diyAB Amp - The "Honey Badger"

I know but for me for some reason it just doesent feel right, though it might be a habit from designing switching stuff where any stray inductance was a liability to the operation of the circuit.

Thin long traces makes good inductors at higher frequencies.

Most current will return from the speaker right to the faston/centertap. The decoupling returns will just pass the ripple back to the center star. I agree , this ripple - current might get high at ungodly power levels.

I won't defend this PCB style as the "best". If I had 125mm heatsinks , I would
layout in Alex's style. My subamp - luxman/DBT is this style. My altec chassis HAS to use a 32mm aluminum L bracket , so my personal amps have to be long/thin. Therefore ,I would have no use for Alex's board unless someone gave me a $300 chassis/ trafo "gift".

The long thin 75mm X 250mm PCB can go on a L bracket or flat. One could even mount it at a 90 degree angle flat without an aluminum bracket. I was given a set of drawings to adhere to , with a end objective to make it as universal as possible.



If wanted , I'll make the other style , whatever is wanted.

OS
 
OS,

I think both pcb's will work just fine. Sure, there are engineering reasons to argue for one over another and in the end each engineer picks the compromises that he/she likes best. I prefer asymmetric layout (and can't resist a good arguement!) but I think you should stick with what you are confident will work well and you've made more pcb's than most of this forum :D

Speaker return currents are an interesting thing. I guess it has to go through the caps rather than the power trafo since the trafo is only 'available' when the rectifier diodes are forward conducting. So the point where the most current flows in the ground will be between the speaker gnd binding posts and the gnd side of the rail caps. Keeping these current paths separate for the two channels is important for channel-channel cross talk control. It's also why I used the speaker gnd binding posts (or close to) as my start earth in the amps I built so far - everything with high current is then all made with big wire except the short pcb traces that take the power rails to the output transistors and the output to the speaker binding posts. After that, the rest is all small beer.
 
Last edited:
Fatter traces....

I fixed it..... 16.7A returns from the caps. 25A Imax for the rails.

2 workarounds ,

-one could also solder wires from star to cap if 16.7A's are not enough :D

- one could only use 1,000uf local decoupling caps in addition to a large external power supply. My first "Mongrel" had 1Kuf@100V with 10mm lead spacing - they will fit these universal capacitor pads.

I am "juggling" the 2 worlds of some who would use this board stand-alone (without external capacitance), and the FANATICS that would hook it to 50,000uf of external capacitance. Hopefully the fatter traces will bridge the gap.

Another factor is that Mark wanted me to keep the burning amp form factor so DIY'ers could swap "burning amps" for real power if they desired.... with minimal rework of the existing DIYA chassis. So , a few compromises were there right from the start.

I must thank all the forum , even DX , for the many examples of the best DIYA amps. Cordell's and Self's books are the "old and new testaments" of amp design even they are quite " interpretable" :D I think this project will hit the performance of the original "blameless" or even exceed it. Faster devices (self uses MPSA's for VAS , tiny MJE's for the drivers), and a much larger output stage with 3 pairs = lower output stage THD. We can't forget TMC ! , Self's present kit has high 5Khz + thd compared to this one.

I think this will have to do (final draft of PCB)

OS
 

Attachments

  • fat trace.GIF
    fat trace.GIF
    69.8 KB · Views: 2,411
I am not picking....really!!

In fact , you might avoid problems like in the apex thread. :D

Make the board bigger ... 200-225mm
Make it square - not really essential , but any total diy'er might not be into cutting (like that).
Make it to accept to247 , not just to-3p (more OP spacing - like the apex sr200 :p ) ... some may like to throw anything they have into this.


Now to the "fine print"

Thermal -
Q1-2/ input pair ... essential, "face to face" thermal compound and heatshrink.

Q3-4/ cascode ... no gain here , not essential

Q5-6/ current mirror ... a little gain here, preferred "face to face" , but not absolutely essential.

Q7-8/2Q- CCS... best to be tightly coupled as the coefficients cancel out to keep current vs. temp. constant. A slightly negative coefficient overall.

Q9-12 - same heatsink for all ,(or at least some heatsink). Since I now run at roughly 9ma with the 68R emitter resistor (r27) , I am glad for the 30-40 sq.CM heatsink I have spanned across Q10,11/12. They get pretty hot without one.

Q13 is under (beneath) where a small heatsink should go.

