For slew rate: see my earlier posts in this thread....
Cmorariu solved this problem adequatly!
You suggest that with a resistor of e.g. 2,5 Ohm the TDA comes out of his healthy working area? And so clipping?
Seems to me unreasonable. If the TDA produces its maximum power (about 68W), the ouput T's produce at least tenfold. So that can not be the clue.....
In the range Cmorariu tests he stays far under this limits, although his TDA becomes (too?) hot.....Cmorariu any smart thoughts?
@zanden30 I'm also in vacation now, so my thoughts may not be so smart.. Please remember again your VOLTAGE triangle, the tip is the TDA output, the bottom is GND, somewhere between is the output (load). Your proposal was to move the output point as close to the tip of the triangle as possible to have the load slew rate closer of the TDA slew rate. This is perfectly right, but you disregarded one thing - the size of triangle which is variable. Size of triangle is given by CURRENT through load and SYNC resistor.
Going to the limit, let's think what happens at crossover - the current is 0, so the triangle becomes one point. Doesn't matter where the point of load is placed, if follows the TDA. If the circuit is able to pass the burden of driving the load to transistors at lower currents this will happen at a small size of triangle, so difference in height (voltage) between load and TDA is not noticeable, load will closer follow the slew rate of TDA. (By contrary if you diminish SYNC the transfer of power will happen at higher currents, and, even if the load will proportionally follow the TDA closer, the difference in voltage between them will be higher).
@Sfthurber explanation is a synthesis of this, I like it: "at higher currents you may see slew rate effects that limits the chip's ability to respond to feedback and clean up the mess".
Dr.Frost found the optimal value of SYNC resistor by listening. Of course this value was dependent of the impedance of his speakers. In fact SYNC resistor depends of many things, TDA slew rate, impedance of load, Ube of power transistors, voltage of power supply, TDA admissible current, etc.
That's why, my enhanced circuit uses an AB class stage overlapped over the classic SYNC-power transistors and makes amplifier's performance less dependent on load impedance and component disperssion.
Going to the limit, let's think what happens at crossover - the current is 0, so the triangle becomes one point. Doesn't matter where the point of load is placed, if follows the TDA. If the circuit is able to pass the burden of driving the load to transistors at lower currents this will happen at a small size of triangle, so difference in height (voltage) between load and TDA is not noticeable, load will closer follow the slew rate of TDA. (By contrary if you diminish SYNC the transfer of power will happen at higher currents, and, even if the load will proportionally follow the TDA closer, the difference in voltage between them will be higher).
@Sfthurber explanation is a synthesis of this, I like it: "at higher currents you may see slew rate effects that limits the chip's ability to respond to feedback and clean up the mess".
Dr.Frost found the optimal value of SYNC resistor by listening. Of course this value was dependent of the impedance of his speakers. In fact SYNC resistor depends of many things, TDA slew rate, impedance of load, Ube of power transistors, voltage of power supply, TDA admissible current, etc.
That's why, my enhanced circuit uses an AB class stage overlapped over the classic SYNC-power transistors and makes amplifier's performance less dependent on load impedance and component disperssion.
Dear cmorariu,
I really cannot follow your reasoning.
The only reason why a too little Rsync will introduce distortion is, as far I can see, because the TDA is in saturation or clipping.
If Rsync = 0, the boost circuit will not work, but the TDA and its feedback loop still do their work, so the TDA will work conform its specs in the datasheet. No distortion above the level of the datasheet should be there!
If you increase the Rsync infinitesimal, the boost circuit will take over some of the current from the TDA at maximum current values. Still there is no reason for additional distortion. The more you increase the Rsync, the more current will be taken over by the boost circuit.
If the Rsync is increased to a value, so that the slew rate of the TDA is too low to follow during the cross over point (as I reasoned before), distortion will be introduced.
This implies: the highest value of Rsync is determined by the ratio of Rload and Rsync in combination with the highest rise/fall of the (maximum) volume/frequency of the signal in the crossing point versus the slew rate of the TDA.
The lowest value only seems, in my reasoning, dependable on the maximum current of the TDA. So: the smaller the Rsync the better, but it implies that the working of the power boost circuit diminishes and less power can be delivered.
This is, alas, the smartest reasoning that I have on the moment about the boost circuit, which I would like to fully understand.
Your improvement with the medium power additional boost circuit is nice. But I think the same result could be reached, if you introduce a quiescent current in the power circuit with 2 or 3 diodes and connect the output of the TDA with 2 big capacitors to the bases of the output power transistors.
Did you try that? In that case parallel pairs of Darlingtons (like TIP142/7) come in view, which could be used to increase the power even further, because the driving current by the TDA is only a fraction of the current necessary to drive the hFE = 15 .. 30 power transistors…..
