| peranders |
I have read the datasheet of TPA6120, slma004 and also slma002 .
Should I have a pad under the IC as described AND must it be connected or can it be left floating?
If I have a groundplane on the solderside, should I have a square hole in it so the normal pads have no groundplane underneath except for those mentioned?
Have I understood it right:
* Hole in groundplane, bigger than the IC
* Power pad on both sides with a couple of vias
* The powerpad connected to -12 V?
I have got samples and really want to test this current feedback amp. Technical data looks promising indeed. :nod:
The reason for I ask is that I have a great respect for 1300 V/us devices. I _may_ make pcb's but I'll start with a CAD design first. This will of cource be a SMD project and I will base it on QRV-04 but I think I'll will have regulators for each channel. This project is only for my own pleasure but I will make a couple of pcb's but this is _maybe_ in a few months. |
|
|
| janneman |
Per-Anders,
From the data sheet:
"Connect to ground. The thermal pad must be soldered down in all Thermal Pad - - applications to properly secure device on the PCB."
and:
"A ground plane should be used on the board to provide a low inductive ground connection. Having a ground
plane underneath traces adds capacitance, so care must be taken when laying out the ground plane on the
underside of the board (assuming a 2-layer board). The ground plane is necessary on the bottom for thermal
reasons. However, certain areas of the ground plane should be left unfilled. The area underneath the device
where the PowerPAD is soldered down should remain, but there should be no ground plane underneath any of
the input and output pins. This places capacitance directly on those pins and leads to oscillation problems. The
underside ground plane should remain unfilled until it crosses the device side of the input resistors and the
output series resistor. Thermal reliefs should be avoided if possible because of the inductance they introduce."
Seems pretty clear to me...
Jan Didden |
|
|
| peranders |
| Jan, I have read this but if you would actually do the pcb, how would it look like? I have some problems to convert this text into a real life pcb. |
|
|
| carlosfm |
Hi P-A,
I have some here too, I'll test them soon.
As a preamp.
Numbers are impressive, but let's see (listen) how it sings.;) |
|
|
| peranders |
| quote: | Originally posted by carlosfm
Numbers are impressive, but let's see (listen) how it sings.;) |
I have good (compared to others!) experience with current feedback amps so I have high hopes on this IC. |
|
|
| janneman |
Per-Anders,
The only part that I think could be misunderstood is this:
"The underside ground plane should remain unfilled until it crosses the device side of the input resistors and the
output series resistor."
I interpret it that you leave the underside ground plane open under the device including the mentioned pins, because that is where you want to avoid capacitance. I would leave it open up to 5mm outside of those pins. I guess it would be OK to fill it under pins 6 to 15 (but also useless I guess). But you are right, the guy could have been a bit more specific.
Jan Didden |
|
|
| carlosfm |
This seams like the trend now at TI.
It's like reading a DIR1703 datasheet.:xeye:
Clear as water.:D |
|
|
| peranders |
One really confusing thing is that it's mentioned that "connect Thermal pad to ground"(page 5 in the datasheet). In the other document it's mentioned that the pad is always connected to most negative potential.
EDIT: This IC has a insulated pad which should be connected to ground. Obviously there are exception for how the PowerPAD is connected and this is not mentioned in the PowerPAD document. End EDIT.
The datasheet doesn't assume single supply. I'll supect a typing error on page 5.
Compare National and their switched regulators. They have very good and clear layout examples. When TI really emphasize the importance of good pcb layout, why don't they show it? |
|
|
| carlosfm |
This is a new product, maby they will release an appnote with PCB layout.
The engineers that are writing that appnote are still decritping the datasheet!:clown: |
|
|
| peranders |
It's indeed :flame: . The datasheet is from March 2004.
I notice also that they are not mentioning input bias current at all?:confused: Page 14 mentions "... with a noninverting gain of -1 V/V". Some typos I'll guess.
What is the max input resistor value for noninverting mode?
