Quad II Forty re-engineering.

Everyone interested in tube amps may have seen old 1950s Quad-II monoblocs and listened to music

amplified by them. They used two EF86 and two KT66 and GZ32 rectifier and could make about
15 to 22 Watts with speakers from 32ohms to 8 ohms. They included the facility for load matching to
optimize the work of the KT66 by means of by altering two links on underside of OPT in the sub-chassis
area, using pliers and a soldering iron, and a brain that had IQ of at least 110.
There were two possible load matches - one for 8 ohms and one for 15 ohms.
Many Quad-II were used with the amp set for 15r0 to power Quad ESL57 which have a benign
impedance variation between about 33r at low F, 8r at middle F and 1r8 at 18kHz, with an average of 15r0
for the main audio power band between 100Hz and 500Hz. In 1950-60, "Normal" dynamic speakers with
cones and domes and voice coils were mostly rated at 16r0 across the AF band, and the Quad amps powered
them well because the sensitivity was usually over 93dB/W/M and little power could make a lot of SPL.
With 16r0 speakers, and OPT links set for 15r0, the amp produces over about 22Watts in class AB1,
because the PLa-a anode load is about 3k8. If the OPT links are changed to match 8r0, then the same
16r0 speakers reflect an RLa-a load of 7k6 and the 18Watts produced becomes nearly all pure classA1
and you then get the best music, but with less maximum Po.

But since 1955, most speakers which people want to use with Quad-II have average Z of maybe 6r0,

with a number of brands under that and over that, and average sensitivity of modern speakers has fallen from
about 93db/W/M in 1955 to about 87dB/W/M so that today's speakers require 4 times more power to reach
the same SPL as in 1955. The benefit is that the modern speaker gives far less speaker distortion and gives a
flatter F response and far less box coloration and far better bass. But of course you have to pay more for this
deal because many speakers on the market are garbage for poor men whose wives hate big floor standing speakers.
Consider a 3 way well made full range hi-fi speaker with all 3 drivers having Z at 6r0.
Crossover filters often make speaker Z low at the crossover band, so 6r0 may reduce the Z to 3r0, and those
speakers should be rated for 4r0.
When powered by Quad-II, and with OPT links set for 8r0, the reflected load to tubes becomes about 2k0
with very high winding losses. Then the Po has a very low amount of initial class A, high THD/IMD, high
noise injection from  poorly filtered B+, and damping factor becomes lousy. Po max is restricted.
In 1990s, the nostalgia about old Quad hi-fi gear remained intact, with most people unable to be rational about
their assessment of the integrity of Quad designs, including the Quad tube gear. Money could be made
by selling the nostalgia, and it could be helped along by including better performance, rather like re-marketing of
the Mini Minor motor car, which was a little buzz bomb in 1975, but in later years became a much bigger
small car with Mini Minor looks.
The amp making capability of Quad was sold to the emerging Chinese entrepreneurs and they engaged
Andy Grove to design a more powerful version of  Quad-II which has two 6SH7 input / drivers and two
KT88 plus a 5U4 rectifier.
This new monobloc was called the Quad-II-Forty, because down a steel hill with wind behind it, it can just
make 40Watts.
The new amp had a schematic virtually identical to the old schematic of 1950s, which goes back to the Quad-1
amps by Peter Baxandall in 1948.
The Forty chassis size is about +12% on each dimension of the old Quad-II. This meant the transformer
cover boxes could contain better PT and OPT, and indeed they do, but it seems the boxes were made before
PT and OPT design was finalized, and when that was the PT and OPT didn't fill  the boxes. Inside the PT
box there are two of the small PSU electro caps. But anyway, I found the Forty made up to 32.5 Watts
when tested with a continuous sine wave, and very little more than what can be had if KT88 are used instead
of KT66 in old Quad-II. The extra Forty power was expected with KT88. In fact, with more thought by
clever minds, the chassis could easily have been slightly smaller to still have plenty more room to fit much
better PT and OPT without wasting space inside boxes.
Most shortcomings of old Quad-II were preserved in the new Forty, and I could easily see why I had
been lumbered with samples to repair which had not been used very much since purchase new.
The new Quad-II-40 had horrid quality printed circuit boards and I concluded it suffered from lack of
prototype upgrading before production. There should have been far more criticism and correction of shortcomings
before mass production, but the Chinese like to make short cuts.