Kind regards,
Jan
I really cannot follow your reasoning.
The only reason why a too little Rsync will introduce distortion is, as far I can see, because the TDA is in saturation or clipping.
If Rsync = 0, the boost circuit will not work, but the TDA and its feedback loop still do their work, so the TDA will work conform its specs in the datasheet. No distortion above the level of the datasheet should be there!
If you increase the Rsync infinitesimal, the boost circuit will take over some of the current from the TDA at maximum current values. Still there is no reason for additional distortion. The more you increase the Rsync, the more current will be taken over by the boost circuit.
If the Rsync is increased to a value, so that the slew rate of the TDA is too low to follow during the cross over point (as I reasoned before), distortion will be introduced.
This implies: the highest value of Rsync is determined by the ratio of Rload and Rsync in combination with the highest rise/fall of the (maximum) volume/frequency of the signal in the crossing point versus the slew rate of the TDA.
The lowest value only seems, in my reasoning, dependable on the maximum current of the TDA. So: the smaller the Rsync the better, but it implies that the working of the power boost circuit diminishes and less power can be delivered.
This is, alas, the smartest reasoning that I have on the moment about the boost circuit, which I would like to fully understand.
Your improvement with the medium power additional boost circuit is nice. But I think the same result could be reached, if you introduce a quiescent current in the power circuit with 2 or 3 diodes and connect the output of the TDA with 2 big capacitors to the bases of the output power transistors.
Did you try that? In that case parallel pairs of Darlingtons (like TIP142/7) come in view, which could be used to increase the power even further, because the driving current by the TDA is only a fraction of the current necessary to drive the hFE = 15 .. 30 power transistors…..
Kind regards,
Jan
thought of a possibly more accessible and thorough way to state the same thought:
slew rate limits the chip's ability rapidly to respond to both -IN and +IN input.
-IN will be particularly challenging in this design given the large amount of added spurious signal products generated in the very non-linear bipolar output stage.
given that slew rate is not a fixed parameter (in the 7294 datasheet it is stated only for a fixed environment with specified voltages and a pure resistive load of 8R), the reality of this design may be more complex than it first appears with respect to the influence of slew rate on fidelity under various loads. doubtless there is a sweet spot load. the datasheet won't reveal it so experiments will be necessary to discover it.
Last edited:
I've got a 5A transformer with 29-27-0-27-29 and .5A 11-0-11 winding. I think I'm gong to try this.
On eBay this looks like a good board to play with. (make an offer, he accepted mine!)
TDA7293 X2 100 100W Digital Stereo Audio Amplifier Board 2 0 with Cable HiFi | eBay
I'll source capacitors and outputs locally and build a separate PS.
These are the only sync resistors I could find, wish I had selection to choose from.
10pcs 5W 5 w Metal Film Resistor 5 5 6 6 2 6 8 7 5 8 2 10 22 47 100 150 Ohm | eBay
I've bought 2 so the idea is to build a single channel, on one board to test. Then try and BTL the other board. I've got a pair of old 15" 16Ohm speakers to parallel. I've got an old CRT scope, and a function generator but I haven't used them yet.
I expect the chips are fake and the caps are rubbish, bit it looks like a fun project.
This interests me. How would I go about making the whole amp into a low pass filter?
On eBay this looks like a good board to play with. (make an offer, he accepted mine!)
TDA7293 X2 100 100W Digital Stereo Audio Amplifier Board 2 0 with Cable HiFi | eBay
I'll source capacitors and outputs locally and build a separate PS.
These are the only sync resistors I could find, wish I had selection to choose from.
10pcs 5W 5 w Metal Film Resistor 5 5 6 6 2 6 8 7 5 8 2 10 22 47 100 150 Ohm | eBay
I've bought 2 so the idea is to build a single channel, on one board to test. Then try and BTL the other board. I've got a pair of old 15" 16Ohm speakers to parallel. I've got an old CRT scope, and a function generator but I haven't used them yet.
I expect the chips are fake and the caps are rubbish, bit it looks like a fun project.
This interests me. How would I go about making the whole amp into a low pass filter?
the pictures show a bridge rectifier on the heatsink between the chips.
so if the circuit is arranged correctly it may work with AC although the specified 32-0-32 may push the rail voltage too high for safe operation.
the board layout has obvious issues.
look where it has the input -- how smart could it be to put the low level signal right in the middle of all the power traces and components?
and look at the gigantic signal and NFB loop areas!
so if the circuit is arranged correctly it may work with AC although the specified 32-0-32 may push the rail voltage too high for safe operation.
the board layout has obvious issues.
look where it has the input -- how smart could it be to put the low level signal right in the middle of all the power traces and components?
and look at the gigantic signal and NFB loop areas!
since the chip is not unity gain stable it likely will work poorly or fail entirely if operated as an active filter. in a low pass active filter the gain declines with frequency asymptotically approaching zero while NFB increases with frequency asymptotically approaching 100%.