Maybe this IC is totally unsuitable for unknown signal sources?
I did a quick measurement and the PowerPAD seems to be isolated from the rest of the chip but reserve myself in this. I may have tested only one polarity with the DMM. |
|
|
| thomas997 |
I just soldered the bottom to a solid copper wire and used it to attach to the perf board.
Chip didnt get hot at all.. maybe it depends on how you run it though.
Had a single 9V supply (wall transformer), and it seemed to clip if I turned my mp3 player to moderate levels..
Used the -2V gain setup.. |
|
|
| peranders |
My questions have been answered here
http://focus.ti.com/lit/ug/slou169/slou169.pdf
Thomas how much offset voltage did you get? Some values are mentioned in the document above but there are no figures in the datasheet (yet).
Do you like it? Does it deliver a "crisp" (according to TI) sound? |
|
|
| carlosfm |
P-A, I have news that this chip is really good.
It's on my list to try soon, after holidays.;) |
|
|
| carlmart |
If this chip is as good as Pavel Dudek suggests, even better than the AD815, then it might be interesting to try it as a line preamp or a buffer.
Soldering is not that easy though.
Carlos |
|
|
| carlosfm |
| quote: | Originally posted by carlmart
If this chip is as good as Pavel Dudek suggests, even better than the AD815, then it might be interesting to try it as a line preamp or a buffer.
Carlos |
That's how I'm going to try.
As a preamp.;) |
|
|
| peranders |
| Carlos, beware of the rather high input bias currents. The seems to be 10 uA! |
|
|
| thomas997 |
| quote: | Originally posted by peranders
Thomas how much offset voltage did you get? Some values are mentioned in the document above but there are no figures in the datasheet (yet).
Do you like it? Does it deliver a "crisp" (according to TI) sound? |
I like it but its not quite powerful enough for my needs.
One measures .10V, the other I guess wrecked it somehow and it measures 1.2V
Didnt really spend a lot of time on them, so there was a bit of interference, but it was very quiet.
Sound was good as far as I could tell.. |
|
|
| carlosfm |
| quote: | Originally posted by thomas997
Had a single 9V supply (wall transformer), and it seemed to clip if I turned my mp3 player to moderate levels..
Used the -2V gain setup.. |
| quote: | Originally posted by thomas997
I like it but its not quite powerful enough for my needs. |
With a single 9V PSU...:rolleyes:
2x gain testing with a portable (usually low output) device, not enough... |
|
|
| peranders |
I'm checking the market for a TPA6120 headphone amp. I will build it myself regardsless of the interst.
http://www4.head-fi.org/forums/showthread.php?t=82139
I'll guess the time schedule is October, November.
Warning: This TPA6120 is nothing for a newbeginner I'm afraid but I may help you to solder just this one. |
|
|
| peranders |
| I sent a message to TI and a real person called me! :bigeyes: Very nice that they pay attention, not like Infineon, nobody at home and when someone answers the telephone everybody else are out...like gone fishing. |
|
|
| carlosfm |
P-A, I received a very nice message on my PM some days ago telling me that TI sells the PCB of the appnote.
In the States it's $49.
That guy bought it, built and reported me the results.
That's why I said I have "news" that this chip is worth a try.:) |
|
|
| peranders |
| quote: | Originally posted by thomas997
I like it but its not quite powerful enough for my needs.
|
How do you mean? What type of headphones do you have? |
|
|
| thomas997 |
| quote: | Originally posted by peranders
How do you mean? What type of headphones do you have? |
Some crappy Sony MDR or something. Not very good.
Its just that the output I have is very low, so I will try increasing the gain and see how it does.
Soldering is a little challenging. If you can find some sort of socket for it, that would help a lot.
Carlos: I tried with 12V as well, seemed to help a little. and yes i need to increase the gain. Just wanted to build it identical to the datasheet first. |
|
|
| peranders |
Thomas, I vote for something isn't otpimized. Weak input signal? Altough with only gain of 2 your ears will hurt and 5-7 Vrms will make your headphones to glow.
An oscilloscope may do some good. Can you drive the amp to clipping for instance. |
|
|
| carlosfm |
| quote: | Originally posted by peranders
Altough with only gain of 2 your ears will hurt and 5-7 Vrms will make your headphones to glow. |
Nah...
With a portable device?
I can bet that the mp3 player doesn't even get to 1V output.
500mv~750mv is common for portables.:dead:
You can't ask much more for a thing that runs with one or two 1.5v batteries... |
|
|
| Upupa Epops |
| Hello again, guys :) . What do you think about this idea : ten of these chips connected in parallel will can easy drive an speaker box and I mean, that it will be more better than any Aleph or Hiraga :D . Any comments ? |
|
|
| Upupa Epops |
| Mea culpa - not Aleph, Zen :clown: . |
|
|
| carlosfm |
| quote: | Originally posted by Upupa Epops
Hello again, guys :) . What do you think about this idea : ten of these chips connected in parallel will can easy drive an speaker box and I mean, that it will be more better than any Aleph or Hiraga :D . Any comments ? |
Mmmmm...:D
Well... if you parallel another 10 and bridge 10+10, we may be talking more serious business.:p |
|
|
| peranders |
| Carlos, I ment of course that the gain must be adjust for the application. If thomas has a weak signal he must increase the gain which is done very easily. |
|
|
| carlosfm |
| quote: | Originally posted by peranders
Carlos, I ment of course that the gain must be adjust for the application. If thomas has a weak signal he must increase the gain which is done very easily. |
That's what I meant on a previous post.
2x gain is low, as he is testing with an mp3 player.
If the intention is to make a headamp for portables, gain should be much higher.
But the feedback resistor must remain as per the manufacturer's recommendation. |
|
|
| peranders |
| quote: | Originally posted by carlosfm
But the feedback resistor must remain as per the manufacturer's recommendation. |
Indeed, becasue this is a current feedback amp where you also control the speed with the feedback resistor. |
|
|
| ericallan |
I cut a small rectangle of copper out - the width of the chip - heated a blob of solder on the centre with a gas torch, then quickly stuck the chip on the molten solder. I then hard wired leads to the active legs of the chip and made up the circuit as per the basic evaluation model.
It sounds very good - my friend has a Chiarra Single Ended Class A Headphone Amp - and he was a bit sickened to hear this almost free amp. |
|
|
| carlosfm |
| quote: | Originally posted by ericallan
It sounds very good - my friend has a Chiarra Single Ended Class A Headphone Amp - and he was a bit sickened to hear this almost free amp. |
Wednsday I'm going on holidays for a week.
Don't make me take the soldering iron to the beach, please.:clown: |
|
|
| Upupa Epops |
| How you can endure it - you must have realy big other inspirations, probably with nice t... :D ! |
|
|
| carlosfm |
| quote: | Originally posted by Upupa Epops
How you can endure it - you must have realy big other inspirations, probably with nice t... :D ! |
Ah yes... forget the soldering iron...:D |
|
|
| Devil_H@ck |
| So, how would one solder this chip to a PCB? Would applying solder to the copper, then putting the chip on & then warming up the opposite side of the PCB work? |
|
|
| peranders |
Today I have started to convert my QRV04 headphone amp to AD8620 and TPA6120. I have great expectations in this IC.
The pcb will be easy to solder, even the "PowerPAD".
