Yes, exacty... but now doesn't matter 😀 'cause I find the problem : The Oscilations !
I made this modifications :
* C3 : to 430pF
* C4 : to 139pF
* in T9,T10 B-C i put an 100pF
* R12,R15 : 56R (now i have the current that quasi said)
- R6:18K and R17=33K remains the same 😀
... and those 50W are back ! loud and clear
but after 50W it begins to distort again a little ....
i think i'll go up with capacitors value.
Do you think it's good to raise C8 a little ? In my PSpice simulation it realy affect the bandwidth if is changed. 😕
I made this modifications :
* C3 : to 430pF
* C4 : to 139pF
* in T9,T10 B-C i put an 100pF
* R12,R15 : 56R (now i have the current that quasi said)
- R6:18K and R17=33K remains the same 😀
... and those 50W are back ! loud and clear
but after 50W it begins to distort again a little ....
i think i'll go up with capacitors value.
Do you think it's good to raise C8 a little ? In my PSpice simulation it realy affect the bandwidth if is changed. 😕
HI Marus,
You can increase C8, try 27pF.
But I think you still have a problem and some of the component changes you have made particularly C4 are only masking a problem not fixing it.
Can you get some genuine MJE340 / 350's and replace the KSE types? Also make sure that there are no unconnected FET gates (ie, an open cct 27 ohm resistor or broken track).
Also try grounding both heatsinks (attach a wire to main ground to each one).
Cheers
Q
You can increase C8, try 27pF.
But I think you still have a problem and some of the component changes you have made particularly C4 are only masking a problem not fixing it.
Can you get some genuine MJE340 / 350's and replace the KSE types? Also make sure that there are no unconnected FET gates (ie, an open cct 27 ohm resistor or broken track).
Also try grounding both heatsinks (attach a wire to main ground to each one).
Cheers
Q
Hi Quasi,
Even though I do not have the level and depth of knowledge you have about amplifiers, I have the same sense that Marus rescent results with added capacitors and value changes is masking the problem. The magnitude of some of the capacitor values concerns me.
I assume you have noticed beyond the obvious distorted waveform, the P-P of the output signal becomes smaller. When I do the math on the output V/input V the gain at the output dedreases. Seems to me the gain is supposed to remain constant with a perfect PSU and within SOA of the output stage devices.
That said, I am wondering if part or all of the problems Marus is experiencing may be related to the zobel network? Either the values need changing or one of the components is faulty perhaps?
I still wondering if part of the problem may also related to the transformers Marus is using?
Regards,
John L. Males
Willowdale, Ontario
Canada
14 September 2007 (23:15 -) 23:42
Official Quasi Thread Researcher
Even though I do not have the level and depth of knowledge you have about amplifiers, I have the same sense that Marus rescent results with added capacitors and value changes is masking the problem. The magnitude of some of the capacitor values concerns me.
I assume you have noticed beyond the obvious distorted waveform, the P-P of the output signal becomes smaller. When I do the math on the output V/input V the gain at the output dedreases. Seems to me the gain is supposed to remain constant with a perfect PSU and within SOA of the output stage devices.
That said, I am wondering if part or all of the problems Marus is experiencing may be related to the zobel network? Either the values need changing or one of the components is faulty perhaps?
I still wondering if part of the problem may also related to the transformers Marus is using?
Regards,
John L. Males
Willowdale, Ontario
Canada
14 September 2007 (23:15 -) 23:42
Official Quasi Thread Researcher
Intersting point John,
Marus what are you using as a load when you are taking your measurements? Have you taken a measurement before the output filter?
Cheers
Q
Marus what are you using as a load when you are taking your measurements? Have you taken a measurement before the output filter?
Cheers
Q
* I don't use KSE.
* there are no unconnected FET gates ( I can measure the bias of every FET)
* when i take the graphics i used :
Yesterday, as i said 😀, i got back those 50W with +/- 81V DC rails. With a little distorsion at this maximum power level. (my sound card output only 0.9 V peak )
At 50W output the rails was 77...78V and with a few watts 79...80V.
I don't see why the changes that i made can be a concern. Those capacitors are put there exactly for this job, to stop the oscilations. Let's ask Quasi what is C2,C3,C4,C5,C6,C8 for ? 😀
I change this capacitors even more (in PSpice simulation) and nothing change in the output wave form. Just the bandwidth has drop a little from 0.76 V (normal capacitors) to 1.28 V (new capacitors), but this is what we want to do. Of course, at higher frequecys the attenuation it's more intense.
