Optimizing TDA7294 Output

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Stanton was fine with posting the info on the board I'm using. It is the same standard design as was posted earlier but with english comments and little larger. I'm adding a picture of the board to be a visual aid for folks like me to understand which segments are being discussed.
 

Attachments

  • JA Kit.jpg
    JA Kit.jpg
    117.1 KB · Views: 2,980
  • TDA_Big.jpg
    TDA_Big.jpg
    118 KB · Views: 2,981
Daniel,Wouldn't the cap bank be ALMOST as easy to construct using an unetched double-sided PCB for each polarity, a la Terry Given? It should then have much lower impedance, right? Or am I missing something? Tom
Sure, that's fine too.
For the photo in post#15, I just found it fairly easy to use 3 strips of 14ga solid copper, a scrap of phenolic, a bit of flux and a ~35w iron.
I'm following along with this data sheet. Are we all based there? So va is arbitrary and derived - not being limited by a posted spec - correct? Daniel, How did you determine the number and size of the caps used in the ps you posted?
The datasheet has terrible quality schematics but good quality footnotes.

For when the transformer specs are unknown but it is known there's an 8 ohm speaker, then the audiometric value is at least 13,200uF (or more) per each rail of a split rail supply, if you want TDA729X, STK or a discrete amp to hit the low notes as clearly as possible. This capacitance will work best if kept full charged, so it remains true that the transformer must have sufficient amperage to simultaneously charge UP the caps while running the amplifiers. Problem: Using a slightly smaller transformer would require a lot more capacitance.

The transformer is charging the caps while the amplifier is discharging the caps.
We need to keep the caps fully charged if you want high quality clear bass.
That's why needed transformer amperage is decided by amplifier watts.
This is the board I'm using now:
Heavy Duty Supply.......
That's just fine.
You can use 3300u caps for smoothing (C1, C2, C3, C4) and 10,000u caps for reservoir (C5, C6, C7, C8).
It has a lot of capacitance, and, that's probably beneficial with that 200va transformer.

Before parts swapping tinkering/tuning with the amplifier, I'd like to see an LED's, amber or red, per each rail on that power board and a drainer resistor per each rail. Parts: 2 led's and 4 of 5.6k resistors. Each LED needs its own resistor. Each rail also has a 5.6k drainer resistor. Of course you could do 2 led's per rail if you wanted to--they'll drain it down 1.9v per rail, shortly after power off. The combination creates a dual-duty drainer and a safety indicator, useful for. . . waiting for the LED's to go dark before tinkering with the amp. The drain is very mild (doesn't interfere with audio performance) and will take a little time for the LED's to go dark. The LED's are inexpensive and they might save your life.
Daniel's nick name should be "The P2P Man" :)
P2P? That's doable.
247247d1320044638-need-help-triple-parallel-lm1875-dynamics-amplifier-power-big-little-dual.jpg

After this photo was made, the drainer/indicator leds have been added, they're amber, and the whole thing lights up. Without those additions, and without load, this supply stays fully charged (fully hazardous) for about a week. The white diode snubbing caps shown per each diode are 3.3n polyester. The red DC cable snubbing caps shown are 47n polyester. Those added polyester caps are unnecessary for your CRC power board. The power supply in the photo shows the estimated minimum 13,200u per rail for solid clear low bass and will do it if the transformer has enough amperage to keep the caps fully charged in all conditions. . . else double the capacitance or a replacement transformer may be indicated.
 
Last edited:
The TDA7294 works best with a cleaner supply to it's input stage. This has been covered in various threads put up a few years ago.
But it is very important to put in protection diodes in the supply pins to ensure that the chip will not get voltages out of the permissible range which could cause it to blow up ! It does sound cleaner this way and the reduction in distortion in not small ! Check out the FFT .

The last time I tested the 7294, it appeared to have more transient impact than the 3886 ! Even at low volume!It's a good idea to try out both and see for onself. Is this better performance or just more HF trash ? Doesn't look like it on an FFT !:confused:
 
Thanks, all useful information. I did install two LEDs on the PS boards when they were used with the BPA150s. Nowhere as sophisticated as what you suggest - just a straight connection.
Since I hadn't used them in months I forgot about the retained energy, and got a quick reminder when a short happened. Luckily, my body wasn't included in the circuit, but that shock hazard has to be constantly in mind. That's one reason I would prefer a board approach over P2P as there is a lot of bare metal exposed in that style.

The time the LEDs on this kit stayed up was one of the first things I noticed. That was another reason I started to use my two lead light bulb dis-charger as I finalized the build.

Lets see what other replies come in on this portion of the project.
 
