Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › the Pimeta-V2 thread
New Posts  All Forums:Forum Nav:

the Pimeta-V2 thread - Page 30

post #436 of 627

The ground channel is still there, fighting against the external ground (EG). Worse, you've increased the EG impedance by 6.49k because the IG-OG short comes from after R1G. This puts EG at a severe disadvantage to OG, which has a near-zero output impedance.


If you moved the left side of that short to the other side of the resistor, you'd still have the ground channel components there, fighting against the external ground.


It may be possible to hack the board to physically disconnect the entire ground channel from the circuit at the same time that you connect IG to OG, but I think it would require modifying traces. It's probably even tougher than you initially think due to the internal ground plane.


A middle ground that might work is replacing that SPST with a 3PDT. Use the extra two poles to open/close the R4G and R7G paths. This would still leave you with BUFG's output fighting with EG, but it wouldn't have a feedback loop wrapped around it to drive its impedance low. With the IG connection moved to before R1G, that might be enough to let EG win the fight for control of the OG node.


EDIT: I don't mean to suggest that you go and redo all that wiring to add a switch. You should first try simply desoldering R4G and R7G. This should be easy, since you've used SMT resistors. Open the preamp switch and try it. If it starts working well, you can consider how to proceed.

Edited by tangent - 4/9/12 at 3:11pm
post #437 of 627

Thanks tangent!


Your comments make absolutely sense - I will follow your advices.  I should have invested a bit more time to study the circuit...




post #438 of 627

Some time ago I got myself a Pimeta board thinking I would try going balanced with it. But as I planned and researched, I soon realized that would be above my skill level. Thus far I've mostly just been a paint by numbers DIYer.


Still, I've long been intrigued with the idea of driving my headphones in balanced mode. So here is my attempt at implementing the PIMETA v2 balanced on a perf board. I tried to follow Tangent's schematic, but obviously my layout is different. I've left out the biasing scheme for simplicity (and because I wouldn't know how to do it balanced). I used two sigma25 PSUs to get +/- 15V.


Noble 20K pot

OPA627 x 4

(LMH6321 x 4)

Cthulhu grounding scheme biggrin.gif

Using only a 2 wire power cord for now since that's what came with the "enclosure" I have.



350 LL 350 350 350 350



As you can see, I removed the LMH6321 buffers. I think they were oscillating badly. Horrible sounds with rhythmic clicking and very hot ICs with them installed. My guess is I either screwed something up or they just aren't suited for mounting on an adapter. Maybe I'll try BUF634 instead?


With the buffer positions jumpered, it sounds fine.


I'll post pics again after the casework is done. But I could use some advice on the buffers in the meantime.



ps Never again will I use those SOIC adapters. I thought the holes would aid cooling. But soldering with that layout is a pain.

Edited by Yoga Flame - 4/24/12 at 4:22am
post #439 of 627
Originally Posted by Yoga Flame View Post

Cthulhu grounding scheme biggrin.gif


Oh, is that what it's called? :)



I either screwed something up or they just aren't suited for mounting on an adapter.


It's broader than that. No 110 MHz part is well-suited to such a loose layout.


The more parasitics you introduce, the more problems you will have at such speeds. Yes, the adapters are part of it, but I'd say the wiring and such are a bigger part.



Maybe I'll try BUF634 instead?


You'll have a better chance of success, particularly if you don't run them in high-bandwidth mode.



Never again will I use those SOIC adapters. I thought the holes would aid cooling. But soldering with that layout is a pain.


Another weakness it has is that they had to use extra-short pads in order to fit between the DIP-8 pads, and even then they've created a few risks of shorts. Yes, it's clearer, but there's a reason both Morsel and Brown Dog decided to go with 90 degree variants instead.


I'd check for shorts between neighboring pins.


I think the clicks you're hearing are due to the the buffer shutdown circuitry. It's heating up for some reason (short or oscillation), hitting its thermal limit, shutting down, cooling down, then starting back up. GOTO 10.

post #440 of 627

Thanks, Tangent! I swapped in some BUF634s and it works well now. smily_headphones1.gif And it was a direct replacement too, with no need for wiring changes especially since I'm not using high-bandwidth mode. That was a pleasant surprise.


