MODIFIED QUAD II POWER AMPLIFIERS
Last updated February 2007.

The Quad II power amps rarely ever moved out of pure class A operation even with the less sensitive Quad ESL57 which has quite high impedance for most of the main signal power range between 50Hz and 1 kHz.

Those of you with a keen eye will realise the pictured amps to the left are not quite like original Quad II amps. This pair of amps have been built using the schematic further down the page.............................

Better technical performance and better sound from Quad II monobloc is possible if the circuit
 is upgraded with modern components which were were unavailable in 1955 such as modern electrolytic capacitors,
silicon diodes, polypropylene coupling caps and metal film resistors.
The use of a couple of LEDS and a couple of transistors to make sure an owner knows what the bias condition is or if there is a fault is prudent in an age where modern people are just not used to the unexpected and perhaps smoky failure of the
primitive amplifiers of the 1950s.
And you NEED to have some added protection because there is no store just down the street where you can buy a
replacement new output transformer or power transformer if you fuse one because
of failure of an old capacitor or tube.

   This is the only ESL57 impedance graph I was given prior to 2007.
   It is fairly accurate.
   In  January 2007, I was asked to examine a pair of Quad II amp, one of which had suffered
   serious overheating of the power transformer in Australia's hot summer weather. After replacing
   the single cathode bias resistor and its bypass capacitor which had shorted with two individual  cathode
   bias resistors for each KT66 and with two 1,000uF bypass caps, the amp failed a second time, burning
   one resistor open and shorting one of the 63V rated 1,000uF caps, after only an hour powering
   the owner's pair of ESL57 speakers.
   The KT66 survived the experience
   because the owner turned off the amp immediately.  I suspected one of the ESL57 speakers
   was faulty and had the owner deliver them both for a check.
 
   One appeared to be quite healthy and would sustain 20Vrms applied without
   any arcing or distortions. Its impedance was very similar to the graph to the left,
   except that the bass peak in Z was at 100Hz and was 36 ohms.
   The other speaker at low level had a similar Z as the graph to the left but
with a dip at 70Hz. Then as the signal voltage was raised above 2.7Vrms  at any F the speaker exhibited a low impedance causing the amplifier
to clip and act as though loaded by a very low Z value.

I then removed the metal mesh rear cover from the speaker panels and laid open the step up transformer which transforms the low voltage
and high current supplied by the power amps to a high voltage and low current. The wire feeds to the panels were all disconnected and
the amplifier was connected with the speaker mains supply turned off.

The signal voltages of up to around  3,000 Vrms were found to be possible at the output of the transformer so I assumed the  transformer was OK.
Then I connected the 3 panels one by one and found the midrange-treble panel connection was the cause of the very low Z
when the input amp voltage exceeded 2.7Vrms.
I concluded that there is a short circuit arcing of the high signal voltage that occurs in the midrange-treble panel,
and that the panel would have to be repaired. This change in speaker impedance was enough to cause one KT66 to become
seriously overheated and operate in thermal runaway mode where it is saturated with about 300mA of anode current instead of
its normal 70mA. The fitted  mains fuse was 1amp, and did not blow, but the mains transformer became so hot
it expelled some of its internal potting compound which appears to be some kind of tar based materail.
What a mess!  This clearly illustrates the messy way in which tube amps can fail because of faulty 50 year old
speakers.

We can only assume Peter Walker was aware of all this  and  took this into account in because the amps were designed before the ESL57,
so the speakers had to suit the amps.

Quad II have a standard output resistance of 1 ohm when set with 16 ohm load matching on the output transformer, but this occurs
at a limited frequency range at at high frequencies the output resistance rises enough to cause considerable attenuation
at extreme high frequencies. ESL57 have a reputation for  being a bit "toppy" when used with a modern amp with low output impedance.
But with the Quad II amp the response on axis is probably benign.

