12W Broskie-Macaulay Amplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi guys, I'm a long time lurker here and I can't even remember if I have any posts, but I'm registered for more than 2 years now and was I lurking even before that.
I wanted to share one of my DIY projects. I hope you will like it and that you will not noob-bash me.:D

First of all, a little about my system. It is a normal Windows 7 PC with Realtek ALC889a running in default format of 24bit, 192kHz (because stupid Windows resample everything anyways)

Yes, the soundcard is on-board, but I don't know if there's better on-board soundcard and I'm just a poor EE student so I can't afford something like Xonar STX. Here's the picture of soundcard RMAA test, something is not quite right because SNR should be just a bit over -100dB, it's around -93 here; could be that this stupid 50Hz harmonic is from power line interference or even more stupid 40kHz harmonic but whatever. It's really good for on-board soundcard... I used loopback cable to line-in for measurement.

An externally hosted image should be here but it was not working when we last tested it.


I have got me a nice set of DIY buschhorn speakers with Visaton FRS8 8ohm from a friend when he made himself a better pair of speakers (I think he made clangton cheap trick something with 20cm visaton speaker and tweeter... can't remember the name)

Buschhorn.jpg


well, I had one of his tube amps but had to return it so I decided to make me a new amp, this was the status of my desk when I was building it. Notice the tube amp, 220V Weller soldering iron plugged into a cheap Lötstation from old crap soldering iron and 1kg of solder:

Mess.jpg


Well, as sometimes I like to read JB TubeCad, I was browsing for something nice, cheap and solid-state and I found this:

Broskie-Macaulay Amplifier

It is John Broskie "modification" of Macaulay class AB amplifier that won “Low-dissipation class-A amplifier” award. He called it Broskie-Macaulay Amplifier and made it use expensive mosfets and huge bias current.
As I can't afford those mosfets and didn't have a huge power supply capable of such high idle currents, I made some of my own modifications that suit my needs (you know what they say - Engineering is the art of making what you want from things you can get.)
I made it use idle current of around 350mA and TIP132 transistors with power supply of 19.25V (18V Zener plus 1.25V LM317) and really cheap NE5532P OpAmps (I plan to swap them for LME49860 when I buy them on ebay).
I had some toroid that was around 150VA and it had +18/0/-18 AC windings so it was perfect for that.
Well, my modified schematic is this one:

Amp.png


For this amp, I made power supply like this: is toroid, old TV rectifier and CRC filter with 1000uF capacitors, 10W 1ohm resistors and 4700uF capacitors. After that, there are three LM317 and 3 LM337 (one pair for each channel and one pair for the OpAmps)
This is the picture of toroid only in old PC PSU casing, there is rectifier and first capacitors in the casing.

Toroid.jpg


After that, 3 wires come out to the amplifier.
As 19.25V @ 350mA per channel can get hot (13.5W at idle, 1W of pure class A operation to 8ohm speakers :) ), I drilled some holes in the copper of old GeForce8800 graphics card cooler and put some thermal pads and paste for transistors.
As voltage regulators and resistors from CRC filter also must put up with 700mA combined and cut DC voltage from 25.5V to 19.25V, they also "spend" around 4.5W so it's around 18W combined for everything. I've cut some aluminium to press on nicely thermal-greased regulators (with pads) and resistors, here's the picture of that (cellphone quality)

Transistors_Only.jpg


Here is the picture of the same thing, but with (C)RC filter for power supply, 5W 1ohm resistors connected to transistors and all voltage regulators with their zeners, resistors and capacitors. Now only OpAmp circuitry is missing. (sorry for cellphone quality again)

NoOpAmps.jpg


And after attaching opamp ciruitry on top of everything. Finally a nice photo; you can even see one lonely LM317 on the left, he's making this old graphics card fan spin at very slow rate (current regulated, I think it's something like 50mA). The whole thing can get quite hot if the fan isn't running.

