Hi everyone
I want to put clip detector on my TDA 7293 amps and don't understand how. Tried to find answer on forum and google it but failed.
Can someone put a simple picture or explain how to make clip detector for that chip?
I want to put clip detector on my TDA 7293 amps and don't understand how. Tried to find answer on forum and google it but failed.
Can someone put a simple picture or explain how to make clip detector for that chip?
Ice, In the December 1989 edition of EA magazine, Rob Evans in his presentation of the PS1 MosFet amplifier, describes a very clever and ingenious distortion indicator. His description may help you to adapt the simple circuit to your needs:
. . . . we have also included a novel overload detector circuit based around transistors Q13 and Q14. While its operation is very straightforward, the circuit utilises a complex effect which occurs within the amplifier during overload conditions. In practice, the circuit will energize LED1 if the amp's distortion rises above about 0.05% - a lower figure than many amplifiers can ever hope to achieve!
The signal at the collector of Q5 (point A) represents the amplifier's overall error or correction voltage. This occurs since the other half of the differential pair Q6 generates the full output signal swing - if we consider the MOSFETs as electrically transparent for the moment. Now if the signal at the collector of Q6 represents the corrected output (that is, less the distortion components), and Q5 and Q6 represent a true differential amplifier, we can expect the signal difference to appear at the collector of Q5 - or in effect, the distortion components which are cancelled at the output - thanks to the overall NFB.
The circuit basically detects transient AC signals, extends the pulse length and illuminates the LED for that period. The signal from point A is AC coupled to Q13 by C19 and R25 (which raises the circuit's input impedance). When sufficient energy arrives to bias Q13 ON momentarily, its collector falls - charging C20. This stored voltage holds Q14 in conduction for an extended period, while C20 is discharged by R27 and the transistor's base current. During this period, the collector of Q14 rises to the full supply voltage (connection B), energizing LED1 via the current limiting resistor R28.
In practice this effect is quite easy to detect with an oscilloscope. If for example, the amplifier is driven into clipping, large spikes are generated at point A as the NFB attempts to compensate for the flattened peaks at the output. So in general, whenever the distortion is high at the output (for whatever reason), sufficient voltage is generated at point A to drive the overload indicator circuit.
Unlike conventional overload indicators which simply sense the output swing, this simple circuit will warn the user of any unacceptable (and dangerous) distortion components - and of course its action is true for any load conditions and supply rail voltages.
An externally hosted image should be here but it was not working when we last tested it.
. . . . we have also included a novel overload detector circuit based around transistors Q13 and Q14. While its operation is very straightforward, the circuit utilises a complex effect which occurs within the amplifier during overload conditions. In practice, the circuit will energize LED1 if the amp's distortion rises above about 0.05% - a lower figure than many amplifiers can ever hope to achieve!
The signal at the collector of Q5 (point A) represents the amplifier's overall error or correction voltage. This occurs since the other half of the differential pair Q6 generates the full output signal swing - if we consider the MOSFETs as electrically transparent for the moment. Now if the signal at the collector of Q6 represents the corrected output (that is, less the distortion components), and Q5 and Q6 represent a true differential amplifier, we can expect the signal difference to appear at the collector of Q5 - or in effect, the distortion components which are cancelled at the output - thanks to the overall NFB.
The circuit basically detects transient AC signals, extends the pulse length and illuminates the LED for that period. The signal from point A is AC coupled to Q13 by C19 and R25 (which raises the circuit's input impedance). When sufficient energy arrives to bias Q13 ON momentarily, its collector falls - charging C20. This stored voltage holds Q14 in conduction for an extended period, while C20 is discharged by R27 and the transistor's base current. During this period, the collector of Q14 rises to the full supply voltage (connection B), energizing LED1 via the current limiting resistor R28.
In practice this effect is quite easy to detect with an oscilloscope. If for example, the amplifier is driven into clipping, large spikes are generated at point A as the NFB attempts to compensate for the flattened peaks at the output. So in general, whenever the distortion is high at the output (for whatever reason), sufficient voltage is generated at point A to drive the overload indicator circuit.
Unlike conventional overload indicators which simply sense the output swing, this simple circuit will warn the user of any unacceptable (and dangerous) distortion components - and of course its action is true for any load conditions and supply rail voltages.
Thank you Andrew, but that is not what I'm looking for. TDA's pin 5 is named clip detector but I don't understand how to use that pin. That is my question. Do I connect LED on it or what?
If you read the datasheet, the Electrical Characteristics table give a hint about the function of the clip detect pin. I reproduce that below, re-organizing for clarity:
CLIP DETECTOR (RL = 10kohm to 5V)
Duty Cycle ( pin 5)
THD = 1%: 10 % duty
THD = 10%: 40 % duty
Also, see this post:
http://www.diyaudio.com/forums/chip...basic-tda7293-pcb-parts-list.html#post1498203
So it seems that the clip output is a logic output. You need to supply +5V to this pin via a 10k ohm pullup resistor. This is actually shown in the datasheet in Figure 2, typical application (layout). On the IC, the leftmost pin is #1, and you can see that pin 5 connects first to "TP", then through a resistor and to a 5V supply. TP stands for test point I believe.
Maybe someone else can chime in on how to make use of the signal at TP, and how you could for instance connect this to a LED or other indicator. I think the current would be a bit too low to drive an LED directly. You could probably use a JFET op-amp as a buffer, connected to TP and driving the LED.
CLIP DETECTOR (RL = 10kohm to 5V)
Duty Cycle ( pin 5)
THD = 1%: 10 % duty
THD = 10%: 40 % duty
Also, see this post:
http://www.diyaudio.com/forums/chip...basic-tda7293-pcb-parts-list.html#post1498203
So it seems that the clip output is a logic output. You need to supply +5V to this pin via a 10k ohm pullup resistor. This is actually shown in the datasheet in Figure 2, typical application (layout). On the IC, the leftmost pin is #1, and you can see that pin 5 connects first to "TP", then through a resistor and to a 5V supply. TP stands for test point I believe.
Maybe someone else can chime in on how to make use of the signal at TP, and how you could for instance connect this to a LED or other indicator. I think the current would be a bit too low to drive an LED directly. You could probably use a JFET op-amp as a buffer, connected to TP and driving the LED.
Maybe this can help you. It from a class G made with TDA7293.
The yellow trace (C1) is the clip detector output, red (C2)it is the output and blue and green are the variable supply lines.
More you can find here: http://www.diyaudio.com/forums/chip-amps/184102-7293-parrallel-bridge-paralel-best-sound.html
The yellow trace (C1) is the clip detector output, red (C2)it is the output and blue and green are the variable supply lines.
More you can find here: http://www.diyaudio.com/forums/chip-amps/184102-7293-parrallel-bridge-paralel-best-sound.html
Attachments
@sesebe:
I read the thread you linked to. You didn't use any external components but just brought pin 5 to a connector/header. So what you show in yellow in the plots is the behavior of pin 5, which is to go low when there is clipping.
@IceTorch:
The document at the link below gives some ways to interface logic circuits with other components, like LEDs. Check out Figure 5-17b, "Interfacing to LEDs using a transistor driver circuit". This simple add on will give you the behavior you are looking for (LED on when clipping). The 150R resistor sets the LED current. The value shown results in 20mA and may need to be adjusted so that this current does not exceed the allowed forward current for the LED.
http://is.iiita.ac.in/study/Digital Systems Design/Chapter05.pdf
I read the thread you linked to. You didn't use any external components but just brought pin 5 to a connector/header. So what you show in yellow in the plots is the behavior of pin 5, which is to go low when there is clipping.
@IceTorch:
The document at the link below gives some ways to interface logic circuits with other components, like LEDs. Check out Figure 5-17b, "Interfacing to LEDs using a transistor driver circuit". This simple add on will give you the behavior you are looking for (LED on when clipping). The 150R resistor sets the LED current. The value shown results in 20mA and may need to be adjusted so that this current does not exceed the allowed forward current for the LED.
http://is.iiita.ac.in/study/Digital Systems Design/Chapter05.pdf
I apologise for being lazy. As CharlieLaub says, it looks to be a logic output, but it is more versatile than that I think. The "test circuit" from the data sheet, shows it as a FET switch to ground:
An externally hosted image should be here but it was not working when we last tested it.