I haven't desided yet if the TPA6120 should run in inverting mode or not. |
|
|
| Upupa Epops |
| Matjans, problem is soldering of POWERPAD, which is something quite different;) . |
|
|
| peranders |
| I haven't tried yet but I think it will go just fine if I just burn thourougly on the "solderside" with the soldering iron. This asumes a pcb with lot's of vias. My pcb will have 18 vias under the metal area. |
|
|
| Upupa Epops |
| Per, I mean that it will be not so easy ( I have not experience with it ), 'cos must be used soldering paste for soldering of pad. Device must be " preheated " in this case by hot air and all thrue holes must have correct diameter - soldering of SMD devices is " big science " ;) . |
|
|
| Nicke |
| Soldering paste,a good flux and a hot air device is all that is needed,not that difficult..BUT perhaps one should practice some first.... |
|
|
| Upupa Epops |
| To Nicke : For profi working place is it certaincly no problem, but we are talking about home work :xeye: . |
|
|
| Nicke |
Upupa Epops:I see your point
We use this at work:Hot air soldering mashine
But what I meant was that it is not THAT difficult.
I wouldnŽt have any problems using a regular hot air tool(donŽt know what it is called in english,"varmluftspistol" in swedish) at home.
What matters is that you should NOT try this without getting used to this and try a it with some junk components first.
And not to heat the components too long... |
|
|
| Upupa Epops |
| Yes, no problem with this machine ;) , except price for this :( . |
|
|
| Nicke |
| quote: | Originally posted by Upupa Epops
Yes, no problem with this machine ;) , except price for this :( . |
IŽll agree...
But it wouldnŽt be that difficult to use a regular hot air gun like this:
http://www.tumstock.se/itemimages/full/0102434.gif
It isnŽt the first thing a beginner in soldering should try...
I would not hesitate,but IŽm used to solder things like this ;) |
|
|
| matjans |
oops my mistake... finally read the datasheet.
it's a useful link anyways... :smash: |
|
|
| Upupa Epops |
| Per, you are strictly using separate PS for each channel - what you will be doing in this case ? This circuit have seemingly separated feeding, but biasing circuity are common. Will you use parallel connection of both channels and two chips ? |
|
|
| peranders |
I have not yet desided if I'm going to use a single or a dual supply. At least I have desided that it wll be an non-inverting buffer with gain and the TPA6120 will go a inverting amp with gain of 1.
I plan to use one AD8620 and one TPA6120. |
|
|
| ericallan |
I like the idea of soldering a copper wire, perhaps mains earth wire, to the pad as a solution. The chip doesnt get very warm at all and I've run it quite loud for extended periods and left it on for hours. I think the important thing is to earth the pad and give at least some conductance of heat. I'm going to build another to go inside an integrated chip amp I've got planned.
I've got some OPA4134 opamps as used in the suggested application on the pdf file - might try to incorporate those as well -any thoughts?
For the power supply, I used a transformer with twin secondaries and fed the outputs into 2 seperate regulators. |
|
|
| carlosfm |
| quote: | Originally posted by peranders
I have not yet desided if I'm going to use a single or a dual supply. At least I have desided that it wll be an non-inverting buffer with gain and the TPA6120 will go a inverting amp with gain of 1.
I plan to use one AD8620 and one TPA6120. |
P-A, why don't you use the TPA6120 alone, NI?
That way you will evaluate this chip.
An AD8620 before it?
Why?:confused: |
|
|
| peranders |
Good question!
The TPA6120 has rather large bias currents which can cause trouble when the signal source is not known and/or you have AC-coupled input. Therefore to eliminate this I chose to have a buffer and when I have it I chose also to have an inverting configuration of the TPA6120 since I have good drive capability. It's not impossible to modify the pcb into a non-inverting, non-buffered version.
The group buy interest at head-fi.org is zero (to hard to build?) but if anyone here feel this headphone amp could be interesting, send me a message and maybe I will set up a group buy.
It's not ruled out that I can solder the TPA6120 but I won't promise anything. |
|
|
| mandat |
What do you say about OPA1632 opamp (marvellous THD and SR specification) as a buffer for TPPA6120 or THS6012?
