[jamont]

Quote:

Originally Posted by peranders skyskraper, have checked out my TPA6120 amp which is alive and working? I tried to create some interest for a group buy but the interest was absolutely zero. Maybe DaKi][er has a more interesting design, who knows?

peranders has a very nice looking board, but it requires transformers that are not available in the US, DaKi][er's board does not have this problem.

DaKi][er, it looks like your board uses the Alps Blue, but if so, it is facing the wrong way.

 

[doobooloo]

Quote:

DaKi][er, it looks like your board uses the Alps Blue, but if so, it is facing the wrong way.

I think it is facing the right way, it is going to use a shaft extension, no?


Also, I think all the lettering is best left to be silkscreened (if this is going to be an option), especially "DtpaKiller" which is taking a substantial chunk out of the ground plane...

 

[JohnFerrier]

I understand the TPA6102 is a great chip. Is the opamp before it necessary?


JF

 

[skyskraper]

peranders, nice board. looks pretty complex. what size smd parts are those? i must have missed it the first time round! sorry man

 

[DaKi][er]

Quote:

Originally Posted by doobooloo I think it is facing the right way, it is going to use a shaft extension, no?

I probably should have pointed that out as that was my intention, it gives the board a good flow from back to front

in the M^3 discussion 1/2 the people liked the idea and 1/2 didn’t, and decided to put that in mine, plus the layout of it sort of forced it to fit like that in the end

nothing hard about getting a bit of rod and putting a coupler onto it and fixing some sort of bushing to the case
Quote:

Originally Posted by doobooloo Also, I think all the lettering is best left to be silkscreened (if this is going to be an option), especially "DtpaKiller" which is taking a substantial chunk out of the ground plane...

while silkscreening looks nice I don’t care much for it, if this does get done with a silkscreen I’ll change the text to go on that layer
the big chunk of plane its taking out i got a split down it to sort of separate the power section from the signal section anyway and the text just sits right in the middle of it (reading too many gainclone board layout threads where their all for star wiring and keeping power and signal separate, while good practice headphones aren’t as high powered to want to go all the way with that)
Quote:

Originally Posted by JohnFerrier I understand the TPA6102 is a great chip. Is the opamp before it necessary?

going through the datasheet of the chip it didn’t explain how it liked to be setup to have a gain and with the current feedback was real picky about what resistor values you used so I stuck with the inverting design with a gain of 1 with 1k resistors and used the opamps before for gain stage following peranders design
I’m no electrical engineer and are learning as i go, so if someone want to educate me for a better way to implement the tpa chip then feel free to (and on that note) -
can someone explain what this 'tombstone effect' is? I don’t really know what your talking about there

 

[Garbz]

/EDIT: PLease dismiss my ramblings.

 

[DaKi][er]

final board layout, been over and over it cleaning little bit up all over it, changed to bigger holes for wires and im happy with it all now

getting 12 made up, 9 are spoken for so far and i'll keep the rest if no one else want any more

if you want one, their $9 US plus postage (which could be as nearly as much as the board, coming from Aus)

thanks for the input guy's cant wait to see (and hear) the finished amp

 

[roadbuster]

Quote:

Originally Posted by peranders This tombstone effect won't go away with 1206's. This is a matter of designing the pads. If you are concetrated when you are doing the soldering this is not a problem. It's better to choose 0805 because they are more common.

When guzzler said, "tombstoned resistor," I don't think he was referring to the
tombstone effect, oddly enough. I think he was referring to the vertically-
mounted resistor which is beside the LED on the 3D rendered board image.

Quote:

Originally Posted by DaKi can someone explain what this 'tombstone effect' is? I don’t really know what your talking about there

When one-end of a device tips-upwards during the soldering process, it is
said that the device has "tombstoned," much like how tombstones in a
cemetary tip-over and are rarely perpendicular to the ground.

When tombstoning occurs, the device may have a very weak electrical and
mechanical joint with the board. If the device has tipped by a very large
degree, there will be no mechanical or electrical connection at one terminal,
and your circuit will fail.

You don't have to really worry about this effect since you're not going to be
producing your board in mass quantities, and you're not going to have an
automated process for soldering the components to the board. Just make sure
everything is connected properly after the solder joints cool.

Now, I have a question for you: why don't you have a high-pass filter at the
input of your op-amps? Without a filter (eg: simple capacitor), you'll be
amplifying the DC component of your input signal. You can always block this
DC offset at the output of your amplifier, but you don't seem to be doing
that, either.

