Member
Joined 2009
Paid Member
Blameless is always around....i was reading a book
And the guy suggested a low distortion unit..... the blameless naturally.
Everybody does it....because it is the best naturally.
Self may be a little bit sad.... no modern schematic could beat it..including Self made schematics (the ones i could see naturally).
regards,
Carlos
And the guy suggested a low distortion unit..... the blameless naturally.
Everybody does it....because it is the best naturally.
Self may be a little bit sad.... no modern schematic could beat it..including Self made schematics (the ones i could see naturally).
regards,
Carlos
Over the past week I've assembled a second card, this time using 3 pairs of MLL4302/4281 outputs on 56V rails. The Vdrop across the driver base resistors is around the same at 0.9mV and 2.8mV. Halving R35 increased the Vdrop as expected however proportionally similar at approx 1.5mV and 5.8mV.
So I'm at a loss now... both amp modules sound great and the DC offset at the output speaker terminal is very low.
=AndrewT;3880196]Vr18 is typed wrong.
I can see the imbalance in the output stages and in the LTP.
I suggest two changes.
1.)
Increase the driver current massively. Try doubling by paralleling another 100r across R35. This will make the drivers faster and may require a small change to the existing compensation. If that works and the drivers are still cool enough, then try adding a third 100r for a tripling of driver current.
Note that the output devices are also showing a big difference in hFE. The sinking and sourcing into the driver circuit is not equal.
2.)
Add a trimming resistor across R6. Try 1k0 and remeasure the difference in LTP collector voltages. Estimate what trimmer value will give an exact balance of the LTP emitter currents.
The reason for the imbalance is that Q9 is not sinking the same current as the Q4 cb link and not selecting Q1, 2, 3, 4 for equal Vbe and nearly equal hFE at the operational current and temperature.
Finally,
Check that all the Grounds are connected to the ground trace to the right hand side of the R37 network. No exception, I mean ALL grounds.[/QUOTE]
So I'm at a loss now... both amp modules sound great and the DC offset at the output speaker terminal is very low.
=AndrewT;3880196]Vr18 is typed wrong.
I can see the imbalance in the output stages and in the LTP.
I suggest two changes.
1.)
Increase the driver current massively. Try doubling by paralleling another 100r across R35. This will make the drivers faster and may require a small change to the existing compensation. If that works and the drivers are still cool enough, then try adding a third 100r for a tripling of driver current.
Note that the output devices are also showing a big difference in hFE. The sinking and sourcing into the driver circuit is not equal.
2.)
Add a trimming resistor across R6. Try 1k0 and remeasure the difference in LTP collector voltages. Estimate what trimmer value will give an exact balance of the LTP emitter currents.
The reason for the imbalance is that Q9 is not sinking the same current as the Q4 cb link and not selecting Q1, 2, 3, 4 for equal Vbe and nearly equal hFE at the operational current and temperature.
Finally,
Check that all the Grounds are connected to the ground trace to the right hand side of the R37 network. No exception, I mean ALL grounds.[/QUOTE]
I was hoping that the imbalance in driver hFE was mainly due to low driver Ic.
But your higher Ic is proving that the hFE is still imbalanced.
The effect this is having is that the upper driver is sinking much more current from the CCS/VAS string than the lower driver is sourcing.
This leads to a drop in VAS current even though the CCS is sourcing a constant current.
As the driver Ic increases, this sinking/sourcing imbalance will get worse. It could reach the point that the VAS is starved of current when high output currents are demanded by the speaker.
Can you remeasure Vr18 to see how close the VAS current is to the 5.53mA passing through the CCS? It should be ~ 5.33mA
Can you measure all the output emitter resistor and base stopper resistor Vdrops.
But your higher Ic is proving that the hFE is still imbalanced.
The effect this is having is that the upper driver is sinking much more current from the CCS/VAS string than the lower driver is sourcing.
This leads to a drop in VAS current even though the CCS is sourcing a constant current.