Yes, these ICs are very fast and DIYer should take a great care during PCB layout design and power supply decoupling. |
|
|
| peranders |
| It may be a good choice but I will concentrate myself on AD8610, just to get at least one "approved" design if you know what I mean. |
|
|
| ericallan |
Peranders - is the OPA4134 opamp a bad choice? TI promise 120db dynamic range using the OPA4134 and TPA6120 together. Have you looked at the Evaluation Model pdf document? I got more ideas from that than the datasheet.
http://www-s.ti.com/sc/psheets/slou169/slou169.pdf |
|
|
| peranders |
Only a single opamp will fit but OPA134/2134/4134 doesn't match the TPA6120 IMHO, you need one the sharpest opamp around for this.
Compare also that AKM and others recommend NE5532 as the best choice for their DAC's! :bigeyes: |
|
|
| peranders |
| Now... is the layout ready. The pcb will have gorundplanes on both sides. More info will come in a few hours or at least this weekend. |
|
|
| carlosfm |
P-A, if you use a buffer, you can use it before the volume pot.
And the TPA in NI mode.
As for the solder pad...
:bulb: why not use a big via (around the size of the pad) on the exact place, to fill up with solder?:bulb:
I can't remember an easyest way to do it.
It should work fine.:angel: |
|
|
| peranders |
I have chosen inverting mode just to get ultra extremely low distortion :)
The PowerPad is made according to recommendations, 18 vias under the PowerPAD. I have also thought about a big hole under the IC. |
|
|
| Upupa Epops |
| Carlos, peranders is professional, not amateur :) - by soldering with paste and hot air by big hole paste flow away. |
|
|
| carlosfm |
I meant a big via almost the size of the pad, and fill it up with solder with... the soldering iron.:D
Of course P-A is doing it the recommended way, but I just gave a suggestion to do it simple. |
|
|
| peranders |
| I understand what you mean Carlos but I want to keep the pcb solderable for wave soldering... just in case. Big plated holes aren't so good when it comes to wave soldering. I think it possible to solder by hand with those 18 small vias. Reports about this will come :nod: |
|
|
| carlosfm |
| quote: | Originally posted by peranders
Reports about this will come :nod: |
Indeed.
I'm making a prototype on veroboard.:D
Listening tests when finished.
I've just started.
Let's see (listen) if :up: or :down: .
:cool: |
|
|
| Upupa Epops |
| If resuts will be not OK, is here big field for shifting this chip to class A :D . |
|
|
| carlosfm |
| quote: | Originally posted by Upupa Epops
If resuts will be not OK, is here big field for shifting this chip to class A :D . |
:D
With a switch to turn biasing on and off.:cool:
Mmmmm... 100ma?:hot: |
|
|
| fit |
| I am currently doing similar project but as buffer I am planning use OPA627 as e buffer - and no caps in signal path... I just wonder why are you using two resistors in PS Reg ? |
|
|
| Upupa Epops |
| Precision PCB - I am applauding :cool: ! Probably it will give too high sound quality, so you must give in front off some " glass " for humanize of sound :D . |
|
|
| Upupa Epops |
| 627 on output ? |
|
|
| carlmart |
| quote: | Originally posted by peranders
The layout is ready now. Notice that the groundplane isn't shown, which are on both sides.
|
That means the so much discussed copper heatsink is part of the groundplane?
A very very minor comment on the regulator output capacitor. It's been some years since I have been using smaller caps at the output of 3X7 and similar regulators, like 100uF or so. Sometimes split 47uF and 47uF between reg output and chip input. Audio quality seems to improve with small caps there, but I am not quite sure why.
So perhaps it might be interesting trying a smaller cap than 470uF at your reg outputs. Instead it pays off putting a larger cap between adj pin and earth.
This is certainly a project to try with batteries too.
| quote: |
A group buy? Anybody interested? This question was put already at head-fi.org but noone seems to dare to build it |
Soldering the heatsink is certainly a concern on this DIY project. But Upupa Epops said pins 6 to 15 might be heat-sinking pins, so maybe joining all those pins and the copper surface below the chip, pre-covering it with a thin solder layer and when applying heat with a solder pen on the pins to solder them that would also solder the IC below pad. Just a thought.
