Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Itsy Bitsy TPA6120 Amp
New Posts  All Forums:Forum Nav:

Itsy Bitsy TPA6120 Amp - Page 2

post #16 of 41
sweet amp
post #17 of 41
just a curiosity, why the soic TLE2426 and not the TO92 package? ..dB
post #18 of 41
Thread Starter 
Originally Posted by dBel84 View Post
just a curiosity, why the soic TLE2426 and not the TO92 package? ..dB
They take up about the same board area, but the SO8 version has three advantages - it only affects the top layer, it offers use of the noise reduction capacitor, and it didn't undergo a perceived shortage a few months ago . I also happen to prefer working with SMD .
post #19 of 41
Thread Starter 
I should really read datasheets before I choose parts. OPA690 is rated for max 12V operation, so it's out.

I've looked around and I think the THS3091 looks to be a good substitute, it has low noise, high slew rate and can drive 250mA. If you're not comfortable using this chip, OPA551 should be a drop in replacement, but may require lifting/cutting the pin 8 trace pulled high for ~DIS on the OPA690. For now I will change the BOM to OPA551, just note that pin 8 is not compatible with the board. Driving FLAG high probably won't break anything, but the trace should really be cut or pin removed. Once I test the THS3091 and find it stable, I will likely make a permanent change due to much improved specs across the board (except possibly noise). If you do choose to use THS3091, R1 should be increased to 1.78K or more to improve stability.

Images of the board aren't quite accurate, since the layout I sent to CustomPCB was not set up with Olimex's annular ring requirements yet. The biggest change is the TO-92 part, which was previously mounted with pins flat on rather than offset.
post #20 of 41
Thread Starter 
I received the boards today from CustomPCB and somehow there is no ground plane on the bottom layer, so they're unusable . I hope they'll do something to rectify the situation, as it'd really suck to be out $100 on useless boards. There's an error or two of my own (PowerPAD area is solder-masked for example), but nothing as serious as the missing ground plane. They also sheared one of the boards incorrectly. How disappointing!

Here's what they look like anyway, with the incorrectly sheared board:

I suppose I could air wire ground, but then what's the point of having a nice small PCB anyway...
post #21 of 41
May I suggest you try BatchPCB.com, they're a bit slow on the turnaround but they are very flexible with what you give them and will do a single PCB for a very good price and the quality is as good as any other top fab place.

If I had of noticed, I would have told you to steer clear of custompcb.com, my first fabed board were through them and while I didn't have ground plane problems and the PCB's I got actually looked like the files i send, the actual quality of the board was pretty poor. The soldermask wasn't your normal one and when i went to wash the boards after soldering in metho it actually dissolved away the soldermask

Though defiantly get them to redo the lot or a full refund, that is completely unacceptable
post #22 of 41
Thread Starter 
So finally an update on this project. I received the redo (done and shipped at CustomPCB's expense), and it is of acceptable quality. The soldermask is 'weird'; it makes a plastic-y smell if you get it too hot, and it seems very weak as there are some spots on my board where it scratched off rather easily. I wouldn't recommend this fab house, but the boards are working.

Due to the (unknown to me at design time, since it doesn't seem to be in the datasheet) rather large input bias current for this amp, it has a pretty serious DC offset as designed. Swapping 5K resistors for the 100K ground reference resistors in the initial design gets the DC offset down to approximately 20mV, which I find acceptable. Input impedance is definitely on the low side, but both my portables and AlienDAC seem to do just fine driving it.

Summary of issues I ran into building this board:
  • 100K resistors should be replaced with 5K to get DC offset to reasonable levels
  • Drill holes for RK097 pot are too small, the part will fit but must be forced fairly hard. I recommend straightening the crimped pins first or they will likely crumple instead of being forced into the hole. I was able to fit the pot properly without modifying the board.
  • I forgot to remove the soldermask from the ground plane under the thermal pad on the TPA chip; you could scrape this away if you wanted to try to solder it down.
  • Pin 8 on the OPA551 should be removed, if you use this chip

Other than that, I think it sounds pretty fantastic considering the size. It's smaller than the last protoboard CMoy I built was, and sounds far better. I don't detect any hint of oscillation, but I don't have a scope to verify this for certain.

If anyone wants to build this up, PM me. I have a few boards I'll part with at my cost + $1 shipping = $9/board.
post #23 of 41
congrats & glad to hear. do you think that there would be any benefit in buffering the inputs? ..dB
post #24 of 41
Thread Starter 
Definitely, that is IMO the main problem with this amp. Input impedance is far too low to present an easy load to most sources, so a buffer would be beneficial. After reading more on other people using this chip it seems it's the best way to use it if you want to use it with any arbitrary source.
post #25 of 41
Error401, I soldered everything into place. I used very arbitrary resistor values, 1 k everywhere except 2 k as feedback resistor, even from input to ground. Still there's about 30 mV DC-offset at max volume. I've only hooked it up to a 8.4V battery, so I can't really tell how it sounds (but it sounds without any flaws, very clear - maybe overly clear, and the slam/speed/timing is very good) and how it works at a proper supply, just wanted to see that everything was allright.