Casework is turning out to be lots of work, as usual. Don't know when I'll be able to do a proper paint job, so here are some updated pics of my build in working condition.






The board is backwards now due to my lack of foresight regarding the XLR input jacks.


post #441 of 627

Here's my completed PIMETAv2+GrubDAC amp:

ampback of amp

Dog's breakfastBattery meter driver board

Power board.


(Fullsize images also here: amp exterior and interior)



  • PIMETAv2 using AD8620/8610 opamps. Everything's pretty standard.
  • Power supply is 10 x NiMH AAA cells, for a nominal supply voltage of 12V.
  • Input is a GrubDAC mini-USB version.
  • NiMH trickle charger moved to its own board. The battery pack is switched between the charger and the amp, to allow the use of non-isolated switchmode supplies. Rccs on the charger is socketed in case I upgrade the cells to higher capacity.
  • Prototyping area on the PIMETA is used for molex plugs for audio output and power to pass to the battery meter.
  • Battery meter doubles as a power-on LED. LM3914 bargraph driver in dot mode, with the range set to around 9-14V. The first of the 3 red LEDs on the bargraph lights up when the individual cells hit 1V.
  • Added a bunch more rail capacitance on a section of proto-board. If nothing else, this makes the battery meter decay in a cool way on power-off.
  • Black anodised Hammond case, with laser-cut acrylic panels from Ponoko. The washers on the case screws are temporary until I can find some non-countersunk screws. It also appears that the top of the case isn't aligned to the panel. The original panels wrapped around the case, hiding this. A lot depends how it's screwed together.


The interior layout leaves a lot to be desired. I didn't plan it too well and ran out of room. The amp and DAC are screwed down, but the rest of the stuff is jammed in and taped into place. I've held it upside down and shaked it and nothing came loose, but it's very much something that will live on a desk rather than being thrown around in a backpack.


I like the sound using the Senn. HD518s in the picture above, but my Audio Technica ATH-M40fs headphones seem to lack punch at the bass end (I find myself upping bass substantially in the EQ). On the whole I'm very happy how it turned out, despite a number of setbacks in the build (blowing up DAC kits mostly).

Edited by geofftnz - 4/29/12 at 6:28pm
post #442 of 627

Nicely documented, Geoff!


On the bass, the amp is probably just keeping better control of the headphone drivers in bass, giving the impression of "less" bass, when it's just presenting what's in the audio file more accurately. This is one of the reasons bass boost circuits are fairly common around here.


You might also look into Burr-Brown chips. You might like the sound difference.


Can you explain the bargraph driver circuit a bit? I guess it's something fairly standard out of the datasheet, but I suspect there are some component customizations. A resistor divider selection at least?


On the stability of the thing, I'd only worry about the battery pack. Everything else should survive anything that wouldn't put your panels in danger, too. I've used Velcro to hold battery holders inside the case. The industrial strength stuff is significantly better than the normal kind; I had to reinforce the normal Velcro glue with hot glue, since the normal adhesive was the weak point. The nice thing about Velcro is that it automatically pads out the space you have. If you can arrange it so it's anchored top-and-bottom, it'll be a lot more stable.

Edited by tangent - 4/29/12 at 6:58pm
post #443 of 627

Yeah, the "lack of bass" might be that. Those specific headphones are marketed as flat-response phones for mixing and production work, so it's not surprising they don't have the subwoofer-on-your-head marketing many other headphones have.


I do have a couple of Burr-Brown OPA2132PA opamps around (thanks TI sample programme!), so my next PIMETA (I originally ordered 2 boards and have everything for another one bar the buffers) will use BBs. While I was waiting for the panels to be lasered, I turned one of the OPA2132s into a CMoy, with a TLE2426/BUF634 virtual ground:



I do have a source of industrial velcro and am well familiar with its stickiness (having tried to remove the adhesive in the past - can't be done - you end up tearing the structure of the velcro first), so will repack the battery when I get the chance.


