| theAnonymous1 |
Seven hours later and I now have a working TPA3122D2N amplifier running BTL :D. I followed the eval board schematic.
This little baby is sweeeeeet. I built up a TPA1517 chipamp yesterday and this little class-d totaly crushes it. I've been running the TPA3122D2N into an 8R bridged load @ 14v for over an hour now and the thing is barely warm. I was running the TPA1517 in the same configuration yesterday and the IC got extremely hot, even after I attached a heatsink. The TPA3122D2N sounds better to my ears as well; very tight clean bass.
Now I just have to build the other channel.:xeye:
1" x 2"
 |
|
|
| e_fortier |
Hi,
Very nice PCB design, I have 5 of these just sitting at home.
Nice to know that it sounds sweeeeeeet, I also built the TPA1517 (class AB IIRC) so your feedback is appreciated.
Once I complete my Amp6 that will probably be the next one in line.
Eric |
|
|
| theAnonymous1 |
| quote: | Originally posted by e_fortier
Hi,
Very nice PCB design, I have 5 of these just sitting at home.
Nice to know that it sounds sweeeeeeet, I also built the TPA1517 (class AB IIRC) so your feedback is appreciated.
Once I complete my Amp6 that will probably be the next one in line.
Eric |
Thanks, I utilized a continuous groundplane on the top layer and it seems to have turned out well. I have to press my ear right against the tweeter to hear any kind of background noise.
It really amazes me how cool this chip runs. It's obviously a night and day difference when compared to a linear amplifier like the TPA1517, but it also runs MUCH cooler than my AMP32. That's surprising considering it technically has lower efficiency.
This design was just to test how the small TI class-d perform. My next project will be with the TPA3106D1. It will be a bit harder to work with, but it doesn't require a differential input for BTL operation and it's capable of more power. |
|
|
| luka |
Hi
NICE PCB design!! like the ground plane |
|
|
| ArjenShenzhen |
Hi there mate,
Cool looking amp, looks like a simple solid solution!
Prehaps usable as a amplifier for portable solutions aswell
Well done! |
|
|
| mobyd |
My samples of these puppies just showed up - Kudos to TI for making a switcher in a dip package. I notice that these are not rated to run 4 ohms with a 28 volt supply - anyone know if it is possible to parallel the inputs and outputs (prior to the pi network) to up the current handling (in mono, of course).
M |
|
|
| theAnonymous1 |
| quote: | Originally posted by mobyd
My samples of these puppies just showed up - Kudos to TI for making a switcher in a dip package. I notice that these are not rated to run 4 ohms with a 28 volt supply - anyone know if it is possible to parallel the inputs and outputs (prior to the pi network) to up the current handling (in mono, of course).
M |
I paralleled the inputs and outputs and it seems to work. If only one input is connected with the outputs paralleled the amp just clicks though. I would test it at 28v with a 4R load for you, but I built the board with 16v capacitors. |
|
|
| theAnonymous1 |
I just finished making my first TPA3106D1 PCB. I've soldered the IC on, but I can't finish the board yet because I'm all out of 1uF 0805 capacitors.
There is a pad under the IC that connects to the top side ground plane both electrically and thermally. I don't have a way or method of soldering PowerPADs, so I just used a dot of thermal compound. I think with all the other pins that connect to the ground plane it should be fine.
 |
|
|
| luka |
Hi
Thermal compound is ok....stop, let me say first: you are crazy(in a good way), those boards are killing, very nicely done!!
