Optimizing TDA7294 Output

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
...............Once again, the Honey Badger discrete amplifier has a gain of 41X, and so our gain of 38X is three times less voltage amplification. .............
No.
the difference between 41times (+32.26dB) and 38times (+31.60) is about 1.08times (0.66dB), not 3times voltage gain.

You have a knack at turning your gobble de gook.............
 
Last edited:
LM7294?? SUPPOSED TO BE ABOUT TDA7294 LOL

The theme of this thread will be adjustments and modifications necessary to take advantage of and optimize the performance of the LM7294 device. The base or starting point will be the following schematic:


View attachment 319665

Due to a wide range of technical understanding, very basic definitions and descriptions of both primary and supporting components should be posted. Hopefully, this "Paint by the numbers" approach will attract and inform both novices like myself as well as those who wish to share their advanced knowledge and experience.
 
mashoni23 said:
LM7294?? SUPPOSED TO BE ABOUT TDA7294 LOL
Yes, the thread is about TDA7294. However, there's no guarantee to get a real TDA7294 from Ebay. So, from Ebay source, I'd rather use the Modular (master/slave) configuration Parallel TDA7293 without ballast, which narrows the possible chips down to either an authentic TDA7293 or a re-marked TDA7294S, both of which work reliably.
No. the difference between 41times (+32.26dB) and 38times (+31.60) is about 1.08times (0.66dB), not 3times voltage gain.
Thanks AndrewT. I will look up the materials again.
And, we're still aiming for lower gain, when practical to do so.
Question: If given a 1v input, what gain factor setting do I need in order for the power amp to reach all the way to 37+37vdc rails?
Hi *danielwritesbac!!What type of Bootstrap Cap do you use?
47uF for the solo board,
or 100uF for the Parallel board.
Preferred make/model includes Panasonic FC and Nichicon FW, although there are many other suitable makes and models.
The voltage tolerance can be either 50v or 63v or higher. A skinny 50v "tallboy" style cap fits the board easiest.
 
Last edited:
Yes i use 22uf, but i mean what type low Esr, esl audio grade or something else? Thank you for rep!!
The amplifier is more fragile if you risk the boostrap function faltering with a too small cap, so for durability, you'll probably want something larger than 22uF. That 22uF is the minimum value, and minimized functionality with maximized risk and therefore minimized longevity is probably Not a suitable design goal.
The minimum value safe in all conditions is 33uF.

You can use as small as 33uF value of Low ESR capacitor if you want to.
It is also just fine to use normal capacitors in the range of 33uF, 39uF or 47uF per chip.
But, let's not overdo the bootstrap cap value either.

Except when under-volted, the TDA7294/5/6 doesn't have protection against either too little or too much. Here's some examples (links):
Under-volting to support a too small bootstrap cap, doable with regulated front end
Under-volting to support a too large bootstrap cap, doable with wide range tolerance
Those are the exceptions.

P.S.
Normal cap of right value is far better than high end cap of wrong value. :D
 
Last edited:
Adjustable gain

The attached schematic shows adjustable gain for precise audio quality control.

Materials:
A 200 ohm 10-turn trimmer (the little blue box with screw on top), which we're going to employ as a variable resistor; and, you might like to have a range of caps to try out for supplementary compensation, 22n, 33n, 39n, 47n, 68n.

Function/benefit:
These materials will let you adjust the amp gain and speed all over the place, and you can adjust the speed separately from the gain. Simply dial it in for the best audio quality. That dial is shown on the schematic attached.

Notes:
At my house, I'd be using a 100 ohm multi-turn trimmer; however, I've put the 200 ohm multi-turn trimmer on the schematic to accommodate people who want to install lower gain. On audio quality, the amplifier will no doubt inform you of the setting it wants, and the trimmer is far more precise than fixed value resistor guesswork, and so you're a LOT more likely to get good results with the trimmer. Even more importantly, need of compensation and also the ideal gain setting varies by operating voltage and load, so I think that the trimmer dial should be mandatory for optimal audio quality. Function of the trimmer is adjustable gain, but benefit is "audio quality dial." I'm sure that we should use it.
 