I tried it first with the TLE as IG and OG = ground channel opamp bypassed. Then I was at ELFA to buy a LMH6642, only to find that they only sell the SOT23, so I tried an OPA690 I thought was fried, but it seems to work allright even though it runs quite warm. Current reading is what's expected. DC-offset is the same. I have a THS4221 to try if the OPA runs too hot at 12 V, but I don't really like the sound of it. I don't think OPA551 is an appropriate match in a high tech amp like this. It sounds slow and mellow. I don't think OPA690 is a very good opamp either, even though it's used in the highly regarded Mini3, because I think my PINT-clone with LMH6654 as ground channel amp sounded better.

I guess a DC-servo is the best option now. There's no room for high quality output caps (and input caps, there's 5 mV at the input), and they would probably degrade the sound more than a servo.

If I get it to work I'll probably box it up in a mint tin with 9 AAA's. 10 won't fit. Or maybe a shortened Hammond J1201 with 10 stripped cells from 9 V's. If I use OPA690 and an OPA2134 for servo, the current draw would be about 35 mA, the use of 250 mA batteries would be OK for semi-portable amp.

Edit: there's no dramatic difference (or really no difference at all at modest listening level with 300 Ohm phones) in sound with 9 or 12 V supply. This is a fine sounding amp, but a bit "overly clear".
post #26 of 41
BTW, if you're planning on using an opamp as input stage with TPA6120 as buffer, you should try ADA4899-1. I can see no point in using some crappy sounding opamp in front of it. It would make your amp become like all of the rest.
post #27 of 41
Thread Starter 
What I ended up doing with my amp was putting an LME49720 unity-gain buffer in front of the amp, and lowered the input resistor to 1k. In my case, this brought the output DC offset to a reasonable couple mA.

I'm using OPA551 in output ground as I'm running my amp at 15V and OPA690 can't handle that voltage. I'd be interested to hear your comments on other ground opamps, but I'm quite happy with OPA551. I'm not that picky when it comes to sonics though...

A DC servo would certainly clear up the problem, and maybe I even have room for one on my buffer stripboard. If you get something going, be sure to post details and maybe I'll try it out too.

Thanks for the comments, and I'd be interested to see if you manage to get a semi portable setup going. Pics!
post #28 of 41

Here are some pictures

More to come if and when I get a servo to work. ELFA doesn't sell SOIC TLE's, that's why I use TO92. You see there are 4.7 Ohm resistors on the output. I'm a cheap bastard, and that's what I had laying around.
post #29 of 41
OK, now it's up and running with a DC-servo, an ugly prototype one. It's an OPA2132 in inverting mode. 0.22 uF ceramic in the feedback loop, because that's what I had at hand (maybe I should make this larger, please help me on this). 470 k from output of TPA connects to inverting in of OPA, and 47 k from OPA out to TPA in. Noninverting input of OPA connects to ground. I changed the 1 k resistor from input to ground to a 10 k resistor to get an acceptable input impedance. The DC-offset is about 2 mV at max volume. The sound seems to be the same.

It takes a little while for the DC to settle, I think this is normal - is it? I've noticed the same thing when using DC-servo in other amps.
post #30 of 41

Oh no ... there's a bug in my amp...

... a Hi-Fi bug.

Now it's only waiting to get boxed up. I haven't decided yet - a pretty C801 or an ugly mint tin with better run time.

It sounds good, very detailed, great slam. It's very low noise as well when using IEM's, but there's better synergy with HD650.

I increased the gain to about 4, still only 2 mV offset.

Edit: I have to listen some more to tell how it really sounds, but I think this a real butt-kicker. I can't understand how you could rank it below Mini3. Try it with OPA690 sometime. I think I have to take back what I said about OPA690 before, but the sound still is slightly edgy (but without harshness and sibilance). The soundstage is great, the clarity is great, the PRAT is great, it's exciting withot being annoying and fatiguing, it's very powerful, it's transparent. Why haven't we seen this opamp in more commercial amps? They're still trying to sell us "new" and "revolutionary" portables with AD823, OPA2134 and other crappy opamps, they should use this one instead. The world is crazy, don't tell me I am!
New Posts  All Forums:Forum Nav:
  Return Home
Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Itsy Bitsy TPA6120 Amp