Balanced IGBT for amplifier output

Status
Not open for further replies.
inertial asked:
Vas: does it means voltage amplification stage?

Yes.

inertial asked:
How is approx the input capacitance of the output stage?


The one that matters is capacitance refering to ground which unlinearly loading VAS as complex impedance, shifts phase of the current and therefore tends the amp to oscilate. In our case are parallel MOS-FET input capacitance plus sum of all parasitic capacitances from electronic parts to ground. Exact value you can only get with measurement of the populated board mounted in the case.


inertial asked:

If I put two 100K in parallel ( or two 82K) the current bias go down, am I right? So more B class than A class, right?

It will still be A class, because VAS bias current will still be enormously higer than MOS-FET input current. Your VAS will be in perfect A class.


inertial asked:

Is this scheme near 180W/8ohm ?

No. Output stage has +/-50V, minus few volts losses on Vsat, minus reserve to clip and you get app. 100W/8ohm rms.
 
Lazy Cat said:
The one that matters is capacitance refering to ground which unlinearly loading VAS as complex impedance, shifts phase of the current and therefore tends the amp to oscilate. In our case are parallel MOS-FET input capacitance plus sum of all parasitic capacitances from electronic parts to ground. Exact value you can only get with measurement of the populated board mounted in the case.. [/B]

Thanks, Lazy Cat . Thus it is approx the sum of Ciss of N and P plus
something more ?



It will still be A class, because VAS bias current will still be enormously higer than MOS-FET input current. Your VAS will be in perfect A class.. [/B]


Excuse me , my question was referred to the power stage


No. Output stage has +/-50V, minus few volts losses on Vsat, minus reserve to clip and you get app. 100W/8ohm rms. [/B]


I do not understand. Is it not sufficent 28V RMS to make 100W/8ohm?
Why I need 38,9 V RMS ? 😕

Thanks for help me to understand
🙂

Cheers,
Paolo
 
Using an LME49810 input stage ... me too 🙂
 

Attachments

  • bigbt_lme49810a.jpg
    bigbt_lme49810a.jpg
    28.3 KB · Views: 1,136
inertial said:

Thanks, Lazy Cat . Thus it is approx the sum of Ciss of N and P plus
something more ?

Yes, plus all artificial little capacitors from separate electronic parts to ground plane (ground islands on PCB to parts, output transistors to heatsink which is at ground potential, etc). At the end when you install amp in metal case the value can easily reach nano Farrad or more from MOS-FET gates (VAS output point) refering to ground.

inertial said:

Excuse me , my question was referred to the power stage

OK, your VAS output is at A class in any case, but the A or B class rank of the amp output stage is depending on the level of quiescent current through output transistors (biasing). When the level of output (load) current exceed quiescent current of the output transistors, your amp cease to be in A class.

inertial said:

I do not understand. Is it not sufficent 28V RMS to make 100W/8ohm?
Why I need 38,9 V RMS ?

:wave: Listen very carefully I will tell this only once:

We have +/- 50V output stage supply, so theoretically we can squeeze out 100Vpp of output swing and that is exactly 156,25Wrms/8ohm. And that is only theoretically without any losses.

In real conditions you have to consider saturation Uce of the output transistors, app. 2V, U-Re, app. 0.5V, 3% of PSU voltage collapse, app. 1,5V and at least 2V reserve before clipping. So now you have max. 88Vpp real output swing and that is exactly 121Wrms/8 ohm. For me that is 100W/8ohm amplifier :nod:
 
No problem at all, because quiescent current through MOS-FET is stable (sum of parallel Ibe currents from one half: NPN or PNP) the only difference you can measure is dissimilar quiescent currents Ice through each output transistor depending on relevant Hfe. You don't have to find 100% Hfe equivalents but let say in 90-95% range of similarity.

From my measurements and observations on ready made units there is no sign of any unstability or unreliability and I have (also many of my customers) different amps with BIGBT output for more than two years. One very strong 400Wrms/8ohm is my workhorse on daily basis 😉
 
revintage said:
Seems to be a problem as dependending on standing current the PNP and NPN draw different current in relation to each other. Are there really any that can be matched?

Matching is going to be near impossible because you have to match four BJTs with its mate and its compliment in different configurations.

