Simple Symetrical Amplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
120w / 4ohms ???
More voltage needed.
Oops.... sorry, i meant the wrong thing. Its about 80-100W for 4ohm (last time was planning on 22V)

It just depends on the resistance of the wires.

But your project to split your amp in several board is affraying. 1/2cm more on the tracks of the power FETS of mine transform-it in a 150w HF oscillator !
Remember that we are on Mhzs and wires (or track length) are inductances.
By example, non symmetry on the output lines and specially if the point where you take your CR and output connector is not in the middle can increase the distortion by a factor of 10 !

I was meaning several board, I know its "afraying" to some extent. But I don't quite what you are explaining. May you tell me the "risk" i'm possessing by doing so, and what precaution I had to made to prevent/minimise that.
(If about precision of circuits, i'm using ruler to draw out. (I'm not printing those circuit ^^)

The reason i'm doing separate board is mainly for the design/layout of my amplifier module. If it is too impractical, then I would change it. (hope to make my amplifier slim and high up, therefore stacking three boards ^^
Hope there are other technique for doing so without having much risk.
 
The reason i'm doing separate board is mainly for the design/layout of my amplifier module. If it is too impractical, then I would change it. (hope to make my amplifier slim and high up, therefore stacking three boards ^^
Did-you mean 3 boards stacked to simulate 3 layers ? If yes, i suppose that working a little on your Layout, and you can use a single one, or at least a double-sided one.
Just take care of the size of the power (and hp) tracks, and reserve-you the ability to add tin on those tracks to minimize their resistance and avoid their destruction in case of short circuit.
As i said, with wires between boards inside the CR loop components is a risk of oscillation that can be very difficult to compensate, and if you can, reduction in bandwidth and slew rate.
 
I would support having transformer, diodes, and initial smoothing cap on a remote board.

But having explored the "several board" scenario for a recent design my conclusion was that better layouts could be realised by having one board for the actual amp circuit with components on both sides.
 
I would support having transformer, diodes, and initial smoothing cap on a remote board.
Yes. It Seems obvious to separate power AC parsitics and electromagnetic fields from the signal parts. And it is better for maintenance too.
Some remarks based on my linear supply experience.

1- A simple cap multiplier stage ("emitter follower" with a voltage reference (zener) in the base)
http://www.esperado.fr/images/stories/SSA-Crescendo/supply.gif
was always the best sounding solution, on all my testings.
Ten time better than a simple rectifier with the same total amount of caps. Keeping the caps equal before and after the regulation stage.
Stabilized PSUs with feed back are not so good sounding.

2- Contrary as audiophile prefers, a single supply for the both channels is far the best. First, when there is a signal on only one channel in the stereophonic signal, it will benefit of the total filtering capability.
Second, when there is big transients, the power voltage will decrease for the both chanels in the same time the same way. It stabilize stereophonic image.

That's said, each amp need additional caps near his rails connectors to compensate the wiring and decrease it impedance. We all know the way to avoid HF paralleling a non electrolytic cap in each critical stage in the amp itself. We can add a little more filtering with electrolytic caps on each board.
Once your main power supply is done, just try several values on each amp board, comparing the two same rails . (compare and amplify the difference with an OP amp), listening to real dynamic music at hight level. The bast value is the one witch gives-you the less signal difference. You will find a point where there is no enhancement increasing the cap value. More than that, you will find the value increased if the cap have is too large, as we approach the separate filtering situation.

Just my two cents.
 
Last edited:
Drawings !

I think the first understanding of yours is correct (following are different). So i drawn a illustration diagram with "mighty Paint".

There are rail-board interconnection and signal-board interconnection separately. (in my idea)
 

Attachments

  • illustration.jpg
    illustration.jpg
    51.2 KB · Views: 442
Did-you mean 3 boards stacked to simulate 3 layers ? If yes, i suppose that working a little on your Layout, and you can use a single one, or at least a double-sided one.
Just take care of the size of the power (and hp) tracks, and reserve-you the ability to add tin on those tracks to minimize their resistance and avoid their destruction in case of short circuit.
As i said, with wires between boards inside the CR loop components is a risk of oscillation that can be very difficult to compensate, and if you can, reduction in bandwidth and slew rate.
I have the idea of tinning (for me actually was bold-layer of solders) for whole 28V rails and output. (whenever the current exceed 1A)
~What is CR anyway ? ~

Yes. It Seems obvious to separate power AC parsitics and electromagnetic fields from the signal parts. And it is better for maintenance too.
Some remarks based on my linear supply experience.

1- A simple cap multiplier stage ("emitter follower" with a voltage reference (zener) in the base)
http://www.esperado.fr/images/stories/SSA-Crescendo/supply.gif
was always the best sounding solution, on all my testings.
Ten time better than a simple rectifier with the same total amount of caps. Keeping the caps equal before and after the regulation stage.
Stabilized PSUs with feed back are not so good sounding.

2- Contrary as audiophile prefers, a single supply for the both channels is far the best. First, when there is a signal on only one channel in the stereophonic signal, it will benefit of the total filtering capability.
Second, when there is big transients, the power voltage will decrease for the both chanels in the same time the same way. It stabilize stereophonic image.