Quasi, if what i say is wrong, please tell me how do i STOP the oscilations ?
An externally hosted image should be here but it was not working when we last tested it.
* there are no unconnected FET gates ( I can measure the bias of every FET)
* when i take the graphics i used :
An externally hosted image should be here but it was not working when we last tested it.
Yesterday, as i said 😀, i got back those 50W with +/- 81V DC rails. With a little distorsion at this maximum power level. (my sound card output only 0.9 V peak )
At 50W output the rails was 77...78V and with a few watts 79...80V.
I don't see why the changes that i made can be a concern. Those capacitors are put there exactly for this job, to stop the oscilations. Let's ask Quasi what is C2,C3,C4,C5,C6,C8 for ? 😀
I change this capacitors even more (in PSpice simulation) and nothing change in the output wave form. Just the bandwidth has drop a little from 0.76 V (normal capacitors) to 1.28 V (new capacitors), but this is what we want to do. Of course, at higher frequecys the attenuation it's more intense.
Quasi, if what i say is wrong, please tell me how do i STOP the oscilations ?
About the output filter L1-R35-C12 : It's cut off frequency is 300Khz.
Can i use 8uH with 120nF to lower cut off to 150Khz ?
Can i use 8uH with 120nF to lower cut off to 150Khz ?
Maybe the 50W limited by distortion is due to overloading the output of the source?marus said:
0.9Vpk input and 28.3Vpk output is a power amp gain of 31.4times.
What ratio of feedback resistors have you fitted?
What Zin resistor is fitted?
What Zin can the source drive?
Marus,
Have you done all your testing using your PC's sound card? I think it's time you gained access to a CRO and had a look at a real waveform and not one re-constructed by the D/A converter on the sound card.
You have posted a peak to peak maximum output from your sound card of 0.9v. This is only 0.32v RMS and well below what is required to drive this amp.
Have you checked the waveform coming out of your sound card?
Also can you please post a diagram showing all the test connections and can you please lable all measuring devices and what type they are.
Cheers
Q
Have you done all your testing using your PC's sound card? I think it's time you gained access to a CRO and had a look at a real waveform and not one re-constructed by the D/A converter on the sound card.
You have posted a peak to peak maximum output from your sound card of 0.9v. This is only 0.32v RMS and well below what is required to drive this amp.
Have you checked the waveform coming out of your sound card?
Also can you please post a diagram showing all the test connections and can you please lable all measuring devices and what type they are.
Cheers
Q
C2,C3,C4,C5,C6,C8
C2 with R1 is a low pass filter.
C3 with R9 reduces the gain of the first stage for high frequencies.
C4 rolls off the open loop gain making the amp more stable.
C5 & C6 do some power supply filtering, shunting HF artifacts to ground. These must be of high quality.
C8 increases the feedback at higher frequencies thus reducing the gain.
Cheers
Q
C2 with R1 is a low pass filter.
C3 with R9 reduces the gain of the first stage for high frequencies.
C4 rolls off the open loop gain making the amp more stable.
C5 & C6 do some power supply filtering, shunting HF artifacts to ground. These must be of high quality.
C8 increases the feedback at higher frequencies thus reducing the gain.
Cheers
Q
Yes, with sound card ...
Yes, 0.9 V peak (not peak-to-peak), that makes 0.64 V RMS and of course is below the required level to drive this amp.
Yes, it's ok.
Yes, next time i test it. Now i'm very tired and depresed. I must lay low for a while ...
Thanks for the explication of capacitors.
Today i raised the value of capacitor.... the oscilations disappierd but the amp sound like ... !@#$.
Last configuration:
PSU : +/-80V (77V at load)
C2: 430pF
C3: 660pF
C4: 233pF
C8: 20pF
T9,T10 B-C : 330pF
Everything else unchanged from the latest Quasi schematics.
That's it ...........
quasi said:
Yes, 0.9 V peak (not peak-to-peak), that makes 0.64 V RMS and of course is below the required level to drive this amp.
quasi said:
Yes, it's ok.
quasi said:
Yes, next time i test it. Now i'm very tired and depresed. I must lay low for a while ...