 

[DaKi][er]

Thanks for the explanation of the tombstone, the led resistor isn’t anything special anyway and as its connected to the unregulated voltage side it could see about 35v across it or more depending on what transformer voltage you use (I got some 12.6v center tapped I’m using) so with 20ma for the led that’s 0.7w needed for the resistor, so you can now fit a 1w in there easily without taking up 1/2 the board

About the high pass filter, I just don’t like them much and caps in the signal path = bad. while I would be struggling to tell the difference most of the time the fact is I can get away without putting one in. my 3 main sources I got so far are a chaintech av710, usbdac and iriver h120 and I know for a fact that all 3 have output caps on them and there is no dc offset with any of them and I haven’t come across or heard about a source from someone else that has any dc offset on its outputs that needs worrying about

 

[Garbz]

I don't use a Highpass filter in any of my designs either. Anything to reduce the ammount of parts which can cause distortion in and signal path is a bonus.

I have a lot of trust in my source equipment not to fail and put out DC.

 

[00940]

Very interesting and compact design.

I'm working with my own stuff right now for the TPA6102 and some sections of your layout are rather helpful.

By the way, out of curiosity, how does the USB dac compare to the AV710 ? I could get one for my secondary pc.

 

[JohnFerrier]

Quote:

Originally Posted by DaKi][er going through the datasheet of the chip it didn’t explain how it liked to be setup to have a gain and with the current feedback was real picky about what resistor values you used so I stuck with the inverting design with a gain of 1 with 1k resistors and used the opamps before for gain stage following peranders design

Well the odd thing is that though the TPA6120 is like an ultimate performance chip, the overall performance of the circuit will then be limited by the performance of the first opamp. Meaning, if I got this right, there is little technical gain on a PPA style amp (where the output section is within the feedback loop of the the first opamp).

*EDIT* Corrected part number.


JF

 

[DaKi][er]

Quote:

Originally Posted by 00940 Very interesting and compact design. I'm working with my own stuff right now for the TPA6102 and some sections of your layout are rather helpful.

i'd like to see what you come up with at the end
Quote:

Originally Posted by 00940 By the way, out of curiosity, how does the USB dac compare to the AV710 ? I could get one for my secondary pc.

cant comment as i only got the usbdac sitting here 1/2 populated and waiting to get the parts with my next mouser order and the blackgate supplyer in Aus to get in the caps for them as well
Quote:

Originally Posted by JohnFerrier Well the odd thing is that though the TPA6102 is like an ultimate performance chip, the overall performance of the circuit will then be limited by the performance of the first opamp. Meaning, if I got this right, there is little technical gain on a PPA style amp (where the output section is within the feedback loop of the the first opamp).

what are you suggesting? (if anything) put the tpa inside the first opamps feedback loop, you think the first opamp will hold the tpa back even if its an ad8610 or opa637?

 

[JohnFerrier]

Quote:

Originally Posted by DaKi][er what are you suggesting? (if anything) put the tpa inside the first opamps feedback loop, you think the first opamp will hold the tpa back even if its an ad8610 or opa637?

I've only glanced at the TPA6120 datasheet, but it looks like it is current feedback (which are in) and can not be used within a global feedback loop (PPA style). Also, if it has a minimum gain of 2 and the 637 has a minimum gain of 5, you have a minimum gain of 10. Personally, I think that is 10x more than a person needs. If I were building a chip amp, I would look at how to use the TPA6120 without the first opamp (maybe do a Google search for TPA6120 schematics). Fortunately for me, I don't need another amp.

*EDIT* Corrected part number.


JF

 

[peranders]

You are quite right about the TPA6120 (not 6102), it's a current feedback amp and it can't be used as a common voltage feedback opamp. One other thing are the massive input currents. This will be an disadvantage in many cases.

DaKi][er, why don't you turn the TPA6120 180 degrees? Now you have the output trace across the inputs, on purpose! Texas recommends even to remove the groundplane under the inputs and you make it even worse. I must also say that I haven't tested how important Texas' recommendations are but I don't any chances with a 100 MHz device. Just check my board, check for the hole in the groundplane and how the output is located.


Check my pictures.
http://home5.swipnet.se/~w-50674/hif...r0_topview.jpg
http://home5.swipnet.se/~w-50674/hif...v07pcbsold.jpg
http://home5.swipnet.se/~w-50674/hif...v07pcbcomp.jpg