Younger brother of the LM3886 - LM2876 in parallel

Member
Joined 2011
Paid Member
Hi all,

Because LM3886 is nowhere to find I got two LM2876s and paralleled them. These are the same Overture series as LM3886 and sound equally good, except not as powerful.

I was able to get 40W 50W into 8ohm and 60W 76W into 4ohm with 24V-0V-24V transformer secondaries.

Updated measurements:

Power supply
24V-0V-24V AC secondaries
33.4V DC at idle

Gain: 25.8dB
SNR: 101.39dB
Sensitivity: 0.835mV

8ohm
19.9Vrms max output
49W into 8 ohms
THD+N: 0.0011% or -98.96dB at 1kHz
THD+N: 0.0023% or -92.58dB at 10kHz
Heatsink barely got hot.

4ohm
17.5 Vrms max output
76W into 4 ohms
THD+N: 0.0024% or -92.4dB at 1kHz
THD+N: 0.0058% or -84.6dB at 10kHz
Heatsink got warm.

4ohm 5W output
THD+N: 0.0013% or -97.61dB at 1kHz

Listening tests:
This board sounds amazing!
Great resolution. Clear highs, warm mids, and powerful low end.
Had no problem playing low sensitivity ELAC speakers.
Very dynamic sound. At max volume got so loud!
I would say this chip sounds 95% of the LM3886.
 

Attachments

  • IMG_1479.jpeg
    IMG_1479.jpeg
    341.5 KB · Views: 203
  • IMG_1481.jpeg
    IMG_1481.jpeg
    374.9 KB · Views: 204
  • 1000hz_4ohm_4.5Vrms_out_V+-29.2V.PNG
    1000hz_4ohm_4.5Vrms_out_V+-29.2V.PNG
    65.9 KB · Views: 170
  • 1000hz_4ohm_15.5Vrms_out_V+-29.2V.PNG
    1000hz_4ohm_15.5Vrms_out_V+-29.2V.PNG
    64.5 KB · Views: 138
  • 1000hz_8ohm_6.5Vrms_out_V+-29.2V.PNG
    1000hz_8ohm_6.5Vrms_out_V+-29.2V.PNG
    65.3 KB · Views: 133
  • 1000hz_8ohm_17.89Vrms_out_V+-29.2V.PNG
    1000hz_8ohm_17.89Vrms_out_V+-29.2V.PNG
    66.1 KB · Views: 148
  • IMG_1474.jpeg
    IMG_1474.jpeg
    405.7 KB · Views: 148
  • IMG_1475.jpeg
    IMG_1475.jpeg
    389.9 KB · Views: 151
  • IMG_1483.jpeg
    IMG_1483.jpeg
    315.9 KB · Views: 175
  • thd_n_1khz_4ohm.PNG
    thd_n_1khz_4ohm.PNG
    27 KB · Views: 54
  • thd_n_1khz_4ohm_5W.PNG
    thd_n_1khz_4ohm_5W.PNG
    26.8 KB · Views: 50
  • thd_n_10khz_4ohm.PNG
    thd_n_10khz_4ohm.PNG
    27.4 KB · Views: 43
  • thd_n_1khz_8ohm.PNG
    thd_n_1khz_8ohm.PNG
    26.5 KB · Views: 44
  • thd_n_10khz_8ohm.PNG
    thd_n_10khz_8ohm.PNG
    27.4 KB · Views: 73
Last edited:
  • Like
Reactions: 1 users
I tried out a Chinese version of TDA7294V from Sanxin, a serious chip maker for medical domain.
It costs $1.5 only and has much better sound than LM3886.
I used it for bellow circuit, with LM3886 is stupid sounding.
The LM2876, although TI selling for $2.5k/1k units, I couldn't find better than $19 a pair.
 

Attachments

  • tda7294 k.JPG
    tda7294 k.JPG
    51.5 KB · Views: 226
Last edited by a moderator:
  • Like
Reactions: 2 users
Member
Joined 2011
Paid Member
I tried out a Chinese version of TDA7294V from Sanxin, a serious chip maker for medical domain.

It would be interesting to compare (and measure) sanxin and original versions. Is it possible that both original so to say and those are coming from the same place of manufacture?
Btw, I really like simple idea of the inverter 2N7000 with CCS - only three transistors. How does it sound and measure?
 
