Optimizing TDA7294 Output - Page 6 - diyAudio
Go Back   Home > Forums > Amplifiers > Chip Amps

Chip Amps Amplifiers based on integrated circuits

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 29th December 2012, 03:49 PM   #51
Jay is offline Jay  Indonesia
diyAudio Member
 
Join Date: Feb 2003
Location: Jakarta
Quote:
Originally Posted by bcmbob View Post
Unfortunately, they are all large and previously used. Is simple off-board mounting OK, or should I attempt to extend the leads ? (I've done that before successfully)

I'm working on the dual transformer setup right now.
Caps have very long useful life (when not abused of course). What Daniel has done here is "tweaking". As he said, every pcb trace has its own parasitic inductance. You need to "tweak" it to find the sweet spot. With 220uF I think I don't want to have more than 4 cm (no technical justification here) from IC pin, and that is with solid wire/trace. Bypass cap is a must (100n direct on pin).

Contrary to "tweakers", more technically minded people will usually prefer technical approach to solve problems. I myself think that both must be employed, where tweaking should be done at the end after more technical approach has been employed. And I also always think about price to performance ratio. What I mean here is, I don't like to invest too much money for quality parts just to get a cheap sound. Discrete amps clearly more enjoyable for me.

If I try to find an optimum output for TDA7294, I will first find the issues to be solved. And I will pick two: power supply noise and input impedance. Work it out and finally tweak it by ears. I will start (actually I have) from here: TDA7294
Attached Images
File Type: png TDA7294_radiolub.PNG (13.2 KB, 638 views)
  Reply With Quote
Old 29th December 2012, 10:09 PM   #52
diyAudio Member
 
danielwritesbac's Avatar
 
Join Date: Nov 2007
amplifier board power caps
Bob those 50v 220u with the "M" are Panasonic, and may do nicely.
The amplifier power caps need to be quite close to the amplifier chip.
Those are about normal size for the amplifier board's power caps.
Maybe see if there's room to add a small drilling to the board (so that normal size caps can fit) without drilling through a trace?
__________________
Tools, Models & Software for DIYClipNipper boostLM1875 TurboPowerful TDA7293 kitTDA7294 pt2pt ♦ My post has opinion.
  Reply With Quote
Old 29th December 2012, 10:13 PM   #53
bcmbob is offline bcmbob  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Bay City, Michigan
Thanks, I'll investigate.
__________________
Bob M.
"Arrange Whatever Pieces Come Your Way."
  Reply With Quote
Old 30th December 2012, 12:57 AM   #54
ilimzn is offline ilimzn  Croatia
diyAudio Member
 
Join Date: Feb 2005
Location: Zagreb
A few notes on the TDA7294...

The schematics of the 'insides' are notional only and not the actual schematic. Someone mentioned an op-amp as a phase inverter for the bottom MOSFET, in actuality this is a rather different circuit.

The output stage uses a combination of current source and bootstrap to enable the 'VAS' stage to produce outputs above the positive rail in order to be able to fully switch on the top MOSFET in the output stage. In some cases omitting this mechanism actually produces better sonical results even though there is a reduction in power. The reason is capacitive coupling of the bootstrap circuit to the VAS recycling any crossover distortion into nonlinear current sourcing for the VAS. It is a subtle difference, and can be avouded by driving the input stage power supply to about 8V above the output stage power supply.

In cases where separate power is provided to the input stage, it is VITAL there is a schottky diode preventing the input stage negative power rail ever going more more positive than the output stage negative power rail. The reason for this is the negative input stage power rail is tied to the chip's substrate (and with it the cooling tab), this should always be the most negative part of the chip otherwise parasitic thyristor structures which normally provide insulation of circuit parts from the substrate latch into conduction turning the chip into a short circuit, resulting in catastrophic failure when full power is applied (chips will literally blow apart). This is only hinted at in the datasheet in the 'high efficiency application circuit' - actually a class G amplifier built around the chip.
The most improvement in distortion comes from providing a regulated more negative rail for the input stage, as the first stage of the chip amp has lower PSRR from the negative rail. Adding a regulated and more positive supply rail for the input stage is mainly useful to eliminate the bootsrtap cap for the output stage - a higher input stage positive rail will automatically provide the extra voltage needed to drive the top MOSFET in the output stage to full conduction.

It is highly advisable to use the chip derated, in a configuration with lower than maximum output power. 30-40W has proven to be completely reliable and optimizeable without ever reaching any of the chip's maximum ratings. In particular, the power dissipation rating is a problem, the case will not dissipate as much heat as the actual silicon might suggest from it's voltage and current ratings. Think of these as a 'reserve'. The chip will reward you with top performance as none of it's protection circuits will ever be activated and muddle up the music signal.

The input stage is fairly low noise, but it's still preferable to keep input and feedback impedances low, better 10 or 20k than 50 or 100k. Keeping the DC resistance on both the + and - input is crucial for low offset, also if somewhat lower amplification factors are used (IIRC 20 or 24 is the minimum for stable operation), I have found the chip's input stage offset to be low enough for the feedback DC nlock capacitor to be eliminated entirely.

