Cambridge Audio Azur 350A Amplifier Repair And Review

The 350A was the penultimate in the Azur line from Cambridge, in my opinion some of the best amplifiers they’ve ever made. The 350 was the first to use a wrap-around lid for improved acoustic damping, along with the same multi-layer isolating feet found on the flagship 8-series, brushed front fascias and the end of support for an internal phono stage module. This amp is strictly line-level only, though it does have a portable device input via a 3.5 mm jack on the front. It also has switching for two pairs of loudspeakers, bass and treble controls, a balance control, electronic input switching for 6 line-level inputs and remote control support for key functions including a motorised volume control.

These amps are some of the biggest bargains in hi-fi currently. You can pick one up for peanuts on the used market, and unlike many current amps they’re not festooned with all of the digital stuff you don’t want or need. Digital inputs and Bluetooth streaming in an integrated amplifier are great; for a few years anyway. They eventually become outdated, rendering half of the amplifier obsolete and limiting your options for running more current outboard equipment, as the prevision of those digital inputs in the first place means less space for old-fashioned, never obsolete analogue connectivity. Perhaps I’m alone in this thinking but I’d like nothing more than to see manufacturers revert their amps to having only analogue inputs. If you must put a DAC in the box, do so on an expansion board and commit to making it user-replaceable down the line.

Climbing down from my soapbox and getting back to the amplifier. This particular example was given to me by its original owner who had had it serviced on the recommendation of a hi-fi dealer. PSA – amplifiers, especially not amplifiers from 2010 that don’t have any obvious faults, don’t require servicing. The old adage “if it ain’t broke, don’t fix it” very much applies. Nevertheless this working amplifier was submitted for a service and was returned with a list of ‘faults found’ including the onboard protection triggering immediately on power up. To add insult to injury, the dealer sold them a Rega io to replace it – downgrade much?

The lid on this unit is secured by a number of screws on the back and bottom. Removing these allows the cover to slide off of the chassis, revealing the amplifier constructed almost entirely on a single circuit board. These are not particularly service friendly as to access the underside of the board requires complete disassembly of the amplifier. However they do use almost exclusively through-hole components, and everything is readily available from any component supplier. Only the microcontroller software is not publicly available, but those are rare to fail and if Cambridge chose not to supply the software the microcontroller could easily be substituted for an Arduino, or custom software produced and a new chip flashed if you had a PIC programmer.

on the rear left is the power supply section comprising an unregulated ±28V split-rail supply for the amplifier stages, a ±15V regulated split supply for the preamplifier and a single 5V regulated supply for the microcontroller. Also note the large toroidal transformer and large reservoir capacitors. These amps have bags of power supply headroom for being fairly low-power designs and they have the bass performance to show for it.

The amplifier section is in the middle and comprises a pair of LM3886 output ICs bolted to independent heatsinks. The remainder of the board contains preamplifier circuitry and the logic for the input selection and remote control. It’s basically a commercial gainclone with a very nice preamp and pretty case, which is no bad thing. Cambridge has used variations of their LM3886 output stage for years, first int eh A5 version2, then in three iterations of the Azur 340A, the later Azur 351A, the Topaz AM10 and the current AXA35.

The preamp is based around NE3352 op-amps and uses decent components including high-tolerance metal film resistors and film caps. It uses a 50KΩ Alps RK16812MG082 logarithmic volume pot. It’s a simple, reliable design that has stood the test of time and should provide decades of service without fault if operated within its limits.

The logic is based around a PIC16F882 microcontroller. A BA6218 motor controller drives the motor of the Alps volume pot and the TSOP1836 handles infrared duties. The inputs are switched by a TC9163AF analogue switch IC which is an excellent choice, with more than enough voltage swing to handle input signals far beyond the 2V RMS line-level standard. The 351A replaced one of the inputs with a 16-bit, 48kHz USB input based around a PCM2704 DAC with integral USB, first seen on the 650A if memory serves.