Interesting: I was also curious about the gain you used, but it seems TI indirectly seems to recommend no gain on this stage. It looks like a buffer, as the gain is set by the previous stage. Perhaps I should say a differential buffer, as that is what the TA seems to be. Pity TI rarely includes an equivalent schematic on their datasheets, as National usually does.
Carlos |
|
|
| carlmart |
| quote: | Originally posted by peranders
This question was put already at head-fi.org but noone seems to dare to build it :D |
Perhaps because it's a project involving only SMD parts? And many of them?
If you re-design it using regular size parts, leaving only the audio ICs to deal with, there might be some more interest. It would be a more DIY version. That's what Pavel does.
Carlos |
|
|
| fit |
| quote: | Originally posted by Upupa Epops
627 on output ? |
no - at the first stage where Per is using AD8610 - OPA627 is stable at gain=1
(I have PCB with SOIC8 mask so I can use any opamp :)) |
|
|
| carlosfm |
| quote: | Originally posted by carlmart
It's been some years since I have been using smaller caps at the output of 3X7 and similar regulators, like 100uF or so. Sometimes split 47uF and 47uF between reg output and chip input. Audio quality seems to improve with small caps there, but I am not quite sure why. |
Actually I would (and do) use just a 100nf cap.:att'n:
And electrolythics+ceramics on the chip.
Not big electrolythics on the reg's output and just 100nf on the chip.
I'm using 68uf+0.1uf on each voltage pin on the chip.
P-A, I'm makin' it, and I'm using small 470uH chokes (instead of resistors) on each voltage, for each pin, to isolate, because I have a LM317/337 PSU for both channels.
I will test this as a line pre, not a headphone amp.
The two applications have different implementation, mainly the feedback resitor has a different recommended value.
The recommended feedback resistor is even different for low or high impedance headphones! :att'n:
I'm becoming a specialist in decoding TI datasheets, one needs to read between the lines!:dodgy: :D
Almost done, but I have to go out now.
Gonna do a listening session with my phono pre and a Linn LP12.:cool:
Later, |
|
|
| carlosfm |
P-A, I've just noticed that you don't use the Cadj cap.
Any reason not to include it? |
|
|
| peranders |
| quote: | Originally posted by fit
I am currently doing similar project but as buffer I am planning use OPA627 as e buffer - and no caps in signal path... I just wonder why are you using two resistors in PS Reg ? |
You don't have to use caps in my design, just omit them if you don't want them.
You mean two resistors in parallel? This was just for convinience. |
|
|
| peranders |
| quote: | Originally posted by carlosfm
P-A, I've just noticed that you don't use the Cadj cap.
Any reason not to include it? |
No not really but a have larger output caps instead. |
|
|
| peranders |
| quote: | Originally posted by carlmart
Perhaps because it's a project involving only SMD parts? And many of them?
If you re-design it using regular size parts, leaving only the audio ICs to deal with, there might be some more interest. It would be a more DIY version.
|
I like SMD simply and it's fun... and not so hard to solder either.
Anyway, anybody here interested in a group buy? |
|
|
| Devil_H@ck |
| quote: | Originally posted by peranders
No not really but a have larger output caps instead. |
Adding Cadj gives a lot beter regulation, try a 220”F cap. |
|
|
| carlosfm |
| quote: | Originally posted by peranders
No not really but a have larger output caps instead. |
It's not the same thing with these regs...
| quote: | Originally posted by Devil_H@ck
Adding Cadj gives a lot beter regulation, try a 220”F cap. |
That's it.
More than 80db PSRR.
But 220uf seams too much...
I normally use 47uf.
On this pre I'm using 68uf.
With more than 10uf the protection diode is needed.
I use both protection diodes.
Oh, btw, the board is just finished.
Gotta test it.
:att'n: Advise: This is not really like some op-amps (627 included, and most fet-input op-amps), you need to use output coupling caps with this chip.:att'n:
I have around 60mv DC offset on the outputs.