Input stage decoupling does not have to be excessive and the chip works with fairly long supply lines to the input stage quite well. On the other hand, keeping all the output stage loops as short as possible is HIGHLY desirable. Output stage decoupling is very important and more so with respect to low ESR than actual capacitance values. Also keep the output boucherot RC cell grounded as close as possible to the star ground point of the local decouplers for the output stage.

DO NOT operate the chip without a heatsink, because the MOSFET output stage standing current is quite high and will cause the thermal protection to activate in a few minutes, producing utterly atrocious performance even at very low output levels

If a separate output protection circuit is used, so that the chip does not have to be put into standby or mute, an external current sink can be used to drive the output stage into partial single ended operation, i.e. a higher standing current / asymetrical crossover point amplifier, or even a full single ended class A amplifier up to some 10W or so. This is a power-OPamp 'translation' of the same technique used to drive regular OPamps into class A operation. By providing an external sink, the bottom MOSFET in the power stage is completely switched off and the phase splitter is inactive, making the top MOSFET operate in pure follower fashion. As long as the complete amplifier is asked to sink only as much current as the current sink provides, and the chip to source up to twice the current of the current sink, the output will remain symetrical and work purely in class A. 3 things to take care of here - firstly. the maximum current sourced from the chip is lowered by the current of the current sink WRT chip maximum rating, i.e. maximum current at which the overcurrent protection in the chip will engage. Secondly, observe the maximum dissipation limit, this is a point where the chip is somewhat deficcient so an implementation like this will have to be seriously derated WRT output power as the current sink's current increases. Thirdly, keep in mind that the chip's standby/mute circuitry does indeed put the chip into standby and switches off it's output, but has no idea there is an external current sink - if the chip is put into standby with the current sink turned on, the sink will gladly pull current through your speaker into the negative power supply, potentially destroying your speakers. In theory one could also control the sink with the standby signal, however because it's virtually impossible to make both come on and off completely in sync, there will be a transient as it turns on and off. The proper complex way to do this would be to turn on the chip, turn on the sink (preferably a soft start function should be built into the sink), then unmute the chip. Reverse operation for switching it off, keeping in mind a power supply dropout might not give you enough time to do a soft stop on the sink!
Although it is possible to use a current source instead, this is not optimal. Instead of the bottom MOSFET in the output stage being turned off, it will turn off the top MOSFET which requires the bottom half including the phase splitter to do all the work, requiring the signal to go through the additional circuitry of the splitter, and also because the bottom half works in common source mode, requiring the NFB loop to do more with a 'longer' and thus more phase shift prone circuit. Unlike this, the top part is a straight follower from the VAS, the quickest and shortest way out the power stage.
  Reply With Quote
Old 30th December 2012, 01:33 AM   #55
bcmbob is offline bcmbob  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Bay City, Michigan
That's a bunch of information. Will require several readings to absorb. What I believe I'm hearing is, along with gootee's posts, full optimization will require a more sophisticated board layout than what is generally used. Is that correct and are you aware of any attempts to design that animal?

I finished the dual transformer setup, but won't do any serious listening till tomorrow morning.
Attached Images
File Type: jpg Dul1.jpg (146.4 KB, 620 views)
File Type: jpg Dul2.jpg (238.4 KB, 861 views)
__________________
Bob M.
"Arrange Whatever Pieces Come Your Way."

Last edited by bcmbob; 30th December 2012 at 01:41 AM.
  Reply With Quote
Old 30th December 2012, 02:22 AM   #56
redjr is offline redjr  United States
diyAudio Member
 
Join Date: May 2011
Location: Danbury, CT
Quote:
Originally Posted by bcmbob View Post
....full optimization will require a more sophisticated board layout than what is generally used.....
That takes the idea of tweaking to a whole, new level. While that may be viewed as a challenge for some, I wonder if the sonic characteristics improvements - in terms of 'real' significant differences - offset a total re-design of the amp? That sounds like a real commitment of time, money and a lot of energy. Just a thought. I'm quietly and passively following this thread though.
__________________
redjr
  Reply With Quote
Old 30th December 2012, 02:41 AM   #57
bcmbob is offline bcmbob  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Bay City, Michigan
redjr, I understand what you are saying. It does appear two divergent paths are emerging. I'm coming here from a LM3886/MyRef background and completely appreciate the years and number of contributors necessary to go from "ChipAmp" to it's current state. Maybe this thread will determine which direction gains the most support.
__________________
Bob M.
"Arrange Whatever Pieces Come Your Way."
  Reply With Quote
Old 30th December 2012, 02:50 AM   #58
Jay is offline Jay  Indonesia
diyAudio Member
 
Join Date: Feb 2003
Location: Jakarta
Quote:
Originally Posted by ilimzn View Post
A few notes on the TDA7294...

