Can you help me understand this circuit (P27 PA)?

[talha]

Hi,

I've been trying to get into discrete amplifiers. I know the electronics and I can derive the output as function of inputs etc. But it's only doable on smaller sections of amplifiers so, I suppose working on an amplifier requires a perspective, being able to see all the blocks at first sight. Right know I'm looking into somewhat complex designs and try to separate them into smaller building blocks in my mind. Here's Mr. Elliotts P27 power amp, I built it in LTSpice and played with simulation. Can someone help me understand this circuit? I have a couple of guesses but would like to hear from you instead.

Thanks in advance

Edit: I currently am reading the book published here, teemuk's, but working on something real is always more rewarding in my experience.

[Enzo]

Hi, I think you have to ask a question before we can answer it.  What do you want to know about the circuit? I mean teemu's book describes the amplifiers.  So your question as it sits pretty much asks what is in the book.

Q7,8 are the differential pair, Q12 is the voltage amplifier.  Q14 is the bias circuit, and everything to the right of that is the current amplifier to drive the load.

[talha]

Sorry I wasn't clear about what I was looking for. Though, your reply helped, I can already identify long tailed pair. What do Q5 and Q6 do for ex? At first I thought they were current mirror but it doesn't really resemble a current mirror, it's wired differently and its on the emmitter side of the LTP.

Also I observed a huge difference in the behavior of the amp when I connect R8 to ground or leave it floating (same thing just changes R7s value), both of which common practices it seems, from what I've seen. Also, R27s value affects the output drastically. Ex. if I wire R8 directly to ground and not to spk- net, the output voltage stays more or less constant with the variable load impedence. I guess this is one of the changes Mr. Elliott made from early version of his amp, but I've yet to understand why and how this improves the circuit. Can you clarify whats going on with the spk- net please?

Edit: I was suspecting that P3A and P27B, both would output contant voltage, much like when I disconnected the R8 from SPK- net, now that I built P3A in LTSpicee I can confirm that (Good thing is my simulation matches real measurements perfectly). So maybe this load dependence thing is just a way to simulate valve amplifiers? Even if so, if I recall correctly those should be constant current drives, which again our circuit doesn't behave like. Does this circuit use some kind of Mixed Mode Feedback? as decribed here: http://sound.whsites.net/project56.htm

Thanks

[Enzo]

R27 is in series with your speaker, and is used to sample current.  This and R8 are a current feedback.  I am sure that is in the book.  If you increase it, you first are adding resistance in series with the load, and also, you are increasing the sample voltage feeding back.  of course it affects the sound.

Q5/6 are a current source for the diffy pair.  Emitter side?  Sure, it is the side common to the pair.

Hopefully others will join us, I am not the best person to lay this out.

[talha]

Thank you, as I said, I am currently reading the book, but identifying everything is not trivial to me. Also, Rod Elliott modified this circuit so that it uses current feedback. I know it's common in guitar amplifiers and I think it has to to with emulating valve amplifier output impedance. But hopefully someone can clarify the issue and lay out the exact decision making process.

[phatt]

So maybe this load dependence thing is just a way to simulate valve amplifiers? Even if so, if I recall correctly those should be constant current drives, which again our circuit doesn't behave like. Does this circuit use some kind of Mixed Mode Feedback? as decribed here: http://sound.whsites.net/project56.htm
Thanks

Yes it implements both Voltage and Current FB or mixed mode FB
If you remove R8 you then you have to lower the value of R7 for same/similar output level.

Yes it kind of,, sorta,, emulates a Valve output stage but on it's own it will likely be very disappointing. (Well it was for me)
You need a Screen grid and an output transformer and a very poorly balanced long tail pair (phase split)
Then simulate some form of sag and natural compression. Not easy to do with basic SS power stages.
Power supply of SS is often quite stiff but someone may know how to do fancy tricks with that,,, but here is a dead simple way,,,,,

Years back I permanently inserted a 60 Watt lamp in the mains supply of a very basic SS no name 20Watt combo that had the most basic discrete power stage and it went a long way to emulating a small valve amplifier sagging and compressing. you do loose a fair bit of wattage in the process but hey if tone is the aim then you put up with the power loss. Far more convincing than Mixed mode tricks 8|
Wish I never sold that little amp

Teemu might see this and hopefully add a more professional answer. 8|

Meantime keep reading ESP pages,, if you are like me you have to read it a few times before it sinks in. lol.
I'd just build his most basic circuit Here; http://sound.whsites.net/project03.htm
Scroll down to *figure 1a* and learn to understand the simple versions before you try to get your head around all the other tricks. I've tested that Amp and it certainly does work just don't run it on higher voltages.

Hope it helps,, Phil.

