OPA1688 Super CMOY, 2x 9V with real ground and headphone relay - PCBs

Ever wanted to build a CMOY using modern parts? :D

This project uses johnc124's (TI's) new OPA1688 dual channel headphone driver IC that can source/sink up to 75mA per channel at extremely low THD+N levels, while consuming just 1.6mA idle current per side. FET input and just 0.25mV (250uV!) DC output offset. I've just received the 4th version of the 4-layer board back from fabrication. I'll list those at-cost in the vendor forum in a few days if anyone is interesting in building one up.

The project is intended to be a "mini NwAvGuy O2 headamp" as much as it is a CMOY. To that end it uses two 9V batteries with a *real* ground. No virtual grounds here, no rail splitter chips. And just as with the O2 Headamp to make that work a power management circuit has to be used to cut off the batteries if one battery "disappears" or if both batteries simply get discharged. Surprisingly easy for that to happen with broken wires, intermittant battery snaps, a shorted cell or in the case of lithium "9V" rechargeable cells the on-cell protection circuit doing a low battery cutoff. If one battery suddenly disconnects a large amount of DC would appear at the amplifier output without a protection circuit like this.

I've updated NwAvGuy's O2 Headamp's Power Management circuit with some really cool new parts, the CT128 optical-mosfet solid state relay from Coto Technology (Mouser #816-CT128). The part has an "on" resistance of just 0.05 ohms! That is AC-wired. Here they are DC-wired which further reduces the resistance to a ridiculously-low 0.0125 ohms. Not a typo, 0.025 ohms per SSR mosfet. AC-wired they are in series, but DC-wired they are in parallel. By way of comparison the mosfets used in the NwAvGuy O2 Headamp have "on" resistances of around 0.5 ohms each, 40 times higher.

Even better, since these SSR's are opto-mosfet parts the internal LEDs are strung in series with a current source and the power switch (on the pot), guaranteeing that the positive and negative power rails go up/down at exactly the same time. In the O2 the mosfets were powered separately by comparators leading to time lags between on/off of the pair, resulting in one power rail up while the other was down for an instant (= a DC output transient in the O2 Headamp causing on/off thumps). In this amplifier the SSR LED string grounds by a single comparator, which in turn cuts off if the total battery voltage drops below 14Vdc (7V per battery), exactly like the O2's PM circuit.

And speaking of comparators... this CMOY uses the brand new TI TPS3701DDCT. Look it up, it is an amazing part. Internal 400mV reference, hysteresis, and just 8uA of idle current meaning it would take 30,000 hours to run down a lithium 9V cell. That is part in the photos on the small adaptor board near the battery wires. Originally I had the part on the back of the board, but given the size I found it best to put it on a DIP6 adaptor board. I have the adaptor boards and connection pins at cost, but I'm also having a run of professionally assembled adapter boards done too. With the DIP6 adaptor every part on this CMOY is through-hole for easy assembly, except the OPA1688 chip itself which is a SOIC-8 SMD.

This Super CMOY has other good stuff from the O2 Headamp, including the input RF filter and blocking capacitors to keep any source DC out of the headphones. Preserves that nice 0.25mV or less DC output offset. With the 3.3uF film caps specified the resulting low-end corner frequency is just 0.9Hz (the low end of the frequency response), below the audible range. The amp has battery polarity protection diodes and diode reverse rail clamps.

And of course it includes a headphone relay - same circuit I've used on the O2 Booster Board - to insure zero turn-on or turn-off thumps. :) Works great. In the first couple of fabricated version of this amp I didn't have the relay, just the OPA1688 output directly. The chip does produce some small on/off thumps by itself without the relay circuit. One of the revisions used two more of the opto-mos SSRs on an adaptor board for the muting relay, which worked perfectly, but in the end I didn't have the board space.

The board mounts upside-down in an Altoids tin, sitting on the 3.5mm jack cases and screwed onto the front of the mint tin with the 3.5mm jack nuts and the pot nut. Adydula here on the forum built up the previous V3.0 and did his usual fantastic job of centering and cutting the holes! I'm using a paper punch - goes right through the thin mint tin metal - then enlarging it from there.