The most improvement in distortion comes from providing a regulated more negative rail for the input stage, as the first stage of the chip amp has lower PSRR from the negative rail. Adding a regulated and more positive supply rail for the input stage is mainly useful to eliminate the bootsrtap cap for the output stage.
Thanks for the info, ilimzn. I'm attaching your circuit to this thread so everyone can see it as an option to optimize the TDA7294 output (by regulating the input stage).

I myself prefer another option to retain the bootstrap cap. And regulation can be done to entire amplifier. Call it a pre-regulation, because full regulation for output stage of class-B amps usually doesn't sound good for some of us. So a simple zener based pre-regulator can be employed. I'm attaching a power supply used by esperado (Christophe) to power his mosfet amplifier (class-B SAA-Crescendo). Note that after regulation he still provides massive capacitance. I have tried simpler than this and I can accept the sound, noting that I'm a proponent of input-stage-only regulation.

The regulator must be simple, as too expensive might not justify the use of chip amp (?). Fortunately we don't want a full regulation for the output stage. What is your comment about this second option?


As for going "cost-no-object" with this cheap chip amp, can you see any benefit? As you have noticed, it is not really preferable to get the maximum power out of it. About class-A approach, I'm afraid an Aleph-J is simpler and sound better. What I can see is the performance of the mosfet. They say it is a DMOS. But I think it s quite good for class-B. So because lateral mosfets are expensive, this can be a justification for those who like the sound of mosfet amps.
Attached Images
File Type: png TDA7294_ilimzn.PNG (93.1 KB, 589 views)
File Type: gif supply_esperado.gif (19.8 KB, 571 views)

Last edited by Jay; 30th December 2012 at 02:55 AM.
  Reply With Quote
Old 30th December 2012, 03:13 AM   #59
Jay is offline Jay  Indonesia
diyAudio Member
 
Join Date: Feb 2003
Location: Jakarta
Quote:
Originally Posted by ilimzn View Post
The input stage is fairly low noise, but it's still preferable to keep input and feedback impedances low, better 10 or 20k than 50 or 100k. Keeping the DC resistance on both the + and - input is crucial for low offset, also if somewhat lower amplification factors are used (IIRC 20 or 24 is the minimum for stable operation), I have found the chip's input stage offset to be low
Yes, I can see the current mirror of the input LTP. With original schematic I prefer lower feedback resistance (10K for example). But low input impedance (issue for the signal source), along with low feedback (issue for speaker damping) will seriously "ruin" a capacitive mosfet output performance. You will see lower distortion, but the sound will be boring. And this is a situation I think where a matching preamp should be discussed also, as part of the design.

But the schematic from radiolub that I posted previously have a high input impedance and to my ears sound preferable than any schematics with simple FB arrangement I have tried.
  Reply With Quote
Old 30th December 2012, 03:49 AM   #60
diyAudio Member
 
danielwritesbac's Avatar
 
Join Date: Nov 2007
Quote:
Originally Posted by ilimzn View Post
It is highly advisable to use the chip derated, in a configuration with lower than maximum output power. 30-40W has proven to be completely reliable and optimize-able without ever reaching any of the chip's maximum ratings. In particular, the power dissipation rating is a problem, the case will not dissipate as much heat as the actual silicon might suggest from it's voltage and current ratings. Think of these as a 'reserve'. The chip will reward you with top performance as none of it's protection circuits will ever be activated and muddle up the music signal.
. . . the MOSFET output stage standing current is quite high and will cause the thermal protection to activate in a few minutes, producing utterly atrocious performance even at very low output levels
Power dissipation is not always a problem since latfet/dmos/powermos is very efficient. I actually use mine at up to 127 watts with 4 ohm speakers and it requires only a 3"x3"x2" heatsink that doesn't even run hot. It is power circuit options that provide the cool. And, that cool came along with clear sound, directly proportional and related. Yes, a hot running TDA7294 is atrocious, so the easy answer is don't run it hot. Mine is simple bus rail, except. . .

This is not 127 real watts as there is an couple of exceptions that work dynamically. I have employed a peak filter from the hardware store in order to slam something other than the output devices. This is 3 conductors of 20ga solid copper thermostat wire (in the range 5" to 8"), twisted and used as the umbilical cable in-between the power board and amplifier board. . . and then the discharge of the 220u amplifier board power caps is not enough to hammer the outputs. I'll talk about that in the next post because bigger power handling capacity was merely accidental, and not the primary reason for doing it.
__________________
Tools, Models & Software for DIYClipNipper boostLM1875 TurboPowerful TDA7293 kitTDA7294 pt2pt ♦ My post has opinion.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
OPtimizing the VBE MUltiplier fglabach Solid State 45 17th September 2014 07:22 PM
Optimizing active crossover atledreier Multi-Way 27 19th August 2011 02:53 PM
optimizing the VBE multiplier hienrich Solid State 80 29th November 2010 10:25 PM
Interesting linestage optimizing ideas PixelPlay Tubes / Valves 5 23rd June 2008 10:18 PM
optimizing for efficiency Dwiel Solid State 13 10th May 2005 09:31 PM


New To Site? Need Help?

All times are GMT. The time now is 03:43 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2