The paint job made the amps look well, and metal work was nice, but tube socket quality is atrocious, and when you analyze

what Chinese did, and what they didn't do, a simple repair could only achieve a little, and to get the best from these amps
one has to remove the PCB and all parts and start all over again, like one does with the old Quad-II from 1950s, which are
full of quality invented by the Quad company accountant, Bertrand Parsimonious.

When you take off the bottom cover of Quad-II-40, you see the PCB is crammed in some places and not other and there

are hot running 390r cathode resistors swaying in the breeze off long leads. The more you look, the bigger the frown you get.
Fig 1.

Now this amp had smoked and blown a fuse a few times and made noise. The Quad-II-40 does have separate

R&C networks for each KT88 cathode. The caps were tiny and barely rated well enough and I replaced both
with the brown colored electro caps at the far rear of sub chassis space. The original green cathode R = 390r,
rated about 7Watts, and in theory should not get too hot even with Ik = 100mAdc, with heat at 3.9W.
But you can see how one 390r has begun to turn brown because of heat, and you can't hold a finger on them
while operating normally.

Quality of the pin gripper forks in ceramic tube sockets from 1990s from China was just atrocious and not

anywhere near as good as NOS McMurdo sockets. Some Chinese made tube sockets are now excellent in 2014,
but before you buy any, check they are OK by buying ONE, then testing with tube in tube out to see of pin gripper
function is as good as NOS.

Notice the thermistor used to slow down heating of the 5U4 directly heated cathode. This has high resistance when cold,

so the 5U4 cathode heating takes more time than the usual 3 seconds to begin emission. There are also 2 paralleled
1r0 green resistors  in series between cathode winding and cathode because the winding is 6.3Vac, not 5.0Vac.
The PT did not have a dedicated 5.0V rectifier winding, which suggests that the PT was chosen from "about right"
cheap generic stock. Probably the OPT is also from generic stock. But the build quality is better than some other much
worse Chinese amps. Total production cost to the Chinese is probably < $200 using what is virtual slave labour,
yet you pay many thousands for a pair in the shops in London or Sydney.

The 5U4 works OK for a pair of KT88, and the high Ra of 5U4 gives enough series resistance to ensure B+

cannot be excessive after KT88 have begun to conduct some 15 seconds after turn on. The slow turn on behavior
for 5U4 is not really necessary; and B+ soars quite before KT88 warm up. Hence the need for the pairs of el-caps in
series for B+. There is ZERO NEED for a tube rectifier, other than to satisfy idiot nostalgia enthusiasts who like to pay
for 1955 technology which does nothing to improve the music. Sure, the audio amp tubes do work well for music, but the
tube rectifiers give ZERO positive contribution. Silicon rectifiers allow for far more reliable working without the heat
wasted by the tube rectifier, plus the Si diodes allow lower Vac for HT winding with a doubler or bridge and far better
natural Vdc regulation than any tube rectifier can offer. The use of high value electro caps is then possible which allows
very low ripple in B+ supplies.

On the right hand end of PCB, there are two small size electro caps in series poking down under PCB into spare space
inside PT box. If these caps need replacing, the PCB board must be lifted out to get access to el-caps and repair is
hugely difficult when it should not be. The box for the choke on chassis top has L > 10H and Rw = 375r, and this acts
 to filter the fixed B+ applied to screens of KT88, and for B+ of input stages, just the same method as used in old Quad-II.
The choke box is bigger than it needs to be, and the anode B+ was not well filtered, also like old Quad-II, so best class
AB operation isn't possible.

Fig 2.