The degeneration resistor does its job and although currents are not identical (far from it), they do add together.

In the mock up Gecko I had to adjust off-set several times before stabilizing at about 5 mV off-set and 63 degrees centigrade on the heat sink.

At cold turn-on the off-set can be as high as 70 mV it stabilizes very rapidly and off-set drops.

Lazy Cat's output stage sounds very different from any other I have ever heard. I am not one to explain sounds with words as it is even more misleading and then debates are never ending.

I have not made any real measurements and only listened in mono.

In the posted circuit I did not decouple the 18 ohm NFB resistor and did so with 1000uF in parallel with 1 uF polycarb.

I am guessing that the different and very appealing sound character could be attributed to the geometric summation of all the individual currents at the output terminal.


Kind regards
 
Just a question on biasing. Lazy Cat in your proposed output devices, should the Mosfets be mounted on the OP heatsink or on its own smaller heatsink?

http://www.diyaudio.com/forums/attachment.php?postid=1690232&stamp=1229728575

If we take Dr.Bora's design linked above, do Q25 and Q26 track, the OP stage heatsink, or the Mosfet driver heatsink (if mounted separately on their own heatsink) or do they track the OP devices by being mounted on top of the device itself?

Similarly on which heatsink are Q18 and Q21 mounted?

Thanks.
 
Just had to play around a little more😀.

With a very simple thermal compensation it seems to behave perfect, slightly negative up to over 150degrees C.

Also changed the input stage to SE as the overtone structure was with to much even overtones.

Even in Class- AB with 200mA Iq/transistor it looks good.

Anyway it sims beautiful, maybe one should have a go.........
 

Attachments

  • bigbtse.png
    bigbtse.png
    85.6 KB · Views: 1,032
Nico Ras said:
... I am guessing that the different and very appealing sound character could be attributed to the geometric summation of all the individual currents at the output terminal.

SOUNDs good 😎

Hi LazyCat, have you tried the LME49810 approach (very easy to do)? ... how did it sound ...
🙂
 
Samuel Jayaraj said:
Just a question on biasing. Lazy Cat in your proposed output devices, should the Mosfets be mounted on the OP heatsink or on its own smaller heatsink?

http://www.diyaudio.com/forums/attachment.php?postid=1690232&stamp=1229728575

If we take Dr.Bora's design linked above, do Q25 and Q26 track, the OP stage heatsink, or the Mosfet driver heatsink (if mounted separately on their own heatsink) or do they track the OP devices by being mounted on top of the device itself?

Similarly on which heatsink are Q18 and Q21 mounted?

Thanks.
Samuel, mount MOSFET-s on the same (main) heatsink together with outputs and Q25&Q26. Also, mount Q21&Q22 on separate small heatsink, Q9 on it's own separate small heatsink as well as Q18 on it's own! Q9, Q18 &Q21 should never be on the same heatsink! In actual unit, it might prove necessary to add small cap in the range of 10pF...33pF between collector of Q21 and base of Q20 (emitter of Q27). R31 is preset-trimer of 1k for BIAS adjustment.
 
KLe said:


SOUNDs good 😎

Hi LazyCat, have you tried the LME49810 approach (very easy to do)? ... how did it sound ...
🙂

Well I have to say that my discrete VAS is certainly more realistic for my ears and long term listening more bearable. LME driver is not bad at all, it gives you more "instant" solution, easy approach to high power, very dynamic, fast and a little bit edgey sound. But certainly better than any D class amp, bacause there is no loss of fine detail and imaging.

I would recommend LME for a good studio monitor driver, strong PA amp and of course home Hi-Fi. This is completely subjective opinion of mine and it depends on what you've heard so far. :hphones:
 
revintage said:
Just had to play around a little more😀.

With a very simple thermal compensation it seems to behave perfect, slightly negative up to over 150degrees C.

Also changed the input stage to SE as the overtone structure was with to much even overtones.

Even in Class- AB with 200mA Iq/transistor it looks good.

Anyway it sims beautiful, maybe one should have a go.........

SE input stage is fine and I like hybrid amps design with output BJTs, especially compact ones as yours.

It can be even build in point to point technicque or on a very nice looking PCB with a little nostalgia look.

I'm a bit anxious about it, can we expect something likely to see in near future? :wiz:
 
Status
Not open for further replies.