The project materials will all be put here over the next few days:


A bunch of spreadsheets that calculate the amount of voltage and current a headphone amplifier needs to supply for various headphones and Sound Pressure Levels can be found here:


Read the "readme" in there for an explanation of the sheet. Worst case, with lithum batteries that are nearly dead (6.4Vdc cutoff) you will get around 4Vrms maximum swing from the Super CMOY. With throw-away or NiMH batteries add a volt to that, around 5Vrms. A single-chip Super CMOY can supply 75mA per channel, while a dual can supply twice that at 150mA per channel. Compare those numbers with the rms voltage swing and rms current your specific headphones need in the sheets to hit 90dB SPL and 110 dB SPL loudness levels.

Click on the arrows in the lower left corner to enlarge any of these photos.


  • schematic.png
    96.2 KB · Views: 6,330
  • schematic.pdf
    33.8 KB · Views: 878
  • layout.png
    764.7 KB · Views: 6,688
  • IMG_4177.JPG
    349.7 KB · Views: 5,448
  • IMG_4178.JPG
    336.5 KB · Views: 5,458
  • IMG_4179.JPG
    327.2 KB · Views: 5,234
  • IMG_4180.JPG
    281 KB · Views: 1,957
  • layout layer 4 route.png
    layout layer 4 route.png
    247.4 KB · Views: 929
  • layout layer 3 Vcc.png
    layout layer 3 Vcc.png
    453.9 KB · Views: 873
  • layout layer 2 Vee.png
    layout layer 2 Vee.png
    434.8 KB · Views: 2,014
Last edited:
Looks interesting, I would like to get one of the boards
I'll send you a PM! It is a fun project. :)

The OPA1688 chip sounds fantastic. I forgot to say in the writeup above that johnc124 and TI did put short circuit protection in the chip, so you are OK plugging and unplugging headphones while it is on (the TRS jacks & plugs always short when plugs go in and out).
BOM, adaptor board, and battery options notes

More project info. :)

Below is the latest Bill Of Materials (BOM), which I have also just posted out at the Google Drive link. The BOM lists both Mouser and Digikey part numbers. I also have added a few Allied Electronics part numbers, but I haven't done an extensive cross reference for Allied.

There a couple of things in the BOM that might be a bit confusing. I have alternate parts highlighted in yellow and optional parts highlighted in green.

One of the alternate parts choices concerns that TPS3701DDCT comparator adaptor board. Again the story there is that chip is **tiny**, a SOT-23-6, but also unique. I've scoured the parts listings for any equivalent that is bigger but there just isn't anything. It is a brand new chip. So... the SOT-23-6 to DIP-6 adaptor board is there to make the thing bigger. Much easier to deal with a DIP-6 package than the tiny SMD SOT-23-6.

The adapter board is something I designed. I tried an off-the-shelf (Aries) adaptor board from Mouser, but it was very wide, wasting space, and used the typical square (non-pluggable) 0.1 center header pins. The one I've made here not only has the chip, but a 0603 sized 0.01uF 100V (the TPS3701 chip can handle 36V) X7R decoupling capacitor which the chip requires. On the back is an unused 0803 pad in parallel with the top 0603 in case someone wants to mess with adding a larger cap in parallel with the 0.1uF. I've created this board to be useful in general, for other stuff. :) Those pluggable pins can plug into a standard DIP-6 IC socket, although the CMOY board holes are sized for them directly to save board space.

So you have two choices. :) You can get the bare adaptor board from me, which comes with the two sets of 3 connection pins at-cost which is $2, and do your own soldering if you feel comfortable with SOT-23-6 parts and the 0603 sized cap. Or the other way is I'll have some pre-soldered and assembled adapter boards available at-cost for $8.50. Pins and parts are already on, just solder it into the 6 DIP holes on the CMOY board and away you go.

HINT: A huge thanks goes out to Adydula here for coming up with an absolutely inspired way to solder that tiny TPS chip, if you should want to do it yourself. Use scotch tape! Works great. I tried picking it up by the sides with tweezers at first, like I do with all SOIC chips, and the tiny chip would just fly off the tweezers 9 times out of 10. I'll bet I have a couple of dozen of those chips in the rug and on the cat by now. But with the tape all is solved. Stick a strip of tape over the top two or so pins, then stick it down over the adapter board, which should be on an anti-static mat for this. Re-stick the tape as needed to line up the 6 leads with the pads, then solder the four free leads. Remove the tape and solder the remaining two leads. Then solder up the 0603 the usual way, but putting solder on one pad, stick in one end of the part, then solder the other end. Then solder up the two rows of connection pins. Best to insert both rows in both the adaptor board and into the CMOY on the other end to get it all lined up.

