AB100 Class AB Power Amplifier

excellent quasi-saturation behavior./QUOTE]

What would "excellent" mean in this case?


First, the much-used KSC3503E. At 15mA Ice, when Vce drops below 5V, Ice starts to shut off. This would be the opposite of excellent quasi-sat behavior in my opinion. I grappled with this recently with a VAS stage that was visibly distorting long before clipping.

Next, the TTC004B. At almost 20mA Ice, the Vce where the transistor starts to shut off is less than 0.5V. This would be excellent quasi sat behavior.
 

Attachments

  • KSC3503E_FamilyOfCurves.jpg
    KSC3503E_FamilyOfCurves.jpg
    45.8 KB · Views: 1,272
  • TTC004B_Family_Of_Curves.jpg
    TTC004B_Family_Of_Curves.jpg
    38.2 KB · Views: 1,267
Official Court Jester
Joined 2003
Paid Member
lemme see ....... 28Vac will give , in ideal world .....39V6dc ........ shave volt or so for losses , counting that Iq is low enough , so no loading down the rail

so , you have 2x37Vdc on disposal

let say that, from these rails, you can squeeze 70Vpp in best case on amp output

divide that with 2 - you're getting Vp

divide Vp with 1.41 - you're getting Vrms

so , we have 24V75rms as result .... and let say that we are using 8R declared load

P=(Urms^2)/Rload = 612.5/8=76W

so , yes , you can , but without calling it "100"
 
lemme see ....... 28Vac will give , in ideal world .....39V6dc ........ shave volt or so for losses , counting that Iq is low enough , so no loading down the rail

so , you have 2x37Vdc on disposal

let say that, from these rails, you can squeeze 70Vpp in best case on amp output

divide that with 2 - you're getting Vp

divide Vp with 1.41 - you're getting Vrms

so , we have 24V75rms as result .... and let say that we are using 8R declared load

P=(Urms^2)/Rload = 612.5/8=76W

so , yes , you can , but without calling it "100"

Thank you very much for all the calculations.
I don't have much technical knowledge, I'm just a practitioner. :)

What kind of transformer is needed (how many volts and amper, center tap vs.) for 100w?
 
Official Court Jester
Joined 2003
Paid Member
sq. root of 800 , multiply with 2 multiply with 1.41 , divide with 2 - that's the half sine

add few volts on that , that's the Vdc rail

add 1 or 2V , divide with 1.41 , that's Vac secondary , one

you need two, of course

you need at least twice VA of xformer , considering channel power , so 200VA

read this: http://www.firstwatt.com/pdf/art_power_supplies.pdf

one funny guy already chew it for ya
 
Last edited:
lemme see ....... 28Vac will give , in ideal world .....39V6dc ........ shave volt or so for losses , counting that Iq is low enough , so no loading down the rail

so , you have 2x37Vdc on disposal

let say that, from these rails, you can squeeze 70Vpp in best case on amp output

divide that with 2 - you're getting Vp

divide Vp with 1.41 - you're getting Vrms

so , we have 24V75rms as result .... and let say that we are using 8R declared load

P=(Urms^2)/Rload = 612.5/8=76W

so , yes , you can , but without calling it "100"

sq. root of 800 , multiply with 2 multiply with 1.41 , divide with 2 - that's the half sine

add few volts on that , that's the Vdc rail

add 1 or 2V , divide with 1.41 , that's Vac secondary , one

you need two, of course

you need at least twice VA of xformer , considering channel power , so 200VA

read this: http://www.firstwatt.com/pdf/art_power_supplies.pdf

one funny guy already chew it for ya

Thanks :)
 
Hi all,

today the boards arrived ... But the customs and DHL wanted to see money and that because of 0.30€ over the exemption limit . :shutup:

So, 10 boards are around 50€ and now a stereo set costs 10€ + shipping to you. If someone jumps out is no problem for me (someone on waiting list). I will calculate the shipping costs at the weekend and contact you.

preview


1. manniraj
2. Cambe
3. HerrErnie
4. yuujin

Please send me via pm your address, thanks!
 
A few days ago I got such a nice "donor" , each side heat-sink to-3 pre-drilled, but only for one pair.
think I could get in there smaller AB100 - let's call it AB 60 as ZM suggested :D.
I will have to paint it white and make a new aluminum or wooden front panel.
 

Attachments

  • DSC00831.jpg
    DSC00831.jpg
    271.3 KB · Views: 1,007
  • DSC00830.jpg
    DSC00830.jpg
    256.2 KB · Views: 932
Member
Joined 2011
Paid Member
C9=0pF is a synonym for DNI. Do Not Insert.

It means that there is a "footprint" for C9 on the PCBoard layout, but your assembly technician is told not to stuff or solder any component in that footprint. The amount of capacitance we want in this position is: zero. No capacitor at all.

One reason you might put this in your design is: for future flexibility. If, God forbid, WhizBang Semiconductor stops making the BL74JSK device, or if they accidentally let their inventory pipeline go empty*, you have a plan. You'll buy the almost-as-good M44D74GN from ErsatzPLC, and compensate for its imperfections by stuffing an 8.2pF capacitor in position C9. This reduces the amplifier bandwidth by 15% but you're still in production. Now instead of ±0.25dB from 2 Hz to 110 kHz, the modified amp is "only" ±0.25dB 2 Hz to 93 kHz. Still quite listenable.

*because your fierce competitor Panasony paid 10% more to buy up every last part in existence

_
 
Last edited:
Member
Joined 2011
Paid Member
There is only one person who knows exactly why Nelson Pass included C9 in his PCB layout, and that person is not me.

I offered a hypothetical scenario including fake names of semiconductor companies, nonexistent part numbers, and completely bogus bandwidth specifications, to illustrate one possible reason why someone MIGHT include C9. It was a hypothetical scenario. A work of fiction. A story from the land of make-believe.

The good news is: you don't need to know why there is a footprint for C9 on the PCB. Just be sure not to stuff and solder anything there, and live a happy life.