Maybe you will now try to make Tpa3120d2
If you have thermal pads, you could use oven, works great |
|
|
| theAnonymous1 |
| Thanks guys. I have a small toaster oven and contemplated doing the thermal pad that way, but I don't have any solder paste and it isn't something available locally. I'm too cheap and impatient to order it by mail. I think it should be fine with the combination of thermal compound and the pins connected to the ground plane. |
|
|
| luka |
Hi
Solder paste? nah didn't you read my thread? I used normal solder, which melted nicely between board and IC, making great termal conduction |
|
|
| theAnonymous1 |
| quote: | Originally posted by luka
Hi
Solder paste? nah didn't you read my thread? I used normal solder, which melted nicely between board and IC, making great termal conduction |
I read the thread, but I'm not sure exactly how you did it. Did you melt some solder to the PCB first, or did you put some of the un-melted solder wire between the board and the IC? |
|
|
| theAnonymous1 |
I managed to scavenge enough parts to complete a board. I like this one even better than the TPA3122D2N. I also love when a project powers on and works the very first time.;)
 |
|
|
| luka |
Hi
Damn that looks nice
Ok let me gide you for any later project you would need this
You will be able to see first pic, I used that as test, some broken fet and unused PCB and do this:
1. Make sure your solder material has a melting point that is not higher than 190C. Most 60%Zn 40%Pb will melt at around 180C. (My was one of them)
2. Heat an oven to about 230-240C (I think I had 230, and was enough, or did I had only 220, it is stil more then enough)
3. Scrape of the solder mask in the rectangle in the middle of the chip position on the PCB with a sharp knife. Make sure you remove all solder mask varnish in this area. (That is if you have factory made PCB, for home you don't have to do anything)
4. Place about 20 mm (3/4”) long bit of solder with a diameter of 0.6 mm or the equivalent amount, bent into a circle, on the cleaned PCB area. Place the PCB in the oven and measure the time it takes before the solder has melted and flowed out on the solder area. (This was for TK2350, so you can use diffrent wire lenght and it doesn't have to be 0.6mm, but my was 0.7mm) It may take about 2-3 minutes.(it was in that time somewhere)
5. Remove the PCB from the oven and let the PCB cool down.
6. Spread some “no-clean” non aggressive solder flux on the solder area. A solder flux pen is handy for this. (I used normal whatever I use normaly, no problems with it, just very small amount)
7. Place the chip perfectly in place on the PCB. Make sure all chip leads are aligned to the solder pads of the PCB.
8. Place the PCB in the oven and allow as much time as before, but add 20 seconds for the chip to heat up. After this time shut of the oven power, open the oven door and gently fan cool air into the oven. (I have waited about 30s, and then slowly and easy took whole board leveled out and put it down to cool)
9. Let the PCB cool down and inspect it.
10. If everything is OK, solder the chip legs to the PCB pads. |
|
|
| theAnonymous1 |
| quote: | Originally posted by luka
Hi
Damn that looks nice
Ok let me gide you for any later project you would need this
You will be able to see first pic, I used that as test, some broken fet and unused PCB and do this:
1. Make sure your solder material has a melting point that is not higher than 190C. Most 60%Zn 40%Pb will melt at around 180C. (My was one of them)
2. Heat an oven to about 230-240C (I think I had 230, and was enough, or did I had only 220, it is stil more then enough)
3. Scrape of the solder mask in the rectangle in the middle of the chip position on the PCB with a sharp knife. Make sure you remove all solder mask varnish in this area. (That is if you have factory made PCB, for home you don't have to do anything)
4. Place about 20 mm (3/4”) long bit of solder with a diameter of 0.6 mm or the equivalent amount, bent into a circle, on the cleaned PCB area. Place the PCB in the oven and measure the time it takes before the solder has melted and flowed out on the solder area. (This was for TK2350, so you can use diffrent wire lenght and it doesn't have to be 0.6mm, but my was 0.7mm) It may take about 2-3 minutes.(it was in that time somewhere)
5. Remove the PCB from the oven and let the PCB cool down.
6. Spread some “no-clean” non aggressive solder flux on the solder area. A solder flux pen is handy for this. (I used normal whatever I use normaly, no problems with it, just very small amount)
7. Place the chip perfectly in place on the PCB. Make sure all chip leads are aligned to the solder pads of the PCB.
8. Place the PCB in the oven and allow as much time as before, but add 20 seconds for the chip to heat up. After this time shut of the oven power, open the oven door and gently fan cool air into the oven. (I have waited about 30s, and then slowly and easy took whole board leveled out and put it down to cool)
9. Let the PCB cool down and inspect it.
10. If everything is OK, solder the chip legs to the PCB pads. |
Thanks Luka.
I'm kind of glad I didn't solder the thermal pad to the PCB now. I made the stupid decision to try and ground the negative input while the board was powered up. It didn't like that and decided to throw the rail voltage at my test speaker.:bawling:
Oh well, 15 minutes and a fresh IC later it is back singing again. I think this amp might work better with a differential input, even though is doesn't need it to work BTL. I will give it a try and report back. |
|
|
| theAnonymous1 |
It seems I have a small inductor heating problem with the TPA3106D1 board. I didn't notice it at first , but they are definitely heating more than the ones on the TPA3122D2N (which don't heat at all really).:o
I know a multimeter isn't exactly the correct tool to trouble shoot a problem like this, but I did measure some differences between the two boards with it.
AC voltage across output inductors:
TPA3122D2N = 1.938Vac
TPA3106D1 = 2.664Vac
AC voltage from output to ground:
TPA3122D2N = 0.003Vac
TPA3106D1 = 0.078Vac
You can see both reading were higher with the TPA3106D1. Both boards were being powered from the same 12v battery.
The only thing I did differently with the TPA3106D1 board is put one inductor on each side of the board and I used 1uF WIMA MKS capacitors for the output filter instead of 1uF X7R ceramics. Could either of those things cause this problem, or could it just be a layout issue? |
|
|
| kenpeter |
A small dab of lead free solder paste is what we use for the pad,
but if you are doing this at home I would definitely use LEAD....
The machine can do lead free just fine, but when it come to the
manual rework. I dread the lead free on the heatsinkless PCB,
where the board itself conducts the heat... The difference from
just enough heat, and burning up the board is a very fine line.
You need something (like hot air) to warm the board from the
back side, and the iron (or slightly hotter air jet from the top)
to do the fine local melting....
I don't recommend thermal compound, cause you will still need
to clean the flux from between the pins, and this will wash out
any grease you may have tried to substitute.
The DIP is so much easier package for the do-it-yourselfer...
But I even the DIP needs to be soldered too, I wouldn't risk
a socket to conduct the heat... |
|
|
| theAnonymous1 |
| quote: | Originally posted by kenpeter
I don't recommend thermal compound, cause you will still need
to clean the flux from between the pins, and this will wash out
any grease you may have tried to substitute. |
Surprisingly it didn't. After I finished soldering the IC I had submerged the board in alcohol forgetting that the thermal compound was there. When I killed the IC and had to replace it all of the thermal compound was still there. I didn't submerge it again after the replacement; I just used cotton swabs to clean the flux.
I wouldn't touch lead free solder if I was given a truck load of it for free.;) |
|
|
| luka |
Hi
I had also no problems of anykind, didn't even clean anything, yet if worked ok |
|
|
| BWRX |
| quote: | Originally posted by theAnonymous1
The only thing I did differently with the TPA3106D1 board is put one inductor on each side of the board and I used 1uF WIMA MKS capacitors for the output filter instead of 1uF X7R ceramics. Could either of those things cause this problem, or could it just be a layout issue? |
Try eliminating the possibilities one by one. Swap out the film caps for the ceramics and see what happens. If that doesn't change anything then try connecting one of the inductors to the board with short wires (~0.5 inch) and move the inductor around.
Very nice prototypes by the way :) |
|
|
| theAnonymous1 |
| quote: | Originally posted by BWRX
Try eliminating the possibilities one by one. Swap out the film caps for the ceramics and see what happens. If that doesn't change anything then try connecting one of the inductors to the board with short wires (~0.5 inch) and move the inductor around.
Very nice prototypes by the way :) |
Thanks Brian. I tried both changing the caps to ceramics and moving the inductors around with no change in temp.
The inductors heat up to around >100F with a room temp of 76F. Is that considered "hot"? Should I even worry about it?
A scope would make a really nice fathers day gift. Would it count as a gift if I buy it for myself?:clown:
EDIT: I forgot to mention the idle current draw is 50mA. The datasheet says it should be 14-17mA. |
|
|
| theAnonymous1 |
Without the inductors connected the idle current draw is 16mA.:confused:
EDIT: When the ground point between the two 1uF caps is removed the idle current is normal. As soon as it is connected it goes back up to 50mA.
 |
|
|
| luka |
Hi
LOOL? that shouldn't happen, bad PCB? you would really need scope right about now |
|
|
| theAnonymous1 |
Oops.:o
Well, I found one major issue that fixed some of the problem. I had the MASTER/SLV pin driven LOW this whole time, meaning it was in slave mode, accepting a clock frequency on the SYNC pin.:cannotbe:
I soldered a jumper wire from the MASTER/SLV pin to VREG and now the current draw is only 32mA instead of 50mA. The inductor temp is down to <90F.
I'm thinking the other problem is my layout. I didn't bother keeping the AGND and PGND seperate. I think I was so tired when I made it that it didn't even cross my mind that I should have. |
|
|
| luka |
Hi
You could probably do better, but 20F is nothing, it is ok |
|
|
| theAnonymous1 |
I finished my new layout. Doing the layout the way I do is very time consuming as everything is laid out manually; no nets at all. It's a royal PITA, but I can maximize the layout for how I hand make my PCBs.
