I started to modify the PCB on my side following your recommendations.
First I disconnected the signal ground from the power ground at the input connector (hoped to break this ground loop) - still hum/buzz.
Next I added four 330nF/100V stacked film caps on the +Vs bus at the proximity of pins 7-13 for each TDA, and four 330nF at the - rail, in the middle of the pins 8-15, I grounded capaitors by 8 individual cables to the point in the middle of PSU caps (I hope the star ground could be defined there) - still hum/buzz.
I have no the earth ground connected to the PCB, there is no case for the moment. I only have 3 wires from the transformer (24V-0-24V) , L/R inputs shorted to the signal ground at the connector through two 680 Ohm resistors, 2 8 Ohm speakers connected at the outputs, TDAs on a reasonable sized heatsink (isolated).
Next I will try to cut the +/- power rails and to put the individual power cables to each TDA directly from the PSU capacitors. Any other ideas ?
First I disconnected the signal ground from the power ground at the input connector (hoped to break this ground loop) - still hum/buzz.
Next I added four 330nF/100V stacked film caps on the +Vs bus at the proximity of pins 7-13 for each TDA, and four 330nF at the - rail, in the middle of the pins 8-15, I grounded capaitors by 8 individual cables to the point in the middle of PSU caps (I hope the star ground could be defined there) - still hum/buzz.
I have no the earth ground connected to the PCB, there is no case for the moment. I only have 3 wires from the transformer (24V-0-24V) , L/R inputs shorted to the signal ground at the connector through two 680 Ohm resistors, 2 8 Ohm speakers connected at the outputs, TDAs on a reasonable sized heatsink (isolated).
Next I will try to cut the +/- power rails and to put the individual power cables to each TDA directly from the PSU capacitors. Any other ideas ?
Hi gootee,
Thank you a lot for your advises.
I had a look and it is rather difficult to place 8x2,000uF on the existing PCB: those +/- rails are rather thin, so even if this step will fix the problem of the humm the PCB will be no more good for a real use: the 8 capacitors soldered at the surface of the thin PCB copper will break it quickly, and I think it will be difficult to use such a fragile mess for a real application - so to me if it works it would be more a proof of concept than the easy fix for the other happy owners of this board
Is there any other pragmatic step to be tried before this, like cut & retrace of the +/- rails, of the gnd, or similar, keeping the PCB still relatively usable?
So far, I've progressed a bit:
- I've added 2x2000uF close to TDAS on the Vs+ rail and 2x2000uF on the Vs- rail connected at the power ground in the middle of 2 PSU caps,
- I've cut the + and the - power rails and I've connected each TDA using two individual power wires coming directly from PSU capacitor legs, I kept the 2x330nF decoupling caps on each TDA but I removed 2000uF caps (no enough space on PCB)
- I've added missing capacitors between pins 1 and 3 on the slave TDAs (I had 1uF film available, the master input capacitor is 2,2uF)
The hum still there, seems insensitive to all those changes and heatsink still become quickly hot.
I stop my experiments for the moment.
- I've added 2x2000uF close to TDAS on the Vs+ rail and 2x2000uF on the Vs- rail connected at the power ground in the middle of 2 PSU caps,
- I've cut the + and the - power rails and I've connected each TDA using two individual power wires coming directly from PSU capacitor legs, I kept the 2x330nF decoupling caps on each TDA but I removed 2000uF caps (no enough space on PCB)
- I've added missing capacitors between pins 1 and 3 on the slave TDAs (I had 1uF film available, the master input capacitor is 2,2uF)
The hum still there, seems insensitive to all those changes and heatsink still become quickly hot.
I stop my experiments for the moment.
Hi,
Does the rectifier are mounted on the board? If you a bridge rectifier mounted on the board removed it and make a power supply out of the board. Remove the two big caps and connect the voltages directly into the holes left by the capacitors. In this way you will find if the noise is coming from the power supply or is it internally.
Does the rectifier are mounted on the board? If you a bridge rectifier mounted on the board removed it and make a power supply out of the board. Remove the two big caps and connect the voltages directly into the holes left by the capacitors. In this way you will find if the noise is coming from the power supply or is it internally.
Better idea. It looked like that the rectifier is mounted on the board. Just removed the rectifier and jumper the connection for the transformer connection.
Use one of the connection for the positive side and connected it directly to the positive side of the capacitor for the plus side. Then use the third one to connect the negative side to the negative capacitor side. I think the middle goes to ground. Now build a bank of capacitors for the neg/pos voltage and connect them where the AC was coming. I think this will take care of your problem. You do not have enough voltage filtering to drive the 4 TDA7240. Hope you understand why I trying to accomplishing by the above procedures.