Amplifier.jpg


Geforce.jpg


You can see it's on top of power supply and there are speakers and a cable with 3.5mm stereo jack directly soldered on it until I buy some nice banana plugs and RCA connectors AND put everything in a nice box.

Closeup of OpAmp board:

Closeup.jpg


As you can see, the "board" was improvised on the fly so I don't have the board picture; above the board you can see zeners and upside-down capacitors for regulators. The regulators themselves are below the aluminium. All wires that have some current flowing through them are fat copper wires from PC PSU. Weak signal wires are thin copper wires from CAT5 cable.

Picture of 3.5mm connector with nice cable plugged in my PC:

Cable.jpg


Before buying CMK222 cable, I had quite a lot of noise, but this cable is AMAZING!
Now for some final RMAA measurements:

Broskie-Macaulay.png


You can see it's almost as the soundcard itself, could be inside the tolerance range because the soundcard measurement is more than a month older than amp measurement so maybe all settings weren't the same and we have a stupid 40kHz harmonic again, but whatever...:D

Only "big" difference I can see is third and fourth harmonic:

Broskie-Macaulay_THD.png


As third harmonic is the "loudest" and it's at -93dB, I highly doubt that anyone can hear it. Second harmonic is the same as fourth, around -105dB.

When I buy a better soundcard and put LME49860 OpAmps instead of NE5532P, this thing will have like no THD.
I think it's already quite nice with THD of 0.0034% and 0.01% IMD. Who's got less than that? :)

This was meassured by a loopback through 1/10 resistor attenuator with speakers connected so there was probably some speaker interference. It was also quite loud, near maximum power so it probably went deep in to AB mode.

Sound - AMAZING.
I can't say that I heard a lot different stuff on the rest of my system, but I had one 3W tube amp and I used friends Marantz model 140 for a while. This one I like the most because it has the most bass and it sounds really natural. It has really good control of the speaker and every instrument can be heard. When I put this in a nice casing and some nice RCA and banana plugs (because it's currently soldered directly to all my cables), I'll take it to a friend of mine that has much better system than me and he can be much better judge than me because I can be very subjective when stuff that I made is on the table.:D

Oh, and why is it only 2x 12W when with 19.25V you could put 46W to 8 ohm?
It's because those voltage regulators are limited to around 1.5 Amp current so it starts clipping at that point. I plan maybe to change those regulators to something stronger but I never need even half of this power so there is no point; horns are really loud and when I throw away those Visaton speakers and put some Fostex speakers in, 12W will be overkill.

Why not use 12V instead rails then? Well, it's because OpAmps can't swing to voltage rail, it starts clipping at 1V or even 2V before hitting rail voltage so It's nice to have some headroom.
 
Quasi N darlington! Who doesn't need that??!

Could you explain that a little? I don't know what that it so I don't really understand what you're saying. Is it good or was it sarcasm?

If it was good, then thank you very much.:)

If it was sarcasm then you probably saw 2$ darlingtons and thought: "what a cheap ****" or: "oh, another darlington schematic, whatever..."
Well, you were wrong if you thought that. I must say that transistors used in this schematic have close to 0 effect on sound quality.
Also, OpAmps and resistors can stay exactly the same no matter what you use in the output stage. You can use any darlington and you can also use mosfets with absolutely no modifications to the schematic. Changing output stage components will have minimum (if any) impact on sound quality because everything is done by OpAmps and any output stage non linearity is automatically corrected by OpAmps.

If you want better sound quality and less THD, you just put in better OpAmps.
If you put in a "popular" OpAmp like LME49720 (or any other OpAmp with THD of 0.00003%), whole amplifier shouldn't have much more THD than OpAmp itself ( rough estimation is that it probably wouldn't have more than 0.00006% for 0.00003% OpAmps) and it should sound just like this OpAmp sounds.

