Class-AB Midrange Amplifier


This stereo power amplifier was built for the midrange in an active three-way system.  It would be excellent for full-range bookshelf speakers too.  Each channel is rated at about 60W into 8Ω and 100W into 4Ω – plenty for an efficient midrange driver.

It consists predominantly of ESP project PCBs.  The amplifier modules are from a stereo P3A (Rev-B) PCB cut down the middle for discrete installation on left and right heat sinks.  There is a P39 soft start for the 500VA transformer, a P33 speaker protection and muting module and the safety ground loop breaker found here – very useful in a multi-amplifier set-up.

There are plenty of Class-AB power amplifier kits and PCBs out there including highly regarded designs such as “The Leach Amplifier” and famous designs by Pass and others as discussed ad nauseam on the audio forums.  Then there are the pre-assembled modules built like computer motherboards including the much hyped Class-D modules fitted with SMDs of “their” choosing with profit as a primary motive over sound quality and serviceability, let alone longevity.  Not interested in following the crowd and just wanting to build something clean and simple that I could repair myself if necessary, P03 stood out from the crowd by a mile.  The simplicity and long standing of the design with minor revisions over the years rather than radical departures into the unknown were also very attractive.

The divided amplifier modules (less output transistors) – modified slightly to reinstate the power supply bypass capacitor arrangement after the cut:

Mica insulators cut from microwave oven parts to reduce the risk of shorts to the heat sinks – especially from jumper wires associated with the added capacitors:

A finished module mounted to a heat sink with stainless self-tappers in holes carefully tapped using an identical screw and kerosene as a cutting oil (a lot more solid than M3 machine screws in die-tapped holes in aluminium) and spring washers above the PCB to maintain pressure on the thermal paste and mica washers:

Thermal transfer was necessary between two of the transistors so I used a bit of thermal epoxy:

The best parts that were reasonably available were sourced including ON Semiconductor MJL4281A/MJL4302A output transistors and MJE15034/MJE15035 drivers.  Electrolytic caps are Panasonic and as suggested, film caps replaced any ceramics.

Speaker protection module.  White coax at left to the loss-of-AC terminal with its shield to ground:

It was recommended that the protection/muting module be powered by the main power supply, but I preferred to dedicate that to the audio circuitry.  Instead I used the 9V auxiliary transformer for the soft start module to power the protection circuit as well. 🙂 Here is the auxiliary transformer with the other and of the white coax connected to its output:

The soft start module is powered by the same auxiliary transformer.  12V DC is taken off the soft start PCB to the protection module’s DC terminal and no voltage-dropping resistors to the 12V DC muting relays were needed. 🙂

Larger than standard capacitor next to the relays for smoother DC to the protection module with relay timing upped to around ½ a second (still quick enough to activate before the mute relays at start-up).  Another sheet of mica under the PCB for safety as mains voltage is present under the ballast resistors and PCB relays.

The soft start and protection ground returns could not be connected to the chassis as that would have upset the ground loop breaker function.  Here they are connected to the zero volt line of the main power supply instead 🙂

The main power supply transformer is a 500VA 25-0-25V Nuvotem Talema brand for approximately 35V DC rails.  It is mounted inside a heavy steel magnetic absorption canister.

For quiet operation, I tried this method of minimising the transmission of any vibrations to the chassis:

Materials of different resilience to suspend the toroidal in the canister.  The large black rubber disc is relatively hard, the white neoprene rubber is softer and the small black disc immediately adjacent to the threaded clamping plate was cut from a very soft neoprene mouse pad.   I figured it was a bit like double glazing with different glass thicknesses to reduce traffic noise – no sympathetic resonances.  I don’t know if it’s effective but I can say that it is reasonably quiet.

Stereo AB amps can benefit from dual power supplies, but this amp is only for midrange where heavy bass demands on the supply are not envisaged, so a single supply was considered sufficient.  The main power supply filter capacitors are on a PCB intended as part of a Class-A amplifier kit (with 22V rails) from another Australian electronics supplier.  All that needed changing were the voltage-dropping resistors for the on-board LED rail voltage indicators.  The capacitors had an adequate voltage rating of 50V.  The power supply is admittedly a little over the top with 60,000 μF of capacitance, but it was inexpensive and saved a lot of trouble – cutting a copper ground plate etc. and should make for a very low noise amp.