Drivers and outputs are too close together... "hotspot" on main heatsink , not ideal.

I did not finish the BOM yet , some resistors need to be bigger (r27 and 23 -1/2w).

You still did not get the ground right ... look very closely at mine. The schema is the same (no component changes) my last board post is final.

I'm just curious , Alex ... do you do layout from mind like me , or do you have some fancy schema-to PCB / auto routing that you use ??? :confused:

PS- next, I will post the real final overview and a workable JPG image to be toner transferred , the design can be done on single sided FR-4 ( connect XYZ) ... WE NEED A VOLUNTEER !! ( I am moving and ran out of material) :(
OS
 
Prototype...

(attached)

BOM , schematic , the overview , a 300dpi (tested on my printer)PCB image , and a list of input devices.

PCB should be 100%. BOM , schema are still in draft form - (not pretty) ... but correct.

PS- single sided , just join the X's , Y's , and Z's (and the OP NFB)with thin wire under board. All J's geta standard jumper. When this is double sided - NO jumpers at all (the board store units).
OS
 

Attachments

  • DIYA-amp.zip
    395 KB · Views: 731
  • final-DIYA-ax-overview.GIF
    final-DIYA-ax-overview.GIF
    67.9 KB · Views: 2,270
Last edited:
Not just simulations.

This is a working amp (X 14) ... (below is "ol' faithful") Silent/humless , no thump , LOUD as hell , very reliable.

Would not pass anything out to the forum that did not pass the "test". :)
Above Artwork WILL work.
OS
 

Attachments

  • AX.jpg
    AX.jpg
    206 KB · Views: 2,461
I was quite pleased when I found them too. At the time they were really cheap as well £1.70 for one and £2.50 for the other. Now they've almost doubled in price, makes me wish I'd bought a few more!

I still see them for 3-4 USD at mouser/digikey , I will add them to the device listing which will be integrated with the BOM.

In the finished BOM will be the Caps , resistors ,diodes , and a wide selection of semi's ..... most I have actually tried in this amp.

A full self- contained LT spice simulation "package" will be available (below- attached) It is fine tuned to show pretty accurately what we are dealing with (nothing - THD wise, below). Make sure "DIYAamp.txt" is in the same folder as "DIYA-amp-LT.asc" (models are there).

- A "Options text" - This will explain all the mods that can be done.
1. - CMC or TMC , what to omit and what to jumper - what values work.

2. - FET or BJT - How to adjust the Cascode/ current mirror to use Jfets at input.

3 - Option for lead compensation and other component changes with a explanation of the advantages/disadvantages.

4 - Advice for single sided Toner transfer construction , this board is backwards compatible for FULL DIY. I'm sure the Board Store will be reasonable, so this might just be for prototyping.

I "Branded" the PCB as requested by Mark , I want to REALLY fine tune and debug the layout for 0 errors , even down to the aesthetics of the PCB.. before I submit Gerbers (No GOLDMUND mess-ups).

An article on this amp/topology would be forthcoming (Help :eek: ... Mark , I'm no editor). Any final comments or suggestions for documentation or parts selection is welcome.

OS
 

Attachments

  • DIYA_classAB_schematic.jpg
    DIYA_classAB_schematic.jpg
    246.4 KB · Views: 2,382
  • excellent.gif
    excellent.gif
    25.4 KB · Views: 2,171
  • DIYA-amp-LT.zip
    15.7 KB · Views: 446
  • final-DIYA-ax-overview.GIF
    final-DIYA-ax-overview.GIF
    68.3 KB · Views: 2,093
Last edited:
MJ is to-3 (this one needs to-247/to-3p) , 21193/4 is too low Hfe for a EF2. My triple (DBT-luxman) is the beast for the to-3p/to-247 21193/4's.

LIST of outputs ....

For the full 150w
NJW0281/0302
MJL4281/4302
2SA1943/2SC5200



Sanken to-3p's (below 1) quite a selection ... any below would work.
150w (marginally)


Drivers -
MJE15030/31 or 32/33 or 34/35 - optimum ON- semi
2sa1837/2sc4793 - best for 8R use , EF2 would have the most total Hfe .

I've tried most of these devices , so I should document all these choices and include them in the book.


OS


What I have are the MJL flat packs, which I think are designated TO-264. Though I believe that I also have some in my stash the 3281/1302 and 5200/1943 pairs.