Last edited:
ashok, I was being polite when I commented on the other thread using words like "mushy", some IM distortion and low headroom. "Junk" sounds as good as anything I suppose and may be more direct,:)

The base sound is what attracted me and it's my hope that people more knowledgeable than me can help identify and correct those shortcomings. I am a strong devotee of the LM3886 designs, but this TDA has a character and warmth I feel is unique and presents a real high value potential.
 
bcmbob said:
"Junk" sounds as good as anything I suppose and may be more direct,
Yes, the datasheet example schematics are dreadful.
However, the footnotes and text is of useful quality.
And the actual component is of excellent quality.

Problem: The datasheet example schematic shows overlarge power caps at the amplifier board, your amplifier's schematic shows undersize 22u power caps at the amplifier board. However, this amp is quite sensitive to small signal power. I think you're going to need to upgrade to some 220u caps (or 330u caps) for the amplifier board power caps. . . before fine tuning anything else.
2. Bootstrap Cap
This is a miller comp amp and that cap is required for function. Your schematic shows 22u from pin6 to pin14. This is slightly undersize. Some of the newer schematics have 47u. According to the datasheet footnotes, upgrading the bootstrap cap to 47u will do slightly better bass clarity.
3. Bypass caps
Adding a tiny cap to an ordinary quality large cap, to create an audiophile quality large cap. Each combination is unique. I suggest trying it at the NFB cap to improve the treble quality.
1. NFB cap
5. Feedback-shunt
6. Feedback resistor
Your schematic shows 47u NFB cap series to 680R feedback-shunt resistor divided with 22K feedback resistor.

First lets figure the gain. Calculator keys 22 / 0.68 +1 = 34x gain.
That's a bit high for this amp.

Next, let's look at some audiometric NFB cap sizes with matching feedback-shunt resistor values.
150u with 3k3, and nearby values work
220u with 2k7, and nearby values work
270u with 2k2, and nearby values work
330u with 1k9, and nearby values work
470u with 1k2, and nearby values work
1000u with 680R and this isn't a good idea because it is hard to find small signal worthy 1000u caps and that much cap discharge can break the input transistors.

Currently, your NFB cap with 47u and 680R, on the schematic is a bass blocker and the source of poorer quality bass. See the datasheet footnotes where they advise that a bigger NFB cap can get you better bass.

Here's some options that don't have a bass blocker effect:
150u with 3k3 fs and 68k fb, 22x gain

220u with 2k7 fs and 75k fb, 29x gain
220u with 2k7 fs and 68k fb, 26x gain
220u with 3k3 fs and 75k fb, 24x gain
220u with 2k7 fs and 56k fb, 22x gain

270u with 2k2 fs and 68k fb, 32x gain
270u with 2k2 fs and 56k fb, 27x gain
270u with 2k7 fs and 68k fb, 26x gain
270u with 2k2 fs and 47k fb, 22x gain

330u with 1k8 fs and 56k fb, 32x gain
330u with 1k8 fs and 47k fb, 27x gain
330u with 2k2 fs and 56k fb, 26x gain
330u with 1k8 fs and 39k fb, 23x gain

470u with 1k2 fs and 33k fb, 29x gain
470u with 1k2 fs and 27k fb, 24x gain
470u with 1k5 fs and 33k fb, 23x gain
8. Output caps
9. Speaker protector kit
Output caps block DC at the speaker jack, about 6600u for an 8 ohm speaker can be made high quality with 3300u||3300u for low loss. This cap simply goes in series with the speaker+, blocks dc, and protects the speaker.
Speaker protector kit from velleman is an alternative and the kit disconnects the speaker if DC is sensed at the speaker jack.
 
Last edited:
The TDA7294 works best with a cleaner supply to it's input stage. This has been covered in various threads put up a few years ago.
But it is very important to put in protection diodes in the supply pins to ensure that the chip will not get voltages out of the permissible range which could cause it to blow up ! It does sound cleaner this way and the reduction in distortion in not small ! Check out the FFT.
Do you have a link or some examples?
 
The base sound is what attracted me and it's my hope that people more knowledgeable than me can help identify and correct those shortcomings.

What you heard can be different from different system (it is very true for mosfet in class-B). The speaker and the preamp is critical. It can sound good and it can sound terrible. If you have a good speaker, almost any amp will sound good. Can you specify your speaker and preamp?

The bass character and "speed" of this amp is close to Hitachi lateral K135/J50. Even with most discrete K135/J50 class-B amps people are divided into loving it or hating it. Most are against this "muddy" sound, that's why in so many blind test, TDA7294 never won against LM3875/3876. And the mosfet amp only shows its most lovely character rarely in rare music such as an orchestra (while what everybody want is that bloody sweet FEMALE vocal :p)

Adding a tiny cap to an ordinary quality large cap, to create an audiophile quality large cap.

I use a bypass cap with 1/1000 of the bigger cap, to avoid "overlap". If bigger than 1/1000, a series resistor with the smaller cap may be needed. OR, as in my TDA7294 amps, I use paralleled caps of the same capacitance.
 
Last edited:
The speaker and the preamp is critical.
Especially in this case. I'd like to see the gain not up in the 30'x's but rather down near 24x, and then push with a nice preamp. That's an inconvenience, but I think it is worthwhile if one wants real hi-fi from this chip.
TDA7294 never won against LM3875/3876.
Both the ideal support circuits and the usage for which these excel is so very different that I'm not sure how we could compare these, as the only way to make any of them "lose" is misuse it.
The TDA729X is further disadvantaged by its terrible datasheet schematic examples, its short gain capacity and its bizzare pinout that often leads to unoptimized PCB layout. Nevertheless, the part itself is of good quality.
If bigger than 1/1000, a series resistor with the smaller cap may be needed. OR, as in my TDA7294 amps, I use paralleled caps of the same capacitance.
Earlier I was speaking of the NFB cap, which is really not different from an Input cap (except that the typical sizes are different). Effective and easy bypass does include parallel same value cap; but it also includes 10n and nearby value polyester as well as 0.47u and smaller electrolytic. 4n7 and smaller ceramic can also work. And, combination bypass can be done.

At the NFB cap, series resistor with a larger cap works. Here's a composite approach: 22u NFB cap with 10n bypass cap. Onto that is 220u+3R3, and that resistor is at groundside. And additional 0.47u electrolytic can be added to the 220u. Basically the smaller signals and vocals will go through the 22u, while the larger signals and bass will go through the 220u. The approximately 243u agrees with a 2k7 feedback-shunt resistor. It is a real example. This approach and either noise gain or anti-miller comp can be added to TDA7294 to use it high gain without a preamp. That is quite time consuming because the amp is not suited to high gain use. And, actually, the higher the gain on that chip, the more likely it is to be a mid-fi and/or make the fine tuning difficult.

But, of course if the amp is set to low gain, there's less gain put on cap peculiarities of any cap at the amp, so extremes of complexity would be unnecessary. . . or perhaps merely "relocated" to the preamp on clean regulated power not affected by jolts from driving a speaker, which makes the preamp a more sensible spot to have some gain. And, if at low gain, the TDA7294 is more likely to do hi-fi. There's many other factors, but low-ish gain probably makes them easier.
 
Last edited:
Bob? What is your power supply voltage?

I ask because there's 3 TDA7294's.
TDA7294S = TDA7293 Cull (all pins are active)
TDA7294V = What you expect (pins 5, 11, 12 do nothing)
TDA7296/clone = Bose table radio or pod amplifier

They've all got TDA7294 printed on the chip!
But, all three have different needs for operating voltage.

The clone, actually pretty little radio/pod amp, is designed to be run from a DC power supply that has an inexpensive 12+12vac (14+14vac max) transformer, latfets at low voltage are extremely efficient and the gigantic 15 pin package is for reducing heatsink expense (rather dramatically). Unfortunately the label is misleading, sounds like junk on TDA7294's operating voltage and overvoltage/overcurrent accidents that destroy speakers, is frequency reported. This is one of the long list of reasons that I mention output caps. The clone/radio/pod chip is hard to identify, but look at Mouser's 250 unit price of $4 each, in comparison to the radio/pod amplifier's typical street price of $1.50. Well, if the fine tuning efforts mentioned earlier, just don't work, then try a much lower voltage transformer at the supply and find out if the amp sings well at low voltage, possibly indicating that you accidentally bought a wholesale priced Bose. :)

Meanwhile, back to output caps and non-fried speakers. I'd advise to create low loss 6600u caps with either 3x2200u or 2x3300u (typical paralleling of caps for low loss). That's for an 8 ohm speaker. And then put this 6600u cap in series to the Speaker + at the speaker jack. This is highly effective speaker protection for blocking chip amp DC accidents. That's my excuse, but there's great fun to be had by choosing a just right value for bass extension enhancement, customized to a given speaker--a reliable and easy way to get deep power tube bass from a chip amplifier. For treble fine tuning, small caps and/or RC's can be used in parallel to the output cap. Whatever you do, please consider speaker protection any/every time you're using a Chinese amplifier kit. My preference is obviously output caps, but if your preference is different, then consider the Velleman speaker protector kit. Speaker protection costs less than speakers.
 
Last edited:
Stanton was fine with posting the info on the board I'm using. It is the same standard design as was posted earlier but with english comments and little larger. I'm adding a picture of the board to be a visual aid for folks like me to understand which segments are being discussed.

Do you have an image of the other side of that PCB?
 
Do you have a link or some examples?