[talha]

Phil, that was a very detailed answer, thanks. To be honest I would use tubes rather than going into so much trouble of emulating them, especially if I'm going to need an output transformer (which are hard to find and expensive here) anyways.

As for the light bulb trick, I believe any resistor with appropriate power rating should cause voltage sag and limit the current. Does dependence of bulb resistance to temperature play a big role in your application?

[phatt]

Unlike a resistor whose resistance is fixed over a broad temperature range but an incandescent filament has low resistance when cold and rapidly goes high resistance when heated. (about 10~15 times the cold resistance)

This idea is well known and used to trouble shoot circuits.
Read the lamp current limiter here for clues
http://www.ssguitar.com/index.php?topic=2093.0

Look for the post by *Roly* on page 4 where he gives the resistance hot and cold for a few common lamps.

IME which is only basic electronics I've found that some of the more simple *Discrete* poweramp circuits actually sound quite good when driven hard. The more complex circuits tend to be far more rigid and refuse to distort until they clip hard which is not guitar friendly. xP

I may have given you a wrong impression about simulation tricks;
When I said;
"You need a Screen grid and an output transformer and a very poorly balanced long tail pair (phase split)
Then simulate some form of sag and natural compression. Not easy to do with basic SS power stages."

It was tongue in cheek 

I was just referring to the difference between Valves and Transistors. A transistor has 3 elements while a power valve has 4 (Screen Grid) and in general there is no transistor equivalent. Also transformers that where used for guitar amplification often had limited bandwidth while SS power stages tend to be much wider.
The common PI (phase inversion) used in a lot of Valve circuits are not well balanced and that adds a lot to the breakup of Rock guitar sound. Most of the classic Marshall distortion happens in the PI.

That does not mean a SS rig cannot reproduce a great sound you just have to approach the while design stage from a different perspective. Some folks become fanatical about getting exactly the same effect but for me it's just getting a few things to happen that get close to the same feel and dynamics and in a mix few could tell the difference. 8|
You can do a lot to the power amp stage in SS circuits (Read Teemu's book) but they are high current circuits and you can blow stuff up  in the blink of an eye if you don't fully know what to tweak.
Guess how I know that,,  :lmao:

Much easier for the novice to work the magic with preamp level circuits and let the high current run clean power.
I use a very simple SS amplifier and all the tricks are done with a pedal board which have a few circuits,, some I have designed over the years and some just bog standard units.
Cost wise this is by far the easiest way to go as you don't rely on the Amplifier so much. In fact I can use other amplifiers if the need arises and still have close to the same sound/dynamics and feel. <3)

For me I'm a Musician First,, so I don't wish to waste the rest of my life learning how to reinvent the wheel. I'll let the experts refine that stuff. 8|

Phil.

[talha]

Actually I sort of feel I'm getting ahead of myself on this. I'm going to build this amp first and see what I can tweak. I also happened to find a board house that allows multiple designs on the same pcb as long as they're separated by only silkscreen. So I guess I'll fit a simple LM3886 amp next to the preamp of P27. Honestly, I think they should sound pretty much the same as far as clean tones go, but I'm rather excited to hear the difference in distorted sound. Is the pa of P27 one of those "simple" and well behaving ss amps? I guess we'll figure it out. This may take a long time, both 'cause of I'm busy and it takes ages to ship stuff here inexpensively. I also couldn't find any sound tests of this particular amp and I'm planning recording one, as soon as I get my hands on those boards.

[phatt]

I'm not which way you are heading but it would be folly to try and tweak a *Discrete* SS power circuit without proto boarding  first.

There is more scope with discrete poweramp circuits but you have to learn a lot first otherwise you will smoke a lot of parts, with LM3886 you are limited to one circuit only but still some cool tricks can be done.

ESP P27 pcb is well laid out and created by someone with a deep understanding of the teck,, so just placing a few parts on a Cad and hitting auto route is likely going to end up as land fill and a complete waste of time and money. (you have been warned,, Hint)

The preamp is not so hard but still if you are new to *Audio PCB* design then there are traps. 8|

I've said it many times,,,, the fastest way to make land fill is print a PCB without first proto boarding the circuit to find out if the darn thing actually does what the Dude tube geek said it would do.
You will learn so much more if you take the time to actually test your ideas,,,,,,, and that time can be quite long.

I've found it's about as long as the short cut but at the end you DO actually understand what you have built and knowledgeable enough to fix any issues that may arise. :tu:

Here is a shot of some of my test layouts.

The power amp is actually based on the ESP power amp P03  http://sound.whsites.net/project03.htm
many different ideas were tested but it did not have enough power for what I wanted. It took about 6 months until I settled on a bigger power stage,, long story maybe another day.
The battery one is a little tone preamp I designed.

So yeah guys ,, Breadboard Everything or pay the price.
Or in the words of Clint Eastwood,,, Are you feeling lucky punk? :trouble