Attachments

  • TDA7294-Adjustable.gif
    TDA7294-Adjustable.gif
    40.3 KB · Views: 418
Last edited:
I have two final questions before ordering the last of the parts.
1) Is the 2.2uf cap on the input a non-polarized electrolytic such as I use for speaker crossover work
2) Is the resistor in parallel with the speaker output 6.8ohm or 6.8kohm. I feel it is the latter but want to be sure. Or should this value be varied depending upon the impedance of the speaker
 
You can use the output RC that comes with the kit. No need to buy more. The value can be varied somewhat, depending on need. Most reasonable values outperform omission by far.

The input cap can be 1uF, 1.47uF or 2.2uF, so I suggest to collect some of the little green Bipolar Nichicon ES series, in sizes, 0.47u, 1u, and 2.2uF to test drive and find out which you like. They're very high resolution performers at low cost.
For input cap there exists other charming options, such as an Elna Cerafine in parallel with a tiny value polyester.

I've used the 0.47uF little green Nichicon ES bipolar electrolytic in parallel with my NFB-shunt cap for enhanced clarity. It is generally trouble-free for use as a bypass cap. A really ordinary 1uF electrolytic cap *may* do a similar job for bypass tasks.

Ah, but don't forget to get multi-turn cerment trimmers (little blue box) in either 100R or 200R value. In three days of soldering effort with resistors, the best performance was from the schematic that shows the trimmer.
 
Last edited:
A collection of boards by the same artist:
TDA7294 solo: TDA7294 65W Mono Amplifier Board Kit 28 | eBay
TDA7293 solo: 1 PC x Assembled Finished 85W TDA7293 Mono Power Amplifier Board Support BTL | eBay
TDA7293 Parallel: TDA7293 x2pcs 170W Mono Amplifier Board Kit 29 | eBay
TDA7293 Triple: TDA7293 x3 BTL 250W Mono Amplifier Kit 26 | eBay

These seem to share one flaw: The feedback-shunt resistor is arranged decoratively/mistakenly, causing a weird metallic tone. The fix: When you move the feedback-shunt resistor and the feedback resistor to under-board, extremely close to the (master) chip, the "sound of the board" influence is subtracted. That indicates the rest of the layout, especially power layout, is working well.

-----------AND-----------

So far, progress on the TDA7293 Parallel amp. . .
It works without supplementary compensation added; however, I'll be trying out some new compensation from Keantoken to see about the resolution and imaging possibilities.

Consider the attached schematic a work in progress as it is not quite finalized. Without using supplementary compensation, it is stabilized by gain and it sounds best stable. It does play rather nicely. A 100R or 200R cerment trimmer (little blue box) is good and with no "wrong" setting, just dial it in for pretty audio (and definitely not an arbitrary gain setting). Suggested trimmer startup value is exactly 50 ohms (if your power board is driven from a 25+25vac transformer like mine and if your speaker is 8 ohms like mine).
Input load is marked "TBD" and plausible range of input load is from 12K to 27K whereby the 12K is more forwards or the 27K is more laid back, so adjust within that range as you like. The plan of adding compensation should broaden the range of the "sweet spot" and make the amplifier easier to adjust and use, and I'm working on that.
This is the one I'm using to refit my receiver, and since the receiver's power circuit is slightly different, including voltage, I'll definitely be needing the trimmer to dial in for audio quality.
 

Attachments

  • TDA7293-proto.gif
    TDA7293-proto.gif
    18.3 KB · Views: 554
Hey Guys, for those of us interested in a new PCB for this project, I did get a reply from Klaus. As with all of us, he has activities and commitments beyond the fun audio DIY projects we all enjoy so much. No doubt when things balance out he will become active again - but that won't be anytime soon. Not a problem.

Daniel has given us much to explore and I for one, need and will take the time to develop that path for now. Anyone wishing to present ideas beyond that, or extensions of Klaus's comments are certainly encouraged to do so.

Thanks All !
 