Last edited:
That sounds exactly like the behaviour of an ideal bass amp. I would think limiting the bandwidth to the required frequency range would increase stability?
perhaps you will find these helpful
TI white paper (a brilliant work IMHO). you can ignore the cascading aspects if you don't plan to cascade to get higher order filtering
http://www.ti.com/lit/an/sloa049b/sloa049b.pdf
DIY thread
http://www.diyaudio.com/forums/parts/188148-whats-current-best-opamp-active-filtering.html
TI white paper (a brilliant work IMHO). you can ignore the cascading aspects if you don't plan to cascade to get higher order filtering
http://www.ti.com/lit/an/sloa049b/sloa049b.pdf
DIY thread
http://www.diyaudio.com/forums/parts/188148-whats-current-best-opamp-active-filtering.html
Last edited:
that "glom" is a typical passive filter
should be easy to find a DIY thread on passive filters for subwoofer amps
and to just close the loop (no pun intended) on the white paper--
you can see that the high-frequency pass through issues will be quite difficult since the TDA requires above-unity gain, which takes up some the open loop gain normally devoted to reducing the pass-through. worse yet, the TDA open loop gain falls of at a relatively low frequency compared to the op amp examples in the white paper. the result is a real mess.
should be easy to find a DIY thread on passive filters for subwoofer amps
and to just close the loop (no pun intended) on the white paper--
you can see that the high-frequency pass through issues will be quite difficult since the TDA requires above-unity gain, which takes up some the open loop gain normally devoted to reducing the pass-through. worse yet, the TDA open loop gain falls of at a relatively low frequency compared to the op amp examples in the white paper. the result is a real mess.
Last edited:
Greetings fellow diyers!
I have successfully build a gainclone based around two lm3886TF chips, which are running from a +- 37v (unregulated, but stiff) power supply, with 30.000μF per rail. The amp sounds fantastic on my 8 ohm speakers, though the heatsink and chips get pretty warm when playing loud.
I always wondered about what it would take to be able to drive 4 ohm speakers whithout the risk of triggering SPIKe or damaging the small plastic chips. When i discovered this thread, i got really enthusiastic and excited, and rushed to try out this very promising idea.
But, being a poor guy, i did not select the particular transistors proposed by the OP, but went for the cheap TIP35c/TIP36c complementary pair (two pairs, one per channel). I modified my circuit per instructions, and moved the feedback to the power transistor emiters.
As it turns out, i get horrible (i think, crossover) distortion! I have tried a lot of values for the Rsync, starting with 6.8 and going up to 3k and down to 0.1 ohms, without much success. When i lowered the value under 6.8 ohms, i had gradually less distortion, but only at low volumes. When the volume was increased, the distortion was again very obvious...
For the record, i used 0.1 ohm emiter resistors, and i have double-triple checked all the connections and i cannot find any fault. What could be wrong? What could i try, in order to make this work as it should? Any ideas? Could it be that the transistor pairs are mismatched, as they are not of the same package? I could not find exact pairs on my local supplier.
Thank you very much for any idea....
I have successfully build a gainclone based around two lm3886TF chips, which are running from a +- 37v (unregulated, but stiff) power supply, with 30.000μF per rail. The amp sounds fantastic on my 8 ohm speakers, though the heatsink and chips get pretty warm when playing loud.
I always wondered about what it would take to be able to drive 4 ohm speakers whithout the risk of triggering SPIKe or damaging the small plastic chips. When i discovered this thread, i got really enthusiastic and excited, and rushed to try out this very promising idea.
But, being a poor guy, i did not select the particular transistors proposed by the OP, but went for the cheap TIP35c/TIP36c complementary pair (two pairs, one per channel). I modified my circuit per instructions, and moved the feedback to the power transistor emiters.
As it turns out, i get horrible (i think, crossover) distortion! I have tried a lot of values for the Rsync, starting with 6.8 and going up to 3k and down to 0.1 ohms, without much success. When i lowered the value under 6.8 ohms, i had gradually less distortion, but only at low volumes. When the volume was increased, the distortion was again very obvious...
For the record, i used 0.1 ohm emiter resistors, and i have double-triple checked all the connections and i cannot find any fault. What could be wrong? What could i try, in order to make this work as it should? Any ideas? Could it be that the transistor pairs are mismatched, as they are not of the same package? I could not find exact pairs on my local supplier.