The fault on this one was entirely due to the ‘repairer’. They had removed the mainboard (why I couldn’t say). In doing so they’d lost the majority of the plastic spacers that support the board and isolate it from the bottom panel. Adding insult to injury they replaced them with metal spacers that were much too large. Metal spacers with a printed circuit board above and metal panel below is a quick recipe for a short circuit, which is exactly what occurred here. 

Amazingly and as a testament to Cambridge’s design, no components besides a fuse in the power supply were harmed. And because I have a bad habit of collecting parts from scrapped units, I just happened to have a full set of the required spacers on hand. These were duly installed and the board checked over before reinstallation.

Another complaint on this unit was sticking pushbuttons. The button caps on this unit are moulded plastic with a rear flange to stop them falling out of the fascia. They are retained by the front panel board and are held in place by rings of springy foam. However, if you use certain cleaning products they can work their way between the fascia and the foam ring and cause it to degrade, turning sticky.

That is what happened here. The foam had all but disappeared, leaving behind a sticky goo that was causing the buttons to catch. I removed the front panel board, removed the button caps and soaked them in a plastic bag filled with soapy water. This removes the foam residue without damaging the plastic, especially if they are silver painted. To replicate the original foam pieces, I used stacked O rings – in this case 8 x 1.9 mm and 12 x 2.4 mm. The smaller ring acts as a spacer, the larger ring placed over it provides a slight spring tension to keep the cap in place but enough movement that the button still operates. If you don’t replace the foam with something, the buttons can slide forward in the fascia and look unsightly, and will also feel very imprecise to press.

I cleaned the controls with a squirt of Servisol for good measure while I had it apart, replaced the fuse and reassembled it for a listen. 

Specs are excellent for a mid-2000s budget amp and will put many a high-end audiophile amp of today to shame, loath as makers or owners of such would be to admit it. Frequency response extends to a subterranean 5Hz and is flat across the audio band, rolling off -1dB at 50kHz. Unweighted total harmonic distortion is 0.15%, 20Hz – 20kHz, at 36W into an 8Ω load and 0.02% at 1kHz. Specified maximum output power is 45W into an 8Ω load with an unweighted signal to noise of -92dB, ref 1W of output.

The shelving tone controls give you ±10dB of boost or cut at 20Hz and ±6dB at 20kHz. The two pairs of speaker outputs are configured in parallel. A single pair with a 4Ω impedance or higher can be used, but two simultaneous pairs must be of an 8Ω or greater nominal impedance. They of course cater too for the pointless exercise of bi-wiring, though why you’d want to do that is a mystery to me.

Cambridge amps of this era were often described as clinical compared to the competition and I’d tend to agree, though not in a bad way. They are detailed and revealing, with plenty of low end when required but not so much that it feels over-emphasised. The circuitry in these amps is so simple that there is little to colour the sound, so what you put in is essentially what comes out – but louder. They can handle reasonably insensitive loads though my AR35BX, a particularly difficult three-way speaker, did present an obvious challenge at higher volume levels.

A pair of AR18LS, however, proved a wonderful match, as did a pair of speakers I recently built containing a single pair of MarkAudio Altair 10.3 full-range drivers. These amps can take some abuse too. I once owned an A5 version 2 with a similar output stage and did on occasion drive it wide open into a pair of Wharfedale Laser 50s. I wouldn’t recommend doing this to any amp and it’s not an ‘apples to apples’ comparison as the A5s preamp has slightly less gain than the preamp of the 350A, but it’s proof of the amp’s ruggedness.

Anyone in the market for an all-analogue integrated amp, especially a second-hand budget model should be looking at these. They are solidly built and very reliable, providing a dealer’s incompetent technician doesn’t carry out any unnecessary ‘service’ work. And the performance is excellent and will show many a current audiophile amplifier the door. £50 will get you a working one, £75 a mint one with a remote if you shop around. Great new budget hi-fi is sadly a thing of the past, but bargains like this are still out there.