PS:
also tried a solen 4.7 poly instead of the panasonic with similar results.
i got best results WITH buffer after C1 (NJM4556 is fine as is NE5532 or 34, OPA2134 or 134 may be a bit quieter with a tube linestage, OPA2227 or 227 may give lower and more stable offset)
BTW, Daniel, i noticed that ESP's 7293 board has pins 9 and 10 tied to v+ so mute and standby are always disabled. is that safe?
also tried a solen 4.7 poly instead of the panasonic with similar results.
i got best results WITH buffer after C1 (NJM4556 is fine as is NE5532 or 34, OPA2134 or 134 may be a bit quieter with a tube linestage, OPA2227 or 227 may give lower and more stable offset)
BTW, Daniel, i noticed that ESP's 7293 board has pins 9 and 10 tied to v+ so mute and standby are always disabled. is that safe?
Last edited:
Doing a little damage to a chip, will still reduce the lifespan, even if it doesn't explode instantly. Also, project 127 has a long list of inappropriate/sloppy, none of which is severe enough to explode it in a single day or even a week. The problem that you saw, has a Lot of company, such as lopside decoupling for poorly stereo imaging and really terrible quality bass since the input is wide open but the amplification is shortchanged, which causes offensive "Poodle-Bark" warm and muddy bass distortion and a few extra tones that aren't impressive either. That still doesn't cause an explosion. However, I don't think it would set any new records for long lasting, with the gain set to minimize stablity. That may not be a concern--I think you're more likely to get tired of it and unplug it well before it wears out. That thing is soldering practice and may serve as a momentary stepping stone on your way to buying an ESP discrete amp project. . . because otherwise, I can't imagine what it is for.BTW, Daniel, i noticed that ESP's 7293 board has pins 9 and 10 tied to v+ so mute and standby are always disabled. is that safe?
Basically, that board is in the "Made for Selling" category, which is significantly different quality than "Made for Using" items (that are scarce). The easiest way to fix project 127 is to throw the board away.
I'd give it a grade of "D" but only because there's worse available.
Last edited:
not to worry. i had no plans to use that board -- just curious about the safety of completely disabling standby and mute.
Mute is before the voltage amp, standby comes after, so it is bad to short those two pins together. It also bad to maximize their current. So, you'll be needing at least two resistors (a resistor for pin9, another resistor for pin10). On my amplifiers I also install the delay cap for standby but not for mute.
to quote myself:
"just curious about the safety of completely disabling standby and mute"
one example implementation of disabling both standby and mute would be to include the two resistors but omit the capacitors. you already omit the mute cap, so my question can be reduced to: is it safe to omit the standby cap.
if so then there exists a safe way for "completely disabling standby and mute"
"just curious about the safety of completely disabling standby and mute"
one example implementation of disabling both standby and mute would be to include the two resistors but omit the capacitors. you already omit the mute cap, so my question can be reduced to: is it safe to omit the standby cap.
if so then there exists a safe way for "completely disabling standby and mute"
Good point Steve.. . .one example implementation of disabling both standby and mute would be to include the two resistors but omit the capacitors. you already omit the mute cap, so my question can be reduced to: is it safe to omit the standby cap. if so then there exists a safe way for "completely disabling standby and mute"
But, unfortunately, I can't answer that question because I haven't tried it. And, There is that one new production TDA7294, that isn't quite right but is durable and could be used for test, if it could be found. That rebellious little thing is missing! So, you'll probably have to test the cap omission prospect yourself.
An alternative option, not yet tested, is to put a zener series to the standby resistor (or possibly instead of the standby resistor???), so that during the power up sequence, as the voltage rises, the mute comes on first with just the resistor, and then next the standby comes on with the resistor series to zener, after passing the zener threshold.
That shouldn't need delay caps for either mute or standby--I'm thinking that the power circuit already has caps and a very slight initial delay, and if that much is enough, then you might not need the standby cap.
