TEA2025B - Heavy noise over V++

chozo4 · 08-10-2014, 06:16 AM

Re: TEA2025B - Heavy noise over V++

.. I'm not sure how spot on my multimeter is. but it could be the PSU over time ended up with poor regulation. After I get back in a month or so from overseas I'll have to look into a potential RMA if it gets any worse as I'm leaving in a week. I just checked the +12 supply rail and it's regulating a bit high at 12.30~12.86v which is close to being out of the 10% spec. I have a 7 year warranty so figure heck with it. Instead of recap just get it replaced and see if that makes much more of a difference.

[hate adding an additional post due to the 90 minute edit timeout x.x sorry!]

chozo4 · 08-10-2014, 04:25 AM

Re: TEA2025B - Heavy noise over V++

Originally posted by stj

instead of filtering the noise, why not try to find and kill the source of the noise.

I know the fan noise is from the two fans on voltage control over the fan headers. The other noises are from the mechanical HDD activity and other miscellaneous signal noise from other case components.I'm being picky mainly since the noise is AC and making its' way through power into the amp and being amplified as it's a single rail amp. While I can take the fans off the headers and put them on the fan controller that leaves out individual speed control through speedfan. The rest of the sources there isn't anything I can do about otherwise.

Originally posted by budm

Do you have scope so you can probe the power supply line and the outputs to see what kind of signal you are getting?

That's one thing I don't have. All I have at my disposal is a cheap multimeter by GE (GE2524) and no other utilities but headphones to listen in on the noise in this case.

budm · 08-09-2014, 10:49 PM

Re: TEA2025B - Heavy noise over V++

Do you have scope so you can probe the power supply line and the outputs to see what kind of signal you are getting?

stj · 08-09-2014, 10:19 AM

Re: TEA2025B - Heavy noise over V++

instead of filtering the noise, why not try to find and kill the source of the noise.

chozo4 · 08-09-2014, 09:48 AM

Re: TEA2025B - Heavy noise over V++

Did another rebuild of the noise isolator due to the issue with output and did more tinkering over the hours trying a bit of everything (also got a migraine that could down a giant). Initially tried a double-pi.. being C-L-C-L-C prior to the transistors and it made very little difference. The main reason the output was so low was due to the higher resistance to base. Downing to 1k caused far more noise in the output too.

Now initially, because of the noisy ground the capacitors in the CLC's were removing only a small bit of the noise. Lower ESR caps resulted in even less noise filtered through which shows higher ESR general purpose caps caused the noise reduction purely from their inherent resistance. I swapped in the bigger transistor (MJE3055T) without much change at all which shows the base current is too low still *shrug*

Went back to square one and started looking more into the transistors and figured that if a 2.2k resistor helped with noise but 1k let more in and provided more current there should be ways around it. Needless to say.. apparently feeding the AC noise from the Emitter output to the base with a Rubycon MBZ 16v1500uf removed a good 80% of the noise and output was higher than before. However, being NPN transistors with a common base ended up causing a small bit of clicking as if a 2 clicks/second pulse was going off.

It then hit me.. wait.. the ground is noisy and we're still getting noise feedback from all this so why not just the decouples to ground? I then for the heck of it tried decoupling the output to the transistor base.. and it made quite some difference. I removed the ground altogether from the filter removing the potential for a ground loop in this case and... the noise is drastically reduced now which made every capacitor but the one disconnected. Output is also up and better regulated now bottoming out around the low-end of 8volts rather than 5v on full load. Might end up returning to the idea of a pi filter after the emitter now and using the transistor base as the ground as that's working good for that purpose.

Then after might consider adding a third transistor (2SC1213) or try it now with just the one MJE again and see how it goes.

Afterthought: This is turning more into a discussion about power supply troubleshooting than anything. Perhaps I should split to another thread or just move this to the appropriate forum?

chozo4 · 08-08-2014, 07:45 AM

Re: TEA2025B - Heavy noise over V++

Originally posted by lexwalker

You may want to consider re-capping that PSU as well (as capacitors do age). The PCI card is attached to the motherboard, which usually has lots of ultra low impedance capacitors (mostly range anywhere from 470uF to 1000uF) along its power rails and as decoupling capacitors at vital sections (can find them often near voltage regulators and big I.Cs)...