I kept the AGND and PGND separate this time. They connect at one point; the thermal pad. Hopefully the 3 hours I spent on this was worth it.
 |
|
|
| theAnonymous1 |
Finished the new board. This one has an idle current draw of 32mA like the other one after I fixed the oscillation problem. I feel better about the layout on this one at least.
 |
|
|
| luka |
Hi
Sick board, you are getting very good at this:D Nice job |
|
|
| kenpeter |
You didn't seem to leave yourself room to drill holes for standoffs.
Many components perhaps too close to the edge for snap track.
What was your plan? Or is this only for the bench? |
|
|
| theAnonymous1 |
| quote: | Originally posted by kenpeter
You didn't seem to leave yourself room to drill holes for standoffs.
Many components perhaps too close to the edge for snap track.
What was your plan? Or is this only for the bench? |
Zip-ties! :D
I don't even have any extra standoffs on hand, so holes for them would do me no good. :o
Looking at the middle pic with the board on its side; I figure one tie around the PCB where the inductors are, and then one around the large electrolytics.
I think I will do dual mono-blocks with these. There are some really small 19v laptop SMPS available I plan to power them with. |
|
|
| kenpeter |
I got a wacky wacky abuse to try... Gotta look through the
scrap for one with an old rev chip or eval board with some
cosmetic defect, nobody will care if I blow one of those up.
If it works, I might then try with a current rev board.
I got a pair of RCA 814 directly heated transmitter tubes.
I want to try them for audio output. Filaments need 10V
at 3.25A, and I hear bad things can happen to one end
of a directly heated filament after extended periods DC?
I am no fan of filament hum to try plain old AC. They do
light up with a bright golden glow like the 32.5W bulbs
they are, with DC on my Agilent bench supply.
Suppose I take a clean ultrasonic 50KHz sine wave instead?
From a Wien Bridge or whatever... And pump that through
one of TI's tough little Class-D amps with a real high frame
rate? The filament would then be about a 3 ohm BTL load?
Wonder if output inductors would be an issue at 50KHz?
TI may have some more appropriate part# to suggest?
I'll discuss this tomorrow (if an engineer has time) before
I move any further. I can still breadboard a Wien Bridge in
anticipation of some future experiment. |
|
|
| przem |
Hi
Currently i'm making pcb layout for tpa3106d1. I have noticed that you connected Vcc to Fault and Mute through a resistor - could you please explain why?
And what's the cap between INP and INN for?
Btw. How is the sound from this little baby? ;) |
|
|
| theAnonymous1 |
| quote: | Originally posted by przem
Hi
Currently i'm making pcb layout for tpa3106d1. I have noticed that you connected Vcc to Fault and Mute through a resistor - could you please explain why?
And what's the cap between INP and INN for?
Btw. How is the sound from this little baby? ;) |
If you look really close you will see that the Fault pin is actually connected to the Mute pin. This way when there is a Fault condition it will pull Mute high and mute the amp. The 100k to ground on the Mute pin is to pull it low when there is no fault condition.
The cap between INP and INN is a 10nf NP0. I was just experimenting and it isn't required.
I really like the sound of this amp. Plenty of punch and very clean. I suggest using it with the lowest gain setting though as the the background noise increases quite a bit with a higher setting.
I haven't built the other channel yet due to a recent move and broken laser printer; plus I need to order some parts from digi-key. |
|
|
| Eva |
| quote: | Originally posted by kenpeter
I got a wacky wacky abuse to try... Gotta look through the
scrap for one with an old rev chip or eval board with some
cosmetic defect, nobody will care if I blow one of those up.
If it works, I might then try with a current rev board.
I got a pair of RCA 814 directly heated transmitter tubes.
I want to try them for audio output. Filaments need 10V
at 3.25A, and I hear bad things can happen to one end
of a directly heated filament after extended periods DC?
I am no fan of filament hum to try plain old AC. They do
light up with a bright golden glow like the 32.5W bulbs
they are, with DC on my Agilent bench supply.
Suppose I take a clean ultrasonic 50KHz sine wave instead?
From a Wien Bridge or whatever... And pump that through
one of TI's tough little Class-D amps with a real high frame
rate? The filament would then be about a 3 ohm BTL load?
Wonder if output inductors would be an issue at 50KHz?
TI may have some more appropriate part# to suggest?
