FFT Analysis with different TPA boards...
Hi
i did an "update" at my DS1054Z
now i try to compare my 4 ohm boards.
TPA3251 EVM, YJTPA3255 am p 5 and amp 6 (ad8599)
pic 1 amp5 (TLo2, epcos and TDK filter )
pic 2 EVM TPA3251
pic 3 amp6 (ad8599-original lc outputs)
graph:
upper pic
is the input 1khz with 200mVrms input both rca, 4ohm load,
lower pic = FFT
i can see the harmonics 1khz , 3khz, 5khz are bigger....
i am not sure if the measurment is correct: Math function /FFT, hanning, trace, centre is about 7,5 khZ, so the 0hz is in the left corner, 1khz/DIV
i set the offset (pruple marker) in the lower window at 38,2dbV - approx 40dbV
low frequencies are about 80-90dbV and so i get about 70dbV in the right corner?
please can some expert look at that?
chris
Hi
i did an "update" at my DS1054Z
now i try to compare my 4 ohm boards.
TPA3251 EVM, YJTPA3255 am p 5 and amp 6 (ad8599)
pic 1 amp5 (TLo2, epcos and TDK filter )
pic 2 EVM TPA3251
pic 3 amp6 (ad8599-original lc outputs)
graph:
upper pic
is the input 1khz with 200mVrms input both rca, 4ohm load,
lower pic = FFT
i can see the harmonics 1khz , 3khz, 5khz are bigger....
i am not sure if the measurment is correct: Math function /FFT, hanning, trace, centre is about 7,5 khZ, so the 0hz is in the left corner, 1khz/DIV
i set the offset (pruple marker) in the lower window at 38,2dbV - approx 40dbV
low frequencies are about 80-90dbV and so i get about 70dbV in the right corner?
please can some expert look at that?
chris
Attachments
additionally i create a 10ohm board
so its a TL02 original opamp with other caps for input and regulators but the coils are selected from all original board to get about 13,5 µH --> this is why i called it 10ohm board....my sonus faber have a high impedance measured by my brother and its simlar to that what you can find in the net.
for this board i use 8ohms load on both channels.
oohh...i forgot to mentoined that the psu during the test and the post before is set at 28V so its nearly to my amp light and actual setup psu
chris
so its a TL02 original opamp with other caps for input and regulators but the coils are selected from all original board to get about 13,5 µH --> this is why i called it 10ohm board....my sonus faber have a high impedance measured by my brother and its simlar to that what you can find in the net.
for this board i use 8ohms load on both channels.
oohh...i forgot to mentoined that the psu during the test and the post before is set at 28V so its nearly to my amp light and actual setup psu
chris
Attachments
I cannot comment on the Rigol FFT - based on 8bit Y-resolution I do not expect high resolution.
Another issue is that measurements across the BTLlegs often deliver smaller amounts of THD than measurements of one leg referenced to ground as with an oscilloscope.
For a quick comparision it would be helpful if you measured the numbers of the ratios of k1/k2, k1/k3 etc
All in all an 100MHz/8bit oscilloscope seems to be a poor instrument to measure audio distortion.
So I recommend looking for some 24bit USB-soundcard that affords one symmetrical line input required for symm bridge output.
Insert a symmetric attenuator of ca 30dB between bridge out and line in.
Install ARTA on your pc. Works on Linux/Wine as well as under win or OSX.
Now you have a fully featured audio measuring station.
Another issue is that measurements across the BTLlegs often deliver smaller amounts of THD than measurements of one leg referenced to ground as with an oscilloscope.
For a quick comparision it would be helpful if you measured the numbers of the ratios of k1/k2, k1/k3 etc
All in all an 100MHz/8bit oscilloscope seems to be a poor instrument to measure audio distortion.
So I recommend looking for some 24bit USB-soundcard that affords one symmetrical line input required for symm bridge output.
Insert a symmetric attenuator of ca 30dB between bridge out and line in.
Install ARTA on your pc. Works on Linux/Wine as well as under win or OSX.
Now you have a fully featured audio measuring station.
hi voltwide
Another issue is that measurements across the BTLlegs often deliver smaller amounts of THD than measurements of one leg referenced to ground as with an oscilloscope.
do you mean i use the a normal scope probe? no.... i upgrade my lab - i use a differential probe.
i use a notebook and i do not want to buy a sound card now.
so the measurements are not usefully? for compare in my lab?
Another issue is that measurements across the BTLlegs often deliver smaller amounts of THD than measurements of one leg referenced to ground as with an oscilloscope.
do you mean i use the a normal scope probe? no.... i upgrade my lab - i use a differential probe.
i use a notebook and i do not want to buy a sound card now.
so the measurements are not usefully? for compare in my lab?
k1k2k3....
Hi voltwide
do you mean k1k2 compare like this..?
i try some calculation with math and cursor function
amp6 and 3251evm at 24Vpsu, 1 kHz 200mVrms input, 4 ohms load
with the cursor position between 1khz and the 13khz i got a difference BY-AY = amp6 of 42,8dbV - EVM board 38,8dbV
so 4 db better
pic 1 24v amp6
pic 2 24V 3251evm
max voltage YJ 40V - EVM is use 34V
pic 3 amp6 40V
pic 4 3251evm 34V
chris
Hi voltwide
do you mean k1k2 compare like this..?
i try some calculation with math and cursor function
amp6 and 3251evm at 24Vpsu, 1 kHz 200mVrms input, 4 ohms load
with the cursor position between 1khz and the 13khz i got a difference BY-AY = amp6 of 42,8dbV - EVM board 38,8dbV
so 4 db better
pic 1 24v amp6
pic 2 24V 3251evm
max voltage YJ 40V - EVM is use 34V
pic 3 amp6 40V
pic 4 3251evm 34V
chris
Attachments
Yes that is what I meant.