While they do age the PSU has been emitting barely any heat that would otherwise contribute to a farmore rapid aging cycle. It's an 80+Gold supply so cap wear should be heavily reduced from lower heat output as it is. I did check though through the vents and so forth too and they seem in good shape visually. Just don't want to kill my 7 year warranty on this thing just yet.

Good call on the PCI slot though - was thinking something like that but wasn't quite sure. Though this board is all small solid electrolytic capacitors from 560uf to lower accordingly. Was a bit surprised to see the downsizing though but the board is quite stable with a good overclock thus far.

Originally posted by lexwalker

Are you using any voltage regulators before that amplifier I.C? For the pi filter, make sure the inductor/choke has a ferrite core (in other words, not an air core type). Any measureable resistance on the GND connection, especially around 1 Ohm or higher, will always create noise. The amplifier I.C can pick up slight voltage difference between GNDs (especially ripples and noise from other devices like fans). You may want to consider isolating the power supply by using a LDO voltage regulator.

I actually don't have any LDO's and instead have LVR's by ST for 5v/9v output respectively and didn't want to reduce the voltage too much. I do have an adjustable linear regulator somewhere but I lost track of it so need to hunt it down again sometime (damn cats!). I did however spend time tinkering on the breadboard with an alternative. It was a Discreet noise reduction circuit apparently called a 'capacitance multiplier'(Img1). Changed it up a bit as I tweaked and got a good result at first with a 2.2k resistor to base and the 16v 680uf Rubycon PK. I did test some low esr caps and such but they actually let a bit more of the noise through likely due to the lower ESR and a noisy ground. I had also followed up with a 15ohm 1watt resistor at the end and got a noiseless result without much voltage loss. The transistor used was an ST mje3055T for the start. The noise reducer ground is separate with the amp itself being grounded via the input ground rather than power ground. Testing has all been done using headphones attached to the amp accordingly.

So while I built it from bread to board... I made a couple more changes...
-> Replaced the 15ohm resistor with a 10.5ohm enclosed ferrite inductor.
-> Decoupled the output there with a 0.1uf molded polyester capacitor.
-> Swapped the MJE3055T with a small TO-92 2SC1213 NPN ~500mA current. Voltage dropped a bit more from that however but its' only a test run for the moment.

I might make a few more changes later on as I put it through the paces but it's working great thus far. In the end I'll see about reducing its' size to handle only what's needed as I will be needing it for the ST LM358N microphone amp I have wired to the 5v USB at the moment.[/quote]

Originally posted by lexwalker

The ceramic is for handling very high frequency ripples and noise, and mainly for decoupling function near I.Cs. For 8.5x12mm sizes, you may want to consider 10x12.5mm size capacitors as well. Panasonic SEPF series (ultra low ESR) have large capacitance and are small enough to fit that category.

Usually in the schematics, the capacitor close to the I.C's power inputs are decoupling capacitors.

Aye, I had a braindead moment when I called it the supply cap. Which while it is a decouple it also acts as a voltage smoother/reservoir cap besides hence my going braindead.

Ironically I was looking over the SEPF series last night while trying to ID the bootstrap caps I had. Was too late to edit in the changes though. Those were UCC PXA's without the logo on them so wasn't too far off the mark. I have been using an SEPC however on the SVR. I vowed to no longer buy any more parts for this amp however. The only exception were to get parts I wouldn't otherwise come across much if at all such as the vishay resistors and the supply decouple. The rest was purely aimed at being considered the salvage amp.. built from all used or otherwise salvaged parts save for the PCB of course.

Though I may have to invest in a decent LDO however as all I have on hand are a bunch of mosfets and linear regulators on these half destroyed projection TV boards. Are there any that you would suggest for the task with the highest possible PSRR? Also, as one has probably noticed... I'm a bit biased and prefer products by ST whenever possible.

(IMG1 Source: http://www.edn.com/electronics-blogs...or-noise-floor )t

edit: After more testing I might need to revise it for sure. While the voltage output was good when idle I didn't check on load. During load with a bass test to push the heat/power draw.. the voltage is dropping out to between 5-6volts while the inductor is getting quite toasty so might need to reduce the base resistor a small bit to increase current flow and/or swap in the MJE. Seems I might end up going with a low dropout regulator soon afterall if I cannot tweak this a bit more. Then again, would reduce size using a single LDO and couple ceramics in comparison.

Attached Files

lexwalker · 08-08-2014, 12:32 AM

Re: TEA2025B - Heavy noise over V++

Originally posted by chozo4

The PSU has been running reliably since I got it about 2 years ago. It was also built by the OEM Superflower which is top notch, providing for distributors such as Seasonic and others. Even johnny guru gave it high ratings for instance... http://www.jonnyguru.com/modules.php...Story&reid=266

What I do notice is that it's mostly feedback noise from the fans and that when everything is idle and quiet - there is very little noise from the 12volt rails. Curiously, the same amp built in on a PCI card doesn't get any of this noise from the 12v rail with everything running full bore and does not have a PI Filter beyond a usual n440 diode. There is no noise with my raidmax SMPS on this amp connected seperately which is over 10 years old with bulging (and one leaking) TEAPOs for example. I just use it for a shortterm benchtest PSU but will recap it one of these days. This further hammers it's not a direct issue with either PSU. So for a PSU with both really bad ripple and low wattage and the ilk normally due to poor caps... not producing audible noise says something.

You may want to consider re-capping that PSU as well (as capacitors do age). The PCI card is attached to the motherboard, which usually has lots of ultra low impedance capacitors (mostly range anywhere from 470uF to 1000uF) along its power rails and as decoupling capacitors at vital sections (can find them often near voltage regulators and big I.Cs)...

Originally posted by chozo4

I tried a basic LC filter rather than a PI filter and that just pulled all the noise back in from the ground line instead. The ground line is a short 18 gauge wire typical of most PSU's so it has no measurable resistance. I will tinker with it again tomorrow with a CLC/Pi filter accordingly and see what I get then. I have a very strange feeling that if I swapped the roles.. putting the amp on the comp PSU rail.. and a source on an external supply there would be no noise the either. Will have to test that theory as well then as it would also explain why my fan controller starts off with a CLC at the molex as well for example.

Wait... that gives me the idea of disconnecting everything from my fan controller, connecting the amp to that for the variable voltage adjustment and already existing CLC. Then see if it helps with isolating the noise as well accordingly.

Are you using any voltage regulators before that amplifier I.C? For the pi filter, make sure the inductor/choke has a ferrite core (in other words, not an air core type). Any measureable resistance on the GND connection, especially around 1 Ohm or higher, will always create noise. The amplifier I.C can pick up slight voltage difference between GNDs (especially ripples and noise from other devices like fans). You may want to consider isolating the power supply by using a LDO voltage regulator.

Originally posted by chozo4

I find they are nifty caps that continue to take a beating and keep ticking. I have one other in my collection sitting around but find they have a nice effect on the feedback. That and.. I was a glutton for their uniqueness and colour. A while back I even used a 25v 47uf green cap from Marcon manufacturered in '84 for the SVR at one point. I kind-of wanted to add a bit of vintage flair to it along the way I suppose.

I've also have some (unused) ancient Marcon capacitors around, mostly 4.7uF and below. Also had some ancient Nippon Chemicons also...

Originally posted by chozo4

I did originally have a 0.1uf ceramic parallel with it but it didn't have any effect so omitted it in my final build. I went with the Rubycon PK as well due to the capacitance and size. There isn't much room as the max space allocated for the power cap on my layout is 8.5Dx12H and the PK ended up being the highest capacitance I could find in an 8Dx11.5H package. Everything else within that size was general purpose 470uf with 3/4 of the ripple or polymers around 270uf maximum from what I could find on mouser/newark. I do have room to increase size of the Supply Voltage Rejection (SVR) cap a bit more however as needed.

The ceramic is for handling very high frequency ripples and noise, and mainly for decoupling function near I.Cs. For 8.5x12mm sizes, you may want to consider 10x12.5mm size capacitors as well. Panasonic SEPF series (ultra low ESR) have large capacitance and are small enough to fit that category.

Originally posted by chozo4

Considering the amp doesn't draw too much, ~650mA ripple is more than sufficient for the purpose for a ~300mA draw when being deliberately pushed hard with bass tests. The schematics actually call for a 470uf normally for the power filter however but don't take into account the amp being pushed beyond typical usage specs. During testing on the last build I figured the heck with it and actually replaced the power cap with both a Teapo SD 16v/2200uf and Rubycon MBZ 16v 1500uf with only very little change in the noise.

Usually in the schematics, the capacitor close to the I.C's power inputs are decoupling capacitors.

chozo4 · 08-07-2014, 11:30 AM

Re: TEA2025B - Heavy noise over V++

Originally posted by stj

try taking a low impedence cap like 220uf and putting it across a fan plug.
see if you hear a change.

the fan is an electronicly pulsed sequence of coils - it probably looks like pure evil on a scope.

I don't have a scope unfortunately but will give that a try to see if there is much difference. The offenders are fans that have a 2A draw on the supply when starting up so makes sense to give that a try.

Originally posted by Th3_uN1Qu3

You have a ground loop. When you connect an amplifier to the power supply of your desktop computer, and you also connect the jack to the soundcard, which has ground in a separate place, you create a ground loop.

Try putting a 10 ohm resistor between the soundcard ground and the amp ground, instead of connecting it directly.

I actually disconnected the input ground to break that potential loop and no go. Normally if I broke the ground loop or put a resistor in place of it I would otherwise get mild motorboating/oscillation but that doesn't happen if I break the input ground in this instance. I further tried keeping the input ground connected and the power input ground disconnected to ensure the power ground uses the input ground instead and the same noise occurred.

Though the one thing I didn't try that I should have like I did long ago when tinkering was to place a capacitor in series with the input ground to break any DC current flow over that ground. I'll have to try that again in this instance and see if there is any positive result from it.

Th3_uN1Qu3 · 08-07-2014, 05:52 AM

Re: TEA2025B - Heavy noise over V++

You have a ground loop. When you connect an amplifier to the power supply of your desktop computer, and you also connect the jack to the soundcard, which has ground in a separate place, you create a ground loop.

Try putting a 10 ohm resistor between the soundcard ground and the amp ground, instead of connecting it directly.

stj · 08-07-2014, 05:18 AM

Re: TEA2025B - Heavy noise over V++

try taking a low impedence cap like 220uf and putting it across a fan plug.
see if you hear a change.

the fan is an electronicly pulsed sequence of coils - it probably looks like pure evil on a scope.

chozo4 · 08-07-2014, 04:53 AM

Re: TEA2025B - Heavy noise over V++

Originally posted by lexwalker

That could be the main cause, a noisy power supply on your desktop. The rest of the symptoms mentioned pretty much points me to that PSU...

Have you tried a pi filter? Also measure the resistance of the GND connection wires. Semi-floating ground problems can create a lot of noise...

The PSU has been running reliably since I got it about 2 years ago. It was also built by the OEM Superflower which is top notch, providing for distributors such as Seasonic and others. Even johnny guru gave it high ratings for instance... http://www.jonnyguru.com/modules.php...Story&reid=266

What I do notice is that it's mostly feedback noise from the fans and that when everything is idle and quiet - there is very little noise from the 12volt rails. Curiously, the same amp built in on a PCI card doesn't get any of this noise from the 12v rail with everything running full bore and does not have a PI Filter beyond a usual n440 diode. There is no noise with my raidmax SMPS on this amp connected seperately which is over 10 years old with bulging (and one leaking) TEAPOs for example. I just use it for a shortterm benchtest PSU but will recap it one of these days. This further hammers it's not a direct issue with either PSU. So for a PSU with both really bad ripple and low wattage and the ilk normally due to poor caps... not producing audible noise says something.

I tried a basic LC filter rather than a PI filter and that just pulled all the noise back in from the ground line instead. The ground line is a short 18 gauge wire typical of most PSU's so it has no measurable resistance. I will tinker with it again tomorrow with a CLC/Pi filter accordingly and see what I get then. I have a very strange feeling that if I swapped the roles.. putting the amp on the comp PSU rail.. and a source on an external supply there would be no noise the either. Will have to test that theory as well then as it would also explain why my fan controller starts off with a CLC at the molex as well for example.

Wait... that gives me the idea of disconnecting everything from my fan controller, connecting the amp to that for the variable voltage adjustment and already existing CLC. Then see if it helps with isolating the noise as well accordingly.

Originally posted by lexwalker

This is an ancient capacitor. Also have quite a few of those, unused...

I find they are nifty caps that continue to take a beating and keep ticking. I have one other in my collection sitting around but find they have a nice effect on the feedback. That and.. I was a glutton for their uniqueness and colour. A while back I even used a 25v 47uf green cap from Marcon manufacturered in '84 for the SVR at one point. I kind-of wanted to add a bit of vintage flair to it along the way I suppose.

Originally posted by lexwalker

This is a general purpose capacitor. I would suggest using ultra low impedance/ESR capacitors such as those found on your desktop machine's motherboard (because generally the PSU used for your desktop machine is of SMPS type). Bigger capacitance is recommended, and try adding a 0.1 uF ceramic or film capacitor as close as possible to the power supply pin of that I.C...

I did originally have a 0.1uf ceramic parallel with it but it didn't have any effect so omitted it in my final build. I went with the Rubycon PK as well due to the capacitance and size. There isn't much room as the max space allocated for the power cap on my layout is 8.5Dx12H and the PK ended up being the highest capacitance I could find in an 8Dx11.5H package. Everything else within that size was general purpose 470uf with 3/4 of the ripple or polymers around 270uf maximum from what I could find on mouser/newark. I do have room to increase size of the Supply Voltage Rejection (SVR) cap a bit more however as needed.

Considering the amp doesn't draw too much, ~650mA ripple is more than sufficient for the purpose for a ~300mA draw when being deliberately pushed hard with bass tests. The schematics actually call for a 470uf normally for the power filter however but don't take into account the amp being pushed beyond typical usage specs. During testing on the last build I figured the heck with it and actually replaced the power cap with both a Teapo SD 16v/2200uf and Rubycon MBZ 16v 1500uf with only very little change in the noise.

lexwalker · 08-07-2014, 04:27 AM

Re: TEA2025B - Heavy noise over V++

Originally posted by chozo4

Just [rebuilt] my small sound amp after practically ruining the layout trying to debug this issue. Normally it works 100% and performs quite admirably when connected to an external source (Old Raidmax ATX-PSU) and is clear. However, when I powered it from the power source in my desktop where it is intended to be placed there is heavy noise from the internals such as the fans.

That could be the main cause, a noisy power supply on your desktop. The rest of the symptoms mentioned pretty much points me to that PSU...

Originally posted by chozo4

- Using an LC setup causes noise in full force again with capacitor connected to power ground.
- Ground and inputs connected only, can hear the power noise.

Have you tried a pi filter? Also measure the resistance of the GND connection wires. Semi-floating ground problems can create a lot of noise...

Originally posted by chozo4

Code:

Rubycon   CE-W  85c  220uf/ 6.3v Feedback

This is an ancient capacitor. Also have quite a few of those, unused...

Originally posted by chozo4

Code:

Rubycon   PK   85c  680uf/16.0v Supply

This is a general purpose capacitor. I would suggest using ultra low impedance/ESR capacitors such as those found on your desktop machine's motherboard (because generally the PSU used for your desktop machine is of SMPS type). Bigger capacitance is recommended, and try adding a 0.1 uF ceramic or film capacitor as close as possible to the power supply pin of that I.C...