, especially the treble. Of the pile of caps I have here, this combo I like alot. I really prefer the direct sound of DC coupling; But this sounds just fine to me!
The meters are Sifam, imported from England. They are true to the BBC spec.: 300ms rise, slow falloff, 1% overshoot. They’re designed for a 600-ohm line and hooked to line level, the needles barely move. So a couple diodes and an opamp boost later I’ve got them isolated and calibrated. O VU is set at 1.228V source signal. This is the proper calibration for dB VU. For volume units, 0dBu is 0.775V (1mW @ 600ohm) but 0dB VU is 1.228V (2.5mW @ 600ohm) or +4 0dBu. Retro and very cool looking – I really like the spade tips on the needles.
The other knobs are for input and record selectors. I think I’ll make yet another FET follower board, this time for a front end of Jan Meier’s crossfeed filter, and use the rotary switches for bass boost and treble cutoff. For now they’re decorative. The knob by the headphone jack was for high/low output impedance; now it’s just an on/off switch for the headphones.
A Dremel tool with a rotary cutter bit is THE tool for cutting out those circular mounting holes for the meters and also the squares for the power rocker and the IEC power jack. It cuts through aluminum like butter. I don’t know if the bit would survive working in steel, but it eats aluminum like nobody’s business. The thing is gorgeous. Thirteen drill holes plus the rectangle power and large round cutouts for the meters – not a scuff or scratch on it. I could slip this into someone’s guitar rack and it might be awhile before they noticed. My most professional looking to date.
You want to know how it sounds? It sounds so good that I don’t mind it not being portable that much. And its not that expensive. Even with the fancy case and meters and EVERYTHING, it’s still under $200.
The Hansen Amplifiers
[Editor: The following amplifiers are built on the Hansen PC boards – a pocket amp variation that incorporates the Meier enhanced-bass crossfeed filter (see the addendum of the pocket amplifier article for more information on the Hansen PC boards). Apheared also shows how to add a current-booster to the Hansen board for driving low impedance headphones like the Grados.]
These things are perfect for fast builds. I did up 3 boards in under an hour, but it took over 3 HOURS to put one of them in the case. Anyone who wants to build cmoy’s pocket amp but doesn’t want to play with layout and protoboard – this is your chance! Get them while they last!
A word of wisdom – they were made for a specific case; USE IT. In the picture above, I am stuffing the boards into Hammond 1593Q cases, which are smaller than the PacTec they were made for. I had to sand down the edges a bit. I got to see the copper layers from the profile, a nice heavy thick layer let me tell you.
When I say it’s smaller, it’s small. On the front panel, check out the LED on the lower left with the leads going *around* the case boss. The pot and the input jack on the other side are right against that boss, and the board fits with 1/8″ or so slack space. Tight fit indeed. The crossfeed setting switch is a multi-pole slide switch hidden in the battery compartment. It’s not something I change much anyway.
If I could just come up with a fast way to case them, I’d build 20 of them ahead of time to waiting friends and family. I’m trying to do them all a little differently since they’re gifts, this one is for my brother.
The next Hansen amp is in a Hammond case: black plastic case with aluminum plates and brass thread inserts, so it will survive repeated openings/closings to change cells every few MONTHS. The board isn’t mounted in place – just the 11 solid core wires from the rotary switch keep in in place. The battery pack has 8 AA batteries for ±6V. Two velcro straps to hold it to a brace/bar that I screwed into the card bosses. Finishing the look are a chrome LED holder, chrome push on/off power, black and gold 1/4″ headphone jack, gold RCA jacks, and black, solid aluminum knobs. Pretty.
This is just a basic amp using the Hansen PCB (no crossfeed), but it sports current-boosting, parallel outputs (like the “47” amplifier) and a 1/4″ headphone jack. There is room in a PacTec HML-9V case.
- Wire the input as you would normally.
- When installing the feedback resistors (R4), only solder the bottom side. Leave them standing vertical.
- From the other side of the feedback resistor location, run a jumper wire to the R11 location that has the trace going to the 2nd amp’s + input. Note orange and brown wires. They are the upper side of R11-R and the lower side of R11-L. You could just jumper across the chip from pin 7 to pin 3, but this is messier.
- Solder the 4 load sharing 47 ohm resistors into S2 OUT R/L and OUT L/R, again standing vertically.