The power supply caps charge noticeably more slowly during the last ~30% of the initial charge up (when a polar cap has surpassed its effective nonpolar range), so for a moderate delay, I think that the proposed zener value would need to be approximately 70% of the V+ rail voltage. For example, 35v rails, *.7 = 25v, but we need to subtract the 5v threshold built in to the chip, so, that's a 20v zener. This usage is hypothetical, so the component values may need adjusted.
When using the zener series to standby, you can tell if the resistor may be omitted by using a thermal probe on the zener plus standard electrothermal calculations to estimate the wattage present (to see if you need to dump some of that current/heat into a resistor or not).
Well, there went with the excessive thinking! I hope it works for you.
Last edited:
without an equivalent schematic for the chip we are left to guessing and trial and error. the datasheet's absolute maximum ratings do not appear to prohibit the types of things we are talking about. but the datasheet fails to mention a lot of the gotchas so this may be just another one of those gotchas.
i'm inclined not to push my luck. i really don't like things blowing up unexpectedly!
i'm inclined not to push my luck. i really don't like things blowing up unexpectedly!
Steve, I think that the method with the zener for standby will probably work and it does adhere to the recommended sequence.
If that is inconvenient, a two resistor voltage divider (70% to ground, 30% to standby) may work. So that is simply two resistors for standby, one resistor for mute. Unfortunately that's not less parts since you're simply replacing the standby cap with a resistor instead. But, it would give you a test without having to buy zeners.
It seems that Rod Elliot has used the TDA7293 without the delay caps, so I'd say that the experiment will probably succeed. If merely removing the sound quality foobar from Rod Elliot's hookup, we could try something as simple as 10k for mute and then 100k (value TBD) for standby and try to find a standby resistor value with significant voltage loss, but not too much. In that case, just a simple pair of resistors is highly likely to work well.
If that is inconvenient, a two resistor voltage divider (70% to ground, 30% to standby) may work. So that is simply two resistors for standby, one resistor for mute. Unfortunately that's not less parts since you're simply replacing the standby cap with a resistor instead. But, it would give you a test without having to buy zeners.
It seems that Rod Elliot has used the TDA7293 without the delay caps, so I'd say that the experiment will probably succeed. If merely removing the sound quality foobar from Rod Elliot's hookup, we could try something as simple as 10k for mute and then 100k (value TBD) for standby and try to find a standby resistor value with significant voltage loss, but not too much. In that case, just a simple pair of resistors is highly likely to work well.
Last edited:
To summarize, that will probably succeed by choosing the standby resistor value lossy enough to cause a significant voltage drop at the standby pin, and optimal would be much like riding along at about 30%~45% of V+."just curious about the safety of completely disabling standby and mute" one example implementation of disabling both standby and mute would be to include the two resistors but omit the capacitors.
Meanwhile, mute still has a 10k of its own.
Hi guys... Took me a week to plow through this entire thread and jot down things interesting for my build... Then it dawned on me it might be wiser to also ask if anyone knows my specific PCB...
I think I have this PCB, but a bad ripoff copy of it 🙂 including wrongly placed silk screen drawing parts...
The PCB and the schematic found in an obscure PDF somewhere. Beware the schematic DOES NOT match the PCB exactly, but I found them both on a single page PDF. I think I reverse engineered the entire PCB when I got them, but cannot find it anymore...
My actual PCB (note the corrections like the orientation for the zener for the NE5532 powersupply)
The numbers of the resistors and the capacitors correspond to these two schematics:
I started populating the PCB a year ago, but started removing some components after reading this thread, like the local (2x10u=)20uF powercapacitors (C6 and C8) and replaced them with 180uF versions. The larger powercaps will probably be 4700uF versions (but I also have/could fit some NOS 12000uF or 18000uF ones, but they are very high).
I also removed the NFB 20uF cap and have 47uF in stock or could go bigger (C2). Not sure if my simple electrolytics would be OK (e.g. Siemens 25V 47uF but I also found (pretty large but fittable) 68uF bipolars iirc).
Just haven't really looked into the NE5532 function yet and although not a complete noob, am not exactly bright enough to see in once glance if this means that e.g. the gain for the TDA should be lowered...
