Tda7293v

I started playing around with this IC and its smaller brother, the TDA7294 about one year ago. The spec sheets can be downloaded from http://us.st.com/stonline/profiles/powerapplications/index.shtml and doing a part number search for "TDA729*" (without the quotes). These devices are power op amps using a bipolar input stage and mosfet outputs.

I found the basic 7294 to have a very pleasant sound, not as open or detailed as the Tripath 104, but reasonably good. It was built according to the plans in the spec sheets. Before building this unit, I searched the Internet and found no variations aside from the spec sheets.

After my experiences with the STK modules (all bipolar stages) described in another thread, I've decided to re-visit this IC power amp using the TDA7293V. This differs from the others (except the TDA7294S is now similar) in that one can create a parallel slave using only the output devices of the amp. In other words, unlike using LM3886 or 3875s in paralllel, where the input stages are duplicated for each device, paralleled TDA7293V use the input stage from one device and run all of the output mosfets from all devices in parallel. This should result in a cleaner setup. Additionally, it may be bridged. Although the spec sheets recommend the abysmal feedback method that is guaranteed meager sound quality. So far I've built up one paralled board and run tests on it. It seems very stable with no tendency to oscillate.

After looking at, I almost had to slap myself on the head. The engineers who designed it had also allowed one to separate the input stage power supply from the output stage power supply. The spec sheets never commented on this. I cut up my board today to decouple the input power supply (pins 7 and 8) using a 100 ohm resistor with 470 microfarads going to ground for each and the distortion was dramatically lowered. This unit has one master and one paralleled slave, to use ST's descriptions.

One thing that was terribly obvious was that prior to decoupling, whenever I changed the signal level while running tests, the distortion tracing would go crazy. This bothered me since music is an ever changing AC signal. After decoupling the input stage, this effect with varying input signal almost completely disappeared. I have to say, I've not listened to this amp yet.

So now I'm reconfiguring the pcbs to account for this, plus a few more changes from the spec sheets (I'll post the pcb once it seems finalized), then I'll build up a balanced bridged amp to compare with the STK. The STK also allows (and their app note encourages) input stage decoupling. The LM3886 series does not permit this. So on paper, the TDA7293V seems intriguing and is in production. Anyone else have much experience with the TDA7293V or decoupling the supplies?

Regards, Robert
 
Today I powered up two monoblocks in a pair of previously used amps with +/-36VDC power supplies as described above and there was a big problem.

The +V decoupling resistor self destructed and took out both 7293s in the first channel I tested at full power. I un-did the mods in the other boards, and re-installed devices in the damaged unit. All is now working fine, but one must not use my previous suggestion for decoupling. I had in the circuit 0.4W 100 ohm resistors but obviously the + leg is doing something else with some of the current. This wasn't seen when I was testing with 40W bulb, so I'm not sure what went on.

As for paralleling two 7293s, the power difference is as follows (36V rails): with one 7294 85W into 8 ohms, 125W into 4 ohms and 30W into 2 ohms. With the paralleled 7293, 85W into 8 ohms, 150W into 4 ohms, 230W into 2 ohms and 215W into 1.6 ohms. No signs of oscillation. Parallel listening tests in a days. I'll tried some bridged versions in near future.

Again, don't decouple the input stage unless you take other measures like maybe a resistor with a much greater power rating or a much lower reistance.

Regards, Robert
 
TDA7294

I have some TDA7294 chips (no suffix like V) and am planning to build an active system, mainly for classical music.

I was a little discouraged to read in one of the thread in this forum that this chip does not sound very good. While I know that such remarks are subjective ("warm", "bright" etc.), what I want to know now is, whether there are different 7294s, with and without suffixes. If there are, are there any differences in their equivalent circuitry?

I had automatically assumed that a DMOS output stage should sound warmer. You think so?

Thanks. Please reply.
 
ST makes a TDA7294S. This 'S' version can be run in parallel like the TDA7293V, but runs at a lower voltage.

I would not say that amps with these chips sound bad. The sound of the parallel/single ended is slightly less open or airy than the bridged STK4048XI or the Rowland model 10 (LM3886) amps--both bipolar--that I described in other threads.

In parallel/bridged mode, the TDA7293V sounds very different. Transients seemed to be accentuated more than with the other amps. Details at the rear of the soundstage seemed more defined, but there also seemed to be more grain or a harder quality to the sound than with the STK4048XI bridged amp.

Since I'm having trouble with this amp at the moment, and since I did not complete my listening tests, I cannot fully comment on the parallel/bridged TDA amp as it sounds in my system. But this amp is definitely not bad sounding. Two TDA7293V in parallel make a wonderfully robust, single ended amp (as I said elsewhere, it easily drove a 1.6 ohm load at full power!) that costs relatively little to put together.

regards, robert
 
so, where can one buy a couple TDA7293 chips

it seems that everyone but me can find these chips in small quantities. I need a couple for a repair job on a sub woofer and all I can find is wholesale houses with a 250 chip minimum. There must be some source in the USA that sells these at a reasonable price and in quantities of 2 to 10. But, what is that magic source??

thanks in advance
 

OddOne

Member
2002-11-12 7:08 am
Earth
I built an aux supply board for my 7293 project, taking advantage of the extra 50VAC center-tapped secondary on my transformer. Hooked it to the signal-stage power pins on the 7293 after rectifying and smoothing the AC to suitable DC. The loaded output was 30VDC above and below ground, whereas my driver-stage power is 44VDC. Grounds for both otherwise isolated supply sections were tied together so that the voltages will reference each other properly.

It didn't work. All the 7293s did was try to dump raw DC into the speakers, tripping my MOSFET-based speaker protector circuit, which immediately pulled the plug on the outputs. :bigeyes:

Apparently the reason the 7293s' specs don't mention isolated supplies for signal and final state is that the ICs do not like it AT all if you're not feeding both off the same supply, even if you do common-ground them.


Looks like other methods of dealing with distortion are going to be needed...

oO
 

OddOne

Member
2002-11-12 7:08 am
Earth
Has anyone tried an LC network to separate the signal power instead of RC? It seems to me that this would be better at damping strange transients while not adversely affecting current flow, assuming one picks the right parts. Plus, you won't be blocking DC to the supply pins if you're not using a capacitor in series with them.

And, since it's pretty obvious that a capacitor in series with the signal-power pins causes the magic smoke to be released, what about L in series, and C bypassing to GND? (IOW, build what is essentially a passive 2nd-order lowpass filter in series with the signal supply pins.)

oO
 

halojoy

On Hiatus
2002-11-05 6:14 pm
Must be a way to solve this.

Possible causes of the blow up.

IC does not like one stage powered up before the other.
What stage would be first I do not know,
I should guess the driver first then output stage.

Mute and standby is activated, deactivated in wrong order.
In the datasheet there is some diagram
that tries to show how this should be done.
with diode and RC-delay.

A combination of both this things.

Remember that big capacitors in power supply can act as time delaying components.
What is the reference for the mute and standby - at what time is that reference and voltage present?
 

moamps

diyAudio Member
2002-10-11 8:15 am
Croatia
www.moxtone.com
Hi, Richard

In past two weeks I made some experiments with three TDA7293 (master + 2 slaves). In one p2p wiring and small heat sink, I have not any problems with PS to +- 45V. But, I find some difference in application notes for this chip according slave wiring.

From:
http://www.ub-elektronik.de/baugr-texte/amp.htm

1. Modular Application (more Devices in Parallel) -

The master chip connections are the same as the normal single ones.
The outputs can be connected together without the need of any ballast resistance.
The slave SGND pin must be tied to the negative supply.
The slave ST-BY pin must be connected to ST-BY pin.
The bootstrap lines must be connected together and the bootstrap capacitor must be increased:
for N devices the bootstrap capacitor must be 22mF times N.
The slave Mute and IN-pins must be grounded.

