completely refurbished, 3 years warranty*
*explanation see warranty disclaimer
- steel case enclosure with aluminium face, black coated in good used condition, only few facile scratches on top and edges
- huge toroidal transformer
- 5 or 6 line inputs including
- 2 tape loops
- 1 input can be configured as line or phono MM
- two pairs of 4mm speaker sockets (biwiring ready)
- power switch at the back
- input switch and volume are the only visible controls
- replacement of al electrolytic capacitors
- new trim pots for output stage bias
- cleaning and sealing of controls
- fixing of design/routing errors in the volume line stage
front: keep it simple
No wonder it came here...
...but astonishing how it performs as soon as the design errors are fixed.
In fact I have two of them meanwhile, one was bought on ebay and one came here by a colleague, who obviously gave in with the crackling volume control. Having no time for new experiments at all, both stood for years. But meanwhile we are two and decided to work it out with both.
Before any testing all electrolytic capacitors were replaced with really good stuff, the bias of the output stage was recorded: one 20mV@2x0.33Ohms, the other 16mV@2x0,22Ohms, both channels similar values. That means the manufacturer provided some 30 to 35mA for the MJ11015/MJ11016 output darlingtons.
Obviously the amplifier runs on a single rail of some 60Volts and provides not less than regulated 52V on a single rail for the low level stages from one special LM317 high voltage type. Thus all stages can run only in AC coupling mode, coupling capacitors are unavoidable. Even the rugged output stage is connected to the speaker by a 2200µF electrolytic capacitor - similar as the ION Obelisks and also e.g. in the Quad 303, the first version of the Creek CAS4040 or the early (non remote) Alto by Audio Innovations.
That has advantages and disadvantages. Neither during operation nor failure usually DC current can damage the woofers, in the Nytech's configuration it's even hard to imagine any short circuit could damage the 30A output devices when there is no DC coupling and some ESR in the way. That ESR also has some advantage, since it reduces the damping factor to a level, that is suitable for the majority of British made speakers of that era (simple crossover, two way, average sensitivity). And the power supply comprises less parts - o.k., having only one capacitor there and no double secondary winding on the mains transformer leads to one coupling capacitor for each channel, that could be avoided with a two rail supply. In this respect the number of parts is not really minimized.
The coupling capacitor reduces the low frequency response, with the given value we have some 9Hz@8Ohms and 18Hz@4Ohms (both at the -3dB point with 45° phase shift). Still this is no real problem, the capacitor would become obvious only where the speaker is really critical, then its function will audible shift from providing power to the load to protecting the output stage.
the speaker outputs were once meant for headphone socket switched and direct, but in this version they are in parallel
Of course the amps both became the "full program", the Alps volume pot and the Lorlin select switch were disassembled and cleaned.
such contacts need polishing and sealing
after polishing the silver platings a special grease was applied to protect the surface from further corrosion.
The Alps volume pot then was adjusted for minimal balance error.
the power stage with new trim pots...
The new Bourns trim pots were adjusted for the former bias values.
...uses rugged MJ11015/MJ11016 output devices
One of the amplifiers was set from pure line to MM phono mode again for the first RCA socket pair.
one pair of cinch sockets can be used alternatively as line input or with the internal MM phono stage.
First a disappointment
The sound test then showed an overall tough behaviour, the amp had some fun factor from the very beginning. But it was a bit too harsh and it had a big disadvantage: the volume pot was always crackling. Not only audible when a open input was chosen, even if you adjusted the level of playing music a bit you could hear drop outs. First I thought something went wrong during the cleaning and sealing of the control, but very soon it was obvious, that this was by design, because it was on both amps. The most probable reason was DC through the Alps' tap connection. But there was a not too big decoupling capacitor directly connected to the control tap, 22µF, nothing parallel. Nevertheless a voltage was measurable at the tap against ground, all the time, not only immediately after switching on the amp. So I could not relate it that to the unavoidable charging effects of the design after switching on, because any pure charging effect has only one singular transient, a e-function that must be (almost) finished after some seconds or minutes - how could that be then?