The TDA7293 parallel turned out to be a bit better performer, I think.
Starting here: TDA7293 Parallel board thread
I'm doing up some build photos like these:

How to relocate the feedback resistor and the feedback shunt resistors
348298d1368403830-tda7293-parallel-kit-ebay-modular-slave-style-no-lossy-emitter-resistors-tda7293-parallel-resistors.jpg



How to fit a large size nfb-shunt cap and antiparallel 1N4007's.
If a 680u is unavailable, a pair of 330u or trio of 220u will fit for 660u.
Not shown: treble bypass, 1u or 0.47u electrolytic, fits under-board
348696d1368566875-tda7293-parallel-kit-ebay-modular-slave-style-no-lossy-emitter-resistors-tda7293-nfbs-antiparallel-047.jpg




P.S.
Coming up soon, next photo up (there-->: TDA7293 Parallel board thread <--link), should be the task of fitting MOAR 220u power caps and a couple of big diodes on such a small board. I believe it works about like this: MBR (0.25v) schottky with 220u or less per rail; OR, MR/MUR (0.47v) fast silicon with 440u or less per rail; OR 6A05 (0.65v) standard silicon with 660u or less per rail. In this case, simple 6A05's from the Radio Shack, seem most useful because they work for All of the examples. However, from the farm, the Radio Shack and 6A05's are 108 mile round trip drive. Sometime this week, I'll probably make that drive.
 
Last edited:
Guys - just FYI. I built in Daniel's circuit on top of my AudioSource AMP-100. Essentially, this is a mono TDA7294 for each side. I want to thank you guys for all the hard work that's being done.

I did want to report back, though, that the "lead lag compensation" circuit is a really bad idea. I took a frequency response curve of the amp and there's a ~+10dB boost at around 20KHz with this circuit in place. No, that's not a typo. Basically this circuit seems to be creating an over-damped filter, and the ringing occurs smack dab in the treble region.

FWIW - I had used a 15 ohm + 68nF.

When I removed this circuit, the FR curve went back flat. I'll post pictures tomorrow when I get back to work and have access to my AP again.

I did verify that the issue is in the amp circuit only. None of the other preamp circuits on the AUDIOSOURCE are presenting the same boost. So, the issue is wholly with the amp.

I would recommend that someone else try this test - it's pretty easy.

With the desired circuit in place, play a 997Hz sine wave. Measure the output from your source and also measure the output from the amp. Then play a 19000Hz sine wave. Again, measure the output from the source and measure the output from the amp.

Essentially, you should see no differences whatsoever in the RMS voltage level for either measurement. If the level is changing coming out of your source, then your source (like a computer) likely has some processing applied. Find a different source, or remove the processing. If the level is changing out of the amp, then something is wrong with the circuit of the amp. This latter case is what I discovered.

Again, I'll post back tomorrow...
 
Last edited:
OK - So here's the data requested. Again, I'm doing this on an AudioSource AMP-100, which is essentially the same as a mono TDA7294 per channel. I've verified that none of the preamp circuits in the amplifier are affecting the signal, so these frequency response graphs are wholly the responsibility of the TDA7294 circuitry chosen.

My circuit matches Daniel's last one posted (on page 43 or 44 I think) with the exception of the fact that I used a 1000uF NFB capacitor instead of a 680uF, and I used a single 732-ohm feedback shunt resistor instead of a 510+220 solution.

Personally, I would remove the lead-lag compensation all together. It is creating an over-damped low-pass filter with significant ringing. I cannot fathom why this circuit would aid the chip in any way. The only reason I can contemplate is if your power supply is extremely noisy and tanking the out-of-band noise is providing some benefit.

However, the power supply on my Audiosource AMP-100 is pretty decent, and the ones being designed by Bob and others are far superior to that. So - again, I say remove it, but you guys can make your own decision. If you want to leave it, I would recommend NO MORE than 4.7nF.

Here is the picture of the 15R + 68nF.
attachment.php


Here is one channel with 47nF and the other with 27nF.
attachment.php


Here is 10nF and 5.6nF. It shows roughly +12dB at 20KHz:
attachment.php


Here is 2.7nF and 1nF:
attachment.php


Finally - here is NO lead lag compensation:
attachment.php
 

Attachments

  • 1.jpg
    1.jpg
    133.5 KB · Views: 641
  • 2.jpg
    2.jpg
    163.7 KB · Views: 640
  • 3.jpg
    3.jpg
    151.4 KB · Views: 735
  • 4.jpg
    4.jpg
    150.9 KB · Views: 634
  • 5.jpg
    5.jpg
    125.8 KB · Views: 650
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.