Custom Audio Amplifier Experiment

[eapen]

Bought an Ebay 350w Class D amplifier. Coupled it with 480vA toroidal transformer & DC filtering circuit.

Initial testing done was on an 20w 8? speaker using a 20? 2w resistance in series (to reduce load). The speaker is silent & when the finger is touched at the audio input noise is heard through the speaker.

 

[spirovious]

Tested with any song?How is it overall?

 

[saheb11123]

Brought from ebay india?

 

[Fantastic]

Curious to know what it cost and how much duty you had to pay.
Sometimes things come through duty free ! It's unpredictable !
Thanks.

 

[kaushik]

waiting for findings and review

 

[eapen]

Bought it from Ebay.com for around 25USD. It's a mono board which uses IRS2092 chip & IRFB4227 mosfet.

As per the specs listed it is rated 350w 4? for DC supply +55v -0- -55v. Transformer is 40v-0-40v 6A AC with DC filtering of 4x 10,000uf plus additional 2x 4700uf on the amplifier board.

The issue is even with less speaker load (20W 8? speaker in series with a 20? resistor) the inductor coil on the amplifier board (between the heat sink & capacitors) is getting hot. I believe since I have given excess VA it is overwhelming for the amplifier to handle.

For a proper speaker test ordered Peerless 12" 4? subwoofer which can handle 400w. Connected everything together again. Seemed fine initially. Ran the finger tip test continuously for 3 minutes. Alas! the inductor coil burned up ! Ho ! these low cost
Ebay Class-D amps are over-rated

Photos with the coil removed -

 

[eapen]

Actually initially I wanted to build a Class AB amp for the sub. In fact I ordered an audio amplifier board (PCB) based on Accuphase Accuphase Laboratory, Inc. transistor amp architecture. It has Class AB Fully Differential topology using low noise JFETs in the pre-driver stage. 3x Transistor pairs 2SA1943 & 2SC5200 or NJW0302 & NJW0281 are used in the final stage.

Wanted to step in the world of Class D amps due to their high efficiency factor & feel the beat. Well, I've felt the heat for now with the inductor coil burned ! Class D amps use a PWM high frequency filtered by the output inductor & capacitor - LC network

 

[spirovious]

The issue is even with less speaker load (20W 8? speaker in series with a 20? resistor) the inductor coil on the amplifier board (between the heat sink & capacitors) is getting hot.

Is 20? resistor really needed? Is that extra resistor causing problem?

 

[IndianEars]

I suspect the Inductor is in Parallel with the 20 Ohm resistor.

Hence the resistor overheated and blew AFTER the inductor / coil went open circuit.

Normally, most of the current at audio frequencies will flow through the coil and not the resistor.

 

[eapen]

As mentioned before, I connected the 20? series resistor to the 20W 8? test speaker to reduce load. With the test load itself the inductor was getting slightly hot to touch.

There is a blue LED onboard which lights up if the voltage rails are correct.

Everything was fine until I connected the 350W 4? speaker. With the excess current draw within 30+ seconds the coil burned up.

The listing says the the 22uH inductor is rated for 20A. I doubt it could take more than 3 amps.

Well I tried repairing the board with 2x compact Nickel Zinc Ferrite 6A 22uH coils in parallel. The MUR160 fast recovery diodes were also shorted (circled white). Replaced those too.

Now a 1-second pulsating beat is heard though the speaker. The blue led doesn't turn on & a 1-second pulsating beat is heard though the speaker.

The issue with these boards is the pre-driver stage is made of SMD components & very fragile making repair a difficult process. Check the underneath.

I guess I'll have to throw this board in the dustbin.

 

[Fantastic]

Don't give up so easily. There must be some way you can fix it , even if it takes time doing it and getting parts. If the board and tracks get badly burnt then it can be trashed !
SMD boards are a pain but often can be fixed if you attempt it slowly and carefully. Nothing is impossible !

 

[omishra]

Guys,

Your Amp landed into oscillations. That resistor make zobal network for class D amp. its required. And to replace coil use red-clear toroide (T65-2 above, T80 like that) to wind 18SWG wire over it - 39-40 turns for 11H.

On eabay search Micrometal toroide cores with red-black(clear) core.

For resistive load use element coil of electric heater which as 200? resistance. Make one end fixed and other variable with crocodile clip. Fix whole assembly over wooden plank. measure required resistance with DMM. If speaker sound need to be heard then connect 100? 3W in series with speaker. Connect this parallel to above resistive load.

Resistive load with burn wattage upto 2KW while little sound audible with speakers.

