Rod Elliot P3A Layout - Critics

Jump to Latest
#121
Ian Finch said:
Put it this way, Tekko - Rod sees a lot of copies of his boards because people think an auto- router in their PC software can do just as good and looks pretty, nice and close together etc. I've made PCBs by hand, artwork and cheap or free software for a lot of years but I can assure you that in hi-end audio and RF work, nothing is as good as someone who knows best spacings and position of parts, track alignments, thermal gradients, EMR patterns and who is designing for A1 performance with the skill and experience for it.

Typically, you can halve THD by smart trackwork in the output transistor traces and feedback routing. You can reduce noise by parts selection and location. Yes, you can learn that there are do's and dont's rules to follow like star grounding paths but do we know how much is necessary for optimum performance of the design? After 10-20 iterations and feedback, you might be getting somewhere. Rod has that experience with his very old designs but I don't think many people see it. It can make that difference between 500 Euro and 5,000 Euro amplifiers and of course, the odds you will ever get paid for any design. There are guys here who are breathtakingly good at commercial work but you see something else when you look inside the best fancy brands.
Click to expand...

For "mission critical devices", I agree that optimal layout for pcb is required.

But for class AB amps, I think the performace gain of very good layout is negligible for normal listening, for me, as long as "other layout's" performance is within -5% or -10% of the measured specs of commercial pcb layout, then it is forgivable...
 
#122
sakis said:
given the chance to talk about it ..... if you willing to design a pcb for a simple device like the P3a and follow all the mentioned rules above, and some others not mentioned in this thread is actually impossible ( trust me on this one been trying for 5 years in my spare time )

you may as well go for a double layer but then you are swimming in unknown waters like proximity where induction or capacitance between traces and layers will be impossible to evaluate properly for people that don't have experience and/or highly sophisticated equipment .

Conclusion is that you have to "keep" the rules but up to a certain point divided by symmetry /looks and all of it underlined with the cost factor ...

conclusion also is that people that exist in the diy often are bad in mathematics ( like me ) so the above equation is almost impossible to solve ....

happy regards
sakis
Click to expand...

Yes, as they said "we do not live in an ideal world", we can't have it all, so even in pcb design, we have to make compromises in the design...
 
#123
I agree with both of you really. The point about P3a in particular, is that it is an old, nasty CFP design from the early days of silicon. These always sound bad when biased at "optimum'" 12mA and it just so happens that with a bit of tweaking, more bias, and care with bias control, this piece of junk sounds like a million dollars. There are commercial amps (e.g. Hitachi) with this high bias scheme too and their owners here still love them so I get to fix a few and play with SQ/bias (fun!).

I am almost retired but I teach some noob kids how these layouts and builds can be done and I know at that level of comparison to Rod's latest board, the differences (apart from parts errors and disasters) is often quite a lot in the SQ. Sure, a lot of guys are just happy to have a cheap, working amp with plenty of power. We can get this anywhere and cheap, too. I like a few designs specially because, despite all the engineering "faults", these do what we would like them to do - just that bit sweeter and more interesting, so we feel great when we listen and all that hi-end talk. I don't judge anybody's work here and I hope your design sounds great when you finish tweaking, jlester.
 
#125
Ian Finch said:
I don't judge anybody's work here and I hope your design sounds great when you finish tweaking, jlester.
Click to expand...

Hi, I already listened to my P3A (the one I posted earlier), and now it's my best sounding amp at home even without tweaks. Thanks 🙂

It beat my current and previous amps using the same speaker: av-735, tda8950, jvc ax-r97, onkyo tx-nr900, LM3875 and LJM L25D

I will try leach amp and techno amp next 🙂
 
Last edited by a moderator:
#126
Here it is 🙂, I replaced the stock amp of JVC AX-R97 with the p3a. Definitely sounds better than the original.

IMG_1572.jpg


PS: the speaker wires are just temporary 🙂 I'll diy another one...
 
#127
Here's my new layout for p3a
Added base stoppers for drivers and output transistors and added VBE CE bypass cap
 

Attachments

  • p3a-6-22.PNG
    p3a-6-22.PNG
    111.5 KB · Views: 842
  • p3a-schem.PNG
    p3a-schem.PNG
    74 KB · Views: 851
Last edited by a moderator:
#128
Your sch is pretty good.
I would add more emitter resistors, LTP, VAS
Limiting resistor in series with VR1.
D.Self resistor between R16 and Q9 collector. As used by quite a few other designers.
R//L on output.
I like your input and your Signal Ground.

