TEA2025B - Heavy noise over V++

[momaka]

Re: TEA2025B - Heavy noise over V++

What bothers me though about the idea of using a TEA2025b in bridge mode in that there is no output capacitor specified to prevent flow of DC through the headphones. In this case should I be concerned or would putting a capacitor on the positive output have any negative impact on output while blocking DC?

Many ICs do not require an output capacitor in bridge mode. This is because you are connecting the load between the two internal amplifier outputs, which are both at 1/2 Vcc (so the net difference between the output terminals on the load is zero).

I also found a pair of full -size 'MCS Series' 3-way studio speakers from back in early 80's made by JCPenny that I eventually intend to use with it too. Tested them and they work fine aside from what sounds like one of the braces inside broken off as shifting one of the speakers causes rattle.

The outside foam on the drivers may be bad. You can change it (i.e. re-foam them) either with a generic kit on eBay, or with paper, some patience, and a drop of creativity, like this :
https://www.badcaps.net/forum/showpo...&postcount=112
It's a bummer that the front mesh is molded and not removable, but if you have to, rip it out. If the foam is completely rotten (like it was in my case above), you will get horrible sound and could even short your amp if the coils inside the drivers scrape too much on the magnets.

The output voltage is normally 1/2 VCC but the gain is adjustable with a resistor in RC Series from the feedback pins to ground. I actually cannot understand the math behind figuring out the actual gain after adding a resistor or the cutoff of changing the capacitor. According to specs it shows R=0, C=100uf for 46db/32hz cutoff yet in my case I used R=562, C=220 to reduce gain and lower the cutoff point for low frequency so as to increase the bass response as well. I'd love if anyone can help me figure the gain/cutoff point result.

Well, at first their formula seemed strange and I couldn't figure it out too, but I think I just did. So they give it out as:

Vout/Vin = R1 / [Rf + R2 + (1 / (jωC1)) ]

So here's how the formula breaks down...
Assuming the feedback capacitor has infinite capacitance, then it is not adding any impedance to that formula - i.e. (1 / (jωC1) is zero.
With that, the formula simplifies to:
Vout/Vin = R1 / (Rf + R2)

Now here is something STmicro didn't mention: R1 = 10 KOhm and R2 = 50 Ohms. I took these numbers from the block diagram on page 1 and just played around with them in the formula above, so that's how I figured it out.

So, with R1 = 10 KOhm, R2 = 50 Ohms, and Rf = 0 Ohms, we have:
Vout/Vin = 10000 / (0 + 50) = 10000 / 50 = 200

Vout/Vin is the voltage gain or Av. It's in Volts/Volts, NOT in decibels (dB). To convert it into decibels, use the following formula:

Av(dB) = 20 * log(Av) = 20 * log(Vout/Vin)
Vout/Vin was found to be 200 above
Av(dB) = 20 * log(200) = 46.0206 dB <-- which agrees with the datasheet.

Next, is the cut-off frequency. Datasheet states with C1 = 100 uF, fc is 32 Hz. Again, I had to do some playing around with the numbers from the block diagram, but I was able to figure it out.
--- It's basically an RC high-pass filter. Here is where our online RC-high pass calculator comes in handy, which can be found here:
http://sim.okawa-denshi.jp/en/CRhikeisan.htm
Plugging in 50 Ohms for the resistance and 100 uF for the capacitance in that online calculator gives 31.83 Hz - again, agreeing with the datasheet.

Alternatively, you can use the formula on page 6 of the datasheet, which is:
fc = 1 / (2 * pi * C * R)
where C is in Farads (F), and R is in Ohms. With stock values, C = 100 uF and R = 50 Ohms:
fc = 1 / (2 * 3.14 * (100 * 10^-6) * 50) = 1 / 0.0314 = 31.85 Hz
.
.
.
Now let's grind those calculations again with your values (but I suggest you pull out a calculator and try them for yourself - it's pretty easy after you do it a few times). First, however, I need to mention that your "R=562" is NOT 562 Ohms... it is 56 * 10^2 = 5600 Ohms = 5.6 KOhms. So with C1 = 220 uF:

Vout/Vin = Av = 10000 / (5600 + 50) = 1.77
Av(dB) = 20 * log(1.77) = 4.96 dB ... which is approximately 5dB gain
^ that's quite below the minimum recommended gain in the datasheet. It's also quite low in general, so you are barely amplifing your audio output. I suggest you bump up the gain to at least 10 dB

Anyways, next is the fc cut-off frequency. Using that RC high-pass calculator, fc = 0.128 Hz.
^for all practical purposes, that is very nearly DC. However, that is only half of what is involved in the overall amplifier response design, because you have the output load and output coupling caps. With 470 uF output coupling caps and Rl = 4 Ohms, the cut-off frequency at the output is 84.7 Hz, as the datasheet suggests on page 6 (actually, they seem to have rounded it down to 80 Hz... which is common for average audio ICs in order to make them appear better).
I see you used 1500 uF Evercon ME (yikes! I suggest you use some other caps, as these tend to go more and more leaky over time - at least that's what happened to the GSC ME on my old Jetway motherboard). With 1500 uF, your cut-off frequency at the output is 26.5 Hz, which makes it much better in terms of bass response.

I've actually never really used a simulator before. I tried SpiceLT but couldn't wrap my head around it at all and I'm more of a trial/error person when it comes to figuring things out.

That's how I started as well. In fact, I still like to go "head-first" and just build the circuit rather than muck around on a computer simulator. But in college, they made us use MultiSim, and I got pretty used to it now, so I find it quite useful.

Watch them steal the build and resell it next...

You know, I wouldn't be surprised. The Chinese are great at copying everything!

As for you asking where I'm going - Bangkok/Thailand for a month due to needing surgical work is all I'd be willing to divulge publicly.

Oh okay. I hope everything goes well there.

[chozo4]

Re: TEA2025B - Heavy noise over V++

The outside foam on the drivers may be bad. You can change it (i.e. re-foam them) either with a generic kit on eBay, or with paper, some patience, and a drop of creativity, like this :
https://www.badcaps.net/forum/showpo...&postcount=112
It's a bummer that the front mesh is molded and not removable, but if you have to, rip it out. If the foam is completely rotten (like it was in my case above), you will get horrible sound and could even short your amp if the coils inside the drivers scrape too much on the magnets.

I'll have to look into that. They sounded fine like said - just it sounds like the wooden brace inside the chassis is loose or the crossover is dangling when the chassis itself is physically moved is what I meant. However, I completely forgot about there being refoam kits available and will definitely have to look into it. I will admit though, at one point I was almost considering just parting them out and reusing the speakers inside to merge into a far smaller chassis while taking the low frequency speakers and putting them down on the floor.

Well, at first their formula seemed strange and I couldn't figure it out too, but I think I just did.

Looks like you plain nailed it. I was mostly confused purely by the jωC1 as I didn't know what the jω was indicating. Considering the fact they didn't specify the internal resistors as well I couldn't get it to match up at all. Not only now can I read it out easier and do the calculations but I learned something there. You did go above and beyond though which I feel bad about however but overall..

Now let's grind those calculations again with your values (but I suggest you pull out a calculator and try them for yourself - it's pretty easy after you do it a few times). First, however, I need to mention that your "R=562" is NOT 562 Ohms... it is 56 * 10^2 = 5600 Ohms = 5.6 KOhms. So with C1 = 220 uF:

Vout/Vin = Av = 10000 / (5600 + 50) = 1.77
Av(dB) = 20 * log(1.77) = 4.96 dB ... which is approximately 5dB gain
^ that's quite below the minimum recommended gain in the datasheet. It's also quite low in general, so you are barely amplifing your audio output. I suggest you bump up the gain to at least 10 dB

They actually are 562ohm resistors specified on mouser. I'd originally used my multimeter to somewhat match them up and they were in actuality 564ohm rounded up considering 1% tolerance.

Taking into account 564ohm the end result would be (rounded down) 24db if I did this right. While that too is below the gain recommendation it provides a suitable amount of gain without much background noise from the source.

I see you used 1500 uF Evercon ME (yikes! I suggest you use some other caps, as these tend to go more and more leaky over time - at least that's what happened to the GSC ME on my old Jetway motherboard).

Of course but on your MB though they were subjected to prolonged high temperatures. These are furthest from the IC and not directly connected to the heatsinking ground or directly connected within a few millimetres to the IC pin traces (unlike the rest) so they aren't subject to any high temps or stress. I picked them because of their effect on the sound output which is very pleasant as I'd tried many other caps in my collection and their max height/size. A couple I tried such as Sanyo WG 16v/1500uf (sounds only suitable for a subwoofer), and OST RLS 6.3/3300&RLX 16v1500 (which sounded nice but too tall!), along with some of my rubycon/nichicons the evercon's were the winner from my testing. They handled being overvolted to 6.5v when I ran the amp over 13.0v but that was cutting it too close for comfort as ~6.7v+ output is their point of suicide over a short time I found when one of the outputs were running high caused one to bulge.[/quote]

That's how I started as well. In fact, I still like to go "head-first" and just build the circuit rather than muck around on a computer simulator. But in college, they made us use MultiSim, and I got pretty used to it now, so I find it quite useful.
You know, I wouldn't be surprised. The Chinese are great at copying everything!

Oh okay. I hope everything goes well there.

hmm.. while I'm in recovery and overseas for a month I might just take a hand at tinkering with multisim then and see what I can do to pass the time then. I won't be a stranger during that month though as I made sure to have internet access paid for during that time.

[momaka]

Re: TEA2025B - Heavy noise over V++

Looks like you plain nailed it. I was mostly confused purely by the jωC1 as I didn't know what the jω was indicating.

j is an imaginary number - it is the square root of "-1".
ω is the radian frequency. ω = 2 * pi * f, where f is the normal frequency in Hz.

Not only now can I read it out easier and do the calculations but I learned something there. You did go above and beyond though which I feel bad about however but overall..

Glad to hear that was helpful

They actually are 562ohm resistors specified on mouser.

Lol, that sure got me tricked. Who thunk of exactly 562 Ohms as a necessary resistance?

Taking into account 564ohm the end result would be (rounded down) 24db if I did this right. While that too is below the gain recommendation it provides a suitable amount of gain without much background noise from the source.

Yup, that is correct. 24 dB is a pretty good gain too.

Of course but on your MB though they were subjected to prolonged high temperatures.

Actually, not really. They were on a cool-running socket A mobo. I had fans blowing on it in all directions too. Moreover, it wasn't just the CPU caps that read excess capacitance (i.e. typical when the internal leakage goes high), but every GSC cap I pulled out.
But yeah, for an audio application, that probably isn't all that critical. I just hope they don't get too leaky for you.

I won't be a stranger during that month though as I made sure to have internet access paid for during that time.

Nice. That should definitely help you keep yourself distracted and pass time faster.

[chozo4]

Re: TEA2025B - Heavy noise over V++

Lol, that sure got me tricked. Who thunk of exactly 562 Ohms as a necessary resistance?

Well... that's the funny part about it. I originally based the resistor value off the soundcard I had that was around 490ohm or so on the feedback. Then over time during rebuilds and changes I would lose the larger SMD resistors I had for it and the closest valus I had were slightly higher each time. So when I decided t buy some low tolerance resistors they didn't quite have many in stock at a reasonable price at 50ppm or lower in the 520-560 range so went with 562. Ended up going with regular through-hole resistors instead for ease of handling and manipulation.

Actually, not really. They were on a cool-running socket A mobo. I had fans blowing on it in all directions too. Moreover, it wasn't just the CPU caps that read excess capacitance (i.e. typical when the internal leakage goes high), but every GSC cap I pulled out.
But yeah, for an audio application, that probably isn't all that critical. I just hope they don't get too leaky for you.

If push comes to shove I might order the only other matching alternative I can find on the market in the same package size. It's the currently now discontinued [ 6.3v/1500uf Nichicon HZ Series ] of capacitors and the suggested replacements//upgrade are the HE/HD series with 20mm height so too tall for the intended application.

Mouser has a bunch on order that won't be shipped till November though so figured I'd best backorder them when I can before they are no longer available. This way if anything happens I have them on hand as a replacement. Also ended up looking into the higher capacitance offered by CDE/CornellDubilier but they seem to be only on special order.

Edit: Got the bugger cut out and cleaned up. Quarter and TicTac container for scale. Though it's so small one can make a tic-tac amp with it. Heck, it would fit an altoid's smalls tin no prob for that matter while using a couple small superflat capacitors in it for power such as a pair of the Maxwell flat 10F caps or the even smaller 1-2 Farad ones.

[momaka]

Re: TEA2025B - Heavy noise over V++

Quarter and TicTac container for scale. Though it's so small one can make a tic-tac amp with it.

That is neat!

I myself am working on a headphone class A amp right now (again). Got it down to one BJT and one MOSFET per channel. If everything works, I might be able to make something quite small as well. However, since it's class A, I'll still need big heatsinks and probably big resistors on the output stage - no way a Tic-Tac box will work for that .
Just tested the design yesterday. It was consuming close to 1.5W, and that's per single channel, too! . I haven't done any other calculations or measurements on it yet, but I expect it to be able to deliver a maximum of about 100 mW or so in 33 Ohms.... so yeah, we are talking about single-digit efficinecies here . But that's just the nature of the plain simple class A beast .

[chozo4]

Re: TEA2025B - Heavy noise over V++

That is neat!

I myself am working on a headphone class A amp right now (again). Got it down to one BJT and one MOSFET per channel. If everything works, I might be able to make something quite small as well. However, since it's class A, I'll still need big heatsinks and probably big resistors on the output stage - no way a Tic-Tac box will work for that .
Just tested the design yesterday. It was consuming close to 1.5W, and that's per single channel, too! . I haven't done any other calculations or measurements on it yet, but I expect it to be able to deliver a maximum of about 100 mW or so in 33 Ohms.... so yeah, we are talking about single-digit efficinecies here . But that's just the nature of the plain simple class A beast .

Oh goodness! Perhaps you can downsize the resistors a bit if you consider wirewound or at least metal film resistors as they are able to dissipate more heat. As for the fets, perhaps see if you can get away with TO-92 BJT's and N-Channel Fets for the job? So far I've been finding that the D965 BJT's thus far are rather nice for the job and don't generate any heat despite pushing 6.5w through it as a feedback noise filter. They are apparently commonly used in disposable cameras which I got a few discarded cases from for free from Walgreens in the area.

I'm curious if you considered showcasing the work in a thread of your own for it? Would be nice to see the progress and results from it. I've considered a Class-A before but been fine with Class-AB as desired compromise.

-> D965 Datasheet attached.

[momaka]

Re: TEA2025B - Heavy noise over V++

Oh goodness! Perhaps you can downsize the resistors a bit if you consider wirewound or at least metal film resistors as they are able to dissipate more heat.

Nah, I like big heatsinks and oversized components. Many of my junk components come from... well junk, like TVs and whatnot. So I don't mind reusing slightly bigger and over-specced components anyways.

As for the fets, perhaps see if you can get away with TO-92 BJT's and N-Channel Fets for the job?

That's what I have so far. TO-92 NPN BJT for the amplification stage and TO-252 N-Ch MOSFET for the buffer. Almost all of the dissipation is happening in the buffer output stage.

I'm curious if you considered showcasing the work in a thread of your own for it? Would be nice to see the progress and results from it.

Well the one I posted above is a work in progress I just started a few days ago, so no schematics or anything yet (I simply sit down and built it like you do ). However, I did built and showcase an all-BJT headphone amp here before:
https://www.badcaps.net/forum/showthread.php?t=32367
Ever since that post, though, I did simplify that design but just didn't post it and scope it like I did with the one in the thread above.
I'm planning to do the same with this one as well. But it would probably have to wait until I start classes in the univeristy again, as that's when I will have access to a scope. Should be in a few weeks from now.

I've considered a Class-A before but been fine with Class-AB as desired compromise.

Class AB is perfectly fine with me as well. However, when it comes to headphone amplifier ICs, some are just plain lousy (like the AN5272 I used in my first headphone amplifier), others are okay like the TEA2025B and TDA1517, and then there's the LM1877 that Th3_uN1Qu3 mentioned, which is much much better specced. So depending on what you pick, the sound quality can vary a bit. While I still use my AN5272-based amp daily, it leaves a lot to be desired, particularly in how much hiss it generates. Then again, it is an amp IC out of a TV board and meant to drive 8 Ohm speakers. But anyways, I digress here (that thread is also on the BCN forums, if interested). My point is that it is very easy to make a good-sounding class-A headphone amplifier without much skills.

[kaboom]

Re: TEA2025B - Heavy noise over V++

I swapped in the bigger transistor (MJE3055T) without much change at all which shows the base current is too low still *shrug*

Don't use a 3055. For this capacitance multiplier (your noise isolator), you need rather high gain and while passing reasonable current from C-E.

Try something like a D44H11 with a 68-100 ohm resistor from C-B.

It's possible that you don't even need the cap multiplier.
You can determine this by isolating each section and testing independently.

First, disconnect the source ends of the shielded cables and connect the center conductors to the shield with 100 ohm resistors. Does this make any difference in the noise?

In a computer, the amplifier's ground, if it's on its own board, and the source ground are not at the same potential. Your "ground" could in fact be sending noise currents, from the MB, into the local ground of the amp, and also coupling into the center conductor of the signal cable.

If it's quiet after this little test, you should be able to use those same 100 ohm (or as high a 560 ohm) resistors to isolate the source.

Solidly connect both center and shield at the source end of the cables. At the amp, do not connect the shield- the return path is thru the amp gnd, into the power supply, then back out of the power supply, into the source stage on the MB. With the center conductor at the amp, connect it to one end of the resistor. The other end of that isolation resistor connects to the "hot" side of the volume control(s).

The idea is to provide a path that's too high an impedance for the loop currents, while still being low enough for the signal.

[chozo4]

Re: TEA2025B - Heavy noise over V++

Nah, I like big heatsinks and oversized components. Many of my junk components come from... well junk, like TVs and whatnot. So I don't mind reusing slightly bigger and over-specced components anyways.

Same here actually - most of my parts if I don't have them on hand are salvaged from various junk. The majority of my transistors, capacitors (Rubycon YK's/0.1~0.22uf Polyesters, etc), diodes, 1/8w resistors, etc are from wrecked boards from a large projection TV someone had out on the curb. Some of the inductors/transformers were literally ripped from the board while the TV itself was smashed apart (was intact the day before). It's a shame the Fresnel lens was destroyed... would have been nice for increasing solarpanel output.

I do say though that for a final build depending on component size, the Radioshack 276-168B Universal PCB is perfect for such things long as one isn't averse to cutting traces to make it much more versatile. It's the exact board I used for my amp and other things while only costing about $3.50each. I buy them on a regular basis and love them - though the traces aren't the most durable for excessive component swapping and heating.

Don't use a 3055. For this capacitance multiplier (your noise isolator), you need rather high gain and while passing reasonable current from C-E.

Try something like a D44H11 with a 68-100 ohm resistor from C-B.

It's possible that you don't even need the cap multiplier.
You can determine this by isolating each section and testing independently.

First, disconnect the source ends of the shielded cables and connect the center conductors to the shield with 100 ohm resistors. Does this make any difference in the noise?

In a computer, the amplifier's ground, if it's on its own board, and the source ground are not at the same potential. Your "ground" could in fact be sending noise currents, from the MB, into the local ground of the amp, and also coupling into the center conductor of the signal cable.

If it's quiet after this little test, you should be able to use those same 100 ohm (or as high a 560 ohm) resistors to isolate the source.

Solidly connect both center and shield at the source end of the cables. At the amp, do not connect the shield- the return path is thru the amp gnd, into the power supply, then back out of the power supply, into the source stage on the MB. With the center conductor at the amp, connect it to one end of the resistor. The other end of that isolation resistor connects to the "hot" side of the volume control(s).

The idea is to provide a path that's too high an impedance for the loop currents, while still being low enough for the signal.

I actually had it resolved with a completely silent output a few posts after the one you quoted from actually. Any reference to power-ground with it would bring the noise back in full force as it would cause a ground loop with the input ground. Any transistor including the MJE did the task quite well but was just a matter of output current and tweaking the resistor value. As you claim the ground from the MB was indeed injecting noise into the amplifier. However, adding a resistor for the power ground to it would still add some noise to it as both the ground and power were both injecting the noise. Removing the power ground and silencing the V+ with the modified capacitance multiplier did the trick. I've also put a diode on the power ground as well accordingly to still have a power ground reference if the input ground were disconnected. Additionally there are no volume controls as can be seen in the posted pictures of the amp and I have the gain fixed with resistors. However, thanks for the input regarding it and do appreciate it!

I have since changed out the MJE with a TO-92 D965 BJT which I linked the datasheet for in the last post of mine. Only change after was increasing the resistance a small bit due to the massive difference in gain between that and the MJE.

Quoted resolution post for reference...

The end result is a minor variation of the 'capacitance multiplier' circuit but the ground is the output instead so it is in no way isolated but rather reduces noise through base feedback. There is only a 1.1v dropout with this method as per the saturation voltage on load. Reducing the resistor more results in little to no reduction of the noise due to breaching the saturation gap.

May try with a lower saturation voltage transistor at another time to be able to increase voltage output more without using an LDO. it only takes 3 components being a 150ohm 1/8w resistor, 16v/270uf Polymer cap, ST MJE2055T BJT. Will look into replacing the MJE with a 1A+ TO-92 package BJT to reduce size as the 2SC1213 using this method isn't providing enough output unless I double them up.

[chozo4]

Re: TEA2025B - Heavy noise over V++

hmm.. resurrecting the thread for updates.

There are 2 pairs of bare traces just before the outputs I can use to mount ta ripple filter on the underside without increasing the size.

Got a hold of some linear regulators I might end up using for it. It's an LM317DCYR by TI in an SOT-223-4 package that I can mount flush with the underside. Being as component space is of a premium there I'm pre-testing off board with a TO-220 LM317T by ST.

Since the only purpose I'll be using the regulator for is ripple rejection/noise filtering rather than regulation I could probably get away with not connecting the ADJ pin I believe and just use the existing output decouple. This will reduce the space requirements considerably.

-----------------------------------------------------------------

edit: Doesn't make as much difference with ripple rejection and would require more components for proper rejection overall. Seems my only option may just be a capacitance multiplier afterall which already does the task well off-board. Which then begs the question...

... in the case of a basic capacitance multiplier for ripple rejection would a Darlington be a better idea compared to using a BJT for ripple rejection purposes? Still pondering this as if I recall the output voltage may sag more with a darlington in comparison with a BJT. If this isn't exactly the case.. then a darlington should allow me to use a much smaller capacitor between the base and emitter than with the BJT as of currently.

[chozo4]

Re: TEA2025B - Heavy noise over V++

Blah, cannot edit post at this point *shrug* Anyhow.. Tip120 worked quite well due to a minimum flat DC gain of 1000 up to 3A/3v (so 1000 gain at 9 watts). However the saturation voltage caused a much larger hit to the voltage output as expected. As promised though it definitely held out well with the gain allowing full noise suppression with even a 10uf capacitor. Could probably go as low as a 6.8uf since it seems I need a about 5k-6k UF worth of capacitance via GAIN*uF. In the case of this capacitance multiplier for filtering there is no ground but instead the caps negative lead is connected to the emitter as the output voltage is always lower than that of the base.

Basing on what I figure so far and tried...
tip120, gain ~1000@3v3a(9w), -1.5~2.0v, 6-8uf~10uf cap (6.8k~10k)
mje3055t, gain ~20@4v4a(12w), -0.7~1.0v, 150~270uf cap (3.0k~5.4k)
*2sd965, gain ~40@?v?A(8w), -0.9v~1.25v, 270uf cap
--> estimated / sharp decline in gain after 1watt

So the search is on for a low dropout/high gain SOT-223 NPN-Transistor that can handle at least 1AMP input current. So far one likely candidate seems to be the BSP52T1G with up to 2k gain @ 500ma/10v (5W) but a 1.9v vBE saturation voltage. Up to 1k gain@8W with under 1v saturation would be ideal for my liking but won't likely find it in a darlington so sacrifices need to be made. Higher the gain the smaller the capacitor the smaller the footprint.

I know I've been at this for a while on and off.. I'm just trying to make this amp a perfect compact all in one solution for my needs without worry of line filtering should I use it in something else at any point. I tend to be too much of a perfectionist unfortunately and a project is never finished until I'm satisfied... this is one of them.

[ben7]

Re: TEA2025B - Heavy noise over V++

Blah, cannot edit post at this point *shrug* Anyhow.. Tip120 worked quite well due to a minimum flat DC gain of 1000 up to 3A/3v (so 1000 gain at 9 watts). However the saturation voltage caused a much larger hit to the voltage output as expected. As promised though it definitely held out well with the gain allowing full noise suppression with even a 10uf capacitor. Could probably go as low as a 6.8uf since it seems I need a about 5k-6k UF worth of capacitance via GAIN*uF. In the case of this capacitance multiplier for filtering there is no ground but instead the caps negative lead is connected to the emitter as the output voltage is always lower than that of the base.

Basing on what I figure so far and tried...
tip120, gain ~1000@3v3a(9w), -1.5~2.0v, 6-8uf~10uf cap (6.8k~10k)
mje3055t, gain ~20@4v4a(12w), -0.7~1.0v, 150~270uf cap (3.0k~5.4k)
*2sd965, gain ~40@?v?A(8w), -0.9v~1.25v, 270uf cap
--> estimated / sharp decline in gain after 1watt

So the search is on for a low dropout/high gain SOT-223 NPN-Transistor that can handle at least 1AMP input current. So far one likely candidate seems to be the BSP52T1G with up to 2k gain @ 500ma/10v (5W) but a 1.9v vBE saturation voltage. Up to 1k gain@8W with under 1v saturation would be ideal for my liking but won't likely find it in a darlington so sacrifices need to be made. Higher the gain the smaller the capacitor the smaller the footprint.

I know I've been at this for a while on and off.. I'm just trying to make this amp a perfect compact all in one solution for my needs without worry of line filtering should I use it in something else at any point. I tend to be too much of a perfectionist unfortunately and a project is never finished until I'm satisfied... this is one of them.

You should just go to an LDO. There are some with only like 300 to 700mV drop at 1A. They also will filter out noise better, and they don't need very large capacitors. Or you could just use a regular linear regulator (like an LM7812) fed by a DC wall adapter. You might need a small heatsink on the regulator though.

[budm]

Re: TEA2025B - Heavy noise over V++

Some info on designing cap multiplier.
http://sound.westhost.com/project15.htm

http://www.radio-electronics.com/inf...er-circuit.php