PCB routing for audioamps

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
In a similar thread someone wrote that for amps they put the output section on one PCB which is mounted parrallel to the heatsink and the IS and VAS sections are on another boad that plugs into the output board at 90 deg. This keeps the signal fairly short, the small signal sections stay a little cooler and the 90 deg angle reduces undesired interactions.
 
Sounds like computer building: motherboard and plug-in boards.
Makes amp building more compact, like the Halcro shows.

Not meant ironically, Mark Levinson had a modular amplifier once, with special sota connectors.
That would make repairs a breeze.

Maybe time for a Rubick amplifier, pcb's in all planes.
 
sam9 said:
In a similar thread someone wrote that for amps they put the output section on one PCB which is mounted parrallel to the heatsink and the IS and VAS sections are on another boad that plugs into the output board at 90 deg. This keeps the signal fairly short, the small signal sections stay a little cooler and the 90 deg angle reduces undesired interactions.

Good idea...but the connections would stillhave to be soldered in for secure connection...particularly the feedback loop!
 
jacco vermeulen said:
Sounds like computer building: motherboard and plug-in boards.
Makes amp building more compact, like the Halcro shows.

Not meant ironically, Mark Levinson had a modular amplifier once, with special sota connectors.
That would make repairs a breeze.

Maybe time for a Rubick amplifier, pcb's in all planes.

May decide to market an amp. consisting of a base in which gain cells of different designs can be purchased and merely plugged in.....

Excellent for subjectivists methinks...

What d'yall think?
 
May decide to market an amp. consisting of a base in which gain cells of different designs can be purchased and merely plugged in.....

The app notes on the LT1116 (Mosfet auto bias) IC have an example that looks like that. An opamp input section on a mother board then plug in the modules 100W at a time. Obviously the app note is intended to peddle LT products, and I seem to recall the liosted specs didn't seem spectactular, but maybe there's a useful idea in there somewhere.
 
Hi, Sam9,

I experienced what you described. Due to space limitation, I have input RCA and speaker output terminal at the same side of PCB, and the result is that the channel that is closest to the RCA input is oscilating, while the other channel (far from input RCA) is not oscilating. Both has exactly the same schematic.

Is there any way to fix this problem without redrawing the PCB?

How does the output trace influence the input trace? Is it via the fiber-glass material of the PCB or via air (inductance-radiation)?

If it is via fiber-glass material of the PCB, maybe I can put ground track between input and output tracks, but if it is ramping via inductance-radiation(air), this ground middle track won't help?
 
If an amp has audible parasitic noise (like intermittent rat sound), where does this comes from? I'm making self-oscilating classD experiment, it drives me bananas :D

Phase Accurate has warned that in classD, the PCB itself is the most important component.

Do you have guidelines for classD selfoscilating PCB? What are the guidelines for drawing self-oscilating classD PCB?
 
Connecting the input-signal to pcb siuting connectors?

Hello,

I like the possibility to mount the pcb itsself to the heatsink and then mount the heatsink into the chassis, and then doing the wirering.

For the power-connections its no problem to find connectors, but no chance for the input signal.

One day I started to implement normal cinch/RCA jacks like these:

http://www.computerkabelversand.de/pictures/picturesbig/2604_Cinch_Einbaubuchse.jpg

on my PCBs. I know there are special jacks for printed boards, but I dont need 90 degree mount jacks.

Looks like this:

http://i2.photobucket.com/albums/y23/Black_Chicken/LM3886/Lm3886_6.jpg

-------------

On my own pcbs its no problem to do it this way (still not being ideal), but other layouts mostly are desinged to solder the inputs direktly to the board, but IMO thats not very practical.

On Quasis Pcbs I am going to use these:

http://www.reichelt.de/bilder//web/bauelemente/C180/RTM13.jpg

->RLP-Version


What is your way to connect the input?


Input appreciated!
Black
 
SM2GXN said:
Thanks for all replies!