Another choice on the BOM is the Solid State Relays. I highly recommend the COTO CT128 parts - fantasticlly low on resistance - but I have also listed an alternate less expensive SSR part that has a slightly higher Ron of 0.0625 ohms when DC wired.

The CMOY doesn't charge batteries, so you have the cnoice of either using throw-away (primary) 9V cells or rechargeables that you charge offline in the manufacturer's charger. A plus here is the manufacturer chargers are almost always quick chargers that will charge up in a couple of hours, vs. 14+ hours for the NiMH trickle charger built into the O2 Headamp.

In the BOM I've listed my favorite choice for "9V" (actual 8.4V, 7-cell) NiMH rechargeables, the 300mAhr units from maha energy (Powerex brand), along with their associated quick charger:

Powerex 8.4V 300mAh (1-Pack) - Maha Energy

ALSO! note that you have a neat second choice here. Powerex makes the same thing in a 9.6V (8-cell) 230mAhr version, which works with the same charger:

Powerex 9.6V 230mAh (1-Pack) - Maha Energy

They look identical, so be sure to check the wording on the battery to make sure you use two of the same thing! These 9.6V units are useful if you have headphones that need the extra volt of output swing. Keep in in mind that yet-another plus of the OPA1688 chip is being rail-to-rail output (RRO) so you will get most of that battery voltage swing out.

Then there is my favorite, lithium rechargeable "9V" cells which run at only 7.2V but have a whopping huge 600mAhr capacity. Which means you can run your CMOY all day long for several days between recharges! And when you do rechare the manufacturer's charger is a 2 hour quick charge. These lithium cells contain a protection circuit board inside of them to guard against over-voltage charging and under-voltage over-discharge, just like the standard 18650 size (laptop batteries) have.

All-Battery.com: Tenergy Lithium 9v Rechargeable Battery & Charger Combos

I'm still working on build instructions, but in general solder all the low-height parts first, like resistors, diodes, and the OPA1688 chip. Then solder the medium height parts like ICs (DIP and TO-92), transistors, LEDs, etc. Finally solder the tall parts like the pot and 3.5mm jacks. Make sure you solder the two 1N5818 diodes that go under the TPS3701 adapter board first, before you solder the TPS3701 adapter board on. Also note the LED just fits between the pot and the input jack. Solder the pot first, then the LED, then the input jack. The "power on" LED has two optional positions. Forward it will stick through the panel, in back it won't but you can make a small panel hole to let the light come through.

And, as always, I **highly** recommend doing all your soldering on a grounded anti-static mat, with grounded wrist strap, and ideally a grounded soldering iron tip. There are anti-stat mat and wrist strap kits for cheap out on eBay.


  • OPA1688 CMOY V4.0 BOM 2.zip
    22.5 KB · Views: 395
Last edited:
Wow. There's so much to like here I'm not sure where to start. Well, almost not sure: I'm interested in a board too.

Nice find on the Coto relays. I've been looking at very similar units from Toshiba or Omron, but those cost way more. That power management circuit also looks impressive. I was looking at fiddling with li-ion packs from RC supply places as a power source, and needed something to prevent undervoltage and imbalance. My attempt would need 4 comparators and a bunch of other bits though, and I'm not even sure it would work! I've been dragging my feet over that because of the cost, but I might have a play with the Coto ones.

On that note, the TPS3701 is another nice find. I've been looking at comparators, but TI has that filed under power management. How did you manage to come across it?

And thanks for pointing out those Tenergy 9V cells. A complete package is actually cheaper than the Maha NiMH!

I have all my parts on order and will build this one ASAP!! I built one of AGDR's earlier version and its a very, very nice little amp that drives my Beyer T90's, and AKG's very nicely.

I have compared to other amps as well like the original o2 and AGDR's ODA and inverting version of the O2. This amp with this chip holds its own against them!! Maybe not as versatile as the ODA but sonically its right up there to my ears!!

Revised BOM and new sales thread with prices

I just revised the BOM to take care of a couple things, attached below and revised out on the Google Drive link now:

* I had forgotten to list the 0.1uF 100V X7R 0603 sized SMD capacitor you will need for the TPS3701 adapter board, if you are soldering it up yourself.

* I've now listed alternate values for R4 and R5 if you are using lithium rechargeable batteries rather than throw-away 9V batteries or NiMH "9V" (8.4V nominal) rechargeables. The latter two work well with a minimum discharge votlage of 7.0Vdc per battery, which is what the values of R3, R4, and R5 (battery voltage sampling divider for the power management comparator circuit) were set to. But for the 7.4V nominal lithium rechareable cells a per-cell minimum of 6.26Vdc is better. The lithium batteries have internal chips that will cut off themselves at 6.0Vdc +/-0.1Vdc. The alternate values of R4 and R5 on the revised BOM set the cutoff voltage to the 6.26V per cell.

* I found a typo on the battery section. The Powerx 9.6Vdc 230mAhr battery was still listed as 8.4Vdc.

And I've set up a sales thread in the vendor forum for all my projects on the forum, starting with this one. :) I have the at-cost prices of the CMOY PCB and the TPS3701 adapter PCB, both bare and assembled, over there now:


Here is another brand of "9V" 7.4V nominal 600mAhr lithium ion rechargeable cell and manufacturer's fast charger:


I prefer getting lithium batteries at battery distributors like this, as opposed to generic batteries on Amazon, because they can supply a datasheet for the batteries listing the discharge characteristics. Here is the one for the battery above.:

(opens PDF)


  • OPA1688 CMOY V4.0 BOM 4.zip
    23.3 KB · Views: 179
Last edited:
khd127 & Lucien: thanks for your interest in the project! I have all the (at-cost for forum members) prices listed in the new sales thread now:


If you should decide to get one just send me a PM! I can get an actual-cost shipping quote for outside of the US.

I think I first ran across the TPS3701 when searching for comparators that could handle at least 22V, the maximum rail-to-rail voltage for charged up NiMH "9V" cells. Very few comparators out there that work at higher voltages it seems. The only other somewhat similar unit was from Linear Technology, but it only goes up to 18V. The TPS3701 had the benefit of that 8uA idle current. In NwAvGuy's O2 Headamp he had to put the comparator chips (NJM2903) on the other side of a DPDT power switch. At 0.4mA idle current it would take a bigger dent out of the battery if left connected. So such problem for the TPS3701! A great chip for battery operated stuff.

I had a DPDT toggle switch for power on the first version of this CMOY board - no switch on the pot and no solid state relays. The switch stuck out one side, making for a more difficult assembly. Yet another benefit of the solid state relays is just the one set of SPST contacts on the pot+switch switches both rails at once. The dual pots with SPST switches are a stock item at Mouser, but DPDT switches would have been a special order. The stock is probably because the typical CMOY that uses a virtual ground only needs SPST contacts.

Yeah I designed one of these power management circuits for an initial version of my Inverting O2 Headamp project here (back when the design still included batteries) that used one comparator per battery, rather than across both like this one (and in NwAvGuy O2 headamp). That upped the parts count quite a bit. This circuit simplifies things and still does a good job.

Alex: Hey thanks again for all your efforts in building up the version 3, this version 4, and the feedback along the way! Your idea with the Scotch tape in soldering that TPS3701 made the whole thing possible. :)
Last edited:
Here is another brand of "9V" 7.4V nominal 600mAhr lithium ion rechargeable cell and manufacturer's fast charger:

Rechargeable 9 Volt Lithium Ion Battery: BatteryMart.com

I prefer getting lithium batteries at battery distributors like this, as opposed to generic batteries on Amazon, because they can supply a datasheet for the batteries listing the discharge characteristics. Here is the one for the battery above.:

(opens PDF)

Interestingly, that pdf led me to look up GN Batteries & Electronics Inc., the makers of those cells. Seems GN also makes a 7.4 V 740 mAh LiPo, but it's more than twice the cost of the 600 mAh HiTech Li-Ion. And that in turn is about twice the cost of a similar Tenergy cell... I wonder how much difference there is between the HiTech and Tenergy.
Interestingly, that pdf led me to look up GN Batteries & Electronics Inc., the makers of those cells. Seems GN also makes a 7.4 V 740 mAh LiPo, but it's more than twice the cost of the 600 mAh HiTech Li-Ion. And that in turn is about twice the cost of a similar Tenergy cell... I wonder how much difference there is between the HiTech and Tenergy.