I have two boards (and a totally different and more straightforward bridged stereo board without uPC1237/NE5532), and will be testing the first one with new V6 MAR versions of the TDA chip on a 2x18.5VAC or a 2x24VAC powersupply. But I do have a small stock of original Singapore versions from 15 years ago 🙂
If anyone's interested I can also post my 15 year old design for the dual powersupply board that blew up a chip and cost me a subwoofer... (40VDC on the poor sucker) 🙂 Maybe you guys can see what I did wrong... But any advice on this board would be welcomed even more, as the switching powersupply is something I will never restart again (especially with the thread about the TDA and powertransistors)
I think I have this PCB, but a bad ripoff copy of it 🙂 including wrongly placed silk screen drawing parts...
The PCB and the schematic found in an obscure PDF somewhere. Beware the schematic DOES NOT match the PCB exactly, but I found them both on a single page PDF. I think I reverse engineered the entire PCB when I got them, but cannot find it anymore...


My actual PCB (note the corrections like the orientation for the zener for the NE5532 powersupply)

The numbers of the resistors and the capacitors correspond to these two schematics:


I started populating the PCB a year ago, but started removing some components after reading this thread, like the local (2x10u=)20uF powercapacitors (C6 and C8) and replaced them with 180uF versions. The larger powercaps will probably be 4700uF versions (but I also have/could fit some NOS 12000uF or 18000uF ones, but they are very high).
I also removed the NFB 20uF cap and have 47uF in stock or could go bigger (C2). Not sure if my simple electrolytics would be OK (e.g. Siemens 25V 47uF but I also found (pretty large but fittable) 68uF bipolars iirc).
Just haven't really looked into the NE5532 function yet and although not a complete noob, am not exactly bright enough to see in once glance if this means that e.g. the gain for the TDA should be lowered...
I have two boards (and a totally different and more straightforward bridged stereo board without uPC1237/NE5532), and will be testing the first one with new V6 MAR versions of the TDA chip on a 2x18.5VAC or a 2x24VAC powersupply. But I do have a small stock of original Singapore versions from 15 years ago 🙂
If anyone's interested I can also post my 15 year old design for the dual powersupply board that blew up a chip and cost me a subwoofer... (40VDC on the poor sucker) 🙂 Maybe you guys can see what I did wrong... But any advice on this board would be welcomed even more, as the switching powersupply is something I will never restart again (especially with the thread about the TDA and powertransistors)
Last edited:
A simple electrolytic is okay for -in coupler, and commonplace for this use.I also removed the NFB 20uF cap and have 47uF in stock or could go bigger (C2). Not sure if my simple electrolytic would be OK (e.g. Siemens 25V 47uF but I also found (pretty large but fittable) 68uF bipolars iirc).
If a coupler is the wrong size, harmonic distortion results and you'd be chasing all over the place trying to find the tone problem. That's why it is vital to begin with the capacitance sized within a useful range.
The size difference will adjust the bass harmonic balance.
The range is approximately 220u~680u.
Smaller sizes will do boom-only muddy bass with a warm sound.
Larger sizes (like 2200u) will do thud-only clear bass with a cold sound.
An even blend of Both harmonics will cancel bass distortion, resulting in useful, practical, high fidelity bass.
The size of this cap also relates to what you allow to pass at the + input cap, such as a larger cap there also requires a larger -input cap. So, to avoid trouble, it will be easiest to use the TDA7294 with a 1u input cap.
Last edited:
Thanks, will dive into my stash shortly. 😀
I will draw up a decent schematic and post that up here, both with the adviced values and the proposed changes.
So feel free to answer, but maybe it's easier to wait for the schematic 🙄
I did reverse engineer the PCB before, but I think I lost it in my harddrive crash last year (Pictures I back up, but not al my projects data, the above schematics were only saved as I had emailed them to someone...)
The NE5532 seems to add another 5x gain. 220pF and 100k parallel to earth as input on the "+" input. And 100k in the feedbackloop to inverting input which has 20k series with 22uF to earth.