Hi Daniel,
Offhand I don't have a link but I plan to find my old board and measure it again. I had looked at the FFT when I used it. I seem to recall seeing a FFT on the Net. Will check that out also. I will post it if I find it or post my own measurements a while later. I need to find the board as I don't have time to build a new board.
Cheers
Ashok
 
Last edited:
Hi Daniel, Offhand I don't have a link but I plan to find my old board and measure it again. I had looked at the FFT when I used it. I seem to recall seeing a FFT on the Net. Will check that out also. I will post it if I find it or post my own measurements a while later. I need to find the board as I don't have time to build a new board. Cheers Ashok
Thank you very much. I'm so curious to see what you were talking about in post #24.
 
About the sound aspect in that post , you will have to listen to a 7294 as explanations will not do justice to the subject.

Regarding the supply pin/ voltage issues, this was covered long ago , maybe 2006 (?) . I'll try to recollect the details. There are two supply pins ( 4 pins if you consider both the + and - rails) on the chip. The input stage and the output stage supplies. The differential between the two pins like +Vs and +Vp and -Vs/-Vp must not exceed 15 volts I think and the -Vs pin must not be more positive than the -Vp rail by about 1V or so. If it does you get a mini bomb !

So ( typically ) a Schottky diode and a zener diode are connected between +Vs/+Vp and -Vs/-Vp pins to avoid such a situation. This is done only if your two supplies ( Vs and Vp ) come from different supply circuits. All circuits that use a common supply don't need these diodes.

Check this link: http://www.diyaudio.com/forums/chip-amps/3857-tda7293v.html

Lots on info including FFT's of performance of the 7294 and the 7293 . Don't have time right now to go through it and put up a summary.
Cheers.
 
It seems that some of this effect could be had with just 2 parts, a cap and a diode. So, a fast silicon diode from pin 13, to pin 7, stripe at pin 7, power connects to pin 13. Pin 7 has a diode drop, and then re-stiffen pin 7 with a cap to ground. At pin 7, some fluctuations/noise has been removed, Right? Well, I'm not sure how to make that symmetrical.

For a more symmetrical approach that works with any chip amp, especially for a stereo build, using 4 diodes to convert to dual mono, just by sticking 2 fast silicon diodes into the V+ output of the power board (stripe away from power board) and 2 fast silicon diodes into the V- output of the power board (stripe towards power board). Viola! Dual mono! Half of the jolts from the left amp aren't seen by the right amp and vice-versa. None of the power circuit of the left amp can be utilized by the right amp, and vice-versa, so although they can sag the power board, they can't directly sag each other's amplifier board power caps. Shields up Scotty! Wow, really cheap dual mono, with just one transformer. Also, that won't be dull no matter how much capacitance is piled onto the power board.
 
Last edited:
Thanks, all useful information. I did install two LEDs on the PS boards when they were used with the BPA150s. Nowhere as sophisticated as what you suggest - just a straight connection.
Since I hadn't used them in months I forgot about the retained energy, and got a quick reminder when a short happened. Luckily, my body wasn't included in the circuit, but that shock hazard has to be constantly in mind.
Supplies:
8 of 10k 1/4w resistors
2 of red or amber LED's

Install simple drainers:
Put 3 of 10k resistors under the board, parallel with C1
Put 3 of 10k resistors under the board, parallel with C2

Install safety indicators:
Red LED series to 10k for V+ indicator
Red LED series to 10k for V- indicator
 
Daniel - That's a lot of options to try. I'll sort through my stash for what I have and add some things to my next Mouser order. Thanks for the definitions and examples. May need further assistance later so I don't do the square plug/round hole thing:)

The power supplies I'm using now are 34.5 VDC. It is not a mono-block as both supplies are fed with one 200VA transformer. Havn't had a chance to try the discrete 125 VA setup yet, hopefully today. As stated earlier, I do have two of Peter Daniels bare Universal PS boards that I can put to use if that's a better option.

Jay - My speakers are the Sunflowers by Paul Carmody. I haven't used a pre yet, but my go to piece is a JC-2 clone from the same vendor. My other good pre is a Carver C-1 and I have a Mackie Satellite that is super clean. I have been using JRiver Media Center through a Mini2496 DAC with Dario Inserra's upgrades to feed the amps. The available SPL is more than suficient, though I think I'm hearing that more gain in a pre and less on the amp may benifit the sound quality. Is that correct?

gootee - attached a photo of the bottom.
 

Attachments

  • DSC00902.JPG
    DSC00902.JPG
    222.6 KB · Views: 1,911
  • DSC00897.JPG
    DSC00897.JPG
    230.1 KB · Views: 1,876
  • DSC00899.JPG
    DSC00899.JPG
    150.6 KB · Views: 1,822
  • 01.jpg
    01.jpg
    139.7 KB · Views: 1,824
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.