The low speaker impedance at HF is not a loading problem for an amp which is ideally matched to 16 ohms.
This is because the amount of musical energy above 7kHz is very low, so nearly all the voltage swing in the tubes is due to music signals between
75Hz and 1kHz and where the amp works mainly in class A, so not a great deal of intermodulation distortion occurs to high F signals which are a small amplitude level, except in the case of really high level cymbal bashings.

When using modern speakers which may be rated as "8" ohms yet  have dips to 4 ohms in their impedance within the 50 to 1kHz range, and whose sensitivity is only 88dB/W/M, Quad II have to really struggle, even when the links on the output transformer are adjusted to suit 8 ohms.
In fact the match is for 9 ohms. There is no real provision for matching the amps to 4 ohms where half the windings of the output
transformer are not used and winding resistance losses become very high.
When modern dynamic speakers of nominal 6 ohm average impedance are use with the amp strapped for an "8 ohm" match,
there is a small small amount of pure class A power before the amp begins working in class AB soon after the pure class A threshold
is passed.
The original amp's power supply noise and biasing arrangements tend to cause unnecessary distortions unless the levels are kept quite low.

The original Quad II monobloc amplifier schematic :-
Part numbers are the same as the original schematic.
 

I scanned a good 50 year old paper copy of the schematic but I still couldn't read all the part values and numbers  so I went over the scan in MS Paint,
saved it as a monchrome image, re-typed all the text and re-drew all the circuit lines as carefully as possible and saved it as a GIF image.
Anyone can fill a printed A4 page and be able to read it easily at the larger size.
I hope my copy is less confusing than the original produced by Quad's Information Dept where ink use was severely rationed for many years after WW2 in Britain.

As mentioned above there are a few problems with original Quad II which limit the performance with modern speakers
and power supply noise in Quad II's rather horible quality power supply creates as much noise and
intermodulation distortion in the signal as the THD created by the tubes at low levels in class A, and things rapidly deteriorate
when it is forced to work in class AB without the benefit of common mode rejection of noise artifacts and spuriea.
 
The quiescent bias anode currents of the output tubes tends to drift as tubes age and these two currents become unbalanced
in each KT66 because of the common cathode circuit of R12 and C5. As tubes age, KT66
grids develop a positive voltage due to minute grid currents at idle, and the chosen value of grid bias R7, R9,
of 680k is way too high, and usually the 680k ohm value rises with age to make things worse. The voltage measures normally across the
680k should be no more than +1V, but I have seen KT66 at near the end of their life with +9V at the grid
and with 90mA of anode current, with slightly red hot anodes. This is disastrous for the music, and such a tube continues to overheat insidiously,
before finally  melting down internally, and perhaps terminally damaging a power and/or an output transformer.

The EF86 input tubes are set up in what is called a floating paraphase  phase inverter, and the distortions from V1 anode are fed to V2 grid to get an unnecessary amount of thd. Its not a serious fault, but in all the old samples I have ever seen where they have not been serviced by an authorized Quad repair agent the output tubes have unbalanced anode currents, and the signal balance to each output tube isn't matched close, and THD can measure up to
10 times more than it should at all levels.

Hence I conclude that if one is to re-furbish such grand but somewhat limited old amps, one could
could improve the circuit behaviour to get a better end result. Not many modern techs would be able to build the
schematic I have tendered to everyone here; the work shown is that of the dedicated specialist.

Allow me to present the minimalist amount of upgrading work to make these amps lees of a hassle to own,
and to ensure musicality, smoothness, accuracy warmth and detail.

My latest most basic Reformed Quad II  schematic, Jan 2007 :-

The component numbers used here don't relate to any component numbers in the original schematic
except perhaps by coincidence.

The existing large 16+16 uF capacitors in a metal box have been removed.

The original Quad II amplifiers have a load of 4k anode to anode including the 10% winding resistances when a 16 ohm load is connected to the OPT when configured for 16 ohms.
24 watts class AB1 is produced but only 21.6 watts appears at the output, because of the winding resistance losses.
Only the first 6 watts is really pure class A. But when 32 ohms is connected there is about 18 watts produced of which only 1.6 watts is lost
to give about 16 watts of pure class A which is understandable because the idle anode dissipation in the pair of KT66.
This is = 333V x 0.13 amps = approx 44 watts, and maximum class A anode efficiency is about 36%.