The output is passed through a 1uF to the non-inverting input of the TDA with a 22k to ground.
Feedbackloop to inverting input is the normal 22k and from there to ground the standard 680ohm and 22uF.
So the total gain seems to be close to 31+5=36 times???
Seems quite a lot...
There's also a 0.15 ohm resistor between output of the TDA and the speaker which I don't understand yet.
I will draw up a decent schematic and post that up here, both with the adviced values and the proposed changes.
So feel free to answer, but maybe it's easier to wait for the schematic 🙄
I did reverse engineer the PCB before, but I think I lost it in my harddrive crash last year (Pictures I back up, but not al my projects data, the above schematics were only saved as I had emailed them to someone...)
The NE5532 seems to add another 5x gain. 220pF and 100k parallel to earth as input on the "+" input. And 100k in the feedbackloop to inverting input which has 20k series with 22uF to earth.
The output is passed through a 1uF to the non-inverting input of the TDA with a 22k to ground.
Feedbackloop to inverting input is the normal 22k and from there to ground the standard 680ohm and 22uF.
So the total gain seems to be close to 31+5=36 times???
Seems quite a lot...
There's also a 0.15 ohm resistor between output of the TDA and the speaker which I don't understand yet.
Thanks, will dive into my stash shortly. 😀
I will draw up a decent schematic and post that up here, both with the adviced values and the proposed changes.
So feel free to answer, but maybe it's easier to wait for the schematic 🙄
I did reverse engineer the PCB before, but I think I lost it in my harddrive crash last year (Pictures I back up, but not al my projects data, the above schematics were only saved as I had emailed them to someone...)
The NE5532 seems to add another 5x gain. 220pF and 100k parallel to earth as input on the "+" input. And 100k in the feedbackloop to inverting input which has 20k series with 22uF to earth.
The output is passed through a 1uF to the non-inverting input of the TDA with a 22k to ground.
Feedbackloop to inverting input is the normal 22k and from there to ground the standard 680ohm and 22uF.
So the total gain seems to be close to 31+5=36 times???
Seems quite a lot...
the gain will be extremely high because they multiply, not add (unless you express them in db).
if you change the 100k to 20k or 30k you'll have an integrated amp that you can drive to full output with almost any line-level source.
depending on your source that 220pf may cause too much high-frequency rolloff.
Last edited:
Oh, I should mention. . . to prevent surprise explode, use 33uF or 47uF bootstrap cap, use gel flux for speedy soldering and heatsink the chip before soldering its pins.
So the total gain seems to be close to 31+5=36 times???
The two amplifying sections are cascaded. So the gain is 31x5 = 150 ! Not 36.
Unless you are talking dB gain. 31dB gain followed by 5 dB gain is 31+5 = 36 dB overall gain. ( x 63.1 gain ).
Somehow you came up with a figure of +36dB from your number crunching.
Whereas 31*5 equates to +43.8dB
Whereas 31*5 equates to +43.8dB

Thanks guys, I think I did contemplate for a second if I should add or multiply, but 31x5 sounded so outlandish that my brain ruled that out as a possibility.
I will certainly not solder another component before posting the whole schematic (only gotten as far as redrawing the TDA7294 in ExpressSCH, family life is very time consuming...)
Edit: quick and dirty (long live Paint) and missing e.g. the uPC1237 and the big 4700/12000uF caps, but a good start for just the audio part I hope...:
Intended changes: bootstrap to at least 47uF and possibly try one of the R2, R3, C2 combos Daniel suggested.
Apart from C6 and C8 (22uF in silkscreen, but I soldered 180uF on the PCB) the above pic is entirely the values suggested on the PCB itself, not changed values according to this thread.

Intended changes: bootstrap to at least 47uF and possibly try one of the R2, R3, C2 combos Daniel suggested.
Apart from C6 and C8 (22uF in silkscreen, but I soldered 180uF on the PCB) the above pic is entirely the values suggested on the PCB itself, not changed values according to this thread.
Last edited:
- Home
- Amplifiers
- Chip Amps
- Optimizing TDA7294 Output