C6 might be better with a series R and then maybe duplicated for Q5. Easier to omit than to add.

Layout.
I hate your separated +ve & -ve supplies.
The decoupling is far too far away from the main current consumers. They should be in a tight loop around the Q7&8 emitters.
 
Last edited:
#130
meanman1964 said:
.... close enough same amp different name
Click to expand...
Well. needing stopper resistors, if that is really the case, does not indicate it's close enough at all. Instability is caused by the inductance of long traces, where they should not be. P3a does not need stoppers, so it would be better, whilst you are redrawing, to shorten traces there as much as practical and decouple as AndrewT suggests.
 
#131
jlester87 said:
Here's my new layout for p3a
Added base stoppers for drivers and output transistors and added VBE CE bypass cap
Click to expand...

hello, your design seems very good, now my question is: if you replaced the transistor that adjusts the quiescent current, I have assembled this circuit and the PCB put a transistor "BC546" as in the original scheme. had some problem with the quiescent current and its fit with the transistor "BD139"? thanks and regards.
 
#132
AndrewT said:
Your sch is pretty good.
I would add more emitter resistors, LTP, VAS
Limiting resistor in series with VR1.
D.Self resistor between R16 and Q9 collector. As used by quite a few other designers.
R//L on output.
I like your input and your Signal Ground.

C6 might be better with a series R and then maybe duplicated for Q5. Easier to omit than to add.

Layout.
I hate your separated +ve & -ve supplies.
The decoupling is far too far away from the main current consumers. They should be in a tight loop around the Q7&8 emitters.
Click to expand...

I think I need to expand the pcb size to accommodate your suggestions. Can you please help/teach me calculate the required values if ever? 🙂 For decoupling I saw some pcb designs like baksa and techno and they are all far away from output trannies so I thought it's ok to leave it there..

BTW here's the updated layout and schema... =)
An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.
 
Last edited by a moderator:
#133
luisitoloco22 said:
hello, your design seems very good, now my question is: if you replaced the transistor that adjusts the quiescent current, I have assembled this circuit and the PCB put a transistor "BC546" as in the original scheme. had some problem with the quiescent current and its fit with the transistor "BD139"? thanks and regards.
Click to expand...

I already tried bd139 as VBE multiplier for P3A (see my previous posts) and I didn't encounter any problem... It is also easier to mount on a heatsink altogether with the drivers.
 
#134
jlester87 said:
....... For decoupling I saw some pcb designs like baksa and techno and they are all far away from output trannies so I thought it's ok to leave it there..
Click to expand...
Yes, using those examples would be a good idea except that generally, EF2 designs don't have many instability (oscillation) problems. However, CFP can be a problem with any added inductance and so it is wise to take extra precautions with the layout. The same goes for triples and any high gain output stage.
My suggestion is not to increase the size, but move the drivers much closer to the output transistor connections. Bolt the Vbe multiplier direct (but insulated) to either driver heatsink for simplicity. Of course, a single heatsink for all 3 semis would be ideal. Keep going, I think you are proceeding very well.

So, did you actually find any oscillation on the scope or otherwise?
 
#135
How about placing a wirelink under the pcb so that decoupling can be placed near output trannies? I removed the base stoppers because I assumed it will be stable.
 

Attachments

  • p3a-v2-6-24.PNG
    p3a-v2-6-24.PNG
    94.7 KB · Views: 811
Last edited by a moderator:
#137
Or I will just make decoupling (near output tranny) optional? Will put them if the amp is not stable.

Is there a scenario where the amp is stable in normal condition and unstable in transient ones?

Is redundant decoupling/bypass ok?

Thanks
 

Attachments

  • p3a-v2-6-24.PNG
    p3a-v2-6-24.PNG
    114.1 KB · Views: 781
Last edited by a moderator:
#139
I may have missed on space for a component or 2 but it seems obvious to me from the the last few PCB patterns, that Q5,6 could both be moved up towards the output transistors - about level with the collector of Q4. That would give more compact and tidier output stage tracks and the collector ballast resistors could also now go closer to make room for this.
 
#140
I already tried it that way, is it ok that the output is nearer the input signal?

Thanks for the suggestion, I'm currently revising the previous one...
 

Attachments

  • p3a-6-24-12.PNG
    p3a-6-24-12.PNG
    110.5 KB · Views: 719
Last edited by a moderator:
You must log in or register to reply here.
Top Bottom