That's said, each amp need additional caps near his rails connectors to compensate the wiring and decrease it impedance. We all know the way to avoid HF paralleling a non electrolytic cap in each critical stage in the amp itself. We can add a little more filtering with electrolytic caps on each board.
Once your main power supply is done, just try several values on each amp board, comparing the two same rails . (compare and amplify the difference with an OP amp), listening to real dynamic music at hight level. The bast value is the one witch gives-you the less signal difference. You will find a point where there is no enhancement increasing the cap value. More than that, you will find the value increased if the cap have is too large, as we approach the separate filtering situation.

Just my two cents.
well... for me, the diagram is complicated, and too much components for my design (need to add one more board), and this separates it from the OPS too much (through connections).

I am planning from initial to have one transformer, one rectifier, and 2 sides of reservoir (stereo). Only reservoir are separated from each other with stereo configuration.
Each rail on board have its own reservoir, so should be no problem. (as I illustrated in picture. 8mF for positive 28V, 2.2mF for postive 20V, 1mF for postive 15V)
I will also add some film caps (0.1uF-1uF) at each rail between rail reservoir and components.
 
Last edited:
XD...... NFB or Global Feedback is more familiar ^^
Yeah... about that... I never thought of (keep thinking about other factors only..)
It needs to travel long way from 1st board to 3rd board, with long path....

If I don't reduce the bandwidth & slew rate, what could I change ? (at design factor also)

The below attached is my design concept. Everything is below the heatsink, and if the toroid is messing with my circuits, I would eliminate it by putting a copper clad board (un-printed circuit board). Would this be helping ? (I guess it should be grounded ?)
 

Attachments

  • illustration 2.jpg
    illustration 2.jpg
    119.1 KB · Views: 432
Last edited:
Esperado, you mention about the oscillation for high slewrate, so I should make the IPS & VAS as compact as possible ? (what about OPS ?)
I think i'm changing my 3-board design into 1-board, if there are too much problem. Is it possible to make 2-board of positive half and negative half ? (but its quite difficult to design and built)

Is it unwise to attach transformer near main heatsink ? (my heatsink is anodised surface aluminum)
 
Stacking boards is not very desirable when thinking maintenance / tweaking.
Not very practicle.
Hm..... really ? in my opinion, multiple board (not stacking actually) in the form of rack is better for maintenance, as you can take out particular board that you want to mess/tweak with only. (of course one piece is better, but multiple in my design doesn't hard)
anyway, its practical ^^ but of course you have some points of maintenance.
 
but with this design.....you do run into ****...It's very very broadband and unless you really compromise it you'll for sure run into oscillation issues..best way is to make a compact tight PCB with short narrow traces.. inductance in the stacking connectors paired with capacitance in your stacked PCB traces is a sure cocktail for trouble..
There is also some thermal and mechanical reasons not to go there, one concern on stacked PCB's is thermal induced mechanical stresses that ultimately leads to solder failure.
The circuit is really quite simple so why not keep it on one board...??
 
Well... my idea is that common ways of amplifier is making is wide, big but not so tall. (alot of ground area used)
My idea is to reduce ground area used by making it higher with 'racking'. So it will be tall, slim & narrow. I think I had to give up though.... way too much trouble i get into ^^
 
inductance is also in the distance...with thick traces you need more component spacing...The ohmic part tends to dampen oscillations...where pure LC components will ring as hell...when doing boards for RF..this was the design guidelines...GND is vital as all is reference here so good GND routing is vital...All is compromises and making good PCBs is a bit like voodooisms black magic...
 
Hi, guitar89
You can simulate any track or wire by a serial resistance, a serial inductance and a parallel cap, proportional to their length. Any single wire, any lengh of printed circuit track will introduce phase turns and attenuation at HF.
Each amp stage, input, VAS, driver, power, is a pole. any wire between those poles will modify their phase and response curve. All those poles are included inside the global loop, and Mr. Nyquist had explained how it behaves concerning stability.
This amp is a very fast one, we are dealing with phases at 3 to 10 Mhz.
More than that, each wire can act as an antenna, and make your amp more sensible to HF electromagnetic fields.
Your multi stack project seems a bad idea, according to those problems, because it increase the length of the wires between each stage where you need to shorter them.
And, as said bobodioulasso, not very easy to tune the value of a resistance or adding a little cap on a board stacked between two others on a working amp during the tweaking period.
Last, as previously said, each connector is a risk for viability.

So, i you want best performance, and less stability issues, stay with a single board for the amp itself, keeping the total surface and length of the critical connections as short as possible.

No problem to fix a transformer on an aluminum Heat sink, as aluminum is a-magnetic.
But keep your transformer far enough from the low level sensible parts amp circuitry to get rid of electromagnetic leakages from the transformer; both for the 50/60 Hz and the hf parasitics that outlet carries. And remember that a coper shielding will do near nothing against those electromagnetic shield. Only special mu metal can help.

The design of a printed board layout is a very complex work, and need a lot of experience, inspire yourself from the ones proposed by "L.C." or "alex mm" at the beginning of this thread.

Just my two cents.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.