The various power rail caps, in a design that has the main reservoir caps followed by RC filter caps for the drivers and IPS/VAS.
As a general comment on circuit layouts, consider the following exercise: Draw a large circle on a piece of paper. This circle will represent the current flowing in a circuit. The circle, or loop, also represents an antenna, for both transmitting and receiving signals. The effectiveness of a loop antenna is directly related to its area, or the area contained within the loop. The area of the circle drawn being its radius squared times PI. Now fold the piece of paper in two roughly along the diameter of the circle, such that the two halves of the circle overlap. To a first order effect, the loop area of the “folded loop” is zero, and the effectiveness of the loop as an antenna is also zero.
The reason ground planes “work” to reduce EMI is that they essentially create folded loops, or more correctly, a loop with a minimal area, by allowing return current to flow directly “under” a given trace. They are not a “magic solution” however that can be applied without thought. Return currents must flow “around” obstructions like component holes and each such diversion adds to the loop area of the signal. If a ground plane has any “magic” it is that the laws of physics are on your side in helping to create the smallest loop area possible for signals and their return paths.
The reason ground planes “work” to reduce EMI is that they essentially create folded loops, or more correctly, a loop with a minimal area, by allowing return current to flow directly “under” a given trace. They are not a “magic solution” however that can be applied without thought. Return currents must flow “around” obstructions like component holes and each such diversion adds to the loop area of the signal. If a ground plane has any “magic” it is that the laws of physics are on your side in helping to create the smallest loop area possible for signals and their return paths.
now we are getting somewhere.
rail caps are on the amplifiers rails - nothing to do with reservoir caps of the power supply; so why are you putting reservoir caps in the mix?
this is why i find your diagram of post 210 confusing and misleading.
i suggest you redo it without reservoir caps, and fully annote it.
Re pcb layout.
To my mind R13 and R14 are too close to Q1 and Q2. for comfort since these resistors have a small level of inductance and varying currents will induce voltages causing radiated fields in parallel with the resistor bodies.
Radiated fields can be reduced by routing the power supply leads and that for speaker out so these are in close parallel with these resistors. This cancellation strategy does does not neccessarily mean making traces under the pcb.
These leads can be routed around the perimeter of the board with gentle curves to navigate around corners with leads secured to anchor points by cable ties.
40 years or more ago I built an EF2 amplifier from an RCA Transistor handbook. I changed a few things in this such as replacing the bootstrap with a CCS and using 30MHz output transistors. I left out the output stability components - these were not needed for my JLH Class A. I auditioned this in our lounge and this sounded good and remained that way until I turned the Television on to watch the news. The Post Office was responsible for dealing with interference so I was quick to flick the switch off. This is still a problem today we have a holiday house at a coastal spot where there is a holiday camp which brings people from neighbouring cities in summer, school holidays to visit the beach. Our house has an external Tv aerial.. This is unwatchable at times if cars or lawnmowers without spark plug suppression are in use.
You might like to think about the work of Thiele and Boucherot on output stability networks.
Well, i've spent a few hours on this now.
I've attempted to do a 'sideways' style design like the wolverine which would theoretically allow me to do the separated ground layout style grounding of the wolverine. However I am not sure that the juice is worth the squeeze.
The trace from the drivers to the outputs goes from 16mm on my previous design to something like 60mm.
The feedback trace goes from 15mm to something like 40mm.
That is before figuring out how to get everything neatly where it needs to go.
So it seems to me that I am best to take a few steps backwards. If the grounding layout on V3.0 is poor then I am best to remove the extra supply rail decoupling and redo the supply rails to come from one side only so that the grounding can be better laid out like on Rod's board.
The extra decoupling was probably a step away the goal of a pcb with the extra stability components and better capacitor options. No point making it harder for myself than it needs to be.
It has certainly been a good lesson though.
The one's I have are 5mm spacing, quite fat though so I have used a wider footprint so allow space.
This is almost identical to Rod's board.
The learning experience of doing a proper layout is certainly time well spent. The major current loops are now much smaller and that is very positive. The drivers could be placed closer to the outputs to reduce both trace resistance and inductance. Another layout "rule of thumb" is that traces should be routed both "to and from" filter caps. A single trace "to a cap" just adds to their equivalent series resistance and inductance reducing their effectiveness. Layouts usually end up being a series of compromises. The more you understand the complexity of the task the more one appreciates the amount of work it takes to create an excellent board.
Stability testing my new Rod Elliot P3A
The title of thread.Are you shure that ...Eliot new pcb design isn't off topic?
P3A is what is BUT it is a stable amplifier as soon as parts that used are the recommended,like any other amplifier.
No one can stop you from designing your own printed circuit. This is why Elliott is showing the circuit diagram to the public. In this case, however, the thread does not have to have a title stability test of P3A.
This clearly implies that there is a stability problem
Is it?
Isn't amplifier stable WHEN recommended parts installed?
If isn't we must recover.
IMO.
Very true. Is is too easy to accuse the design in case of build failure and propose useless workarounds such as pcb rework.
It's funny how quickly these things escalate.
Some idiot builds an amp using a bunch of non-standard parts he read some other idiot somewhere on the internet say 'sound better'.
Amp blows up and here we are: idiot trying to fix the design that wasn't followed properly in the first place...
FWIW, I do have some db139/140 on order to do the design as specified.
Ill post the results when they arrive.
The same comment applies to the cermet trimmers I would move them away to avoid the perils mentioned in my recent post on radiated fields and wire would resistor inductance.
The whole layout is to fit the heatsink which restricts the pcb dimensions.
This may be ideal in such as high power amplifier where the power supply takes up a large space inside a standard size case. The heat sink in this photo looks outsized for the power level for Class AB working.
The way these wire wound resistors are arranged I don't see any simple route to run supply and output leads in parallel. That would be possible if these were to be oriented at 90 degrees to where they are now i.e across the short length of the pcb.In that way there will be a straight line for the bunch of output and power supply leads to kill two birds with one stone.
Last edited:
No more words,a complete test.https://www.diyaudio.com/community/threads/p3a-comparison-table-long.164756/page-19
Silicon Chip magazine published a "Blameless" style amplifier in December 2021. The BD139 and BD140 were listed as alternatives to the primary driver transistor choices - this with part numbers for Jaycar and Altronics provided the power transformer was limited to 25 V-0-V25 secondaries. With regard to those resistors I mentioned you could stack three 1 Watt carbon film resistors to get 0.333 Ohm non inductive.
I don't remember seeing any CFP amplifier magazine projects in recent times however I kept some pages from one in Silicon Chip magazine from August 1998. This is a Class A 15W item. The image for this is attached showing the form of pcb layout for low distortion. There is a circuit diagram and a pcb copper pattern. I can scan the pages if you are interested.
The driver transistors in this image are BC327 and BC337 and the outputs MJL21193 and MJL21194.
The lead out pattern for the BC types is different to BD139 and BD140 and they have lower Vc specs. I don't see these as an option in the P3A. If BD139 and BD140 are to be used there would need to be a modification to the pcb. There are higher rated transistors with the same lead out pattern as these BC's but these are only a little bit better than BC639 and BC640.
The driver transistors in this image are BC327 and BC337 and the outputs MJL21193 and MJL21194.
The lead out pattern for the BC types is different to BD139 and BD140 and they have lower Vc specs. I don't see these as an option in the P3A. If BD139 and BD140 are to be used there would need to be a modification to the pcb. There are higher rated transistors with the same lead out pattern as these BC's but these are only a little bit better than BC639 and BC640.