Thanks for the explication of capacitors.
Today i raised the value of capacitor.... the oscilations disappierd but the amp sound like ... !@#$.
Last configuration:
PSU : +/-80V (77V at load)
C2: 430pF
C3: 660pF
C4: 233pF
C8: 20pF
T9,T10 B-C : 330pF
Everything else unchanged from the latest Quasi schematics.
That's it ...........
Re: Post #2230
Marus,
Regarding your Post #2230, have you noticed that the input waveform has artifacts of the distortion so clearly obvious in your example images of the Post # 2316 THD-1 and THD-2 waveforms and hint of the same distortion in the THD-0 example waveform?
I would suggest you put one quasi module back to the original, if you have not already, and try a quality input source such as preamp so you can control the input level to the modules under test and connect a CD player to the pre-amp and just play some music you know and see how it sounds ignoring the fact one channel of the input signal is missing for a single module test.
If you happen to have a quality Audio Test CD that has sample signals of different levels and frequencies then listen and scope what happens with those signal tests. If you can borrow a quality signal generator that would be great. I would suggest running the same waveform tests and frequencies you had been trying with the soundcard. I would also run a triangle and square waveform tests as well. The square wave will almost certainly determine if there are outstanding problems and assist in you in finding the cause of the problem.
The distortions you have experienced to date do appear to be related to the soundcard, be it the card's output impediance, quality, software/driver and/or waveform file quality being used to play through the soundcard. One should always use a WAV file that has been created as a WAV file. MP3, and even OGG, files are lossly based sound file formats that will detele what these file format standards feel is not required to playback a music file. I know as I have run quality waveform tests through these different file formats and know the difference in what each file format is capable of. I did so to see the differences on the file format qualities. OGG files were much better than MP3, but WAV files are superior to OGG and MP3 file formats. This difference to the WAV file format is not a file format issue. It is simply the nature of lossly file formats. Lossly file formats will always be missing parts of waveforms.
What soundcard are you using? The one built onto the motherboard? If so, generally the embedded sound chip of a computer motherboard is not all that great. I have no extensive experience with consumer soundcard quality. I know there is one consumer based soundcard, Revolution, by M-Audio that is very good quality. M-Audio professional soundcards are great. Perhaps other members with more knowledge and experience with soundcards, supporting software, drivers, et al can offer you guidance if you cannot borrow a quality signal generator or find a quality Audio Test CD.
Regards,
John L. Males
Willowdale, Ontario
Canada
15 September 2007 (08:37 -) 09:52
15 September 2007 09:59 Added text link to posting. jlm
Official Quasi Thread Researcher
Marus,
Regarding your Post #2230, have you noticed that the input waveform has artifacts of the distortion so clearly obvious in your example images of the Post # 2316 THD-1 and THD-2 waveforms and hint of the same distortion in the THD-0 example waveform?
I would suggest you put one quasi module back to the original, if you have not already, and try a quality input source such as preamp so you can control the input level to the modules under test and connect a CD player to the pre-amp and just play some music you know and see how it sounds ignoring the fact one channel of the input signal is missing for a single module test.
If you happen to have a quality Audio Test CD that has sample signals of different levels and frequencies then listen and scope what happens with those signal tests. If you can borrow a quality signal generator that would be great. I would suggest running the same waveform tests and frequencies you had been trying with the soundcard. I would also run a triangle and square waveform tests as well. The square wave will almost certainly determine if there are outstanding problems and assist in you in finding the cause of the problem.
The distortions you have experienced to date do appear to be related to the soundcard, be it the card's output impediance, quality, software/driver and/or waveform file quality being used to play through the soundcard. One should always use a WAV file that has been created as a WAV file. MP3, and even OGG, files are lossly based sound file formats that will detele what these file format standards feel is not required to playback a music file. I know as I have run quality waveform tests through these different file formats and know the difference in what each file format is capable of. I did so to see the differences on the file format qualities. OGG files were much better than MP3, but WAV files are superior to OGG and MP3 file formats. This difference to the WAV file format is not a file format issue. It is simply the nature of lossly file formats. Lossly file formats will always be missing parts of waveforms.