Some measurements later it was clear, that the currents polarity changed, it wasn't DC but LF, the line stage was a ultra low frequency oscillator sweeping up and down between 10V and 30V on the other side of the capacitor, at the very input of the line stage at the first transistors base.
I couldn't imagine that was designed that way on purpose, so I decided to extract the schematics from that part of the board. Here we have the result: the line amp was really an LF oscillator.
schematics of the line amp as I found it - a "no go"
And while I was analysing the board, I could imagine what went wrong during the routing: some connections were made to the wrong track, I marked it on the photo. At its bottom end the smoothing capacitor for the inputs quiescent point voltage was not connected to signal ground as usual, but to the negative feedback (emitter) of the input stage, thus it drifted around and could not fulfil its stabilizing purpose. Also the filter capacitor at the input was set on that wrong track - an then removed.
If you cut the track at the "X" and then connect the left part of it to the ground track above (arrow), everything works fine. Also the input filter can be fitted again to avoid radio frequency and transient distortion, I've chosen 47pF which makes at least an upper frequency response of about 100kHz.
the routing error that must be changed
Still the values were not perfect, the loading of the smoothing capacitor trough 2x330kOhms in parallel would always last 150s for 70% of the end value, so this was divided by ten to 33kOhms providing a theoretical 15s time constant. To speed things up I added a 27V zener diode in line with a 330Ohms resistor, that quickly loads the capacitor almost to the final value and then releases. And on switch off it discharges the capacitor as soon as the supply voltage is more than its forward voltage below the capacitor's. The result is an almost immediate, stable quiescent point without avoidable extra noise and a fast switch off.
resulting schematic with all tweaks
Finally the input capacitor at the volume control tap was changed from 22µF electrolytic to a 1µF WIMA MKS02 foil capacitor, the low edge of the frequency response here is 0.5Hz now which is far below the response of the output stage and absolute uncritical, the former response down to 0,02Hz made no real sense. The foil capacitor makes much less distortion itself and its biggest advantage is the virtual missing of leakage, now the tap is absolutely DC free, neither a charge change nor a leakage of the capacitor disturbs the controls work.
Immediately after switch on all that theory proved right, of course the capacitor must charge for some time and that still sends some DC current through the control after switching on and reminds you to the unsymmetrical supply design as it also does e.g. at a Revox A78. But long before the output bias has found its final value not the slightest trace of the former crackling remains, absolutely clean behaviour compared to the former state. Now it is as it should have been ever since.
the modified line amp - top view
Second sonic test
Some may say there is just a minimal difference, all the same character, just some little details improved and some disturbance removed.
That's not how I see it. I think the resulting version of my two example amps shows for the very first time their real potential.
After fixing the design errors at least two things were absolutely striking:
- the absence of former distortion and
- the stability of the stereo image
Both are easy to explain. Distortion at the controls tap is a not so seldom case, here the permanent existing but also polarity changing DC load made a very annoying grain effect, this is very close to crossover distortion of a Class B MOSFET amp, just prohibiting any relaxed listening. The crackling during volume changes is the least problem compared to that.
The other point is the drifting of the line stage. If there is a constant change of the amplification conditions, you would not even expect a stable imaging from that.
As I already learned from the Musical Fidelity A1 line stage, where the original design suffers from almost the same effects, with the improvement of that part the sun rises and the really well made output stage comes alive.
Still the amplifier tends to be one with good punch and quick going, the difference now is that it can also play relaxed and easy.
overview with open case, picture taken before fixing the line amp problems
Golden lemon ranking
You know I like British amp designs, but meanwhile I can list lots of UK design errors. Until now Musical Fidelity stuff made in the 80s and beginning 90s lead that list by far. But this one here also has the capability to take the affected amp out of any serious competition.
Remember building hand made Hifi - as Nytech mentiones on the 250's case - is an obligation rather than a gimmick. Only few small companies that kept neglecting that obligation survived the industrial take over approaches of the new millennium. I don't know how exactly the story of ION, Nytech and Heed went on since that amp was built, but it possibly could have been better without such golden lemon trophys.
And these samples NOW can be proud on their "hand made" for the very first time.