Money also matters. That's the point of DIY, isn't it? Well, this one cost me 40$ plus parts I have salvaged plus connectors when I buy them. Even if it was built with no salvaged parts, it would still cost me around 70$ plus box, heatsinks and connectors. How much does commercial amp with THD like this cost? I would be also interested in other DIY amplifiers with THD as low as this one.

There is an error in my "heat" calculation. The 18W I calculated is only for positive rail. When you add the negative, you get 36W.
If you want less heat, you just drop idle current to 100mA and it will be 10W or you can drop it to 50mA and it will be 5W. I wouldn't go lower than 50mA though.
 
Your further explanation explained why this is very useful. Remember how the Quasi N channel was considered useful. I have been trying to find a good quasi-N-darlington amplifier. Yes, I have read the original article, but they were no darlington.

Darlington amplifiers are usually either in-stable or not good sounding. This one looks different. I have plenty of small size n-darlington and opamps that I will use for active speakers with your circuit. Thanks again!
 
Yes, it's a really nice amplifier and it only gets better with good opamps. Maybe I made a mistake when I wrote a lot about how I did it and not very much what I did so I'll try to do that now.
You will need an input capacitor that I didn't put on schematic and you will need something like 100k resistors after that capacitor to positive and negative rail so you don't get DC offset at output.
You could rewire it to inverting mode to avoid those resistors, but capacitor is probably mandatory.
This is "updated" schematic with idle current measurement when R16 is 12k.

Amp2.png


If you want full class AB operation, you can put smaller R16 (closer to 10k like other resistors next to it).
DON'T make it too small because if you by accident (or resistor tolerances) make R16/R17 ratio smaller than R13/R14 ratio, you'll probably burn a transistor and possibly a loudspeaker if you connect it to try out your amplifier.

And there are some simulations if you're interested, there is small power class A simulation, class AB simulation and AB crossover distortion enlarged, it is a really small distortion but if you don't like it you can put 18k in place of R16 resistor and have 1.4A idle current. It will get VERY hot and spend even more electricity. :)
Green line is current through loudspeaker
Red line is top transistor and blue line is bottom transistor.
While in class A operation, neither transistor completely shuts down.
When bottom transistor is swinging hard, top transistor swings less but it never shuts down completely.
When bottom transistor current drops to zero, amplifier exits the class A operation and bottom transistor shuts down while top transistor starts swinging harder to compensate.
And the best part in all of this is that transistors don't do all that by themselves, opamps make them do it.

Push-pull class A operation (it is asymmetrical, but it is push pull):
Class_A.png


Class AB operation:
Class_AB.png


For all kids that don't understand why would anyone want a pure class A amplifier when it can be mistaken for a radiator from central heating system or something like that, you can take a look at this close-up picture of crossover distortion (when transition from class A to class B is happening):
AB_Crossover.png


Length of the distortion is around 1us and it makes it 1MHz so you can't hear it and your speaker will most certainly ignore it. Also I don't think that it looks quite this big and obvious in real system. With better opamps (faster slew rate and better CMRR) it should be invisible. This is a simulation of 1$ opamps so you cant expect them to work as good as 500% to 1000% more expensive ones.:)

You can safely put other opamps in it, but read datasheets for their safe operating area and maximum ratings so you don't put 18V opamp in 20V position.
If you have any comments or questions, I would be happy to hear them.
 
I would edit, but I can't or don't know how.
This came to my mind somehow - you don't put 100k resistors to top and bottom rail because if one of the rails start collapsing, you will get DC offset. You should put one resistor to ground and that's it. You should also use corresponding capacitor, we don't want to build a highpass filter (33k resistor and 10uF capacitor should be fine)!
It is how I did it in my amp and I don't even know why I did it the other way now, lol.
It was probably because I was working on another opamp circuit yesterday and it used only positive rail so two resistor version got stuck in my head, sorry about that.:)
And as I said before, you can make an inverting circuit for input opamp (like the one in JB tubecad) so you don't need that resistor. I don't have any practical experience on experimenting with this stuff so I don't know what version sounds better.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.