What soundcard are you using? The one built onto the motherboard? If so, generally the embedded sound chip of a computer motherboard is not all that great. I have no extensive experience with consumer soundcard quality. I know there is one consumer based soundcard, Revolution, by M-Audio that is very good quality. M-Audio professional soundcards are great. Perhaps other members with more knowledge and experience with soundcards, supporting software, drivers, et al can offer you guidance if you cannot borrow a quality signal generator or find a quality Audio Test CD.
Regards,
John L. Males
Willowdale, Ontario
Canada
15 September 2007 (08:37 -) 09:52
15 September 2007 09:59 Added text link to posting. jlm
Official Quasi Thread Researcher
It's nothing wrong with my sound card. I'm sure of that. 100%
I have a Genius 4.1. It's old but it's good. Not like M-Audio but enough. And ... for my taste, mp3 it's good enough too 🙂
It seems to me a little bit exagerated to say that an amp can make a distorsion because of the quality of the audio mp3 files...
Now, I have an amp with TEA2025B and it performs extraordinary on my [PA800+PCT300] speakers from P.Audio with this Genius 4.1 sound card.
The only thing that i can blame is the quality of my transistors.
Anyway .... can i make a good preamp with OPA4134 ? (I have a few samples )
I have a Genius 4.1. It's old but it's good. Not like M-Audio but enough. And ... for my taste, mp3 it's good enough too 🙂
It seems to me a little bit exagerated to say that an amp can make a distorsion because of the quality of the audio mp3 files...
Now, I have an amp with TEA2025B and it performs extraordinary on my [PA800+PCT300] speakers from P.Audio with this Genius 4.1 sound card.
The only thing that i can blame is the quality of my transistors.
Anyway .... can i make a good preamp with OPA4134 ? (I have a few samples )
marus said:
Marus,
Why do you feel there maybe some quality issue with your transitors? You are using OnSemi MJEs and those should be of excellent quality. Do you have some concern about the other transistors?
Regards,
John L. Males
Willowdale, Ontario
Canada
15 September 2007 11:34
Official Quasi Thread Researcher
What a shame ...
An externally hosted image should be here but it was not working when we last tested it.
A shame indeed.
Marus,
How is the small heatsink mounted. Is being held up by the transistors only? I it touching any of the components underneath?
Is any part of the PCB touching the main heatsink? From your photos the left hand side looks very low.
Have you tried grounding the heatsinks?
What material are you using for the transistor insulators?
Do both of your boards behave the same way?
Cheers
Q
Marus,
How is the small heatsink mounted. Is being held up by the transistors only? I it touching any of the components underneath?
Is any part of the PCB touching the main heatsink? From your photos the left hand side looks very low.
Have you tried grounding the heatsinks?
What material are you using for the transistor insulators?
Do both of your boards behave the same way?
Cheers
Q
Re: Post #2335
Marus,
Thanks for posting the pics of your two well made NMOS350 modules on on their heatsinks. The pics shows me something of concern I should of noticed before. It appears you have at least C7, VR2, C6, C5 and source resistor R30 thermally coupled to the T6/T7/T9/T10 heatsink. With T8 is mounted under the PCB into the heatsink it is also possible there is also heat being radiated from the PCB in the T6/T7/T9/T10 heatsink area onto T8. The heat from the T6/T7/T9/T10 heatsink mounted so close to the PCB as it is with at least the noted parts thermally coupled or in such close proximity of the T6/T7/T9/T10 heatsink is not good both in terms of part lifespan, but also in terms of the throwing off the expected design behaviour of those noted components being operated at a much higher thermal rating and in some cases well above the designed thermal operation of the components.
Also of serious concern it appears in addition to the above noted components thermal coupling concerns the T6/T7/T9/T10 heatsink is heating, if not roasting, the components R19, R21, R22, R2, R14, R17, C8, R13, R16 and R20 that appear to be under the T6/T7/T9/T10 heatsink when using the quasi PCB layout. Many of these components under your T6/T7/T9/T10 heaksink are related to bias, feedback and driver stage of the amplifier. This excessive heating of these components under your T6/T7/T9/T10 heatsink
I would change the heatsink for T6/T7/T9/T10 to be well above the PCB. Then I would return all the parts to the same values as they were originally and see if this will fix the problems you have been having.