Not much, but I decided to put the caps in.
Let's see what this delivers as a pre.
It's late now, so I can't crank up the volume.
More tests tomorrow.
I'll leave it :hot: in' in.:D |
|
|
| Devil_H@ck |
| C'mon, show us some sexy SMD PCB pictures :). |
|
|
| ericallan |
| quote: | | Anyway, anybody here interested in a group buy? |
Just me it seems. I think the soldering will put people off, I've done it but my eyes were crossed afterwards. The powerpad bit needs to be made easy for group buyers. |
|
|
| fit |
in PSU (317/337) I use 100uF Cadj (Nichicon VZ) and main cap is 6800uF Chemicon LXZ (+2 protection diodes) and 33uF Oscon + 100nf as decouplers - I am going to try both opa627 & OPA 1632 as buffer - still have hope no caps in signal path
going back to soldering - friend of mine have nice airheater :D so I asked him to do the job - these SMD parts makes me crazy... |
|
|
| carlmart |
| quote: | Originally posted by ericallan
Just me it seems. I think the soldering will put people off, |
Not the soldering. The SMDs.
Carlos |
|
|
| carlosfm |
It's playing.
Low volume, it's late, wife and kid are sleeping.
First impressions... this seams really special.:nod:
Exciting.:cheerful:
Everything seams very natural.
Widely separated, space, detail, smooth, dynamics, bass "slam" (I love this word:D ).
Let it :hot: -in, tomorrow I'll listen more carefuly.
It's very promising.:happy1:
I'm writing this on the next room to my main system and the sound is invading me.
Excuse me, gotta go!:spin: |
|
|
| Devil_H@ck |
| quote: | Originally posted by peranders
Ok, I have changed the design a bit, added Cadj but I choose 100 nF. |
You should really make that bigger. Try 22”F. I'm searching for a test that shows the influence of different sized caps in the Cadj position, I'll post the link when/if I find it.
Here's the article: http://www.tnt-audio.com/clinica/re...s_noise2_e.html |
|
|
| fit |
| quote: | Originally posted by carlmart
Not the soldering. The SMDs.
|
SMD in this case have a great advantage - you can shorten feedback path which makes application much more stable... and SMD 0,1% low noise resistors are easy available even in Poland :) |
|
|
| carlosfm |
After a day of :hot: -in, now I have around 10mv DC-offset before the output caps.:dodgy:
Now I'm considering removing the caps...
The sound is different from my OPA627+BUF634 pre, but I don't know if it's better.
Although very good, there's a certain "smoothness" to the sound that on some recordings seam to rob some of the immediacy in the midband.
The treble is very good, smooth but very detailed.
The testing conditions right now are not the ideal.
I have a test pot wich is a cheap carbon thing...
And i'm listening with the output caps, but I'm gonna test now without them.
I don't have caps in the signal path on my OPA627+BUF634 pre and it gives me 0.1~0.2mv DC-offset.:cool:
I'd like that this TPA6120 did the same thing...:xeye: |
|
|
| Devil_H@ck |
Pictures! Pictures :D!
* Another constructive reply to this thread. * |
|
|
| Upupa Epops |
| I have another question ( for good judging of sound is one day too little ;) ) : Have you any control of this amp by high speed scope ? I am asking about stability, 'cos in dataseeet is many warnings about this :xeye: . |
|
|
| dickiegeorge |
Carlos,
I too would like to see a photo.
I've got one of these chips waiting to go, and would like to bread board a circuit to check it out as a preamp buffer like yourself.
If youve got any more tips on using this pesky beast i am all ears :)
For instance - how have you powered it ? - what caps you for bypass ? - how did you solder the pad ?
Sorry for so many questions :rolleyes:
thanks
Dickie |
|
|
| carlosfm |
I don't have a digital camera.:bawling:
Let me finish the film of my holidays.:D
Anyway, one day is of course not enough.
