Interesting experiments! Why do you think, isn't this done already? Efficiency? Risk of fire? Cost?
//
//
H
HAYK
Historians remain perplexed. Why the egyptians, the ones who built pyramids, navigating up and down the Nile with sails thousands of years, to hoist the boom used a single pulley, did not figure out multiple pulley system and hoist with less manpower? What the Greek discovered and used to built mega temples.The same question is the wheel in the new continent.
I found many PhD thesis on Class D with PhD level mathematics, even one suggesting to use BD mode with digital PWM source, but remains his answer unanswered, how to overcome the crossover nonlinearity. If you follow how I reached to the conclusion to add just a resistor, as if I circumnavigated the planet to find the answer was just behind my shoes. This technique puts a definitive brake for GaN in classD
I found many PhD thesis on Class D with PhD level mathematics, even one suggesting to use BD mode with digital PWM source, but remains his answer unanswered, how to overcome the crossover nonlinearity. If you follow how I reached to the conclusion to add just a resistor, as if I circumnavigated the planet to find the answer was just behind my shoes. This technique puts a definitive brake for GaN in classD
H
HAYK
This circuit doesn't have output inductors to accumulate bias current.
Bellow is a high level analysis of BD mod. Look how they turn around the problem remaining blind for the simple solution.
Bellow is a high level analysis of BD mod. Look how they turn around the problem remaining blind for the simple solution.
OK - I see that now - I just purchased a very low price amp with that chip. It uses inductors. But perhaps one could take them away all together... I'll check the paper - thanks!
//
//
H
HAYK
H
HAYK
To check if your amp sounds good, find on YouTube Franck'S My Way and listen if the metallic voice of Franck sounds golden or garbage. This is high frequency harmonic distortion test. To test the resolution you need vocal group, as Pink Floyd the wall, Turtles, Happy together, Requiems of Verdi and Mozart. If you hear dirty sounds, curse the rubbish MP3, It helps you sleep better.
H
HAYK
The TPA3255.
The 325X series are similar to TPA3221 with advanced loop filter added with error corrector. The patent bellow I think is the one implemented. The circuit shows adding an offset on either branch of BTL to apply positive or negative offset. This good news means in SE mode the error corrector cannot function hence disabled. If I look the distortion curves, indeed in SE mode is higher, even higher than the TPA3221 in BTL mode. What I find strange is the power capabilities of this chip. On YouTube I see with 2X68W for 8 ohms and 1X90W in PBTL. Above, the chip goes into protection due to overheat. The amp I ordered recommends to use it with 36V, I am disappointed for the price $22 cost where the TDA7498e board does the same for $12.
The 325X series are similar to TPA3221 with advanced loop filter added with error corrector. The patent bellow I think is the one implemented. The circuit shows adding an offset on either branch of BTL to apply positive or negative offset. This good news means in SE mode the error corrector cannot function hence disabled. If I look the distortion curves, indeed in SE mode is higher, even higher than the TPA3221 in BTL mode. What I find strange is the power capabilities of this chip. On YouTube I see with 2X68W for 8 ohms and 1X90W in PBTL. Above, the chip goes into protection due to overheat. The amp I ordered recommends to use it with 36V, I am disappointed for the price $22 cost where the TDA7498e board does the same for $12.
Attachments
H
HAYK
A NZ fellow measures the TPA3255 with 45V supply and gets clip light off at 25Vrms that is 35V peak giving nothing but 72W. How is this possible. The DS recommends 53.5V insane volts, where I got wrong? 6:25
Known problem of almost all TPA32xx, especially on cheap cards... At 25 dollars...
I moved on.🧐✌️
I moved on.🧐✌️
H
HAYK
H
HAYK
There is something fishy in this story. With 25Vrms on 8 ohm is 3.125Arms. There are four switches on each channel of 0.1 ohm max. this gives max dissipation of 7.8W. Say that such size of heatsink cannot cool it bellow 125° has certainly error in filter design.
H
HAYK
In waiting the TPA3255, I started converting the CS8683. This Chinese chip is very scarce to find, only one type of board available with two inductor flavor. The chip alone, I can't find, the mono board cost me total, about $6. Low voltage electrolytics need to be changed as this chip is designed to function with 36V. The frequency needs to be adjusted by shorting un applied R2, I unwind 18 turns the inductors to be 10uH. The ground linking the two output capacitors are cut on both sides of the board and a pair of 100 ohms/2W in series or 390 ohms in parallel is inserted to ground the capacitors.
On Chinese forum, I found an unclear image of the evaluation board, as TPA3116 that it imitates, the gain can be reduced by adding in series resistors to the inputs as the gain is 40 times or 32db.
Those who like to read numbers, this chip can go down to 0.005% THD+N at frequencies bellow 1khz.
I am hoping that it has the same extraordinary transfer function as the Chinese TPA3110, presuming to be the same manufacturer.
On Chinese forum, I found an unclear image of the evaluation board, as TPA3116 that it imitates, the gain can be reduced by adding in series resistors to the inputs as the gain is 40 times or 32db.
Those who like to read numbers, this chip can go down to 0.005% THD+N at frequencies bellow 1khz.
I am hoping that it has the same extraordinary transfer function as the Chinese TPA3110, presuming to be the same manufacturer.
Attachments
I did this with one of my 3118 boards and can confirm what I encountered years ago, having this same idea: Beginnung with 10R series resonance, THD increases dramatically vs series resistance. There is a reason nobody applies this "ingenious" mod.