Esoteric Audio Research 509
I’ve had the EAR 509 power amplifiers since 1985. The 509 is unusual as are most of Tim deParavacini’s designs. Two cascaded long tailed pairs for the input/phase splitter which provides near perfect balance followed by the driver and output stage – both wrapped up with the output transformer similar to the QUADs and Macintosh.
From the EAR website:- “He developed a unique output transformer/tube interface circuit called ‘balanced bridge mode’, in which all the electrodes (except the control grids) have their own separate windings on the thirteen-section, biflar wound, output transformer. In addition, the amplifier has no overall feedback, something of a de Paravicini trade mark. Technically this amplifier proved that tubes are capable of performing equally as well as transistors in a laboratory, with a specification that included a power bandwidth of 9-85,000 Hz, -3dB, and also proved that tubes are capable of better things than the ‘retro’ sound some manufacturers look for.”
“The 509 was designed by de Paravicini to have no sound; trying to sum up the 509 is hard, it does not offer the warm ‘comfortable’ tonal quality which so often mark most tube amplifiers out. They are remarkably clear, transparent, with firm realistic bass, effortless top end, detailed and very alive.”
As the circuit is not in the public domain, I’m reluctant to show it in full but to get a good idea check the TVA10 circuit which is almost identical up to the output stage.
Although I’ve had these amps for a long time, I’ve never figured out what they really sound like. Everything and anything that changes upstream can be heard so their resolving power must be huge. Most people would say they are analytical sounding but they are also very musical devices. In other words they have yet to be identified as the limiting factor in my system and so have survived virtually intact with no circuit changes other than that used by E.A.R.
These in combination with the Acoustats make a fantastic combination and due to their no feedback design off the output winding, can handle the weird load of an ESL with ease – very few amplifiers can claim or manage this which is why most are put off the Acoustats. Get it wrong upstream of this pair and they will tell you. Unfortunately, most will shoot the messenger(s) and get something that agrees with them – fair enough, this is about enjoying yourself and the music after all.
Check this for more info on E.A.R.
I spoke to Tim about the main modifications to the circuit as they are how he implemented the 519 circuit. Although Tim said it wouldn’t make any measureable difference he didn’t say it wouldn’t improve the 509 either. Only the brave disagree with Tim and as its his baby, I left alone after that.
The 509 input has a level control and for flexibility. As the first stages are DC coupled, this input arrangement ensures and connected equipment with a DC offset won’t upset the amplifier. It wan’t required in my case so removed the 2.2uF and 0.22uF caps, the level control and the input socket which I thought poor.
I put in a better socket and wired direct to the input tube grid circuit on the PCB. The new I/P impedance will be 100K.
Note: the grounding around the input has to be modified as well IF the socket is changed as its directly connected to the chassis. Also the original socket can come loose with time.
Screen and Bias Supply
The main PSU difference between the 509 and 519 is how the screens and bias/front end is powered. Although the transformer is the same for both, the 509 PSU only uses only one half of a center tapped winding, half wave rectified for the bias and negative rail for the input stage. The screens are taken from the main HT transformer center tap.
The 519 PSU uses this winding full wave rectified. With the center tap earthed, the negative rail supplies the bias and input stage; the positive rail supplies the screens. This makes a remarkable difference to the sound of the amp but as that was a long time ago, I can’t remember what it did but it was good!
How anyone got to bias the originals I’ll never know. It involved removing the bottom and twiddling these carbon single turn presets which were hyper sensitive to any change. I replaced these with 10 turn cermet trimmmers and drilled holes in the top casework for access. Later models had an LED circuit for bias indication and adjustment could be carried out without opening it up.
Somehow while adjusting the bias, the user had to hold multimeter probes on the PCB which involved much stress and not a small risk of electrocution – both hands in a high voltage tube amp is not healthy practice. I mounted two (red and black) 2mm sockets to take multimeter probes on the case rear and a DPDT switch wired to the bias reference resisitors and these sockets.
Biasing now is a doddle – plug meter in and adjust one half. Flip the switch and adjust the other.
From what I can gather, a weak point is the original screen/grid bias circuit. If a new set of output tubes are put in to replace a really old pair and the amp not rebiased quickly, they can take out the screen resistors and that can melt some of the other components. If you’re lucky the fuse will blow. This is more likely with the old type screen supply as the tubes seem to take time to settle down and so draw about double the idle current even when biased correctly hence the 10 minute warm up required before biasing. It is therefore best to wind back the bias before the tubes are replaced and then wind up the bias. I haven’t seen the mkII circuit but I guess it’s like the 519 and I’ve had no problems with it in this configuration. This is good practice anyway and so only happens when people don’t know what they’re doing.
A good feature is an input mute switch – real handy when messing about with interconnects and the like – I just put a switch across the input socket. Of course the original potentiometer serves the same purpose – I just prefer it not to be there.
Bias adjustment presets viewed from top of amp
Bias switch and meter probe sockets on back of amp
input mute switch where input level control was.