As the driver Ic increases, this sinking/sourcing imbalance will get worse. It could reach the point that the VAS is starved of current when high output currents are demanded by the speaker.
Can you remeasure Vr18 to see how close the VAS current is to the 5.53mA passing through the CCS? It should be ~ 5.33mA
Can you measure all the output emitter resistor and base stopper resistor Vdrops.
Last edited:
Rail diodes are not only to create a virtual second supply to input circuit
It has not the "voltage stabilizer" function we had in early amplifiers...it is there to make the amplifier safe, as we are DIYers...we are not professional and some guys that assemble amplifiers are not skilled related practice..then they can invert and damage the input stages or Vas.... not a good idea to remove....this was not a good idea.
Diodes applied also not a good idea...if you are nearby a radio station then you will detect there Radio frequency signals..it is not safe, unless you apply a capacitor draining RF to ground in such circuit.
It is all variations around the same theme.... and can result positive or not....observe the performance listening and also observe the harmonic distribution, the harmonic ratio.... this is important as you can make it sound very bad if make mistakes and increase to much the second harmonic (too much body in the sound) or if you increase too much third harmonic compared to the second...watch this...this is very important.
You will see how it is easy to ruin a design not checking, after each modification, it's effect in the harmonic distribution.
I am glad you already have interest in this amplifier...it is really very good..and sound very good...but all depends of the "tuning" you make.... increase the compensation capacitor to 150pf or change the degenerative and you will see the "hell on earth" effect.
The increase of driver current was tried...had not resulted better....resulted the same..the difference was more heat into the driver transistors forcing me to install them into the main heatsink..... and i did that....but.... i perceived latter on, was a stupid mistake i did.... this amplifier can become very hot and people does not buy or install really huge heatsinks (exception exists ... Terry is one of them)... so...amplifier becomes hot..and this overheats the drivers...and this harm a lot the thermal stability.
Many things are design solutions guys... to calculate...other are dinamic stuff connected to real world operation and torture.... only believe in your solutions IF..you test in the real world and watch and feel the consequences.
Using 56pf or 68pf the sound became beautiful in the high end...BUT...sometimes we could see instabilities in the waveform... something intermittent..comes and go..... with 82pf was more reasonable... 100pf is safe...but everything has a compromise...you tune here and generates danger there.
Build your modifications...do not believe in simulator and calculations only...go to the real thing...and observe what happens....watch the scope and distortion meters all the time.... check fourrier step by step.
The most important...check your beliefs and do not let our own pride to fool us...one channel standard and the other your tweak..then compare and change speakers to compare once again...change audio cables and compare once again... check it..never believe.....belief is not science.... i feel is not science.....goes to the best test you can make..use ears..listen!
About the guys that goes making suggestion, the majority really wants to help..but some of them have not made these modifications in their homes...they just "believe" this may be good...and is asking you to the the "slave" to test for them their beliefs....these ones that suggest, ask them to show pictures of their job of modification..ask them to show their amplifiers...no way...the do nothing crew is lazy..wants YOU to work for them to test THEIR ideas... they do not even show their faces to us....do not show amplifiers...they may be a computer or something alike.
https://www.youtube.com/watch?v=jVNG41DqXPI
regards,
Carlos
It has not the "voltage stabilizer" function we had in early amplifiers...it is there to make the amplifier safe, as we are DIYers...we are not professional and some guys that assemble amplifiers are not skilled related practice..then they can invert and damage the input stages or Vas.... not a good idea to remove....this was not a good idea.
Diodes applied also not a good idea...if you are nearby a radio station then you will detect there Radio frequency signals..it is not safe, unless you apply a capacitor draining RF to ground in such circuit.
It is all variations around the same theme.... and can result positive or not....observe the performance listening and also observe the harmonic distribution, the harmonic ratio.... this is important as you can make it sound very bad if make mistakes and increase to much the second harmonic (too much body in the sound) or if you increase too much third harmonic compared to the second...watch this...this is very important.