Use one of the connection for the positive side and connected it directly to the positive side of the capacitor for the plus side. Then use the third one to connect the negative side to the negative capacitor side. I think the middle goes to ground. Now build a bank of capacitors for the neg/pos voltage and connect them where the AC was coming. I think this will take care of your problem. You do not have enough voltage filtering to drive the 4 TDA7240. Hope you understand why I trying to accomplishing by the above procedures.
Thanks for suggestion, I'm out for a week so I will continue once back. I will have a look at the signal ground which is rather strange - maybe some mistake during redo of the PCB passing from the TDA7293 to the 7294 (I suppose it was like that because of the label on the PCB):
https://picasaweb.google.com/114577861556394297323/TDA7294ChinaJoke#
https://picasaweb.google.com/114577861556394297323/TDA7294ChinaJoke#
It seems that you need to discard and replace the old rectifier.
So, look at Mark's rectifier in photograph #4
Synergy - LM3875 Gainclone Amplifier (Chip Amp)
No diode noise, and so there's less fluctuations.
You can use a similar rectifier. The 25a 35a square unit is strong enough
The DC inputs to your board are easy to find, under the largest caps.
It should be fairly easy to upgrade to an off-board rectifier.
You can mount it securely and use thermal paste to help it stay cool
Add a pair of 2200uF caps, securely, to your new rectifier.
Those caps are why it needs to be cool.
The two new 2200uF caps will get "slammed" with ripple, but the caps on your amplifier board won't see that.
So, if you do it that way, the power cable entering your amplifier board has only DC.
Its almost "peel-n-stick" easy.
So at least one big noise can be relocated.
EDIT:
To reduce heat and other misbehavior, put at least a pair of 330uF (or nearby value) as closely as possible to the chips. That is in addition to, not instead of, large caps. Substitutes of far off values will make much heat. TDA7294 needs an available pre-drive size power circuit close to it, exactly like a discrete amplifier. Its not a gainclone.
So, look at Mark's rectifier in photograph #4
Synergy - LM3875 Gainclone Amplifier (Chip Amp)
No diode noise, and so there's less fluctuations.
You can use a similar rectifier. The 25a 35a square unit is strong enough
The DC inputs to your board are easy to find, under the largest caps.
It should be fairly easy to upgrade to an off-board rectifier.
You can mount it securely and use thermal paste to help it stay cool
Add a pair of 2200uF caps, securely, to your new rectifier.
Those caps are why it needs to be cool.
The two new 2200uF caps will get "slammed" with ripple, but the caps on your amplifier board won't see that.
So, if you do it that way, the power cable entering your amplifier board has only DC.
Its almost "peel-n-stick" easy.
So at least one big noise can be relocated.
EDIT:
To reduce heat and other misbehavior, put at least a pair of 330uF (or nearby value) as closely as possible to the chips. That is in addition to, not instead of, large caps. Substitutes of far off values will make much heat. TDA7294 needs an available pre-drive size power circuit close to it, exactly like a discrete amplifier. Its not a gainclone.
Last edited:
If the gain is set within the recommended range, I would try a low-pF cap in parallel with each of the feedback resistors, in case the circuits are oscillating at a high frequency. You will probably have to experiment with different pF values. You should be able to immediately see if that helps, by looking at the output with your oscilloscope. The wide fuzzy trace should become narrower or more-focused. And then the chips should become much cooler. And maybe the buzzing will even be improved.
P.S. Each pair of the new DC rail wiring you installed should be twisted tightly together, all the way from one end to the other end. Ideally, they should follow their ground-return paths very closely, so you don't have any enclosed loop area, which basically forms an antenna and allows any time-varying EM field (such as your AC) to induce a time-varying current in the loop. That phenomenon (see Faraday's Law) is fairly likely to be a cause of at least part of the hum. i.e. There are probably PCB trace pairs that should have been kept very close together but were not. Note that it works both ways, i.e. it can make a receiving antenna or a transmitting antenna. So all high-current pairs should also be twisted or kept very close to each other.
I forgot whether or not the inputs signal's ground-return conductor is kept very close to the signal conductor. It should be, and should stay that way all the way to the input resistors at the amps, and should not be connected to ground before that. Then the grounds for the input resistors should go to star ground and not share any length of conductor with any other type of ground return.
P.P.S. Per others' suggestions, it would probably tell you a lot if you could hook up a different (off-board) DC power supply.
P.S. Each pair of the new DC rail wiring you installed should be twisted tightly together, all the way from one end to the other end. Ideally, they should follow their ground-return paths very closely, so you don't have any enclosed loop area, which basically forms an antenna and allows any time-varying EM field (such as your AC) to induce a time-varying current in the loop. That phenomenon (see Faraday's Law) is fairly likely to be a cause of at least part of the hum. i.e. There are probably PCB trace pairs that should have been kept very close together but were not. Note that it works both ways, i.e. it can make a receiving antenna or a transmitting antenna. So all high-current pairs should also be twisted or kept very close to each other.