The battery meter circuit is the basic one in the datasheet:

from Analog Devices datasheet for LM3914

The circuit I ended up building on the breadboard is written down at home, but the basic principle is that one trimpot is used as a voltage divider to take the ~12V supply voltage down to something usable to feed in as a signal. The other trimpot uses the internal 1.25V reference to set the lower limit of the bargraph. These two settings interact, so it takes a bit of adjusting to get the low-high range you want. Out of the 10 LEDs on the bar, I wanted the top couple to indicate a full battery (1.3-1.4V/cell) and for the 3 red LEDs at the bottom to represent a "charge now" indication at 1.0V/cell.


I'll dig up my bit of paper when I get home and post the circuit. I'd modify it slightly if I did it again because one of the trimpots is only using the last 5% of its travel.


Thanks for the feedback!

post #444 of 627

Here is the circuit I drew for the battery meter. Like I said, if I was doing this again I'd add a resistor in series with the scale set trimpot (on the V+ side) in order to make the pot range more useful.


Both pots are (I think) 50k linear, R2 (LED brightness) is 2.7k, LED power bypass is 4.7uF electro. It draws somewhere around 5-10mA in dot mode (this config). Bargraph mode (pin 9 to V+) will draw more, depending on how many LEDs are lit.


Edited by geofftnz - 4/30/12 at 1:20am
post #445 of 627

.It fits!! It's a tight squeeze but I managed to fit the 60340, and I didn't have to cut the Pimeta board.

I got the Serpac model without any battery holder to minimize any cutting. I still needed to remove most of the ledges at the top and bottom of the cover opening:



Instead of mounting the board in this side of the case, I mounted it in the other side so that the board sits lower in the case 


tenergy 000.jpgtenergy 007.jpg

Originally Posted by mfuerst View Post



I had to go for a 453048 spec, I don't believe the 603040 will fit into the case if you pack 4 cells together with the PCB, the cabling, the insulation, etc.




post #446 of 627

My next Build is to try and squeeze the Pimeta in a Hammond 1455B case which looks a lot better than the Serpac case.



Its about the same size as the Serpac but the height is only 19mm.

The inside of the Hammond has ridges to support a circuit board but can't be used because the board will sit too high to fit potentiometer ( plus some of the traces on the Pimeta are too close to the edges). Power will be supplied by the same 60340 li-ion cell I used before but using 3 cells instead.



 I removed one set of the ridge allowing the Pimeta board to fit.




I was going to place a sheet of mylar between the case and the Pimeta board. 

But since the buffer chips are electrically non-conductive does anyone see a problem with allowing the chips to rest on the bare aluminium case? 



post #447 of 627


Just finished putting together my second Pimeta, this is a gift for my little sister so I used the input caps and current limiting resistors in place. I built this out of spare parts i had laying around. I used an 823 opamp in the LR channel, and a 627 for the ground. She is a bit of a bass head, and I found the 823 to be lively sounding with good low end, so think it will work pretty good. I did have to order caps for the C6 position, as well as another alps POT for the BB. cant wait to get it boxed up and shipped off to her.

post #448 of 627

more pics:IMG_2053.JPGIMG_2054.JPG


Power caps are MUSE, two 330uf and one 470uf. Still need to built a power supply, right now I am running it off of a 12 volt AA battery pack.

post #449 of 627
Originally Posted by H22 View Post

I used an 823 opamp in the LR channel


It's nice to see that op-amp getting a little publicity. Its one serious weakness — low output current — totally doesn't matter in a PIMETA v2.

post #450 of 627

I wish you would quit making that statement without added qualification - at higher supply V the AD823 has more current capability than many "generic" op amps


the data sheet shows multiple tables of specs for a wide range of supply V - if you only read the 5 V spec you get the wrong impression




At TA = 25°C, VS = ±15 V, RL = 2 kΩ to 0 V, unless otherwise noted.



Short-Circuit Current


Sourcing to 0 V






Sinking to 0 V






for driving 300 Ohm cans you could be better off with the bare AD823 from +/-15 V supply without buffers eating V headroom

Edited by jcx - 7/2/12 at 12:35pm
New Posts  All Forums:Forum Nav:
  Return Home
Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › the Pimeta-V2 thread