You will see how it is easy to ruin a design not checking, after each modification, it's effect in the harmonic distribution.
I am glad you already have interest in this amplifier...it is really very good..and sound very good...but all depends of the "tuning" you make.... increase the compensation capacitor to 150pf or change the degenerative and you will see the "hell on earth" effect.
The increase of driver current was tried...had not resulted better....resulted the same..the difference was more heat into the driver transistors forcing me to install them into the main heatsink..... and i did that....but.... i perceived latter on, was a stupid mistake i did.... this amplifier can become very hot and people does not buy or install really huge heatsinks (exception exists ... Terry is one of them)... so...amplifier becomes hot..and this overheats the drivers...and this harm a lot the thermal stability.
Many things are design solutions guys... to calculate...other are dinamic stuff connected to real world operation and torture.... only believe in your solutions IF..you test in the real world and watch and feel the consequences.
Using 56pf or 68pf the sound became beautiful in the high end...BUT...sometimes we could see instabilities in the waveform... something intermittent..comes and go..... with 82pf was more reasonable... 100pf is safe...but everything has a compromise...you tune here and generates danger there.
Build your modifications...do not believe in simulator and calculations only...go to the real thing...and observe what happens....watch the scope and distortion meters all the time.... check fourrier step by step.
The most important...check your beliefs and do not let our own pride to fool us...one channel standard and the other your tweak..then compare and change speakers to compare once again...change audio cables and compare once again... check it..never believe.....belief is not science.... i feel is not science.....goes to the best test you can make..use ears..listen!
About the guys that goes making suggestion, the majority really wants to help..but some of them have not made these modifications in their homes...they just "believe" this may be good...and is asking you to the the "slave" to test for them their beliefs....these ones that suggest, ask them to show pictures of their job of modification..ask them to show their amplifiers...no way...the do nothing crew is lazy..wants YOU to work for them to test THEIR ideas... they do not even show their faces to us....do not show amplifiers...they may be a computer or something alike.
https://www.youtube.com/watch?v=jVNG41DqXPI
regards,
Carlos
Attachments
Last edited:
That indicates that VAS Ie is down to 2.06mA !
Is the post78 pic still wrong?
Where has all the 5.53mA gone?
Your stated measurements show just 0.2mA of imbalance in the sinking and sourcing to the drivers.
Attached is the voltage overlay with this new board with 3pr outputs on 56V rails and 50R between the driver emitters. R18 dissipation is extremely high!
Attachments
Hello Carlos, thanks for dropping by.
I haven't installed the rail diodes because I wanted to keep this as simple as possible.
At the moment I've got the drivers mounted on a separate heatsink I made from aluminium roofing flashing. But since I have a very large main heatsink - not as big as Terry's but close - I'm going to mount them to that instead.
I haven't used a simulator - I'm working with real boards, changing parts and taking measurements - also listening to the sound on a speaker.
I'll take some pictures later today and post them in this thread.
I haven't installed the rail diodes because I wanted to keep this as simple as possible.
At the moment I've got the drivers mounted on a separate heatsink I made from aluminium roofing flashing. But since I have a very large main heatsink - not as big as Terry's but close - I'm going to mount them to that instead.
I haven't used a simulator - I'm working with real boards, changing parts and taking measurements - also listening to the sound on a speaker.
I'll take some pictures later today and post them in this thread.
R18 dissipation is 0.661^2 / 330 ~ 0.0013W = 1.3mW
Did you really mean R18?
Driver dissipation ~ 1.57W. If it stays cool with the existing flashing sink, then leave it as is.
If it is too warm, increase the sink dissipation.
If you opt to increase the driver bias current to >=50mA, then move them to the main sink.
The output bias voltage has been increased, Vre=48mV.
R10 current now 5.7mA
R12 current now 5.84mA
base current difference to drivers = 0.17mA
CCS current - base current diff = 5.67
The "lost" current is now down below measurement tolerances.
Reset output bias voltage and then
Did you really mean R18?