I'll discuss this tomorrow (if an engineer has time) before
I move any further. I can still breadboard a Wien Bridge in
anticipation of some future experiment. |
Put a capacitor across the filament, an inductor in series and apply a square wave. The waveform at the filament will be sine like. You will have to choose component values carefully so that the LC filter removes enough high frequency components and the filament resistance when hot provides enough damping to the filter. That's all. Electronic ballasts for fluorescent lamps work much in the same way. |
|
|
| przem |
| quote: | Originally posted by theAnonymous1
If you look really close you will see that the Fault pin is actually connected to the Mute pin. This way when there is a Fault condition it will pull Mute high and mute the amp. The 100k to ground on the Mute pin is to pull it low when there is no fault condition.
|
I have seen that MUTE should be connected to FAULT (datasheet says that) but didn't know about the resistor. As I can see you have also connected GAIN to VREG thru resistor - can't it just be connected directly?
Thanks for your answers! |
|
|
| theAnonymous1 |
Yes, GAIN0 and GAIN1 are connected to VREG via 100k resistors. This is to pull both GAIN high when no jumpers are installed (giving the highest gain setting of 36dB). Installing a jumper will pull the relative GAIN pin low changing the gain configuration.
If you plan on only using a single gain setting you don't need the resistors. Say for instance you want the lowest setting (20dB), you can simply connect both GAIN to AGND and leave out the 100k resistors to VREG.
Have you taken a look at the schematic in the evaluation module datasheet?
http://focus.ti.com.cn/cn/lit/ug/slou191b/slou191b.pdf
|
|
|
| przem |
I looked at it only once, some time ago and just forgot about it ;>
Thanks for reminding ;) |
|
|
| przem |
| Ok, finished. Looks similar to theAnonymous board but there aren't much things that you can change. Coils are on Top layer, probably I'll put them on bottom. |
|
|
| theAnonymous1 |
Nice. Are you going to have the PCB made by a board house?
I know there isn't much room to work with, but you might want to consider a jumper from SDZ to ground and then a 100k resistor from SDZ to VCC. With SDZ permanently pulled high there is a slight turn on/off noise, plus it takes a few seconds for the amp to power off because of the energy stored in the large onboard supply caps. I was going for the simplest layout I could make, so I left out the shutdown option.
The on/off noise and delayed turn off don't bother me that much, but I know others are more critical of those things. |
|
|
| przem |
Nope, I'm going to make it in home but currently I am abroad so I don't know when would it happen :)
Thanks for the SDZ advise, I hate delayed turn off so I will add a switch.
Now it's time for TAS5424 and TPA3004D2 ;)
Did you build them? |
|
|
| theAnonymous1 |
| TAS5424 is beyond my skill level as it requires an I2C interface. The volume control in the TPA3004D2 seems interesting, but the chip is a bit underpowered. I would rather just use TPA3106D1 with one of the small PGA2320 boards I have. I have some TPA3001D1 I might make a layout for some day. |
|
|
| Möter |
Hello guys, :)
I want to have some TPA3122 pcb's professionally made.
The TPA3122 is in BTL configuration and used in combination with a THS4131 fully differential opamp in order to perform single ended to balanced conversion.
Because I'm quite new to electronics I'd like to have experienced people to take a look at my schematic and future board layout, before I spend money on boards that won't work in a desirable fashion. ;)
Also: should I connect the power ground of the 24V for the amp with the power ground of the +-15V for the opamp?
The schmematic is attached. Thanks in advance for your advice. :D |
|
|
| theAnonymous1 |
| quote: | Originally posted by Möter
The TPA3122 is in BTL configuration and used in combination with a THS4131 fully differential opamp in order to perform single ended to balanced conversion. |
Möter!!!!!!!!!!!!!!!! :smash:
I just started a layout for this exact same thing! Mine is going to be slightly different; if it even works at all. If it does work, then I will have the board powered with a single rail for both IC's AND it will be DC coupled. :devilr:
EDIT: BTW, why not use the TPA3106D1 instead? This way you only need one IC.
The only reason I'm trying the THS4131 + TPA3122D2 is because I have quite a bit of them. |
|
|
| Möter |
HAHAHAHA What an coincidence. :cool:
That sounds really good, I was already wondering how to get dual 15V rails from a 24V supply. :rolleyes:
I'm not using the TPA3106D1, because I haven't yet SMD soldered in my life, so I don't want to start with something more difficult than 0805 and SOIC packages (which are probably hard enough). |
|
|
|