- Install jumper wires to all 6 vertical resistors.
- Bring the 3 left wires to the left output jack pin, the 3 right to the right output jack pin.
Let’s say you like your Hansen-Cmoy and its crossfeed, but it just doesn’t have enough oomph to drive your 12 ohm Sony F1s very well, or small speakers, or any low-ohm thing. This amp will cure what ails you – if you don’t explode something, or cause it to oscillate, or pick up noise, etc. The output stage has 3 paralleled current-boosting opamps. Seriously, this modification is for entertainment purposes ONLY! Anyone that does this and screws up 4 OPAs and wants to complain about it, well, it sucks to be you. DO NOT DO THIS UNLESS YOU KNOW WHAT YOU’RE DOING!
- Start with the (-) input pins (2 and 5). Bend them up 90 degrees, then bend them towards the front.. or more specifically towards the output pins.
- Bend up the output pins (1 and 7) up 90 degrees flat so they touch the bent over (-) input pins.
- Bend (+) input pin 3 up 90 degrees. Bend (+) input pin 5 WAY up, and curl it over the top so the tip is pointing at (+) input pin 3.
- The two pins left alone are the power pins (4 and 8). Align these puppies on top of your already installed chips, and with a HOT iron tack-solder the power pins together. Do it quickly!
- Desolder the R5 jumpers and install wires there on both sides. From the front hole of R5 (beside pin 1 on the opamps), that wire will go to the bent-out pin 3, and then hop to the curled-up pin 5.
- Get your load sharing resistors handy (you’ll need 6). Trim one end closely on 4 of them, and solder one to each (output pin)/(- input pin) junction. Solder the last two into the OUT R and L positions, standing vertically.
- Take the 2 resistors from the rear chip (left) and the wire from the rear R5 location, and tie all these to the top of the OUT L resistor. Do the same for the right.
Your Hansen-Cmoy will now do ±100mA, and you didn’t have to give up your crossfeed, but you just doubled the quiescent current (i.e., halved your battery life). It sounds just dandy.
Tip on using a potentiometer for Hansen PCB. There’s no need to run all those extra wires to the board. It just makes for a tight mess. Wire the inputs to the potentiometer and wire the wipers to the bottom set of holes for the pot location. Wire the input ground and the pot ground in their respective spots. This will also leave free one ground hole, for grounding switches/chassis to the board.
The Mint Box Amplifier
We’ll call this one “Apheared’s C8H10N4O2.” Penguins are my mints of choice, pharmaceutical grade caffeine. I eat them all day, everyday, to cover my double-ristretto, black espressos breath and other assorted caffeinated products. I love the stuff, and Penguins are the BOMB.
It has a dual 9V battery power supply, a single quad opamp (OPA4134) configured with a gain of 16 and current-boosted outputs. The circuit board just a piece of plain protoboard, approx. 30mm x 40mm (smaller than the batteries), has 6 capacitors and the 12 resistors all standing vertically. The local power supply bypasses are probably overkill with the 100nF caps being 15mm from the supply pins. The LED power indicator is mounted on the top cover (in the penguin’s belly). I was going to use a micro 1mm LED for a penguin eyeball, but it didn’t look as cool or as “DIY.” The power switch is a mini toggle at the back of the circuit board.
You can’t use a 3-pad IC protoboard; there’s too much slack space. You have to use a bare protoboard and point-to-point wire it. There’s no “wire” in it; just creative lead bending before soldering. You have to lay it out a few separate times to make sure it’s going to fit before you trim the board, and you have to be exact (~3mm) in your lead trimming. Be prepared to desolder the ENTIRE thing if you mess up; I went through a foot of desolder braid.
The biggest problem wasn’t the board though. It is hard as hell to punch/ream/drill the tin without crushing it. Also the paint job on the disposable candy tin comes off quite easily, if you nick it against something. This one is perfect; you should see the other two that I futzed up first. And, oh, by the way, the tin is conductive. When you stick in batteries to test-fit (they get REAL HOT), keep the battery tabs off the metal (tape over them)! 3M/Scotch 130C Linerless Rubber Splicing Tape. The niftiest stuff I’ve found is this high-voltage splicing tape. It’s 2″ wide and about 1/8″ thick and insulates the inside of the tin. Yes, it’s very expensive. Pulling a piece off the roll requires both hands and a foot.
c. 2001 Michael Shelton.