I forgot whether or not the inputs signal's ground-return conductor is kept very close to the signal conductor. It should be, and should stay that way all the way to the input resistors at the amps, and should not be connected to ground before that. Then the grounds for the input resistors should go to star ground and not share any length of conductor with any other type of ground return.
P.P.S. Per others' suggestions, it would probably tell you a lot if you could hook up a different (off-board) DC power supply.
Last edited:
I've found some time so I revisited this amp last night and I was happy- I solved the hum and the oscillation problem (at 95 %
Now the module is quite and plays nicely the music, and it sounds actually very well.
To summarize my activity I've returned back to the original state and I've played this time with the grounding. I've cut the two PCB horizontal ground traces to several segments and I've connected each of them via the individual isolated wire to the center star ground (in total 10 individual 0.2mm magnet wires). I've payed attention to twist all wires together.
I've disconnected also the input ground at the connector place, it is connected on the top side and also on the bottom side of the pcb so you have to pull out the green input connector to be able to disconnect it.
The hum was almost gone, but the oscilloscope still showed still HF oscillation at all 4 TDA7294s outputs. I've added two 1uF/63V Polyester capacitors at the end of the power PCB traces, returned by individual wires to the same star ground. This has stopped all residual oscillations. Now the TDAs stay cool, even when playing music at the moderate listening level (I'm using the same heatsink as before).
There is a residual hum which I can heard with may ear at 10cm from the speaker cone (~90dB/W speakers). Haven't had a time to try other advices from this thread yet.
So definitely a very bad PCB design but could be cured by a little rework. The result is a bit messy pcb, thought, ....
Now the module is quite and plays nicely the music, and it sounds actually very well.
To summarize my activity I've returned back to the original state and I've played this time with the grounding. I've cut the two PCB horizontal ground traces to several segments and I've connected each of them via the individual isolated wire to the center star ground (in total 10 individual 0.2mm magnet wires). I've payed attention to twist all wires together.
I've disconnected also the input ground at the connector place, it is connected on the top side and also on the bottom side of the pcb so you have to pull out the green input connector to be able to disconnect it.
The hum was almost gone, but the oscilloscope still showed still HF oscillation at all 4 TDA7294s outputs. I've added two 1uF/63V Polyester capacitors at the end of the power PCB traces, returned by individual wires to the same star ground. This has stopped all residual oscillations. Now the TDAs stay cool, even when playing music at the moderate listening level (I'm using the same heatsink as before).
There is a residual hum which I can heard with may ear at 10cm from the speaker cone (~90dB/W speakers). Haven't had a time to try other advices from this thread yet.
So definitely a very bad PCB design but could be cured by a little rework. The result is a bit messy pcb, thought, ....
Thanks guys
I've added some photos of the fixed amplifier on the
https://picasaweb.google.com/vladimir42000/TDA7294ChinaJoke?authuser=0&feat=directlink
I hope it's easier to follow than my broken explanations from the previous message
I've added some photos of the fixed amplifier on the
https://picasaweb.google.com/vladimir42000/TDA7294ChinaJoke?authuser=0&feat=directlink
I hope it's easier to follow than my broken explanations from the previous message
I put 2.2 ohm and 100nf as zobel.It hum a lot.I then Increase zobel cap to 200nf ,it reduces the hum a bit.Further more.i replace 22k pot to 10k good quality pot.All hum just goes.
One more surprise observation I got regarding zobel resistor.I was putting 10ohm and 100nf for my TDA 2040 .After replacing 10 to 2.2ohm I got superior upper frequency .
There is many tweaks for Lm3886 as sub woofer amplifier ,but TDA7294 as a sub woofer amplifier ,there is a little discussion .Anyway I put 220mf for feedback cap,Bootsrape cap to 220 mf,input cap to 10mf non polar.2200mf ,and 100nf near the Ic.I dont wanna put any rf filter coz I dont want any resistor before the input in series.Fillters are don from preceding circuit .
One more surprise observation I got regarding zobel resistor.I was putting 10ohm and 100nf for my TDA 2040 .After replacing 10 to 2.2ohm I got superior upper frequency .
There is many tweaks for Lm3886 as sub woofer amplifier ,but TDA7294 as a sub woofer amplifier ,there is a little discussion .Anyway I put 220mf for feedback cap,Bootsrape cap to 220 mf,input cap to 10mf non polar.2200mf ,and 100nf near the Ic.I dont wanna put any rf filter coz I dont want any resistor before the input in series.Fillters are don from preceding circuit .
Concur!
TDA7294 is (internally) a driver chip pushing fets, and this device is well able to work itself half to death and ruin its audio fidelity while successfully amplifying interference (instead). It has some specific needs if you want hi-fi and low heat. The only time it does hi-fi is when it is running cool AND only doing audio. Then it is really a treat!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.