Building a headphone amp (LME49990 + LME49600)

[schwat]

1. Preparing the etching solution

   WARNING: Muriatic acid is highly corrosive and WILL damage any metal surface it comes in contact with. It is also dangerous if it comes into contact with skin or if you inhale the fumes. Wear appropriate safety equipment (goggles, gloves, etc) and do this in a well ventilated area away from metal objects you do not wish to potentially destroy.

   The only two ingredients required for the etching solution are muriatic acid (HCl) & hydrogen peroxide (H2O2). The etching solution is simply a 2:1 ratio of H2O2 to HCl. For these boards I started by pouring 2/3rds of a cup of H2O2 into my plastic container & then added 1/3rd cup of HCl. Then I put the lid on and gave it a swirl to mix. At this point the liquid is completely clear.

   

    

2. Etching

   After your solution is ready go ahead and drop in your prepared board being careful not to cause a splash. Almost immediately the liquid will begin to take on a greenish tint as the copper combines with the solution to form cupric chloride which is a very vibrant green color.

   

   After a little bit the solution will be very bright green & you will be able to see the copper starting to disappear first around the edges of the board. You can see that happening to the edges in this picture. Once it gets to this point it starts to go pretty quickly so keep an eye on it. For a board this size with fresh solution it only takes about a total of 5 minutes if even that long. Leaving it in longer than necessary will eventually destroy the copper under the toner.

   

   After the etching is complete toss the board in a bath of water to wash off any excess etching solution. 

   

    

3. Finishing up

   All that's left at this point is removing the toner that is covering up the copper. Using acetone & a paper towel it wipes right off.

   

   All finished up except for drilling holes

   

   You can get pretty impressive resolution using this method. The pads in this picture are for a little SOIC-8 surface mount op-amp. 

   

    

[schwat]

Success! Well...sort of. 

 

Finally got V 0.1 up and running after a little bit of debugging. Had made an error when copying my schematic over from multisim to eagle pcb & switched up my V+ & V- pins on my LME49990s. 

 

Nothing a little bit of swearing & judicious application of an x-acto knife couldn't sort out though. This is why you prototype things

 

Unfortunately I didn't take pictures after completing assembly & initial debugging because my phone battery died but I'll take it back apart tomorrow & take some.

 

 

Here's the pic showing where I screwed up in the schematic:

 

 

 

Edit: Took a pic of the top side at least. Blue wire is the jumper fixing the swapped power rails. 

 

[Lauen]

This is really cool, and makes me want to try it myself, but it isn't very easy getting stuff in Norway.. and I suck at soldering, don't even know where my soldering iron is anymore..

[schwat]

Been working all day on a new layout for v 0.2. Going for a 2 layer board this time with all surface mount components except for the audio jacks & pot. Even found a SMT replacement for the 680nF polypropylene cap in a 1206 package.

 

I significantly improved grounding & was able to move my bypass caps much closer to the power pins. Also after spending some more time re-reading the datasheets for these components I decided to up the bypass capacitance a little bit & went ahead and added some 1uF caps in parallel with my 0.1uF caps.

 

Going to get some 2nd opinions about the new layout from some people at work who actually know what they're doing then after some final tweaks going to print up another batch of boards Monday evening. Should have v0.2 built up completely on Tuesday. Might have a chance to take some very basic measurements for v0.1 but we'll see.

 

Here's what I've got so far:

 

Composite:

 

 

Top layer:

 

 

Bottom layer:

 

[Avratz]

I spy a Rampant IPA.

 

You misspelled Hwat though

 

[schwat]

I think I've got a decent solution for my output protection circuit. This circuit includes a time-delay when the board first powers on before the relay closes but it should open almost instantly upon power off. Will still need to settle on values for timing & for the relay & other components but this is just a kinda proof of concept. Will also need to test it in circuit to see how quick the relay actually opens on power loss because both my rails have a bit of capacitance on them.

 

So the idea behind this circuit is that there is a double-pole double-throw relay that controls signal going out to the headphone jack & will delay connection at first power on as well as disconnect the jack if I lose either rail.

 

Below the two channels of the headphones are being represented by some LEDs that signal when the relays close & there is power going.

 

The blue switches (S3 & S4) are manually controlled and let me cut power to either rail for the purpose of this simulation. When +15V (S3) is connected capacitor C1 is charged through resistor R1. This is my timing circuit & I can change how fast it charges by swapping values on either R1 or C1. I set the voltage C1 has to reach before the relays turn on by a combination of D1 (8V zener diode) &  Q1 & Q4 (2 NPN transistors connected as a darlington pair). The voltage C1 has to reach is a determined by adding the zener voltage of D1 (8V) + the diode drops of Q1 & Q4 (0.7V ea) for a total of 9.4V. 

 

After C1 reaches 9.4V Q1 turns on Q4 which allows current to flow through the relay causing it to close & connect the headphones.

 

The time delay circuit & headphone relay are both powered off my positive rail but to make sure my jack disconnects if I lose my negative rail I put another single pole relay (S5) in between my two transistors & ground. If I lose -15V rail my transistors on the positive rail lose their connection to ground & no current can flow so the relay opens.

 

So should be able to do this circuit with 1 diode, 2 transistors (or 1 darlington pair), 1-3 resistors, 1 cap, 1 DP-DT relay & 1 SP-ST relay. 

 

Anybody think this is a good/bad idea for this circuit? I'd love to hear some opinions from people who are smarter than me

 

[ShearMe]

 I'd love to hear some opinions from people who are smarter than me

 

 

You're the one printing your own circuits.... 

[schwat]

Got some boards etched for V 0.2. Just waiting on my parts from mouser to show up at my desk. Have to love mouser, I placed my order at 6pm yesterday and paid for ground shipping & the parts arrived in receiving at 10am this morning after shipping clear across Texas.

 

So in this layout not only have I separated the current return paths for all my power I have also separated the current return path for small signal (input) & large signal (output). All my power return paths are on the ground plane on the top layer of the board & the signal return paths are split into two ground planes on the bottom with all 3 planes being tied together at the back of the board near where power connects.

 

 

Top side:

 

 

 

Bottom side:

 

 

 

Two completed boards. I made two because I was worried about alignment issues since Im trying to line up all these holes on the top & bottom side. Turned out damn near perfect, can actually see light through the holes which means they're lined up close enough to make work. Still need to drill everything out but way less holes on this version than previous one.

 

 

 

Here's hoping I didn't mix up V+ & V- again I'll post pics when I get this thing all built up later this evening. 

[schwat]

Got V0.2 up and running the other day but just hadn't had time to post any pics. Happy to report I didn't screw anything up in layout this time & it worked as soon as I powered it on. I did notice a slight hissing (white noise) with my more sensitive cans & at first I thought this may have something to do with layout but it was actually being caused by my increased input resistance.

 

When ordering SMT parts I decided to try upping my input resistor in my high-pass filter to 100k which dropped my cutoff freq down to about 2Hz but more importantly reduced phase error & made everything nice and flat. However I also learned all about resistor thermal noise & how it is tied directly to the resistance of a part & is not affected by resistor material. 

 

So it turns out that every resistor generates thermal noise & it is directly proportional to the resistance (bigger resistor = more noise). And upping my input resistor from 30k to 100k raised my resistor noise on the input from about 3.1uV to 5.68 uV, almost doubling it. And since my gain was set so high (arbitrarily picked until I could take measurements) i was multiplying that 5.68uV by 11x giving me 62uV of noise on the output which apparently is audible with my more sensitive headphones (couldn't hear it with my DT990s, but clearly audible with my ATH-M50s.

 

Did a bit of testing to determine a more appropriate gain setting & it looks like I can get away with a gain between 2 & 4 for almost every application, even driving my higher impedance cans off a low power source (like a cell phone). So reducing my gain from 11 to 2 dropped the noise on my output by 5.5x (gain factor) which dropped it below an audible level on my sensitive cans.

 

By reducing my gain I am able to use a larger portion of my input pot (volume control) which means im feeding in a larger input signal to my op-amps which will improve signal-to-noise ratio simply by keeping noise the same & increasing signal amplitude. 

 

I plan on increasing input capacitance from .68uF to 2.2uF & decreasing my input resistance even farther to further reduce noise. As stated it's currently inaudible but I'm sure it's measurable, so once I get my hands on that DSA and I can take some proper measurements I'll be able to do a bit more tweaking. 

 

However so far so good & I'm extremely happy with how this revision turned out after tweaking the gain & dropping the noise. It is definitely meeting my design goal of "sounds good to me" and I don't think I'll be able to do any more improvements to it til I can take some actual measurements. It's already about as good as I can hear.

 

Still have a bit more work to do though. Have to finish deciding values for my output relay circuit as well as mock up my PSU. Got my hands on a 15VAC (RMS) wallwart which is like 21V peak so should be about perfect for testing my rectifier & regulator circuit. 

 

Here are pics of V0.2 completely assembled & put in an enclosure:

 

Top side of PCB: 

 

 

Bottom side of PCB:

 

 

 

Mounted in enclosure. Enclosure is covered in Kapton tape to prevent accidental shorts since this PCB is bare and not covered with soldermask. Kapton tape is like electrical tape on steroids.

 

 

And finally w/ the enclosure all closed up. This case used to house a ethernet switch but is now living a way cooler life

 

[schwat]

Been extremely busy at work lately so haven't had much time to play around with the amp but I've finally started to take some measurements.