My view is that Quad II amps should NEVER be used on their 16 ohm outlet match, even with ESL57 connected, and that means using the 8 ohm OPT configuration even though that unfortunately results in winding losses of around 17%. There is also a way of making a 4 ohm configuration, by taking the output from the point Q on the OPT but the windings from Q to T cannot then be used at all and leakage inductance and winding losses dramatically increase for fairly poor performance.
In order not to confuse people with the complexity of wasteless transformer impedance matching, Peter Walker settled for the simpler method of a maximum of just two OPT links for 8 ohm, and one for 16 ohms, and the load adaptability of Quad II was forever compromised.
Despite the high winding losses with 8 ohm configuration, the amps work OK, although they become
only conditionally stable, instead of unconditionally stable.
If pure capacitance loads were applied, HF stability was assured with the additional phase and gain shelving Zobel network
of  R7 and C4 from V3grid to V4grid.

The bottom part of the above schematic shows the active protection provided by the group of modern solid state
parts which have ZERO effect on the sound because all they do is monitor the KT66 dc cathode bias voltage,
and if this voltage rises from approximately +33V dc to 52Vdc, then the SCR is trpped and the amp is turned off.
The green led shows all is well and the amp protection is OK and the amp is turned on,
and if the SCR is tripped, the relay switches off the mains power to the main power transformer.
Then the green led goes out, and the red led is turned on, which tells an owner something is wrong.
Such circuits work far better and more reliably than a fuse, and the above is arranged so the ac signals
at the cathodes are filtered out by C12,13, and R24, 25, so that slow ac signals won't
make the scr trip. The ac never hurts the tubes; its the dc bias flow that we must worry about.
This protection circuit will work if someone connects a completely shorted speaker or cable and turns the volume up well,
since the dc cathode voltage will rise with gross ac overload.
But partially shorted speakers or those of 2 ohms may cause the tubes a slow un-noticed
drift to thermal runaway, and then the anodes may over conduct dc which will then be detected and the amp is safely
turned off.
After the amp has had a temporary fault that causes the turn off, it can be reset by simply turning the amp off at the
Control amp, and back on after 5 seconds at least.

If the amp again switches itself off, the fault must remain, and technical service is needed.
Additional bias balance indication could be added, but its slightly more complex as shown in the more eleborate
modification to Quad-II amps shown in the next schematic....

The component numbers used here also don't relate to any component numbers in the original Quad II or any other schematic
except perhaps by coincidence.

In the power supply section, the ac HT is rectified with silicon diodes through current limiting 47 ohms, R19, R20 into 100uF, C9,
to get a B+ voltage = approx + 410V.  The 200 ohms of R18 and R18A and C10, 470 uF
form an RC filter to obtain a well smoothed B+ for connection to the CT of the OPT.

V5, GZ32, is retained in its socket for appearances only and was tried as a slow turn on series diode with an equivalent resistance of about 90 ohms in
series with about 100ohms to give an added B+ turn on delay but the delay is negligible, and the B+ still soars to +450V
before being pulled lower by the anode currents because the KT66 turn on a lor slower than the GZ32, so this idea of having the
GZ32 as a hot series diode isn't worth the trouble. 
If there was some ability for a 1 minute delay to B+ turn on, the benefit would be that the gettering in the tubes would have a better chance to absorb gas molecules before they are stripped of electrons to become positive ions which bombard the cathode and reduce cathode life and emission.  Since the GZ32 gives almost no useful delay when used as a rectifier or as a series diode, one can leave the GZ32 entirely disconnected.
There is not a greatly extended tube life due to delayed turn on and audiophiles will replace tubes before emission fades badly,
so delayed turn on isn't really needed.

As the above circuit now is, the ripple voltage at the OPT CT is reduced from 17vrms in the original 1950s amp with GZ32 and 16 uF capacitors to less than 0.1Vrms, and the CT is operated from a low impedance supply since the 470uF has a much lower Z than the 16 uF caps used in the original amps.
So a lot less rail noise is able to enter the signal during all classes of operation.
The B+ soars to +460V at turn on but settles back to about 380V at the OPT CT after 20 seconds with about 60mA per output tube. 
450V rated caps will cope OK with this operation condition.

In this Quad-II amp KT90 were used in lieu of the KT66 since I wanted to get a lower amplifier output resistance and slightly lower THD
and better ability to cope with the 4,000 ohms a-a load value when 8 ohms is connected.

The two KT90 cathodes are individually biased with R15 & C7 and R16 & C8 networks of
500 ohms and 1,000 uF respectively. Under dynamic music conditions the very slow time constant
of the bias circuit prevents much movement of the cathode bias voltage.
The balancing of cathode bias is automatic and as the tubes age their current imbalance is
negligible compared to the original biasing with a common Rk. In the original amps if one tube decides to conduct more idle current than the other, a larger voltage appears across the 180 ohm common Rk,
and the other tube tends to conduct less idle current because of the bias change, making the imbalance in DC for each half of the OPT even worse. It isn't unusual for me to find that old Quad II amps have one tube with 40mA of idle current, and running cool while the other has 90mA and runs hot, with cherry red hot anode spots.
Nothing sounds worse....Individually biasing each cathode of the output tubes usually
makes a pair of ill-matched old tubes balance well enough and a few more years of life can be had.

The original Quad II amplifiers have a load of 4k anode to anode including the 10% winding resistances when a 16 ohm load is connected to the OPT when configured for 16 ohms.
24 watts class AB1 is produced but only 21.6 watts appears at the output. Only the first 6 watts is pure class A.
But when 32 ohms is connected there is about 18 watts produced of which only 1.6 watts is lost
to give about 16 watts of pure class A which is understandable because the idle dissipation
in the pair of KT66 = 333V x 0.13 amps = approx 44 watts, and maximum class A anode efficiency
is about  36%.

My view is that Quad II amps should NEVER be used on their 16 ohm outlet match, even with ESL57 connected, and that means using the 8 ohm OPT configuration even though that unfortunately results in winding losses of around 17%. There is also a way of making a 4 ohm configuration, by taking the output from the point Q on the OPT but the windings from Q to T cannot then be used at all and leakage inductance and winding losses dramatically increase for fairly poor performance.
In order not to confuse people with the complexity of wasteless transformer impedance matching, Peter Walker settled for the simpler method of a maximum of just two OPT links for 8 ohm, and one for 16 ohms, and the quality of Quad II was forever compromised.
Despite the high winding losses with 8 ohm configuration, the amps work OK, although they become
only conditionally stable. I placed R17 and C11 across the output to help keep the amp stable at HF if
pure capacitance loads were applied, and revised the global NFB network, R3, R4, and C1,
and added a zobel of R9 and C6 from V3grid to V4grid.

Instead of a paraphase inverter I have the two EF86 set up as a true long tail differential pair
with common cathodes and current sink R12 to -38V at C14. There is a -390V supply derived from the HT winding with two silicon diodes to charge the R23, C15, R22, C14 network.
The balance of drive voltages was found to be excellent, and a solid state constant current source for the two EF86 cathodes was unneccessary.
R10 & R11 and R13 & R14 and the 10k pot allow for KT66 bias current balance adjustment.
The range of voltage adjustment applied to each output tube grid is from 0V to -8V which is enough.
The condition of bias balance or any tube fault is monitored by the 2 LEDS which should glow at equal brightness when Ik for each tube is within 5% of being equal. The bias adjustment is easy to access through a chassis hole and a thumb nail can turn the head of the 3 watt wire wound 10k trim pot. The bias balance indication amp with a couple of small bjts derives its power from a rectified quadrupler supply working off half of the  6.3V heater voltage.

