Class-A Tweeter Amplifier


This stereo tweeter amplifier is for an active bi- or tri-amped system.  It uses PCBs made to the “Silicon Chip” circuit, but departs from their recommended build in quite a few ways.  It is alleged to be a Class-A design of 20W, but only about 10 – 15 of those are really in Class-A, making it useful as a tweeter amp but not much else!  A much better Class-A amplifier which I built later is here.

A complete kit including everything needed to build an amplifier was available from a local supplier, but I didn’t want the pre-amp/volume control module, the junky steel chassis with its under-sized heatsinks or the standard transformer, so purchased just the linear power supply module and the mirrored left and right amplifier modules and took it from there.

I really liked the PCB layout and especially the wide spacing of the output transistors to spread the hot-spots across the heat sinks.

The modules came in grab bags of mixed quality components including ceramic capacitors intended for the filter sections of the audio circuit and junky electrolytics.  The ON Semiconductor MJL21193/4 output transistors, Toshiba low noise input transistors and ½W metal film resistors etc. were fine, but I substituted 105°C Panasonic electrolytics and polypropylenes, 1W metal film resistors (for a couple of carbon ones) and used closer tolerance Dale 5W driver/output transistor emitter resistors and used mica washers with thermal paste instead of the standard silicone pads etc.

The aluminium chassis is from a Chinese knock-off of an MBL power amp, but the quality is good with well positioned ventilation slots and much bigger heat sinks, and removing the stupid silk screened logo from the front panel was easy enough.

I chose a “Nuvotem Talema” toroidal instead of the standard one.  It has a slightly higher VA rating (at 225) and a slightly lower voltage output (at 15).  The standard Chinese transformer had magnetic flux band.  The Nuvotem doesn’t and the summed quiescent current drawn by the amplifier modules is around 2.25 Amps (compared to just 50 mA or so for a Class A/B design), so the transformer will radiate a fair magnetic field even when the music is not turned up.  The audio circuitry needed  to be shielded from this, so I went one step further and purchased a heavy steel shielding canister for it:

Although a small gap is provided, I added a rubber channel to ensure no possible contact between the peripheral skirt and the chassis floor which would result in a “shorted-turn” via the centre post that passes through the toroidal.

Intended as a super hi-fi amplifier after an active crossover for delicate tweeters wired directly to the outputs without a series protection capacitor, a DC protection circuit was essential.  I didn’t really like the standard protection module as it had an onboard muting relay and a large PCB intended to be installed at one corner and connected to long and asymmetric spreaker wires.  I preferred to use 15 Amp contact-rated Omron muting relays directly alongside the output posts, so used an ESP P33 module designed to have them off-board.  It is easily tuned for tweeters by changing a few capacitors and resistors.  I chose 1 kHz as a reasonable compromise.

The protection/muting module could have been driven by the main power supply, but I preferred to dedicate that to the audio circuitry and use an auxiliary transformer for the switching.  And although I wouldn’t normally bother with a soft start for a small toroidal it can’t hurt and ESP’s P39 uses an auxiliary 9V transformer to power its timing circuit and relay coils.

9V AC converts (on the above soft start module) close enough to 12V DC which is perfect for standard muting relays, so I thought it would be tidy to power the soft start and the protection/muting module with the same 9V auxiliary transformer.

Being of Class-A design, the amplifier’s heatsinks were expected to run at up to about 25-30°C above ambient, but anything over about 75°C or so would be asking for trouble, so I attached a 70°C thermal cut-off switch to each heat sink a fair way off the output transistors.  These are in series on the 9V auxiliary circuit to the soft start and have a shielded lead to the “loss of AC detector” terminal of the protection/mute module.  If either switch goes open (over 70ºC), the mute relays isolate the speakers immediately and the main transformer switches off to power down the whole amplifier. 🙂

Also, because the amplifier is intended for use in a multi-amp set-up where ground loops are generally expected, I included a safety ground loop breaker identical to the one used on the Class-D amp here.

A few minor features:

The 60,000 μF of filter caps had to go on stilts to clear the fused IEC inlet socket/EMI filter:

A serious flaw of the PCB design was in their provision for onboard RCA sockets instead of normal terminal blocks.  Anyway left with little option, I preferred this method of connecting the input coax which works very well with Teflon-insulation as is doesn’t melt under the heat of the soldering iron:

That doughnut of solder attaches the braid to the PCB ground trace.

And the output transistors are very close to the edge of the modules, so I used a scrap of a mica at each edge as a barrier between the solder joints to the heat sink:

Output inductors wound with Teflon plumbing tape between windings and heat shrink over insulation tape to hold them together:

A 5mm flat top LED in a 5mm hole:

A Schurter stainless steel 2-pole (A and N) power switch:

A 100nF X2 polypropylene cap across each diode of the bridge rectifier:

Ferrite suppressors on the mains:

Four layers of heat shrink around Neutrik RCAs (binding posts have same treatment) to ensure against chassis grounding:

Binding posts (Made in “U.S.A.” which I think is a factory in China :lol:):

Grounding wires from the soft start controller and protection module to the main PSU at the input end of the power supply caps:

And tall rubber feet to allow for good airflow to the case floor vents and heat sinks:


The power supply, soft start and protection modules worked perfectly with positive and negative rail voltages equal to 2 decimal places, but each amplifier module would draw no current.  After much head-scratching, double checking for mistakes and waiting for a reply from the kit-supplier (which never eventuated), I traced the fault to somewhere around a couple of back-to-back BD139 driver/voltage multiplier transistors (from those “grab bags” – who knows what else might have gone wrong had I used all that junk) and removed them in favour of some better quality ON Semiconductor parts and the modules came to life.  🙂

After following the test procedure to warm up and calibrate the modules for correct quiescent current draw, the DC offset at each channel measured a miserly 2 mV (well under the target of 50 mV) and after an hour of running, the heatsink temperature rose only by 10°C with output transistors barely warm to touch!  This is awesome for a Class-A amplifier having passive cooling!

As a full-range test, I hooked it up to some old Italian Chario bookshelf speakers with a C-J PV10AL preamp in front of it – bypassing the speaker protection relays which would have muted the speakers due to low frequency music content.  Dead quiet!  No hiss or hum at the speakers whatsoever.  I was worried about the possibility of a 50 Hz mains hum, but there was none.  With ear to drivers the only hiss detectable was from the pre-amp when turned up without a CD playing (valves).

I played a vocal CD and it sounded fine to the point of being uninteresting which may be a good thing.  It certainly did not sound bright and cheerful like the little CJ Class-A valve amp of similar power output seen at the top left in the next photo, but by comparison it’s way superior to the cartoon/MP3 sound of my Class-D amp.  Of course there would be no comparison to its bass capabilities!  The sound is incredibly clean, uncoloured and natural.

Done.  🙂


I gave this amp to my brother for his birthday.  It is hooked up to the tweeters of some Legend Acoustic speakers and he thinks it out-performs his Audio Research power amp.  That I seriously doubt, but it’s a nice compliment anyway.  🙂

4 thoughts on “Class-A Tweeter Amplifier

  1. Well done, I wish I knew how to do that stuff, must be very satisfying. I’m in the process of designing and soon building a set of 3way OB active speakers, needles to say I need to power them somehow. Your Class D & this project look like a great way to power the Bass & tweeters, but are beyond my talents.


  2. I thought they were beyond mine too, but you can get a lot of help on one of the forums – the ESP one.

  3. Search Google for “DIY Audio Power Amp Toroidal Transformer Cover”. Several sizes are available.

Leave a Reply

Your email address will not be published.