Driver dissipation ~ 1.57W. If it stays cool with the existing flashing sink, then leave it as is.
If it is too warm, increase the sink dissipation.
If you opt to increase the driver bias current to >=50mA, then move them to the main sink.
The output bias voltage has been increased, Vre=48mV.
R10 current now 5.7mA
R12 current now 5.84mA
base current difference to drivers = 0.17mA
CCS current - base current diff = 5.67
The "lost" current is now down below measurement tolerances.
Reset output bias voltage and then
Last edited:
The gain of the outputs is part of the cause of the unbalanced driver base currents.
If the upper drivers and the upper outputs both have a lower hFE then the current the driver sinks from the CCS will be higher than what the lower Driver puts back in.
In the extreme this can starve the VAS transistor, which is designed as a constant current amplifier.
5% and maybe even 10% difference in hFE is tolerable, but it looks like some of your devices have an hFE that is more than 100% greater than complements that supposedly "nearly match".
If the upper drivers and the upper outputs both have a lower hFE then the current the driver sinks from the CCS will be higher than what the lower Driver puts back in.
In the extreme this can starve the VAS transistor, which is designed as a constant current amplifier.
5% and maybe even 10% difference in hFE is tolerable, but it looks like some of your devices have an hFE that is more than 100% greater than complements that supposedly "nearly match".
This is just simulation so there is no difficulty with simply playing with the transistor models and the number of pairs in circuit to see what occurs and identify the problem.
Presumably, you have the option of using Cordell models instead, if every model in your library proves to have a similar inherent disparity in complementary properties.
Presumably, you have the option of using Cordell models instead, if every model in your library proves to have a similar inherent disparity in complementary properties.
Hi Ian
This is not simulation, I've built and measured two boards!
Today I replaced the mje15030/31 pairs in one with mje15032/33 hoping for some improvement, but alas the difference was minimal.
I'm curious to know whether this problem I'm facing is normal or the exception. I tried increasing the output bias from 26mV to 50mV and 100mV but the proportionally the differences remained similar.
This is not simulation, I've built and measured two boards!
Today I replaced the mje15030/31 pairs in one with mje15032/33 hoping for some improvement, but alas the difference was minimal.
I'm curious to know whether this problem I'm facing is normal or the exception. I tried increasing the output bias from 26mV to 50mV and 100mV but the proportionally the differences remained similar.
Last edited:
Apologies for not having taken my own advice and read all recent posts more carefully.
A couple of things come to mind and they have been mentioned already. The resistances around the VAS are unusually small but current low - something is amiss here. I would be looking to establish 6-12 mA bias in the VAS. You have 5.7 which is OK but the current should be gushing here with Re of only 10R.
R18 should not be getting hot, your voltage drop gives 2 mA, much as Andrew T replied. What is going on with measurements if it is hot? Is it really 330R? Are R17, Q9 functional?
I thought to tack on a link to one of D. Self's own genuine blameless commercial designs for a TAG-Mclaren 250x2r as a reference. Have a look at some key value selections. It is also a 3pr, output stage but these are MT200 size Sanken LAPTs. Be aware that rail voltages are higher at 65V and some parts of the circuit are for protection as VI limiters (Q20,21) and VAS transistor protection (Q16). (8MB PDF)
tag mclaren 250x2r power amplifier schematic Download page :: Schematics Unlimited.
Last edited:
more..
Specifically the value for R18 may be too small, as AJT draws attention to it too. Some advice given early in the thread wasn't true of Self's blameless designs, at least. I think your first estimate was closer to the appropriate figure for biasing Q9 . Simulate first, I haven't estimated any particular VAS bias current increase.
Specifically the value for R18 may be too small, as AJT draws attention to it too. Some advice given early in the thread wasn't true of Self's blameless designs, at least. I think your first estimate was closer to the appropriate figure for biasing Q9 . Simulate first, I haven't estimated any particular VAS bias current increase.
Member
Joined 2009
Paid Member
Here you go
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.