The image at the top left of the page shows the amp balance LEDS with equal brightness.

With KT90 there is 25 watts AB into 8 ohms with slightly more into 4 ohms, and Rout = 0.78 ohms.

Some folks would hang me from an old oak tree after hotting up a Quad II amp like I have, but I did give some thought to what might happen in a fault situtation such as a shorted or saturated KT90 tube, bias failure, gross overload input voltage with a shorted output etc.

The LEDS show when there is a bias problem since they glow unevenly, so an owner knows if a tube is playing up, and a shorted or overloaded output also will make the LEDS glow unevenly and fllicker.

I was frightened that the KT90 may over stress the output transformer and power supply if one or both were to fail and become saturated with current, resulting in a maximum of about 1Amp dc from the power supply.
But such a failure would just blow a mains fuse providing it is a slow blow 0.5A, which I tried instead of the
original 2A fuse shown used on the original amp schematic.

An intermediate level fault could be more damaging if a mains fuse does not blow.

I tried placing a 1,000 ohm resistance from the OPT CT to ground to simulate a serious but intermediate fault event. This caused 350mA of dc B+ current to flow from the power supply, and the mains 0.5A slow blow fuse blew after a 3 second delay. There is NO need for extra fuses in series with each cathode circuit of each KT90, and the trouble of maintaining them.
With the output short circuited, and the input turned up high into gross overload, the 0.5A mains fuse will not
blow, since the power supply anode current in the seriously overloaded condition only increases by just over twice to 250mA, which isn't enough to make such a low value mains fuse blow.
Such a sustained fault like this is very unlikely to occur, because either silence from the shorted output or gross distortion of sound woulds alert an owner that something is wrong, and he/she would turn things off and investigate.

In this pair of amps I didn't provide any means of automatic turn off of the B+ for the whole amp if cathode dc voltage rose too high
as in the Jan 2007 schematic. There was not a large amount of spare space available for an auxilliary power tranny
and extra parts on a 'protection' board. I felt that the owner would keep a fairly close eye on the bias balance
leds which will tell him if something is wrong.

The total equivalent resistance of the B+ power supply at the CT is equivalent to 560 ohms looking back towards the rectifiers, with 200 ohms of that being the series resistance I have used to mimic the resistance of the replaced GZ32.
The 200 ohms is rated at 20 watts and will not fuse open because the 0.5A mains fuse will blow first.
With a fault where an output tube conducts 5 times the idle dc current, Ia = 300mA, the OPT CT supply voltage will be pulled down to 292V, anode voltage will be 273V, heat dissipation in the OPT will be 6 watts which is OK and no worse than would occur with KT66, and cathode voltage will be 150V, leaving 123Vacross the tube so its dissipation will be  36 watts and well within the rating of 52 watts for KT90, wheras a KT66 might be destroyed by such an event, since its rating is only 25 watts.
Should the fault persist the cathode bypass cap rated at 63V will fail to become a short circuit after 10 minutes,
and more current will flow and the amp will definately blow its 0.5A mains fuse.

The slow blow 0.5A fuse has proved to be quite reliable, and it survives the in-rush current at turn on.
I tried a 0.5A fast blow fuse ,and it fused immediately at turn on, suggesting that mains input current exceeds 0.5A for a short time and thus will blow a fast acting fuse. The slow blow fuse will tolerate fast surges maybe up to 1.5A, but will fuse when the AVERAGE current goes over 0.5 Amps and stays there for some time.

The Quad II set up as the above schematic draws 88VA so with 240V mains the input current
average is 0.366A. This is less than in the original Quad-II which has the GZ32 heaters consuming about 10 watts more.
If a fault occurs with 300mA of idle current in a failed tube, the power draw from the mains goes up to about 160VA, so average input current = 0.66A so the 0.5A slow blow mains fuse WILL blow.