From elsewhere in the data sheet:
ABSOLUTE MAXIMUM RATINGS: V5 Clip Detector Voltage Referred to -VS 120 V
CLIP DETECTOR: Duty Duty Cycle (pin 5) THD = 1%; RL=10K to 5V 10 %
THD = 10%; RL = 10K to 5V 40 %
This suggests that in it's simplest form, all you need is a supply, a series current limiting resistor and an LED. . . . but maybe you will be unhappy with the LED brilliance . . . I think I would want more than a faint flicker at 1% THD.
. . so, (you're going to roll your eye's at this point) if you are up to a little experimenting, look back to the little Q13 / Q14 circuit I first put up . . . if you replace Q13 with another PNP transistor like Q14, and tie R26 to your supply (B - what are you using for Vcc 40V) instead of to earth, and connect point A to the Pin-5 Clip output, you might be able to extend the time and brightness for a lot less than 1% THD. The only other thing to consider might be to vary up or down the input impedance R25; and the final LED series resistance will depend on you Vcc and the current capacity of you LED. Have fun experimenting.
I posted a long winded reply, but it isn't up yet. Try ths small portion from the circuit I put up yesterday:
An externally hosted image should be here but it was not working when we last tested it.
Rob's paragraph on how the circuit works is still relevant.
CharlieLaub
Thank you so much, I'll try it next week. Can you explain what this means:
THD = 1%: 10 % duty
THD = 10%: 40 % duty
Does that mean that at 1% THD led will shine at 10%? I'm having hard time understanding that.
Thank you so much, I'll try it next week. Can you explain what this means:
THD = 1%: 10 % duty
THD = 10%: 40 % duty
Does that mean that at 1% THD led will shine at 10%? I'm having hard time understanding that.
This explains what duty cycle means:
Duty cycle - Wikipedia, the free encyclopedia
For instance in this case, 10% duty cycle means that that 10% of the time the signal at pin 5 is at 0V and 90% of the time the signal is at 5V. In general the signal may be cycling back and forth between 0V and 5V sort of like a "square wave" but with different amounts of time spent at the "high" and "low" states following each transition. Its not clear to me how often a transition may happen, but it only needs to be faster than the eye can resolve to look constant so I wouldn't worry about it.
You may still need to experiment a little. The value needed for the series resistor limiting the current through the LED, will depend upon you LED and Power Supply voltage. I may even be wise to start with double that number until you see what happens.
There is no need to apologize. BTW I could have sworn that this post wasn't up when I wrote mine.
That would be my next question.
This is very similar to what CharlieLaub posted in post #4.
I would add a series resistor between this circuit and the drive line out of the amp IC.
The amp IC clip det output is an open drain output...connects to ground when on...10%, 40% etc of the time.
Last edited:
I'm following this too, as I always wanted to add this to my chipamp. Keep up the good work and report back how the real circuit works. 
What is Vcc?
Hfe of BC556 is about 150 so there should not be a big requirement for current.
But you don't want the "off" leakage of the clip output turning on your LED driver.
. . from a quick internet search:
"What VCC stands for in electronics - The Q&A wiki
wiki.answers.com/Q/What_Vcc_stands_for_in_electronics Cached
What does Vcc stand for on a SIM card? Vcc is for positive supply voltage. Why does vcc voltage called vcc? I believe that Vcc stands for Voltage Common Collector." . . .
Also"
But you don't want the "off" leakage of the clip output turning on your LED driver"
. . . . that is why I agreed it was a good idea to include a series resistor on the input.
It may be built so that you can resistor lift the mute ground, remove the mute cap (or greatly reduce the value of the mute cap so the discharge doesn't blow the clip detect circuit), and then tie the clip detect pin to the mute pin to cause partial muting during clipping. This is just a guess. I have no idea if that will actually work.
Alternative:
Use the LED driver circuit (somehow reinforced to withstand the cap I'm about to add) to drive an LDR optocoupler instead and add a 10u or so cap parallel with the opto's LED to cause about a 20ms delay (so we don't cut off dynamics), then employ this very much like the clipnipper circuit.
Alternative:
Use the LED driver circuit (somehow reinforced to withstand the cap I'm about to add) to drive an LDR optocoupler instead and add a 10u or so cap parallel with the opto's LED to cause about a 20ms delay (so we don't cut off dynamics), then employ this very much like the clipnipper circuit.