 

[eapen]

I read about of IRS2092 & IRFB4227 implementation in class D amps. There seems to be a synchronization issue between the chip & mosfet pair if the pre-driver stage is not implemented properly (mosfet gate capacitance affecting switching frequency). Next time I might try with IRS2092 driving IRFI4019 or IRFI4020 - mosfet (dual mosfets half bridged topology) which are specifically designed for audio amplifiers. I have postponed Class D experimentation due to the complexities involved.

Back to class AB - TDA7294 (left) 100w vs. LM3886 (right) 68w - two of the famous Gainclone chipamps for testing

 

[eapen]

Testing chipamps with SMPS pair - one for each voltage rail. Speaker is Peerless 6.5" S16NI. Extra capacitance of 4700uf x2 & 0.1uf x1 per rail were added to reduce line ripple.

SMPS used - ERD 12v 3A x2 first & then 19v 3.5A x2

 

[eapen]

The TDA7294 board made by XD218 (China) has NE5532 in the input stage & uPC upc1237 speaker protection & 3A diode x4 rectifier with 6800uf capacitor x2 regulator stage whereas the LM3886 made by Wintek (India) is a 'no frills' board.

The TDA7294 is designed to be connected AC 28v-0-28v transformer supply

To test it with SMPS I had to bypass the relay of the TDA7294 board since it is 24v & engages for input volt >25v only.

On playing music, the TDA7294 I felt emphasizes the base (could be due to the additional input stage) whereas LM3886 the mids & highs.

With 24v (12v x2) 3A both sounded the same volume whereas with 38v (19v x2) 3.5A the TDA7294 sounded louder with more speaker excursion. Also at 38v 3.5A LM3886 dissipates a lot of heat even at low volumes more than TDA7294.

To check the susceptibility to DC power fluctuation I cycled the power on/off 3 times in 5 seconds. The TDA7294 was blown instantly whereas nothing happened to LM3886. I even switched off one SMPS rail & still no problem for the LM3886. It is much more robust than TDA7294

To check whether the SMPS & bypassed speaker protection circuit caused the failure of TDA7294 , I had a 28v-0-28v transformer custom made & connected to another TDA7294 XD218 board. The speaker protection circuit worked fine with a 3 second initial power on-delay & music played fine. However when the input power was rapidly cycled as before it blew !

 

[omishra]

@eapen, you seems to be fitting the category FAB (Fearless Amp Builder)
Nice but costly experiments. I try to avoid holy smoke, but you seems enjoying it.
:clapping::clapping::clapping:

 

[eapen]

Lol, even I do fear turning amps into 'decorative fireworks' with the flip of a switch !

Usually chipamps have inbuilt protection against power fluctuation (within a range), short-circuit & high heat. That's why I tested like that.

Regardless of the fact I like the sound signature of TDA79x series - punchy base with soft treble, am unlikely I will tread in their path again due to instability issues. Though the TDA7294 had ST Microelectronics logo on them it I doubt its authenticity.

Another factor I noticed is the speaker on/off pop noise is more for TDA7294 whereas LM3886 is silent. Sometimes the off pop noise is too loud to being irritating.

 

[saheb11123]

Lm3886/4780's are made like a tank, usually if you connect the outputs too it wont do anything/nothing will happen as compared to discrete designs. Tda7294 is alltogether another category only, it doesn't have the features of lm3886, it has higher distortion and the sound signature is not subtle and "cool" as of the lm3xxx/4780 series. But tda7294/3 is great for a sub amp.

 

[eapen]

I had bought the Peerless speakers, LM3886 amplifier board & power supply board (shown earlier) through diyaudiocart. The power supply board was enhanced by replacing the 4,700uf x4 capacitors with 10,000uf ones.

Also picked up:
1)Low pass frequency filter by Wintek. It is a fourth order filter using 3x TL072 op-amps with cut-off frequency at 110Hz.

2) speaker protection circuit by Lumin using uPC1237 which protects the speaker against amplifier output DC, over load & short circuit.

Tested the C1237 speaker protection circuit with a 0-12v power supply. Switches on the speaker after 4 sec delay after power on. An extra connection goes into the transformer AC section to switch off the relay immediately during power off suppressing pop off noise.

 

[eapen]

Another Class AB 350w high power amplifier testing. Sourced from ebay.in member id: ajithjose0 of Indo Power Electronics, Kerala

ajithjose0 | eBay

At an initial glance one would notice surprisingly low parts count for a discrete Class AB amplifier of this power - with no components underneath the PCB. Tested with 40v-0-40v transformer, 10,000uf x4 capacitor & Peerless 12" 400w subwoofer.