Excellent pcb Jens!
Very symetrical, guess that a double sided board play a big role and that you spent some hours to make the final finish, real nice artwork. Bad thing with double sided board for most hobbyist is that it is a bit tricky to make them. Does your amp get less noisy by using a groundplane like the one on your board?
Good points you've made 1-5 they are already added to my library :)



Symetricality is one of many keys -- and an important one at that and I certainly concur.

It is also very helpful to keep the radiated energy from the power transformer away from the input circuitry. Bypass capacitors as near as possible to the pins as possible and without long legs to ground (elsewise they become a radiator.)

I spoke with one IC power-opamp firm and they said that there was a difference in measured THD% stats when they twisted the power supply cables --

My recommendation -- keep a notebook and date it. Every time you make a change in the circuitry or layout note the changes. (ok. I will admit to being a franco-phile as I use the Fontaine notebooks).
 
Ex-Moderator
Joined 2005
lumanauw said:
Phase Accurate has warned that in classD, the PCB itself is the most important component.

Do you have guidelines for classD selfoscilating PCB? What are the guidelines for drawing self-oscilating classD PCB?

The PCB layout is just as important as the components for class d, especially with higher switching frequencies, voltages, and currents.

One of the keys of class d layout is keeping loop area small wherever larger currents (with high rise times) are flowing (i.e. the traces between the supply caps and the mosfets and between the mosfets and the inductor).

Grounding is also very important (particularly with unbalanced configurations). You definitely don't want those high currents flowing through or near signal grounds. That can introduce problems that can be very difficult to find and diagnose.
 
To move tracks for long distances a ground plane is useful. (Long distance) is relative to frequency.
 

Attachments

  • star.png
    star.png
    8.6 KB · Views: 148
  • transmission line.png
    transmission line.png
    12.9 KB · Views: 144
Last edited:
When it comes to PCB layout, think in-terms of stars and transmission lines. Combining stars and transmission lines is also a way to go. small capacitors between Ground plane and positive and or negative supply eliminate frequency notches for boards with power layers. If used the Ground layer should occupy the largest surface area. The layer carrying the positive or negative supply or both should limit their area to where they need to supply power.

For those tracks that may have a negative interaction with the ground plane via capacitive coupling, route them on ground plane with sufficient clearance.
 

Attachments

  • combined.png
    combined.png
    5.2 KB · Views: 135
Last edited:
true transmission line design is not relevant to audio power amps - the very highest frequency you find in MOSFET amps is <10 MHz loop gain crossover frequency, 1 ft of wiring would ~= 500 MHz wavelength so even the 1/10 wave criteria for starting to worry about transmission line phenomena isn't met

as mentioned earlier routing large, fast changing current paths and their returns toether reduces inductance, mag field coupling to other parts of the amp - but this doesn't happen spontaneously at audio frequencies - the frequencies are too low to form "image" return current paths as in high speed digital ground plane use

audio currents are mostly resistance controlled - you have to explicitly route the returns to minimize radiated field, ground planes are useful in the low level circuits if you have arranged the gnd Hierarchy to keep large, distorted, "dirty" current return paths from flowing through it
 
www.hifisonix.com
Joined 2003
Paid Member
jcx, yes, I agree with these points.

One point to note is that twisting the supply lines and ground together in a class AB amplifers brings little advantage, and in some cases may make matters worse. The supply rails will be conducting big, half wave rectified currents at the signal frequencies and related harmonics. On the amplifer side that is not conducting, with twisted supply lines you can get coupling into the small signal stages that can cause problems. A good trick is to feed you amplifer with a 1KHz signal and look at the output on a spectrum analyzer (sound card is ok). Then move your supply, speaker and ground return lines around, looking for the least contamination.

Syn08 talk about some of the struggles he had wrt power supply hookup and layout in the <1ppm PGP. Check the PGP website out - some good engineering there.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.