I actually have a set of those "9V" 740mAhr lithium polymers, and they work great! I bought the charger + 2 battery combo last year for the $50:

2 Bank 9 Volt Battery Charger with 2 Batteries: BatteryMart.com



Lithium polymer is supposed to be less susceptable to exploding if abused than lithium ion. Usually - at least for high power LED flashlights, which is another hobby of mine - the LiPo has a lower voltage. I was suprised to see the voltage the same as the lithium ion. Their charger works with both. You are right, the big negative there is price!
johnc124 - thanks! You and TI get the credit on the excellent chip. It really does sound fantastic. If an amp doesn't sound good I'm the first to flag it, but your chip is tops. Really shows that a lot of design thought and testing went into it.

I have a question I've wanted to ask about the preferred voltage gain resistor values. I've kept your 47pF compensation cap from the datasheet, but the BOM default right now is a voltage gain of 2x using 2K for both the feedback resistor and ground return (just the standard non inverting equation of course, [[1+ 2K/2K] = 2]). So that is a 4K total load on the chip output with no headphones attached, which at the battery maximum of 11V peak ("9V" NiMH cells right off the charger) would be 11/4K = 2.75mA, not even a blip on the radar for OPA1688. But the (higher) 2K values should help reduce AC battery drain a bit, although at the cost of a bit of Johnson noise over lower values like 1K's in both positions.

Long story short... what would be your recommendation for resistors for a Vgain of 2x. given the +/-"9V" power rails? :) Then same question for a Vgain of 1 (purely a current buffer), would you recommend leaving off the ground return resistor (R1 in your datasheet schematic) and simply jumpering the feedback resistor (which would short the 47pF) to form a votlage follower, or would it be better to use Rf and instead use some large value resistor (say 10x Rf) for R1 (with a net result of 0.9 voltage gain rather than 1.0). Seems like the later could/would add Johnson noise from the large resistor. And finally same question for a voltage gain of 4x. Any thoughts here are appreciated!

I was looking at the datasheet today yet again and the graph below on the first page just stuck out as a thing of beauty, after a few years now of designing headphone amps! Just look at those *real world* headphone resistances: 16R, 32R & 128R. The typical historical op-amps used in CMOYs have maximum datasheet loads of 2K or more for whatever THD+N numbers are listed, and usually maximum load capacitance of 100pF. They are designed for line-level gain stages, not driving headphones. The OPA1688 is the real thing for driving real-world headphone loads, including the 500pF cable/headphone capacitance. Keep in mind the graph is done with +/-5Vdc power rails too! In this CMOY we have 6.26Vdc - 11Vdc depending on battery type and battery charge levels.


  • OPA1688 datasheet page 1.JPG
    OPA1688 datasheet page 1.JPG
    87.8 KB · Views: 764
Last edited:
TPS3701DDCT adapter board soldering photo-build - part 1

Here is a photo-build showing how to build up the TPS3701DDCT adapter board, if you are soldering it yourself. :)

Mousing over these photos will bring up the file names with the photos numbers I'm referencing below. Clicking on the arrows that show up in the lower left corner of the photos will zoom them up.

* photo 4181 shows the relative sizes of the parts vs. a US dime. Top and bottom of the adapter board, the TPS3701DDCT chip, and the 0.1uF 100V X7R 0603 MLCC chip (Mouser #81-GCM188R72A104KA4D). The blue background is a grounded anti-static mat.

Note the two pin 1 dots on the top (rightmost board in the photo) of the adapter board! There is a small one for the chip near the lower right pad in this photo, then a larger one for the pin 1 connection pin in the lower right corner of the board.

Remember that C2 on the bottom isn't used, at least for the OPA1688 CMOY project here. Those 0805 pads are in parallel with the 0603 capacitor on the top of the board, just for generality if someone wanted to add a larger decoupling cap in parallel.

* 4184 & 4185 show Adydula's tape trick in action. A piece of tape is placed over the top couple of pins, leaving the bottom 2 - 4 pins exposed. Then the chip is picked up with the tape and the tape used to center the chip over the adapter board pads. It only took me 1 additional repositioning this time to get a perfect pin to pad alignment. The first time I did this it took me about 8 un-sticks and re-sticks of the tape on the vise to get the hang of it. :) As with all surface mount parts havng the chip pins *perfectly* align with the pads is a key trick to making the soldering go well. Don't hesitate to re-position the chip as many times as you need to get those leads centered on the pads. I'm not using any additonal flux on the SMD pads here, as I've seen various videos do on the internet. I've never found the need for it. The flux in the solder I'm using has always been sufficient.