 

I found this article about headphone amp measurements from NWAVguy & have been using that as a reference. I should be able to get my hands on the DSA card & take some noise & THD measurements later this week.

 

So far I've only measured the following:

 

DC Offset: <1.5mV

 

This was measured using a calibrated Keithley 34401 6 & 1/2 digit DMM. The amp was powered on and left to warm up for a couple hours during work so everything was completely settled by the time I took the measurement. 

 

Left channel: 1.1016mV

Right channel: 1.2013mV

 

I measured this with no input connected & just for science tried it again with my PC & my phone connected at the input (this should make absolutely no difference due to DC blocking caps on my input)

 

Also planning on implementing DC offset protection with an integrator circuit which would actively monitor my output & adjust accordingly. This will eliminate the need for DC blocking caps on the input. After I get a chance to play around with this circuit some more in simulation I plan on building it up & dropping it onto my V0.2 amp for testing. Hopefully will have enough time for that this week.

 

Also will be interesting to see measurements w/ the DC blocking caps vs the active DC protection.

-----------------------------------------------------------------------------------------

Output Impedance: < 0.1 ohm

 

This was measured following NWAVGuys recommendation of driving the amp with a 100Hz sine wave to an output voltage of 100mV & then connecting a load & measuring the output voltage again then doing some math. To calculate the output impedance you use this formula: Zo=(Rin*(V no load - V load))/Vload.

 

With no load I input a 100Hz sine wave & adjusted my volume knob until I saw ~100mVpk (206mV pk-pk measured). Next I connected a 10 ohm resistor jig I made with a stereo connector & measured the voltage again. This time I got 204mV pk-pk. This 2mV difference is most likely just measurement noise, I will need a nicer scope & a proper signal generator to get a more accurate measurement. However using this measurement we calculate Zo = (10(.206-.204))/.204 = 0.098 ohms. 

 

Since the measurement was made at the load & the cable resistance wasn't taken into account a portion of that 0.098 ohms is cable resistance. Again I'll have to repeat this measurement with a better source & better scope but these measurements are good enough to prove the output impedance is very low. 

-----------------------------------------------------------------------------------------

Maximum output w/ mobile phone input:

 

So to determine the required gain for this amp I have to consider my source & the required power of my headphones. Generally you want the gain to be as low as possible since all your downstream noise gets amplified by your gain, so setting the gain too high = more noise. I currently have my gain set to 3 which gives me a pretty good balance between my high impedance 250 ohm cans & my more sensitive 32 ohm cans. In the final version I plan on having at least 2 selectable gain settings. This will allow for the higher gain needed to drive less sensitive cans & also let you switch to low gain to keep noise down w/ sensitive headphones.

 

Found a frequency generator for my phone, set it to 1kHz & cranked volume as high as it would go. Cranked output on amp as high as it would go & took measurement with a scope.

 

Phone measured 1.8V pk-pk & output measured 5.48 pk-pk. Gain = Vo/Vin = 3.04 

 

 

 

 

So using the measured voltage from my phone as 0.9V pk I did some calculations to figure out gain & output power for my two cans. Top section is for my DT990s & bottom section is for my ATH-M50s. Charts show output voltage, current & power in both pk & RMS for various levels of gain.

 

My DT990s are spec'd as 100mW maximum power rating but doesn't say if this is rms or pk so I'm assuming pk. This means to hit 100mW pk I need to be outputting 5Vpk (3.53Vrms). To accomplish this i would need a gain of 5.5 with an input of 0.9V.

 

 

[MayflowerElectronics]

I had that same scope at school, it was really good. The autoset is awesome for when you want to be lazy

[schwat]

I had that same scope at school, it was really good. The autoset is awesome for when you want to be lazy

 

Yea I love that little scope & its more than enough for most of the testing I do since I mostly work on products that output DC. Plus it's isolated so I don't have to worry about blowing something up if I clip the ground lead in the wrong place

 

I have another one that has probes that cost more than this entire scope but it stays locked up unless I'm doing something like digital signal integrity testing that requires way higher bandwidth. I can just picture the look on our digital engineer's face if I let all the magic smoke out of that one while playing around with my headphone amp....

[Lauen]

so nothing new then schwat ?

[BoredErica]

In case some of you have not seen this and are interested:

http://www.diyaudio.com/forums/headphone-systems/229934-version-o2-desktop-amp-oda.html

[schwat]

so nothing new then schwat ?

 

Recently relocated for work and haven't had much free time so this project kinda got stuck on the back burner. I hope to have more time to start working on this again after end of summer. About the only progress since last update is I swapped the orange led for a blue one lol. 

[Lauen]

Recently relocated for work and haven't had much free time so this project kinda got stuck on the back burner. I hope to have more time to start working on this again after end of summer. About the only progress since last update is I swapped the orange led for a blue one lol. 

ah, I see. Well I'm looking forwards to your next update!