Thank you very much for any idea....
Attachments
I finally got the appropriate transistors (two pairs 2SC5200 / 2SA1943) and substituted my previous transistors, hoping for my setup to work as it should.
Well, it seems that nothing has changed... I still get horrible (I mean really bad, unbearable) crossover distortion, even at low volumes. That is with 6R8 SYNC resistors.
While i was experimenting, i observed some interesting behaviors...
First, when i increased the volume (through my computer soundcard), at some specific level (quite different for each channel), the speaker made a weird sound (like woooooSH) and then the woofer cone was sucked in, which tells me that it was recieving the full negative rail.... It wouldn't come back, even if i decreased the volume level, so i immediately had to switch off the amp.
Secondly, while i was experimenting with the SYNC resistor, i decided to completely omit it, leaving the BE junctions completely open. Well, the distortion at low volumes had almost completely dissapeared, with only some minor high pitch harmonics being barely audible, but when i increased the volume, the same thing happened again (Woooosh and woofers sucked in).
As always, i have double checked all the connections. No error, no shorted collectors to heatsink, nothing wrong was found.
I really want this to work!
Any ideas?
Well, it seems that nothing has changed... I still get horrible (I mean really bad, unbearable) crossover distortion, even at low volumes. That is with 6R8 SYNC resistors.
While i was experimenting, i observed some interesting behaviors...
First, when i increased the volume (through my computer soundcard), at some specific level (quite different for each channel), the speaker made a weird sound (like woooooSH) and then the woofer cone was sucked in, which tells me that it was recieving the full negative rail.... It wouldn't come back, even if i decreased the volume level, so i immediately had to switch off the amp.
Secondly, while i was experimenting with the SYNC resistor, i decided to completely omit it, leaving the BE junctions completely open. Well, the distortion at low volumes had almost completely dissapeared, with only some minor high pitch harmonics being barely audible, but when i increased the volume, the same thing happened again (Woooosh and woofers sucked in).
As always, i have double checked all the connections. No error, no shorted collectors to heatsink, nothing wrong was found.
I really want this to work!
Any ideas?
Continuing from my previous posts...
While experimenting on the above setup (with 6R8 SYNC resistors and 8 ohm speakers), it occured to me that i should place some reverse diodes across the chip output and the +&- rails, in order to eliminate the "woofer sucking" phenomenon. So i placed them and this particular problem dissapeared! But the HORRIBLE crossover distortion remains.
There is no distortion at all at low volumes (less than about 1watt), but when the volume is increased (to the point that the power transistors should be kicking in), the distortion also kicks in, and it's BAD!
One very interesting phenomenon that i observed while experimenting, is that while the amp was playing loud (with the horrible distortion), my desktop computer monitor is flickering!!!!!! The louder i go, the worse it gets. The image quality drops substantially and i can see flickering lines, waves and artifacts, especially on darker areas of the image! My monitor never does that at any other time. NOW THAT IS WEIRD !!!!!!!!!!!!!!!!!!!
I sense some kind of interference or EM radiation being picked up. I hope someone over here with advanced knowledge of electronics could come up with a sensible explanation, which hopefully could lead to a solution on my problem.
Thanks for any input!
While experimenting on the above setup (with 6R8 SYNC resistors and 8 ohm speakers), it occured to me that i should place some reverse diodes across the chip output and the +&- rails, in order to eliminate the "woofer sucking" phenomenon. So i placed them and this particular problem dissapeared! But the HORRIBLE crossover distortion remains.
There is no distortion at all at low volumes (less than about 1watt), but when the volume is increased (to the point that the power transistors should be kicking in), the distortion also kicks in, and it's BAD!
One very interesting phenomenon that i observed while experimenting, is that while the amp was playing loud (with the horrible distortion), my desktop computer monitor is flickering!!!!!! The louder i go, the worse it gets. The image quality drops substantially and i can see flickering lines, waves and artifacts, especially on darker areas of the image! My monitor never does that at any other time. NOW THAT IS WEIRD !!!!!!!!!!!!!!!!!!!
I sense some kind of interference or EM radiation being picked up. I hope someone over here with advanced knowledge of electronics could come up with a sensible explanation, which hopefully could lead to a solution on my problem.
Thanks for any input!
I think that the amp begins to oscillate at some combinations of loading and output currents.
That oscillation (if it happens) could be sending a lot of EMI back into your computer.
That oscillation will also change the bias conditions and that could be the cause of the output offset.
Disconnect the speaker and do some proper tests.
That oscillation (if it happens) could be sending a lot of EMI back into your computer.
That oscillation will also change the bias conditions and that could be the cause of the output offset.
Disconnect the speaker and do some proper tests.