* 4188 shows the results of the soldering effort outside of the taped area. Yep, both sides solder bridged. Don't worry about it a bit. You will get that nearly every time due to the close spacing of the chip leads. We will easily take care of it with some solder wick later on.

* 4190 shows the results with the tape removed and the remaining 2 leads under the tape soldered. Lol, now all 3 leads on the left are bridged. Lets take care of that...

*... 4191, 4192, 4193, and 4194 show the solder wick I'm using and the results from putting the end over all 3 bridged leads and briefly heating with the soldering iron. As 4192 shows it really isn't a lot that has to come off. Then repeat with the other side and 4194 shows the now perfectly soldered-up chip.

* photo 4194 above shows something else too. I've put a blob of solder on one of the 0603 capacitor pads. 4195 shows the cap being picked up by the tweezers and 4196 is the result of heating that blob and sticking in one end of the capacitor. After I took this photo I used my finger nail (also using a grounded wrist strap here) to push down on the cap while re-heating the end to get it to lie flat on the board.


  • IMG_4190.JPG
    233.9 KB · Views: 736
  • IMG_4188.JPG
    182.8 KB · Views: 737
  • IMG_4184.JPG
    240.2 KB · Views: 760
  • IMG_4181.JPG
    363 KB · Views: 763
  • IMG_4191.JPG
    233.1 KB · Views: 292
  • IMG_4192.JPG
    186.4 KB · Views: 277
  • IMG_4193.JPG
    339.2 KB · Views: 265
  • IMG_4195.JPG
    312.3 KB · Views: 281
  • IMG_4194.JPG
    215.8 KB · Views: 284
  • IMG_4196.JPG
    226.8 KB · Views: 287
Last edited:
TPS3701DDCT adapter board soldering photo-build - part 2

* 4198 shows the other end of the cap soldered, allowed to cool a bit, then back to the first end to touch it up.

* 4207 shows the connection pin strips for the adapter board placed in their holes on the CMOY board to get things lined up. They are not soldered in yet here. One end of these pins has the plastic bar closer than the other end, and the other end has the short extension piece. In this photo I've place the two strips in both ways to show the difference.

The leftmost strip in this photo is in the "wrong" way, with the plastic bar down toward the CMOY board and the extension pin area sticking up. The rightmost strip is in the "correct" way, with the extension pins down toward the CMOY board. Notice that the adaptor board just fits over the two SB140 schottky diodes used for battery polarity protection. One item in the build instructions is the importance of getting those diodes centered as much as possible over their PC outline. Solder one end of a diode first, check for alignment, then re-heat that one and and move the diode as needed before soldering the other end. The diodes should touch each other as shown to leave maximum space for the adaptor board pins.

BTW the pluggable connection pins are Mill-Max 342-10-164-00-591000 (Mouser #575-6415691 for a strip of 64 that can be cut as needed. I supply them with the adapter board though.

* 4208 now has the adapter board set down over the connection pins. Neither end of either connection pin strip is soldered yet. An important thing to note in this photo is the "pin 1" lettering on the CMOY board near the bottom right connection pin hole for IC1 (IC1 on the CMOY board is the TPS3701 adapter board). That lines up with the pin 1 dot on the adaptor board. An even easier way to check orientation is just make sure the 0603 capacitor on the top of the adapter board is toward the battery wire holes on the CMOY board.

* 4209 shows the adaptor board end of the pins soldered now.

* 4218 and 4219 show the completed adaptor board lifted off the CMOY board. :) The side shot 4219 shows the extension pin end of the connection pins vs. the other end a bit more clearly. I would recommend getting the entire rest of the CMOY board soldered up first before soldering in the adapator board, just to lessen any chance of damage.


  • IMG_4198.JPG
    239.3 KB · Views: 270
  • IMG_4207.JPG
    304.7 KB · Views: 310
  • IMG_4208.JPG
    307.5 KB · Views: 277
  • IMG_4218.JPG
    262.7 KB · Views: 252
  • IMG_4219.JPG
    275.2 KB · Views: 243
Last edited: