The 0.1 uF and 22 uF capacitors that are connected to pins 6 (PWR) and 1 (FB) need to have their bodies connected very close to where the pins enter the chip's body. No long leads! About 1 mm or less between cap body and chip body would be best, but it might be safer to leave just enough chip pin length to clamp the small tip of long-nose pliers on the pin (e.g. use rubber band around handles), just where it enters the chip body, to act as a heatsink while soldering so close to chip.
But mainly, you also really do need to have a 220 uF or larger electrolytic capacitor connected from pin 6 to ground (in parallel with the 0.1 uF). Your schematic does not show those. They HAVE to be used, there, to prevent oscillation, which you are obviously experiencing. And there should be one large electrolytic for each chip, since they need to be connected as close as practical to the power pin.
Try both of the pin 6 requirements, above, first. If those don't fix the oscillation, then try the pin 1 requirement above. If those don't work, then try adding a 1K resistor in series with the input, as the last thing before the chip input pin and connected very close to where the pin enters the chip body. If none of those work, then maybe it's high frequency bursts so try two 1K or so resistors in series, in series with the input signal, connected to/at the chip input pin, with at least a few hundred pF cap to ground from between the resistors.
All of the above apply to both chips.
Cheers,
Tom
Edit: A photo could be very helpful.
P.S. Verify that your output Zobel networks have the correct component values, i.e. 2.2 Ohm resistors and 0.22 uF capacitors.
But mainly, you also really do need to have a 220 uF or larger electrolytic capacitor connected from pin 6 to ground (in parallel with the 0.1 uF). Your schematic does not show those. They HAVE to be used, there, to prevent oscillation, which you are obviously experiencing. And there should be one large electrolytic for each chip, since they need to be connected as close as practical to the power pin.
Try both of the pin 6 requirements, above, first. If those don't fix the oscillation, then try the pin 1 requirement above. If those don't work, then try adding a 1K resistor in series with the input, as the last thing before the chip input pin and connected very close to where the pin enters the chip body. If none of those work, then maybe it's high frequency bursts so try two 1K or so resistors in series, in series with the input signal, connected to/at the chip input pin, with at least a few hundred pF cap to ground from between the resistors.
All of the above apply to both chips.
Cheers,
Tom
Edit: A photo could be very helpful.
P.S. Verify that your output Zobel networks have the correct component values, i.e. 2.2 Ohm resistors and 0.22 uF capacitors.
Last edited:
hm... the pin 6 connection mentioned, I DO have 2 x 220uF parallel with power and ground. I even also added a super big reservoir capacitor (2.2MF), but it still have same problem.(Although increase sonic quality, going to solder later)
I'd tried to make the cap at pin 1 at close as possible, its within 1cm, but 1mm sure is unachievable. I will try the resistor plan, and i think it will work effectively(probably) because by using volume control VR, the noise will be reduce, if sufficient then it will reduce.
Maybe i know to theory, how it works ? (using resistor, and why add a cap when a series resistor used)
I'd tried to make the cap at pin 1 at close as possible, its within 1cm, but 1mm sure is unachievable. I will try the resistor plan, and i think it will work effectively(probably) because by using volume control VR, the noise will be reduce, if sufficient then it will reduce.
Maybe i know to theory, how it works ? (using resistor, and why add a cap when a series resistor used)
will try out and report after i'm free and get a chance to do that.
But for sure, same netbook, different amplifier have different result. All DIY amplifier (made myself with following schematic ONLY will have noise, but different pattern, probably different input impedance.
But using commercial build amplifier (both lousy and good speaker), I don't have any problem so far (maybe undetected).
220pF cap to ground, series resistor, 220pF cap to ground, series resistor
Its an RC filter to shunt noise to ground. The attempt is to make the resistor unable to pass high frequency noise.
tried, I think not problem of high frequency and resistor only reduce the noise, but also reduce the quality of sound (frequency cutoff).
Anyway, I have problem in bass, or should i say it was easy to reach clip, even its not loud. (I even tried using 22000uF Reservoir, 2,200uF also won't work much (increase alot compare to no reservoir capacitor) but really not much different between 22,000 and 2,200uf reservoir, so i think is not problem of reservoir capacitance(not enough energy). Any idea whats the problem ?( most probably lies in chip.
Anyway, I have problem in bass, or should i say it was easy to reach clip, even its not loud. (I even tried using 22000uF Reservoir, 2,200uF also won't work much (increase alot compare to no reservoir capacitor) but really not much different between 22,000 and 2,200uf reservoir, so i think is not problem of reservoir capacitance(not enough energy). Any idea whats the problem ?( most probably lies in chip.
220pF cap for shunt. . . Can you hear pitches as high as 44kHz? Gosh, that would really be a curse because all CDs would sound like burglar alarms, and a tripath amp about like a church bell in a tumble dryer for hours. lolz! No, the little shunt cap won't affect the audio band.
Have a look at some amplifier datasheets and amplifier schematics and projects and see where to put the 1k resistors. The 1k would be the only thing touching the amp input pin--put the stuff that was previously on the input pin onto the opposite side of the 1k. Here, it does nothing but drop excess current. Elsewhere it makes accidental tone control.
Last edited:
For me, the amplifier isn't going to clip about loudness i heard, even the power supply is not high. Seems it is driving less than 4watt, and clip at 11V is funny (its true that my power supply voltage drop to roughly 11V when louder, when current consumption over 0.7A)
I think i'm going to isolate this chip, since it has many problem and not designed for home application (cannot adjust gain, etc)
I was going to dump it aside, but previously I brought it and was not cheap. Each channel chip at the price of one MJL3281.... I personally prefer transistor, haha !
But i have another chip That i brought (TDA2009A), seems a bit more better, and its labeled Hi-Fi (maybe during that time)
http://www.datasheetcatalog.org/datasheet/stmicroelectronics/1457.pdf
Any 'PRE-caution' i need to take care of, especially building aspect ?
Oh yes, of course, with TDA2009A
For Stereo amp to 2 speakers it wants ~23VDC power
For bridge amp to one 8 ohm speaker it wants 18VDC power
For bridge amp to one 4 ohm speaker it wants 16VDC power
Significant amperage is required.
The datasheet is an interesting read!
Its not difficult to read And it has interesting amplifier applications.
That should be good. Have fun!
For Stereo amp to 2 speakers it wants ~23VDC power
For bridge amp to one 8 ohm speaker it wants 18VDC power
For bridge amp to one 4 ohm speaker it wants 16VDC power
Significant amperage is required.
The datasheet is an interesting read!
Its not difficult to read And it has interesting amplifier applications.
That should be good. Have fun!
Sounds like your power supply was far too weak. That was probably the whole problem.
Anyway, just so you know, the reason for suggesting that a resistor be the last thing before the input, and for it being close to the pin, was that I wasn't sure where that input went, inside the chip. If it happened to connect to a negative amplifier input that resembled a negative opamp input, then even stray capacitance from a length of wire or pcb trace could make the amp unstable. So the resistor was meant to isolate the input pin from any stray capacitance, which wouldn't hurt anything in the case of a positive input.
Then, adding a 220 pF capacitor to ground, downstream from the resistor, would create a low-pass RF filter. But in case the pin needed to be isolated from capacitance, per the above, I suggested putting it between two resistors.
I would shoot for a low-pass RF filter cutoff frequency of at least 220 kHz to 440 kHz. The cutoff frequency should be at 1 / (2 x Pi x R x C), except that when placed just before an input pin of an opamp with a feedback loop you have to double the C value that you would use if there was just one R, in order to keep the same cutoff frequency. Not sure, with this chip.
I suggest ALWAYS using a low-pass RF filter, for every amplifier stage's input.
Anyway, just so you know, the reason for suggesting that a resistor be the last thing before the input, and for it being close to the pin, was that I wasn't sure where that input went, inside the chip. If it happened to connect to a negative amplifier input that resembled a negative opamp input, then even stray capacitance from a length of wire or pcb trace could make the amp unstable. So the resistor was meant to isolate the input pin from any stray capacitance, which wouldn't hurt anything in the case of a positive input.
Then, adding a 220 pF capacitor to ground, downstream from the resistor, would create a low-pass RF filter. But in case the pin needed to be isolated from capacitance, per the above, I suggested putting it between two resistors.
I would shoot for a low-pass RF filter cutoff frequency of at least 220 kHz to 440 kHz. The cutoff frequency should be at 1 / (2 x Pi x R x C), except that when placed just before an input pin of an opamp with a feedback loop you have to double the C value that you would use if there was just one R, in order to keep the same cutoff frequency. Not sure, with this chip.
I suggest ALWAYS using a low-pass RF filter, for every amplifier stage's input.
Haha, I'm agree that it is an interesting chip, going to try build it within next week.
I personally will go for stereo amp, using 22-24V supply approx. It also has low idle current than tda7240 (double mono make it more !)
Power supply far too weak ? I don't think computer power supply aren't that weak, maybe this amplifier consuming too much. (only car battery enough to feed it !)
About the low-pass RF filter, it is a RC network ? interested how to design one (for further use of other amplifier)
What if I didn't use this and let the RF to go through ? Will it cause oscillation, degradation of sound quality(frequency response), induce noises (hum/buzz), or overheat by giving other loads ?
If I were to make a low-pass and high-pass that only allow 20Hz to 25kHz sound roughly, what is the advantage or disadvantage of doing so ?
Then how should I make the input within audio bandwidth or at least nearly audio bandwidth (20Hz to 20kHz) ? at least this makes the it works less unnecessary works.
An R in series followed by C to ground makes a passive first-order low-pass filter with -3dB "cutoff" frequency at f = 1 / (2 x Pi x R x C), and 6 dB per decade slope after that. Put the cutoff freq at least 10x higher than any frequencies you don't want to be affected. Pi is approximately 3.14. R is in Ohms, f is in Hertz, C is in Farads.
Similarly, a series C followed by an R to ground makes a high-pass filter, with -3 dB frequency from same equation and with same roll-off slope.
You can cascade multiple stages to get steeper slope, but would probably want to put unity-gain (i.e. gain=1) opamp buffers after each of them, in which case you might as well download TI's free Filterpro software and use "active" opamp-based filters. But you can usually cascade one high-pass and one low-pass without needing a buffer.
Use polypropylene film capacitors (or teflon, or maybe polystyrene).
Typically you would put your high-pass f(-3dB) at 2 Hz or less and your low-pass at 220 kHz or more.
Google.com is your friend.
Similarly, a series C followed by an R to ground makes a high-pass filter, with -3 dB frequency from same equation and with same roll-off slope.
You can cascade multiple stages to get steeper slope, but would probably want to put unity-gain (i.e. gain=1) opamp buffers after each of them, in which case you might as well download TI's free Filterpro software and use "active" opamp-based filters. But you can usually cascade one high-pass and one low-pass without needing a buffer.
Use polypropylene film capacitors (or teflon, or maybe polystyrene).
Typically you would put your high-pass f(-3dB) at 2 Hz or less and your low-pass at 220 kHz or more.
Google.com is your friend.
Yes, they would be in series.
I would put the high-pass first, and then put the low-pass for RF right at the input pin or just before the input resistor, if there is one. Just remember that if it's a negative input pin for an opamp or chipamp, you need another resistor in series with the pin, to isolate the capacitor from the pin.
You could make steeper slopes but there is no need to do that, for this type of application, and it might even cause problems. And there are always trade-offs, to get the sharper cutoff.
However, just so you know, every time you cascade another simple first-order filter stage, the slope in the stop-band doubles (in dB per decade of frequency). BUT, you can't really just cascade those types of passive filter stages in series, UNLESS you use buffer amplifiers in between them. Basically, though, there are much better ways to make higher-order filters, with steeper slopes and with better control of their other characteristics.
See
ADI - Analog Dialogue | Op Amp Applications Handbook
which is a very nice free downloadable book from Analog Devices.
Also, as I already mentioned, Texas Instruments (ti.com) has a free download called filterpro. It designs nice opamp-based filters for you, up to 6th-order in the free version, I think.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.