The purpose of the circuit is to simulate open loop SE triode sound of my 2A3C by low cost chip amp. The problem with chip amps and many bad designed amps, is the low NFB measured @20khz. The feedback with low NFB, generates a distortion that occurs when multiple voices of same note sing simultaneously. It can be large vocale, as the requiems of Verdi or Mozart for the worst case, or large Jazz orchestras. Instead of hearing a pleasant beat, you hear distorted sound. To avoid such distortion, either the open loop should be higher than 20khz to avoid phase shift or the NFB must exceed 60db@20khz. To have the requiem of Mozart inspiring spiritual emotion, the amp must have celestial voice and it gets with 70db. The purpose of the MOSFET is to generate second order harmonics as SET and pull the NFB from 24db to 70db @20khz. The MOSFET and the I/V chip now is corrected 200 times more @20khz than before. The distortion of LM3886 is not modeled, but the noise is and it is reduced by 170 fold. In other words the circuit cleans the nonlinearities of the chip and replaces by that of the MOSFET.
The sound of the Sanxin TDA has a very beautiful texture at midrange that reminds me the 45 rpm records.
Never buy original TDA from Ali because they are originals but defective.
 
  • Like
Reactions: 1 user
Member
Joined 2011
Paid Member
Got new device in the mail and now I am able to measure distortion with greater precision.

Output power graph is for 4 ohm load.

At 5W into 4ohm load THD+N is 0.00144% or -96.8dB
 

Attachments

  • THD_vs_POWER_1khz_4ohm_29.6V_LM2876x2.png
    THD_vs_POWER_1khz_4ohm_29.6V_LM2876x2.png
    14.4 KB · Views: 66
  • SNR_1khz_4ohm_29.6V_LM2876x2.png
    SNR_1khz_4ohm_29.6V_LM2876x2.png
    27.6 KB · Views: 68
  • THD_1khz_4ohm_29.6V_LM2876x2.png
    THD_1khz_4ohm_29.6V_LM2876x2.png
    24.6 KB · Views: 64
  • THD_1khz_4ohm_29.6V_LM2876x2_v2.png
    THD_1khz_4ohm_29.6V_LM2876x2_v2.png
    28.6 KB · Views: 59
  • THD_10khz_4ohm_29.6V_LM2876x2.png
    THD_10khz_4ohm_29.6V_LM2876x2.png
    27.9 KB · Views: 58
  • THD_5W_1khz_4ohm_29.6V_LM2876x2.png
    THD_5W_1khz_4ohm_29.6V_LM2876x2.png
    27.2 KB · Views: 62
Last edited:
Member
Joined 2011
Paid Member
I had a chance to swap transformer for 24V-0V-24V AC secondary - gives 33.6V DC.

Output power increased to 80W into 4ohm at THD+N <0.1%
and 50W into 8ohm at THD+N <0.1%
 

Attachments

  • THD_vs_POWER_1khz_4ohm_33.5V_LM2876x2.png
    THD_vs_POWER_1khz_4ohm_33.5V_LM2876x2.png
    15.3 KB · Views: 57
  • THD_vs_POWER_1khz_8ohm_33.5V_LM2876x2.png
    THD_vs_POWER_1khz_8ohm_33.5V_LM2876x2.png
    15.4 KB · Views: 51
The chip has no inscription, just black probably to let the vendors write what suits best for their commercial.
if that the case then ,how come it have ST logo."write what suits best for their commercial" if that true then they will put their own name/brand on it just like other.

now tell me do you really have that ''Sanxin made TDA7294".


just because you read the "sanxin" on Aliexpress sellers post ,it didn't make it true that it is really a their product.

Screenshot (76).png
 
There are many companies called Sanxin, no chip makers, the medical device company does not make chips.
So your information is wrong.

There are some reputed Chinese companies, who put their name on their products, like CRC Micro and Unisonic Taiwan.
Those are more reputed here than the ones marked with others names like ST.
 
With naked eyes, I see no inscription, it should be erased. This vendor is the only one giving the choice of its source original or Sanxin. A year ago, it was TDA7293 with free shipment now it is TDA7294 with over $3 shipment.
If I chose Sanxin, to be sure to get the counterfeit because I got counterfeit TDA2050 for $1.5 in lot of 10 pieces, they were newly engineered with unity gain stability and had exceptional high frequency character. I hoped this chip is also from same source and it reviled to be also high quality sounding.