And from:
http://eu.st.com/stonline/books/pdf/docs/6744.pdf

2. Modular Application (more Devices in Parallel)

The master chip connections are the same as the normal single ones.
The outputs can be connected together without the need of any ballast resistance.
The slave SGND pin must be tied to the negative supply.
The slave ST-BY and MUTE pins must be connected to the master ST-BY and MUTE pins.
The bootstrap lines must be connected together and the bootstrap capacitor must be increased:
for N devices the bootstrap capacitor must be 22mF times N.
The slave IN-pin must be connected to the negative supply.


Different is connection of MUTE pin.

I made wiring according first manual, friend of my was trying second apps, and this don't work.

This chip is very sensitive; I made some experiments with using only output FET-s in slave mode for something like "unity gain gain clone" but without success until now (one fried chip).

Maybe my evidence can help you.

Regards
 

Antorian

Member
2003-04-29 12:07 pm
Moon
I'm a newbie. (Sorry for my poor english) :goodbad:
I expect doing Amps (mono) with TDA7293 in modular applications.
So my kestion is:
Wich Impedance must be placed (for the speaker) when using 1,2,3 or n chip in slave?
Is there a relationship between that? (I presume that Power dissipation is a fact that cannot be understimate).
Thank you.


moamps, thanks for the Mute advice
:nod:
 
Hi,

I think that configuration with master and 2 slaves is maximum. With this configuration you have about 3x6A=18 amps peak current, on 2E speaker this is huge power.
You can connect any impedance down to 2E, but must think about power dissipation on chips when choose PS voltage.

My plan is building some active 2way speaker with Seas drivers and one 7293 for high, 2 parallel for bass-midle with overall power about 30-40W, no more.

Regards
 
Well, I am about 3 years too late here, but anyway:

TDA 7293/94 have relatively poor PSRR, which is why a separate well filtered power supply on the signal section can improve things quite a lot.
Why does it blow up if you connect the signal power via a RC filter to the output power? Well, the reason is rather simple if one knows a bit about IC design:
Looking at the pinout, it becomes obvious that the -Vs signal power pin is connected to the heatsink tab. More importantly, this is the substrate of the IC, which needs to always be the MOST NEGATIVE point in the IC otherwise various structures in it become conductive or even behave as latched-up thyristors, rendering the whole IC a big short circuit. Result: the IC literally explodes.
With a simple RC filter, this will happen on initial power application because the -Vs pin will be more positive than the -Vp pin. A simple cure is to put a (preferably schottky) diode from -Vs to -Vp alowing the filter cap off -Vp to charge 'immediately' with the main power supply.
This is quite easy to see on the 'high efficiency' application diagram, although, they used fast diodes instead of schottky. Many get confused because that is a class G amp, the diodes are there for power supply commutation - however, they also insure that -Vs is never less that 1 diode drop WRT -Vp, ditto for +Vs and +Vp. It is obvious from that schematic that the TDA7293/4 indeed can work with separate Vs and Vp voltages.
It is highly probable, though, that there is amaximum difference in Vs and Vp. Given the power supply voltage difference in the high efficiency application (20V) it is safe to assume that the DMOS process used to manufacture the chips produces similar maximum gate to source voltage limitas as for regular power MOSFETs, about 20V. If separate power supplies are used, care should be taken that Vs never becomes over 20V more than Vp.
 
ilimzn , thanks for your explanation of the principles of working of this IC :)
I am going to try to make my TDA7294 working from separate supplies for signal and power parts. My intention is to power the signal part from separate transformers with some 10V higher voltage, than power part (+-27V now). Also, to make sure, the substrate is always the most negative part of the circuit, I will put quite a lot of bypass capacitance in signal supply, a little oversized traffos(around 40-50VA per channel, for fast initial charging of capacitors), and higher value bleeding resistors (than on power part), so that when powered off, the Vp would become zero faster, than signal part.

What do you think about such arrangement? Will it work?

And finally, could you explain please the following:
keeping +Vs a minimum of 7V higher than +Vp also dispenses with the bootstrap capacitor...