Hello, I designed a new PCB. It works very well
Source: http://www.videorockola.com/proyect...ificador-monofonico-de-100-watts-version-2-0/
Source: http://www.videorockola.com/proyect...ificador-monofonico-de-100-watts-version-2-0/
Attachments
Last edited:
Looks very similar front end to Hitachi datasheet circuit for lateral mosfets.
Has darlington bipolar output instead.
I would have added a Vbe multiplier instead of 2 diodes.
Has darlington bipolar output instead.
I would have added a Vbe multiplier instead of 2 diodes.
It is similar topology to Hitachi circuit
and very simplified version of early op amp.
On a positive note it is favorite topology of mine
and also love Quasi Outputs as well.
I will assume Q1 , Q2 current levels
are way to low to help stabilize the amplifier.
And Q3, 4 and 5 currents are also way to
low. Simple balanced Vas can just be all matching
devices. Tip42 or assume C suffix or TIP42C
for 100 volt devices. Poor choice
and would need higher current to be linear.
For Q3/4
Current is likely no higher than 2ma
R8 needs parallel capacitor to stabilize the circuit.
Around 6.8 to 10n.
Which is why I assume C6 / C7 1n !! were added to stabilize
the amplifier. If current levels and 2nd gain stage
were correct such massive compensation would not be
needed.
The slew rate above 10k frequency signals
would be rather horrible.
Very great idea, lovely topology
and seemingly very compact nice PCB.
Current levels, Transistor choices and stability compensation
not well done.
Transistor output bias is kept very low class B
This can help Current sharing with quasi output with no
Baxandall Diode. Otherwise with higher bias, diodes for bias
dont bother me as much as others, but need to be placed on main
heatsink for thermal tracking. This amp has high emitter values
and very low bias. So likely not much of a issue.
But for A/B operation thermal track the diodes or use VBE multiplier.
And Quasi output needs baxandall diode.
Circuit will be more linear and without it and normal A/B bias
AC current will all go mostly to one NPN output.
Looking at input resistors/gain resistors and very low current
for all gain stages I assume high DC offset on output and noted
stabilization issues.
Tip 42 of course will get bashed is horrible device.
But if used on 50 volt rails/ 100 volt supply. Should note C suffix
devices which are 100 volt.
and very simplified version of early op amp.
On a positive note it is favorite topology of mine
and also love Quasi Outputs as well.
I will assume Q1 , Q2 current levels
are way to low to help stabilize the amplifier.
And Q3, 4 and 5 currents are also way to
low. Simple balanced Vas can just be all matching
devices. Tip42 or assume C suffix or TIP42C
for 100 volt devices. Poor choice
and would need higher current to be linear.
For Q3/4
Current is likely no higher than 2ma
R8 needs parallel capacitor to stabilize the circuit.
Around 6.8 to 10n.
Which is why I assume C6 / C7 1n !! were added to stabilize
the amplifier. If current levels and 2nd gain stage
were correct such massive compensation would not be
needed.
The slew rate above 10k frequency signals
would be rather horrible.
Very great idea, lovely topology
and seemingly very compact nice PCB.
Current levels, Transistor choices and stability compensation
not well done.
Transistor output bias is kept very low class B
This can help Current sharing with quasi output with no
Baxandall Diode. Otherwise with higher bias, diodes for bias
dont bother me as much as others, but need to be placed on main
heatsink for thermal tracking. This amp has high emitter values
and very low bias. So likely not much of a issue.
But for A/B operation thermal track the diodes or use VBE multiplier.
And Quasi output needs baxandall diode.
Circuit will be more linear and without it and normal A/B bias
AC current will all go mostly to one NPN output.
Looking at input resistors/gain resistors and very low current
for all gain stages I assume high DC offset on output and noted
stabilization issues.
Tip 42 of course will get bashed is horrible device.
But if used on 50 volt rails/ 100 volt supply. Should note C suffix
devices which are 100 volt.
Last edited:
Yeah. I really wish people would STOP using TIP41/2 in there. You can get proper driver transistors, even from China.
@WhiteDragon can you draw a better schematic with appropriate components? It would be great.
Regards
Regards
@NanoFarad
Yes I could.
Would it be for similar rail voltages / power levels of 50 volts.
And also this is BJT output section, OP shows Quasi Complementary output topology.
Would you like Quasi output for similar Balanced Vas frontend / Hitachi
Or should I just make output section Complementary.
Im a fan either way, and would just use higher speed devices and Baxandall diode
for proper Quasi Output.
Maybe could make separate thread, this is amplifier I desire and would hope to
bring in somebody to do actual board.
Yes I could.
Would it be for similar rail voltages / power levels of 50 volts.
And also this is BJT output section, OP shows Quasi Complementary output topology.
Would you like Quasi output for similar Balanced Vas frontend / Hitachi
Or should I just make output section Complementary.
Im a fan either way, and would just use higher speed devices and Baxandall diode
for proper Quasi Output.
Maybe could make separate thread, this is amplifier I desire and would hope to
bring in somebody to do actual board.
The design and topology should remain as original as possible but with quasi complementary(BJT) outputs and with better performance.
+/-30vDC to +/-50vDC is ample for most of the situation i think.
+/-30vDC to +/-50vDC is ample for most of the situation i think.
I can and have made some changes
Stability is tradeoff as always with THD numbers / slew rate etc etc.
I lowered differential degen R20, R15 to 180, was high at 470 before to get such High Ft output section
to be stable. I was trying to keep slew rate high but raised C5 to 22p for not much difference and got THD1
distortion rating. To be where I would expect it to be with this topology. Around .001 to .008 typical
Also had to get current of T7 and T8 to match, otherwise typical Quasi behavior
and you can have stability issue at high frequency, or AC current sharing of output
transistor can be uneven. And with Quasi even if DC offset is pretty as can be.
AC current sharing is not equal in output section, Its just how good are you, and how close
you can get. Any ways by changing R3 and R26 and R17. To different values to balance
current of EF and CFP sections. And this also resulted in more stable high frequency.
Here is new schematic with DC current values shown
And yes Differential is using only 1ma NOT 2ma like normally seen.
And yes it has Degen like not normally seen. Because it is 35 Mhz output
devices and it is stable amplifier.
THD1 (1 kHz) with 1 volt RMS input .004%
With good Quasi , yes strong 2nd Harmonic profile
THD20 ( 20 kHz) with 1 volt RMS input .05%
Second Harmonic still high with typical higher 3rd harmonic at higher frequency
Square Wave at 22 kHz
Sine clipping behavior at 1 kHz
As you see lower Quasi rail clips well as expected with added
Baxandall Diode. Slight Rail stick on positive portion is not Output
Section. That is typical of Hitachi style balanced Vas.
If R7 10k is raised to 12k or 15k then rail stick is reduced/ removed.
But THD and Vas balance will be pure garbage.
No way to change and typical of simplified topology.
You would have to add mirror and current source.
Or about 4 to 6 more transistors. So = Ignore
AC transfer
Stability is tradeoff as always with THD numbers / slew rate etc etc.
I lowered differential degen R20, R15 to 180, was high at 470 before to get such High Ft output section
to be stable. I was trying to keep slew rate high but raised C5 to 22p for not much difference and got THD1
distortion rating. To be where I would expect it to be with this topology. Around .001 to .008 typical
Also had to get current of T7 and T8 to match, otherwise typical Quasi behavior
and you can have stability issue at high frequency, or AC current sharing of output
transistor can be uneven. And with Quasi even if DC offset is pretty as can be.
AC current sharing is not equal in output section, Its just how good are you, and how close
you can get. Any ways by changing R3 and R26 and R17. To different values to balance
current of EF and CFP sections. And this also resulted in more stable high frequency.
Here is new schematic with DC current values shown
And yes Differential is using only 1ma NOT 2ma like normally seen.
And yes it has Degen like not normally seen. Because it is 35 Mhz output
devices and it is stable amplifier.
THD1 (1 kHz) with 1 volt RMS input .004%
With good Quasi , yes strong 2nd Harmonic profile
THD20 ( 20 kHz) with 1 volt RMS input .05%
Second Harmonic still high with typical higher 3rd harmonic at higher frequency
Square Wave at 22 kHz
Sine clipping behavior at 1 kHz
As you see lower Quasi rail clips well as expected with added
Baxandall Diode. Slight Rail stick on positive portion is not Output
Section. That is typical of Hitachi style balanced Vas.
If R7 10k is raised to 12k or 15k then rail stick is reduced/ removed.
But THD and Vas balance will be pure garbage.
No way to change and typical of simplified topology.
You would have to add mirror and current source.
Or about 4 to 6 more transistors. So = Ignore
AC transfer
Last edited:
Greetings, I would like Mr. WhiteDragon if I could simulate something similar with a mosfet N output but keeping it very simple like the original Hitachi and in that same voltage range between 35v and 50v, as simple as this Kenwood
Attachments
Nice bit of analysis of the design. Interesting read.I can and have made some changes
Stability is tradeoff as always with THD numbers / slew rate etc etc.
I wonder about the lack of any SOA protection though. Seems a bit risky that user error may cause failures.
BTW what sim tools are you using? They look interesting.
Last edited:
Simulation / Spice tool used is TINA_TINice bit of analysis of the design. Interesting read.
I wonder about the lack of any SOA protection though. Seems a bit risky that user error may cause failures.
BTW what sim tools are you using? They look interesting.
from Texas Instruments
and Agree , Output inductor could be added
and also " emitter" resistors are already in the " wrong"
place. Because for VI limiter to be added.
They would need to be in position shown. And flyback diodes
added.
Far as thermal, I love 6 packs of output transistors.
This case I just doubled it from 1 output pair to 2 pairs.
For thermal and even 2 ohm loads. Just go 3 pairs.
I dont like to rely on the superiority of modern devices.
Surprised nobody mentioned 350 volt 4281 transistors
being " overkill" for 50 volts rails.
But if you were boiling the outputs at 150c
I believe SOA for 4281 is still 100 volts.
So 3 pairs and VI limiters be good.
For MI ( musical instrument) or live audio
DC protection relay would be mandatory
in my book.
Last edited:
Thanks for the info.Simulation / Spice tool used is TINA_TI
from Texas Instruments
I agree Vceo is not normally a factor in today's devices. It is fT and the the SOA that matters most for audio output devices.
The MJL4281A you used has good SOA and a impressive fT number. Seems a excellent device to use.
I have been using Pspice since it was introduced around 1985. So know Pspice well.
However the Pspice simulation engine has not been updated in recent years to improve Pspice's weaker ability to converge complex circuits.
So now I am looking for alternate SIM software.
Tried Lt-spice but did not find it compelling enough to switch.
I am going to try out Tina as your SIM reports look good.
- Home
- Amplifiers
- Solid State
- Quasi Complementary Amplifier New PCB