Piggyback quick connectors for the speaker returns to the zero volt line at the PSU output end – just cleared those small caps:

Some constructors complain that their DIY power amps produce humming or buzzing noises in their speakers, so apart from using the steel tank for the main transformer I did anything I could think of to reduce the chances.  For example, where any signal wire had to cross a power cable I spaced them and ensured that they passed close to a right-angle.  Also, the mains AC had to come from the back panel IEC/EMI filter to the front power switch, so I used some Belden shielded mains cable and tied its drain wire to the chassis at the common grounding point under the power supply filter caps:

And to suppress any switching noise from the bridge rectifier I did the usual thing with X2 PP 100nF caps on piggyback lugs:

Testing and calibration:

The power supply was first tested before connection to the amplifier modules.  The soft start and mute modules worked fine and rail voltage measured at ±35V, but there was a slight noise in the auxiliary transformer, so I shut it up with these rubber “engine mounts”:

Then, rather than testing the amplifier modules immediately with the ±35V rails and “safety resistors”, I decided to buy an old dual output bench power supply to slowly ramp up the voltage to the modules with current limited to around 100 mA (with the bias current trimmers turned right down on each module).

Bird woofer ↑.  That 22Kg beast (the bench supply :smile:) is great because not only does it look like it came out of the Time Tunnel complex, but its two supplies can be set up (by adding straps and resistors across the rear barrier strips) as Master with an auto-tracking Slave :shock:.  With these connected in “Auto Series” configuration –  in which a common ground (zero volt line) is provided between them –  the Slave output provides a negative voltage which tracks the magnitude of the positive voltage at the Master supply output.  :mrgreen:

With those wires ↑ connected to each amp module in turn and the voltage slowly increased to 35V, there was no current over-draw (i.e. it made it to 35V within the set 100 mA limit) and each amplifier’s output was near enough to zero.  Great – no shorts or mistakes so far. 🙂  I then installed the modules along with their heat sinks into the chassis, connected them to the permanent supply and followed ESP’s instructions to adjust the quiescent current to 100 mA.  The DC offset at the right channel output measured an incredibly low 6.3  mV (target was 100), and the left measured a miserly 17  mV.  No dramas at all.  🙂

The finished amplifier (and scene stealer again):

The amp is to partner with this Class-A tweeter amp so the Chinese knock-off chassis is identical – with the exception of the ridiculously “borrowed” German brand logo which wouldn’t convince anyone, but I left it in place this time as it looks pretty. 😎 Please excuse the pixilation – I didn’t want to upset the trademark owner.  😛

.. and the back:

That may look like a can of spaghetti electrics, but it is actually very carefully laid out (more or less). 🙂

Listening impressions:

OK I have not listened to it full range, but as a test, I installed it after the active crossover in the upstairs system to power the midrange and tweeters (150 Hz up) and what an astonishing surprise  –  this amplifier sounds seriously convincing – as good as or better than any brand name solid state amplifier that I have heard.  It is more revealing than the earlier Class-A amp, and not near as laid back.  It’s better than the little C-J valve amp that I used to like and with a wonderful sharpness and bite on the music.  I can only imagine how good it will sound on a fully active system!

And after half a day of continuous playing the heatsinks are cold.  Very happy indeed. 🙂

Next project will be a near-repeat of this one, but based on ESP project P3B – a true 25 Watt Class-A amplifier (now here).


Foot Note:  A big thank you to the guys on the ESP forum for pointers along the way while constructing this thing! 😀

6 thoughts on “Class-AB Midrange Amplifier

  1. Once again very nice! I am starting to detect some attention to detail over aesthetics in this one. Pity Rod did not allow for those bypass caps since there is space just above the other one. Hope your brother is happy as he should be.

  2. Thanks Craig. My guess is that the board layout might have been compromised for the standard (non-cut) version as a long jumper wire had to extend right down to “gnd” on one half. Provision for this link would have to be made as it crosses two wide power supply tracks. And really the output transisors run so cool that cutting and mounting on two heatsinks is unnecessary. I only did it as I had that chassis at hand. Thanks for your nice comment. 😀

  3. Dear Ian, you have very taste in laying the project that you build . I consider you as an expert in building Solid State gear . Congratulation once again. Regards Elson .

  4. Thank you Elson! Of course my work cannot compare with your valve amplifiers. You are always very kind to me. 🙂

  5. That’s a German stainless steel vandal proof 250V double pole latching type switch made by Schurter and is available from RS as well as Element14.

Leave a Reply

Your email address will not be published. Required fields are marked *