So we should try to stay with the 0.5A slow blow fuses. If nuisance blows do occur, a move to 0.7A slow blow may be needed but I doubt it. Fast blow fuses would need to be rated at 1A+, but would not offer protection against bias failure resulting in a mild 300mA fault current.

The original Quad II amps had 2A fast blow fuses, giving little protection against damage to OPT etc if there is bias failure as mentioned above.

KT90 are perfectly interchangeable with KT66 and draw the same idle current, but produce a maximum of 30 watts instead of KT66's 22 watts.  KT90 heater current is 1.6A instead ot KT66 at 1.3A but its OK because these amps were designed to have add on tubed tuners etc
which are rarely ever used now.

The musical performance includes tighter bass and more controlled and detailed treble, 
so I have to say KT90 in Quad II sounds better imho.

On the left image you can see the IEC shielded mains socket that replaces the Bulgin original.
The old Quad II recessed 4 mm banana sockets were replaced with something from some retired HP test gear
because the originals had cracked plastic from years of speaker cables being wrenched sideways.
The new signal input socket is Canare 75 ohm RCA which replaces the 6 way "jones" socket used in the original amps. The terminals are mounted on a white fibreglass panel.
Shielded interconnect cables from a  preamp should only be used since the speaker output cables are close
to the amp input. I did try using my unshielded dual foil cables which were close to the speaker cables
and no oscillations occured, probably because the live input cable is tied to the low impedance
of the cathode follower in the preamp. 

The right image has the whole system in trial use before  sending it off to my customer, although the
GZ32 are not plugged in to keep it looking original.

A much modified Quad 22 control units is to the left of the power amps and is fully described in my page on Quad 22 mods. The separate preamp power supply is on the shelf below the two amps.
For each power amp I used a blue "on" LED and red mains on switch placed in an aluminium panel
to cover holes for original mains voltage settings, something not needed since the 240V mains
in Oz rarely varies, and is always over 240V unless it is a cold night in July. There is never any need to employ
the other two lower mains voltage settings.

For those who worry about THD, here is a graph....

The above graphs are drawn on logarithmic axis for both THD and output voltage
The test is for the reformed Quad II and since I don't have Peter Walker's test results he made for an original amp in 1955,
I cannot say how my altered Quad compares.
As you can see, at the onset of clipping, both KT66 and KT90 produce about  1% THD
at 21 watts and 24 watts into 8 ohms respectively.

Above about 5Vrms there is a somewhat more rapid rise rate of THD as the amps move into class AB,
and the curves with all their kinks are typical class AB tube amp measurements.
But at clipping the THD does not matter as much as what is happening
at normal levels of 1 watt, or 2.83Vrms where the KT66 gave 0.04% and KT90 gave 0.02%.
At 1/4 watt, or 1.41Vrms, KT66 gave 0.024% and KT90 gave 0.014%.
This is where our focus on the soloist is intense, but really, anything below 0.05% is not too bad,
considering tube amp artifacts are less objectionable than solid state's.
The trend for the KT90 to have about 6dB less, or half the THD of the KT66
continues below 0.5V, where it became difficult to measure THD accurately since
0.01% of 0.5Vrms is only 50uV, and noise tends to swamp THD measurements.
So below 0.33Vrms output the graphs have been guessed; it is of no importance.

I also tried measuring THD with loads between 4 ohms and no load at all.
With a 16 ohm load connected to the the amp set for 8 ohms, at a watt the THD is 6 dB less than with 8 ohms,
and with 4 ohms its about near twice what the 8 ohm produces but at low levels of below 2 watts
which covers much listening for many people, any load down to 4 ohms is OK, and KT90 use gives around 1/2 the
THD of KT66 for all loads.

The measured output impedance with KT66 in my circuit is 1.2 ohms, and with KT90
it is 0.9 ohms approx. I also tried russian 6550EH which gave similar THD to the russian KT90EH,
and Rout = 1.0 ohms.

