TDA 7293 -- done right ?

well the direction is not yet fixed.
My idea to make this "stackable" is because :


1.) I want the board (s) to be as small as possible.
2.) It should be scalable and as flexible as possible
3.) It should be possible to use the slave-boards with another IC as driver, using pin 11
So the necesary decoupling caps must be populated on each board.
4.) The input components only on the master board of course and the feed-back resistor
soldered directly on pins 2 and 14 to avoid trace inductance.




in short as universally usable as possible while not sacrificing performance. 'Of course stacked boards using only wires perpendicular to them are not exactly service-friendly but in a well-designd and rugged circuit that should not be too big a problem especially for DIYers for whom to to it without too much money is important!?

The term universal caught my attention and my reply is related just for that thought.
I can visualize a single board with 4 amplifier channels of either TDA or LM series. They can be single or dual. The board will also accomodate ADAU1701 DSP as a pluggable daughter board and all the needed power supplies on the same board. Such a board can be used for virtually any number of applications.
I can see only one challenge: The processor board needs 3.3V supply and one needs to carefully analyze the power sequence to avoid turn on/off noise either by using the MUTE pin of the IC or using relays and some extra circuit logic.
 
graph1:
75 Ohm Generator out connected to a 50 Ohm cable terminated with 50 Ohm at the amp-input, measured at the amp input by a 10:1 probe

Much better. The edge of the square wave is much cleaner now. I'm still puzzled a bit by the slight slope around 15-20 us, though. It could just be scope probe compensation.

looks good to me, except the 8 Ohm||10nF measurement that shows some ringing...

Yeah... That's worth investigating further. Try more capacitor values in the 1.0 - 100 nF range. When I designed the HP-1 headphone amp I noticed that it would happily oscillate in the 150-200 MHz region with 100-200 pF load. This wasn't seen with 1 nF load or 47 pF load. My point is that it is possible to have holes or capacitance ranges where the amp is unstable even though it's stable outside that capacitance range.

Another thing i did not expect is the distortion of amps input when adding caps to the output...

I noticed that too. I suggest playing with that in simulation to see if you can figure out where that comes from.

what is the ideal cable to connect the amplifier to the loudspeaker from the technical sight? (assuming 10ft / 3m length)

That sounds like a topic for a separate thread. Personally I would go with the heaviest gauge wire you're willing to deal with.

Tom
 
The ringing at 10nF||8 Ohm is about 1.25MHz, so assuming it is a resonance it has to be an inductor of about1.6µH that is part of the resonance-circuit. Also assuming 10nH per mm of board route, it has to a be a board route of about 160mm in length that is the origin of the resonance. The only route with that length is the power supply for the TDA...
 
a new board

I did some work...

this should do better than the last one.

again: comments welcome
 

Attachments

  • composite-version_2_sch.png
    composite-version_2_sch.png
    139.1 KB · Views: 437
  • composite-version_2_top.png
    composite-version_2_top.png
    110.9 KB · Views: 437
values updated

the configuration shown in the last post works! Attached the schematic with the actual values.

But still the startup is problematic. The time for the TDA to come out of standby/mute is too long, therefore there is a DC-Level of 20V at the output for a few seconds...without the Cs at the standby and mute input the time is much shorter (several ms), but is still too long. I took the standby/mutes supply from the regulated voltage and think this should be changed to the TDAs supply. Hopefully this unmutes the TDA before the regulated power comes up and the error amp can work into a closed loop. I will see if this works...

This problem should show up with all composite amps where the error amp is connected as an integrator, and probably someone has already solved this problem....hints and ideas welcome:worship:
 

Attachments

  • composite-V2-actual-values.png
    composite-V2-actual-values.png
    135.1 KB · Views: 355
TDA7293 Stby & Mute

Try using the R-C circuit for Stby and R-(R//D)-C for Mute as prescribed in the datasheet and powered from the amp supply instead of regulated supply. The start-up should now be silent. If your circuit exhibits noise during shut-down, then let us know, we will take that up separately. I hope that this helps.
 
TDA7293 Stby/Mute

It is possible that the feedback loops need an un-muted TDA7293 to settle down. The output DC could be the result of the loops setting down. So we may need to think of using a relay at the output to keep the speakers disconnected till the loops settle down. The NC terminal can be connected to a dummy load to absorb the turn-on transients while the speaker be connected to NO terminal. Need to think if we need anti-parallel diodes at the amplifier output to ensure that the amplifier output never exceeds the supply rails due to load impedance. I hope this helps
 
Connecting the standby and the mute pins directly to the positive power-rail is a solution that works (for me). As the TDA no longer mutes when connected that way, the feedback loop is closed right up from the start. The measured artefact on the output is now much smaller and short in time and well above the audible audio band. A connected speaker is quiet at startup and shut down with the input shorted.
That works for me, espacially as it saves boardspace - all parts for the standby and mute circuit are no longer needed
 
sometime it helps

sometime it helps to look at a circuit in another way.

As the step responses signal shape from the amplifier, especially when Cs are connected at the output, reminds me on the typically step response of an PI-Regulator, one can see the composite amplifier as such. Having in mind, that the step response of an PID-Regulator is even better (the step sill is a step), this should be tried. Could this work? The simulations i ran so far say: yes, this will work. So it is time to do another prototype - this time hopefully the final one...
 
The Cinch connector for the input i want to use has the same pinout, at least i hope so.
And the BNC for the output is for some speaker-cable test. BNC is specified for 3A continuous (at least by some of the manufacturers) and will therefore fit to my standard listening levels. But if this don't work or doesent sound good for my ears there is also the possibility to change the output to a standard connector (terminal with screws).

Yes I am interested in your comment, but as i wrote "this is intended to be..." i just want to do some minor work on the board - moving the parts a bit to get the board a little smaller - and to separate the input ground from the regulator ground will be a good idea, iguess...
 
Disabled Account
Joined 2002
BNC for outputs... I had to scratch my head. One second thoughts, if they can handle 3A they seem to have the benefit of less soldering and less connections compared to the PCB-MKDSN-wire-Speaker banana terminal-male banana connector. In fact it is more elegant.

It is not elegant to use BNC for both input and output as one will certainly plug a signal cable to the speaker output one day.
 
Last edited:
Just an idea :
the TDA can be put into slave mode and then be driven from pin 11.

'Then one only uses the power-output-stage inside it.
so if I undertand this correctly :

Why not use an opamp followed by a discreete VAS-stage to drive that power-stage or parallel pwerstages? Like an LME 4562 ?


That would avoid the TDA´s complete input-circuitry and , overall feedback assumed , imprint the op-amps characteristics at least as far as the TDA´s powerstage will allow
Better slew-rate very likely , and very likely better THD an noise as well


To test this one can use a modulus286 (or other top class amp) for example to drive a TDA configured as slave then measure the output of that. I would do it but do not have the necesary equipment.
 
Last edited:
in other words take a discrete amp , throw away the input and output stages and substitute the LME or the OPA for the input and the TDA for the output keeping only the complementary VAS


with paralleled TDA´s rails up to 50 volts should be feasible


and use OPA´s
 
Last edited: