A few pics of my Zen Revisited
Schematic: Same as published by Nelson Pass with the exception of
It has a fabricated
heatsink, running two parallel Zen Revisited stages per channel
to cope with my 4ohm speaker balls - double current demands over
an 8 ohm rig. Each mono block is dumping 400W - 2 Zen stages each
running at 4amp bias, 50v rail. This combo can manage 50W RMS
into 4ohm before clipping. With thanks to Nelson for the tips
along the way and making this information available to us who
love to mess with this stuff. Oh, and it sounds superb too.
I'm happy to be emailed with any queries or comments not answered here - a lot of time and head scratching has gone into this - it would be a shame not to share it....
Bi-amp/wire update: So I tried bi-wiring from each of my 'strapped' gain stages to the speaker instead of strapping at the amp and single wiring. Wow - even over appx 2ft it sounds better - seemingly more in control than before - cleaner highs.
The Zen is mated with a Bride of Zen pre-amp
This
is a wonderful little pre-amp!! Note the low load impedance seen
by the Bride (500 ohm appox) due to the 2 strapped Zen Revisited
stages doesn't appear to hurt the Bride, although it does clobber
the max gain a bit - see Nelson's bit re the gain being also
dependant on the load resistance. When the volume isn't at max,
the Bride's gain comes up as the attenuator buffers it from the
Zen's 500ohm. I calculated resistor values for a stepped
attenuator taking into account the Zen's i/p impedance - but
forgot to factor in the Bride's gain drop at max volume. I don't
find the 3dB steps I implemented for my 11 way unit too course.
It's an ex WW2 rotary switch (just HAD to use it), it has superb
big silver (?) plated contacts, from some radio gear I think.
It's clunkyness complements the Zen philosophy brilliantly!
Feel free to mail me, and don't forget to enjoy the music along the way...
(Click on the pics for more detail)
Introduction.
The black thing behind the speaker (Gallo
Nucleus Solos) is a mono block. The PSU and preamp are
remote. Each mono block is two Zen stages in parallel - strapped
at i/p and o/p - literally, but o/p strapped at the speaker
terminals to benefit from bi-wiring - see below.
Top
view - shows the two adjacent semi-circular heatsinks -
one per Zen stage. (Back of a speaker ball at bottom of pic).
They are two completely seperate constructions. The inner 3 fins
of each are split, forming a vertical channel down the middle in
which the electronics sits - see other pics. The MOSFETs are
positioned centrally in these channels on the outboard side of
the 12.5mm plate (I used TO3). Speaker wires come straight out of
the top (not bi-wired in this pic - see comments above). Input
comes in from underneath (not visible here) - strapped together
on the way. Fan (not visible here) is below, blowing upwards, see
pic below. The two silver vertical bars visible are the unpainted
tops of the two rows of Al spacer bars which form the thermal
pipe between all the sink fins. Each row of spacer/fins is
clamped together by 3 long M8 studs. The dome nuts are visible in
this pic at the 6:30 & 5:30 positions. The ends of the big
o/p caps are visible in the middle - I used a single (bypassed)
10,000uF from a previous project. The wooden cradle supporting
the sinks via M8 bolts can be seen at either side. It's something
to grab hold of when moving this animal too - each mono block
weighs about 25Kg!
Top
view closer. The thick plates (12.5mm Al) carry the
MOSFETs and get over the problem of all that heat in a
concentrated area. This works brilliantly. Suppose these plates
only need to extend as far as the 'thermal pipe' spacers to the
rest of the sink - their thickness is wasted after that. The
speaker outputs are strapped at the 'chocolate block' connector
in this pic - now bi-wired and strapped at the speaker terminals.
MOSFETs: I used EC-20P16 and EC-20N16 devices (250W 160V) by Exicon, from Profusion in the UK. The are lateral devices that claim to have thermal characteristics that discourage run-away. I believe these 250W devices have two paralleled devices inside, so effectively twin versions of the 125W devices in one case. They certainly seem to work in this design, and seem pretty rugged too. Availalable in TO3 or a flat plastic format (like the IRF devices, but I think it's a little larger).
Thermal protection: I mounted 50C N/C thermal switches adjacent to the MOSFETS on the same Al Plate. This gives a reasonable margin for those hot days (yes, even in the UK) but will cause complete system shutdown if a fan fails or someone drops a sweater on top of an amp. CAUTION: Remember those school physics lessons and how well matt black surfaces radiate heat so well? The good old principal of duality says it works in reverse too - an amp whose heatsinks are sitting in strong sunlight will seriously compromise (even reverse) the 'sink performance. I was reminded of this when fin painting and putting them under a roof window to dry - they got a LOT hotter than they do working as sinks with the amp.
MOSFET
leads coming through from back. The two bolts in centre are for
the thermal switch. Have to be careful with insulating the bolts
when going through this thickness of plate. Also pre-solder
extensions to the MOSFET leads and sleeve insulate them. Note
cleaned unpainted surfaces left for thermal joints. Simply put a
spacer down & paint around it.
MOSFETs and thermal switch between them. Cable ties holding big
caps.
A word on testing - DOUBLE CHECK the MOSFETs are properly mounted and tightened down - they get extremely hot extremely fast otherwise!
Detail of point to point wiring and humungous yellow ICW i/p cap.
The blk lft cap is the o/p, the blk rt cap the PSU smoothing (the
others are in the remote PSU). Final decoupling is (rh) orange
2.2uF polystyrene and 47uF elec below - nice and local to where
it matters. The whole lot is star (more or less) earthed,
including the plate, adjacent to the end of the wt coax i/p lead.
(+) Speaker cable comes in here too, and (-) direct to o/p cap
terminal and f/b point. The big caps are sitting on bits of
bicycle inner tube to insulate them (thermal) from the plate and
hold them off a little too.
Instability problem:
I previously had the sinks grounded via rather a long wire, instead of right at the star earth point as above. This was actually OK when I had a little electrolytic 47uF as i/p cap. Changing this to the big ICW film cap immediatley resulted in some nasty HF instability on the top portion only of the (+Ve) output peaks, under heavyish load (but well before clipping). I suspect this has to do with the extra lead lengths to wire this physically big cap in. Also as per original design there was no gate resister in the gain MOSFET. Instability also changed on touching the heatsink. Yuk.
Instability solution: (thanks to those who helped me out on the forum on this)
1) Add 220ohm gate resistor to gain MOSFET (just like for the current source). Site this as near the gate lead as possible.
2) Ground sink directly at the star earth point and dispense with previous ground scheme.
3) Pay proper attention to PSU decoupling, particularly HF and make sure it is very local to the rest of the electronics. See note above re 2.2uF and 47uF caps.
General view of completed stage before fin assembly. Speaker
wires (blk/red) go out of what will be top of assembly - straight
up to ball terminal nearby - no long speaker cables here! See pic
below.
A
fin awaiting assembly - this one's adjacent to the main plate -
centre hole for thermal switch - it's thicker than the spacer
bars.
The above pics show one Zen stage and the general layout - one of the PSU electrolytics at the bottom, MOSFETs and other bits in middle, o/p cap at top. All strapped (cable ties) to the heatsink main plate. Rest of PSU stuff including another pi filter output electrolytic is in remote PSU box.
One channel - minimal speaker cables bi-wired from the 2
paralleled stages.
Shows the fan below - a stripped down 30cm room fan unit running
ever so slowly (two in series does the trick) so it's virtually
silent. This uses original motor - at a tiny fraction of it's
capacity - could drop the whole thing 10cm with a smaller motor -
would it still be so quiet I wonder?
The PSU
The PSU is built into a separate box (fan cooled - I messed up on specifying the transformers. The air comes out of that slot formed by the top/sides not quite meeting the front panel) - also houses the PSU for the Bride and an aux 12V PSU for fans, relays, meter back lighting etc. (thanks to Nelson for the inspiration on back lighting - pity they came out wt in the pics - they're a nice cool red!).
Each pair of (paralleled) stages is run off it's own PSU delivering 50V DC at 8amps. Each PSU has the original 2mH choke, but I beefed up the caps a little - 2*10,000uF on the Pi filter i/p, then 1*10,000uF in the PSU box on the PI filter o/p, and then a further 10,000uF local to each Zen strapped to the heatsink - see the pics. I cannot hear or scope (may be beneath noise of local FM transmitter) any hum at all at the output. (It's oh-so-easy to get an earth loop - beware, before blaming the PSU or low ripple rejection on an amp if you have a hum problem). I modelled the PSU with the DuncanAmps PSU Designer2 - don't decall the theoretical ripple figures. (Measured ripple on the 50V rail is around 450mV Pk-Pk 100Hz, but there's another 10,000uF local smoothing 20cm of cable downstream of this measurement point - it may be a little lower therefore). PSU2 was able to let me fine tune the transformer spec to get the o/p volts spot on under load - good stuff! The chokes were custom made by Sowter Transformers in the UK. These guys were extremely helpful and had no problem with me ordering just two. They run totally cool. 35amp bridge rects are on their own little heatsinks (Apprx. 3*1.5 inch. 1 inch fins)
All PSU cables will be put in black sheathing to make a neat high-WAF umbilical. PSU has separate On and Off using levered micro-switches poking through 2 small holes in the front panel - looks cool. Dab of Red/Green paint on the lever ends completes the job. Two switches - to implement On latching via a relay, and Off/total shutdown in event of over heating - the N/C thermal switches are in series with the Off switch. There's another strapped to one of the big toroids. I recycled and back lit two VU meters as bias current meters (a la X series). These measure the total PSU output current per channel. Tweaked to read 1/2 scale at 8amps. Looks great.
The Bride is in yet another box (MDF of course) with custom crafted triangular black MDF knobs. See bride.htm. Whole set-up has turned out pretty neat, with an unprecedented high WAF! Must be doing something right.
Turn on - turn off thumps 'n stuff
There are none to write home about. Certainly surprised me, but I did put in 4 'Surge Gard' SG39 NTC thermisters (wired in series) in the feed to the transformer primaries. Forget who makes these (Farnell p/n 606-777) - check out http://www.rhopointcomponents.com/powerp.htm. These take perhaps 15secs to fully warm up and are 12ohm each when cold. From cold start the speaker cone moves a little - sort of movement when playing bass at mid volume, but ever so slowly, absolutely no problem. An opposite but similar 'shaped' movement occurs on switch off. Thinking about this, the Bride stabilises quite quickly - I think way before the Zen PSU has had a chance to charge all that C in the Pi filter, so there's no issue with DC thump from the Bride being seen by the Zen. This charging takes place with a severely slugged feed to the transformers, so the DC volts to the Zen rises quite slowly, (haven't measured it), hence the minimal reaction at the speaker. So this is all great - nothing too complicated. I rather suspect that less C and no Surge Gard would bring problems. BTW, four Surge Gards are used as advised by the makers (samples - thanks guys) to match the energy (Joules - remember them?) - of the caps being charged in the PSU. Sounds plausible to me.
Pete
So has this been of use or interest to anyone? Please take a moment to mail me if so...
What's next? Probably time for a lttle Zen Variations parts 2, 4 ... See http://www.passdiy.com/amps.htm