                                     ------------------------------------------------------------------                             

Below is a schematic of the first reformed Quad II I attempted in 1999, using KT88 in triode,
plus triode input tubes instead of the EF86 pentodes.

Part numbers for this schematic have no intentional resemblance to those used in the other Quad II schematics.
Here I have used KT88 in Quad II amps set up in triode with +410V supply, and fixed bias
of about 55mA per tube. Without the voltage drop across the cathode bias circuits and across a choke in the
in the B+ rail, and because there is no tube rectifier, and because there idle current is lower for fixed bias operation, the B+ is much higher than in the original quads, at about +390V.
The higher B+ voltage suits triode operation because of the limitation of triode anode voltage swing in class AB1.

This schematic gives 20 watts in triode class AB1 and measures very well although the amount of NFB
is less than the total of about 20dB used in the original Quad II.
The local cathode FB in the output stage is not high because the triode gain is only half that when used with fixed positive bias
applied to the screens as in the originals.
So instead of 8 db of CFB, there is effectively only 3dB. When trioded output tubes are used, just enough local cathode
FB exists to compensate for the winding resistance losses in the OPT.

The total applied NFB is about 15 dB. I used 12dB of global NFB because earlier tests using EL34 in triode with only 6dB of global NFB gave lack lustre dynamics when compared to my client's other 10 watt amp with 6GW8 output tubes in ultralinear with about 16dB of global NFB.
The KT88 delivered the kind of vibrant and accurate dynamics that he was looking for.

You just cannot use Quad II amps without global NFB because the output resistance will be too high
even with trioded output tubes. With the set up as I have it the Rout was less than 1 ohm.

By the way the original Quad II amps with KT66 and without their global NFB and only the local CFB have Rout = 9 ohms. With KT88 in triode, and without global NFB the Rout = 5 ohms and way too high for any speaker. The global NFB reduces this down to a reasonable figure.

There is a 10k balance pot to balance the idle current, Ia, in each tube. The value of R21 is not shown since it has to be trimmed to get the right value of 55mA per tube. The balance is set by turning a pot shaft so each LED is equal brightness.

When the GZ32 was removed, the space was used for capacitors I had acquired at that time.
The amount of capacitance can be larger as I have indictated in the March 2006 and Jan 2007 schematics, and don't need to be
placed above the chassis.
Since the triode operation does not require a well filtered fixed screen supply as in the original Quad-II,
The 20H screen supply filter choke was replaced with a 2H choke of much lower dc resistance and placed
so that it filters the plate supply applied at the OPT CT, but without the voltage drop when a resistance is used as in the 2006/7 schematics.
V2,V3 form the differential driver LTP amp and have a constant current source, CCS, MJE340 transistor in the commoned cathode "tail".
The transistor's extremely high finite collector  impedance is actually above many meg ohms, so it is impossible for the transistor to colour the sound.
The CCS  ensures that if the load resistances are equal for the LTP triodes, output voltage will be exactly equal, and even if the triodes are grossly unmatched.

Input is a 12AT7, but could be 12AX7 or 12AU7 or 12AY7 with circuit mods to cathode, anode and NFB
resistors so that the gain and amount of NFB applied is appropriate.

This modification sounds very well even with 4 ohms speakers, but note that the OPT link settings
are for 8 ohms on the taps Q, R, S, T.

Some have criticized me for using KT88 in Quad II amps because the heater power needed is greater.
But its only a small margin over what KT66 use and in any case the Quad II has been designed for use
with add on tuners and preamps, something not always used.
A pair of amps with this mod has been used every day since 1999 in a house at Cooma, NSW
where summers are not cool.  In these two amps I drilled lots of holes in the bottom cover and reversed the position of plates holding the power tube sockets to get more air flow through the bottom of the amp and up around the output tubes to help keep the amp cool in Oz summers where 35 degrees C indoors isn't uncommon in January.  Because there is no GZ32, and that the tubes draw less idle current than standard, there is also
less heat generated in the power transformer.
These two amps have now not needed any service for 8 years now it is 2007.

Back to index page