Examples of how others mounted their T6/T7/T9/T10 heaksink:
1) quasi
2) TomWaits (See also Post # 592, Post #601, Post #622 and Post #800)
Ok, it is getting late and I have only scanned up to Post #1150 and I need to pack in soon and get some sleep.
Regards,
John L. Males
Willowdale, Ontario
Canada
15 September 2007 (14:35 - 15:40)
15 September 2007 (15:40 - 15:50) Unable to directly link to the pics references of Post #2302 jlm
15 September 2007 (19:00 - 19:20) Still unable to directly link to the pics references of Post #2302 jlm
15 September 2007 (20:50 - 22:00) Found free site to upload pic references of Post #2302 and re-edit all url references of Post #2302 to point to site uploaded pic references to. jlm
(15 September 2007 22:00 -) 16 September 2007 02:50
Official Quasi Thread Researcher
Marus,
Thanks for posting the pics of your two well made NMOS350 modules on on their heatsinks. The pics shows me something of concern I should of noticed before. It appears you have at least C7, VR2, C6, C5 and source resistor R30 thermally coupled to the T6/T7/T9/T10 heatsink. With T8 is mounted under the PCB into the heatsink it is also possible there is also heat being radiated from the PCB in the T6/T7/T9/T10 heatsink area onto T8. The heat from the T6/T7/T9/T10 heatsink mounted so close to the PCB as it is with at least the noted parts thermally coupled or in such close proximity of the T6/T7/T9/T10 heatsink is not good both in terms of part lifespan, but also in terms of the throwing off the expected design behaviour of those noted components being operated at a much higher thermal rating and in some cases well above the designed thermal operation of the components.
Also of serious concern it appears in addition to the above noted components thermal coupling concerns the T6/T7/T9/T10 heatsink is heating, if not roasting, the components R19, R21, R22, R2, R14, R17, C8, R13, R16 and R20 that appear to be under the T6/T7/T9/T10 heatsink when using the quasi PCB layout. Many of these components under your T6/T7/T9/T10 heaksink are related to bias, feedback and driver stage of the amplifier. This excessive heating of these components under your T6/T7/T9/T10 heatsink
I would change the heatsink for T6/T7/T9/T10 to be well above the PCB. Then I would return all the parts to the same values as they were originally and see if this will fix the problems you have been having.
Examples of how others mounted their T6/T7/T9/T10 heaksink:
1) quasi
2) TomWaits (See also Post # 592, Post #601, Post #622 and Post #800)
Ok, it is getting late and I have only scanned up to Post #1150 and I need to pack in soon and get some sleep.
Regards,
John L. Males
Willowdale, Ontario
Canada
15 September 2007 (14:35 - 15:40)
15 September 2007 (15:40 - 15:50) Unable to directly link to the pics references of Post #2302 jlm
15 September 2007 (19:00 - 19:20) Still unable to directly link to the pics references of Post #2302 jlm
15 September 2007 (20:50 - 22:00) Found free site to upload pic references of Post #2302 and re-edit all url references of Post #2302 to point to site uploaded pic references to. jlm
(15 September 2007 22:00 -) 16 September 2007 02:50
Official Quasi Thread Researcher
Re: Post #93
Quasi,
Regarding Post #93 the hv reg nmos amp.pdf. The values for R5 and R6 are not indicated. Do you recall what the R5 and R6 values were?
Regards,
John L. Males
Willowdale, Ontario
Canada
16 September 2007 03:04
Official Quasi Thread Researcher (About to be sleeper)
Quasi,
Regarding Post #93 the hv reg nmos amp.pdf. The values for R5 and R6 are not indicated. Do you recall what the R5 and R6 values were?
Regards,
John L. Males
Willowdale, Ontario
Canada
16 September 2007 03:04
Official Quasi Thread Researcher (About to be sleeper)
Hi John,
I almost forgot about that cct. R5 & R6 can be 10K - 15K and 1 watt.
The cct was developed early in this project when I was going to use split rails for this amp. But the idea never worked properly so this bit got left out.
By the way the hyperlinks are just marvellous.
Cheers.
Quasi
I almost forgot about that cct. R5 & R6 can be 10K - 15K and 1 watt.
The cct was developed early in this project when I was going to use split rails for this amp. But the idea never worked properly so this bit got left out.
By the way the hyperlinks are just marvellous.
Cheers.
Quasi