That's why I won't scavenge my OPA627+BUF634 pre while I'm sure that I found a better thing.
I'm also going to test the AD815.
The TPA was made for headphones and for low impedances.
The datasheet is very vague in applications to drive high impedances (like a line preamp).
Anyway, I could "decode":eek: from the TI datasheet that for high impedances you need a 2k feedback resistor, or a little more.
Even for 600 ohm headphones they advice a 2k feedback resistor.
I used 2.2k, and I have a gain of 3.4x, more or less a standard pre, like my other one.
Now it's time to make a board for the AD815 and compare.
Let's see...:scratch2:
:att'n: I like to prototype with veroboard before considering doing a PCB. Any mod that's necessary is easily done, like adding the output caps, wich I decided not to remove.:att'n:
P-A, I must say that you must test this chip before offering a group buy of a PCB. |
|
|
| carlosfm |
I discovered why I have around 10mv DC-offset on the output.
My test pot is 10k.
What's more, DC varied with the position of the pot (louder=more DC).
Reading the TI appnote slou169 posted here by P-A, it is making sense now.
Please read the topic 4.4 - Input bias voltage, at page 24 to 25.
4k is low as an input impedance, I need more.
What's more, without an input buffer, wich I don't want to use, it will affect the log pot.
I'm thinking on a solution without adding an input buffer.
Maby an input cap, followed by a resistor to ground.
In the end, with low DC (let's say below 10mv) I will remove the output caps. |
|
|
| carlosfm |
:scratch:
Mmmm...
I had 10~12mv on the output, with the pot at minimum position.
On the max, it went to more than 100mv.
I've put caps on the output, and no DC.
But I had some DC on the input, wich was affecting the pot too.:xeye:
Yes, this chip has DC on the input.:att'n:
Testing caps on the input and 2.2k resistors to ground after them, I could remove DC on the input, but on the output it got a little higher.:xeye:
The position of the pot doesn't change DC on the output now.
I went as low as 620 ohms:eek: , and now I have around 18mv DC on the output.
It works fine.
I will remove the output caps.
The input caps are 68uf/16v Philips electrolythics, that I'm gonna bypass with some good 100nf Siemens film caps I have here.
Btw, I'm sure that the PCB on the appnote, with those components, has DC on the output (and not small), and on the input...
If this chip needs an input buffer I'll just forget it.
If it delivers after solving all the issues, it will still have to prove that it sounds better than my OPA627+BUF634 pre. |
|
|
| carlosfm |
It's on the :hot: -in stage.
I'm gonna take a break for a few days, enjoy some music, laugh reading some audio magazines:devily:...
Let's see how this sounds in the end.
:up: ? :down: ?
:scratch2: :tilt: |
|
|
| carlosfm |
Yes, I just received a confirmation.
TI's evaluation module has more DC on the inputs and outputs than I have.
And that's with a gain of 2, while I have a gain of 3.4.
DC changes with the position of the pot, although with a 50k pot the change is smaller.
I have my sources.:D
I've solved these issues on my pre, let's see how it goes.:angel: |
|
|
| dip16dac |
| It shouild be connected to a zero ohm source to have no DC output voltage. The 4.02 K ohm resistors are only there to limit the DC offset when the source is disconnected or turned off. I connected it directly to a soundcard output and used the computer volume control and had only a couple mV DC output voltage. Sounds best with +/- 15V power. |
|
|
| fit |
ok - my app works :) - I know you like to see poor soldering 8)
here it is the pic:
no caps in signal path, output offset -0,1mV and +0,3mV
two 627 as input buffers
plays great - I am not real audiomaniac but it was just worth the job |
|
|
| peranders |
Good job and the soldering doesn't look so bad. :up:
I think have made the right decision to include a buffer because of the large input bias currents of the TPA6120. With this bufer you can connect any source without having problems with offset voltage.
What supplyvoltage do you use? How hot does the IC get? |
|
|
|