The results k1/k13 close to 40dB are way too bad -
and obviously due to the 8bit resolution of your scope. I am happy with the same Rigol Scope than yours - but neither for audio noise nor for THD measurements. You cannot measure realistic THD values with such limiting equipment.
So I bought an used EMU-tracker on e-bay for about 50€, built the attenuator and can measure THD down to 0.001% or even better.
Hi Forest,
just got the LRS-150-36 and I lowered the voltage down a bit to 35v and the power is more than sufficient. In terms of sound quality I think the china board is good. Clean sound, transparent. the bass is good on my Mordaunt short aviano 1 speakers. This board deserves a good and decent enclosure I must say.
It is so tempting to change the stock op amp (TLO72) to opa2132ua and LM2575 TO LM2575HVS-15/NOPB as what other guy is saying in the feedback in aliexpress that it will work correctly. Can anyone justify his claim.
Thanks.
just got the LRS-150-36 and I lowered the voltage down a bit to 35v and the power is more than sufficient. In terms of sound quality I think the china board is good. Clean sound, transparent. the bass is good on my Mordaunt short aviano 1 speakers. This board deserves a good and decent enclosure I must say.
It is so tempting to change the stock op amp (TLO72) to opa2132ua and LM2575 TO LM2575HVS-15/NOPB as what other guy is saying in the feedback in aliexpress that it will work correctly. Can anyone justify his claim.
Thanks.
Hi Ferchin79
This is the normal answer by aliexpress and its correct. the amp is working.
my experience is that the TL02 could be fake - i have 2 board where the labe looks origin.
2 boards...the blue v2 looks fake
so changing of SMD is tricky but possible. Do you have the correct equipment to do that? at this small board everything is really very tight and closed together
...the LM2575 is working but the label looks not original TI
change the caps at the regulator. normally they are the same as the inputs -put it away! and use low some 100-220µF /25V electrolytic caps. -change the coil "marked" with220 --> 22µH to a real 220µH --> labeled 221this give better sound .... lower noise
you still want to have the LM2575HV version? you bought the LRS 150-36 and its fine.
chris
This is the normal answer by aliexpress and its correct. the amp is working.
my experience is that the TL02 could be fake - i have 2 board where the labe looks origin.
2 boards...the blue v2 looks fake
so changing of SMD is tricky but possible. Do you have the correct equipment to do that? at this small board everything is really very tight and closed together
...the LM2575 is working but the label looks not original TI
change the caps at the regulator. normally they are the same as the inputs -put it away! and use low some 100-220µF /25V electrolytic caps. -change the coil "marked" with220 --> 22µH to a real 220µH --> labeled 221this give better sound .... lower noise
you still want to have the LM2575HV version? you bought the LRS 150-36 and its fine.
chris
First thanks...its not TI its NS for the LM2575
on the pics the opamps are wrong side --> but the order is:
PIN
4 -3-2-1
5-6-7-8
normally the label of the opa you can read and then the counting is like this:
8-7-6-5
1-2-3-4
chris
preparation for 8 ohm filter
Hi
for my setup for 8 ohms output filter = 10µh / 1µF .
the coils are about 11µH measured.
i changed the output terminals because of the size of the new coils and the order of coils and caps.
i use the coils from würth 7443631000. TDK caps
have fun...
chris
Hi
for my setup for 8 ohms output filter = 10µh / 1µF .
the coils are about 11µH measured.
i changed the output terminals because of the size of the new coils and the order of coils and caps.
i use the coils from würth 7443631000. TDK caps
have fun...
chris
Attachments
blue board V2 evolution 1
blue board V2 evolution 1
just the blue/gold cap after the regulator directly at the 220 coil. it is labeled as 16V/220µF....its very small...i put it out....
pic 2
i try to made a pic about the fake TL02 at the version blue 2
bye
chris
blue board V2 evolution 1
just the blue/gold cap after the regulator directly at the 220 coil. it is labeled as 16V/220µF....its very small...i put it out....
pic 2
i try to made a pic about the fake TL02 at the version blue 2
bye
chris
Attachments
so the original coils are good enough for the "low" power of the living room for listening???? - i will build another board with better opamp and better output filter coils for compare..
as i wrote in #499
i changed the to the "ultimate board"
AD8599 and the Epcos Coils. (4R 7µH/0,68µF)
the distance to my TPA3251EVM board is getting closer but not so smart/ quiet as the EVM board.at all my SQ test i compare additionally to my reference board = TPA3251EVM and it always was the best
....but....
the sound with the AD8599 is realy very very nice. its stange the EVM is quiet and sounds perfect but always i here the sound over the AD8599 its getting little bit more "noisy" but the saxophone and instruments a very real...hmm...
noise ? regulator?
will search for the regulator noise and change the original decoupling caps
never the less i build a board with a different OPAMP = its the BB OPA1602...its not ready but the SQ-check will be soon...
chris
as i wrote in #499
i changed the to the "ultimate board"
AD8599 and the Epcos Coils. (4R 7µH/0,68µF)
the distance to my TPA3251EVM board is getting closer but not so smart/ quiet as the EVM board.at all my SQ test i compare additionally to my reference board = TPA3251EVM and it always was the best
....but....
the sound with the AD8599 is realy very very nice. its stange the EVM is quiet and sounds perfect but always i here the sound over the AD8599 its getting little bit more "noisy" but the saxophone and instruments a very real...hmm...
noise ? regulator?
will search for the regulator noise and change the original decoupling caps
never the less i build a board with a different OPAMP = its the BB OPA1602
...its not ready but the SQ-check will be soon...chris