Fig 2 shows what I ended up doing with two Quad-II-40 monoblocs.
PCBs are removed to rubbish bin. Connector strips installed, using 10mm x 8mm hardwood rods with 4guage c/s brass plated
cupboard hinge screws as terminals at 10mm c-c. The soldering heat cauterizes the timber and releases pressure in 2mm drilled holes.
But screws remain well fixed and in 500years integrity will be fine. The wood strips are well varnished. Wima caps are glued
to chassis with Selley's Silicone 401, acetic cure, good for 200C, and to last indefinitely. All tiny sized R were replaced with 3/4 or
1W metal film. Wima polypropylene coupling caps were used to replace existing which could be polyester, or goodness knows what.
The input  V1, & driver V2, V3 have entirely different schematic to original and produces at least 1/4 of the THD as the original
amp. I added an extra choke for PSU to filter B+ applied to OPT CT in anode winding. While 2 original el-caps were retained,
others with larger C values were added. DC is applied to input tube heaters.
There is an active error protection board added, left hand side, and its auxiliary PT for protection is at rear right hand
side. near the added black painted choke for B+. Notice the 1.6mm dia copper wire 0V rail running
above tube sockets.
Fig 3.

Fig 3 is the new amp schematic.
Notice the revised operating conditions for KT88. Originally, the KT88 have quite high Ia = 100mA and Ea = +360Vdc,
so each KT88 has idle Pda = 36Watts which is getting up and one may find one might idle at 37W and the other at 33W
if the tubes are not matched. I have Ea = 375Vdc, and Ia at 68mAdc, so Pda = 25.5Watts. Rk = 630r, not the original 390r.
The KT88 and tube rectifier are very happy with my easier working conditions.
In Australia, the mains voltage can often measure 250Vrms, and this makes the HT voltage higher and hence B+
higher than it should be, so the Iadc in KT88 is higher, so the product of Iadc x Ea is a higher than wanted number of Watts
and KT88 can run too hot. The screen voltage is NOT regulated. I thought of adding shunt regulation of Eg2 which could be
at +350Vdc, about 50Vdc below the anode B+ voltage. Its not absolutely necessary for what will always be mainly be
use in pure class A1.But the lower the Eg2, the lower the Ek needs to be, so Rk could be less for the same value for Iadc.
KT88 have high Eg2 ratings, but that does not mean the Eg2 must be as high as the Ea. 

The owner had purchased a box full of  NOS 6SH7. I tested over ten of them and found half were gassy and noisy,

or highly microphonic or had all 3 defects. They had been made for the NZ military before 1944 and in unopened riveted cardboard
boxes. Rivets and some tube pins had turned green with age after storage in a damp shed for so long. Like so many tubes made at
that time, they cannot be expected to work well for long or acceptably because of gassiness.
In pentode mode and used similarly as in old Quad-II and with very low Ia, 6SH7 do not offer better performance than the EF86.
The 6SH7 has outstanding gm when Ia is say 5mA+, hence very high gain, but is low gm with low Ia. But there is a trade off between
low gm and high RLa and high gm and low RLa. The high RLa between anode and B+ of 180k limits the possible Ia to about 1mAdc
where gm is low. The tube data curves show gain of about 100 is possible with resistance loading. The pentodes in original are set up for
paraphase to make two phases for KT88 grid drive.
Much better performance is to be had with 6SN7 as an LTP  to replace one 6SH7, and wiring the remaining 6SH7 as a triode. The 6SH7
makes a very nice triode, with low Ra and gain = 25+. But I only used the 6SH7 in triode because the owner had a few good ones
among those he'd bought, and among those already in the amps as purchased - which were red painted NOS from maybe 50 years
ago. The circuit would work better with a paralleled 6SN7 used instead of trioded 6SH7 for V1 input. Gain of 6SN7 would be about
4dB less than 6SH7, so input voltage would be about 1.2Vac with the amount of global NFB kept constant.

So if anyone wishes to improve Quad-II-40, DON'T use any pentodes at input or driver, use only a suitable octal triodes and the most

suitable is 6SN7. You could always remove the 2 input octal tube sockets and install 9pin mini sockets mounted on a small
round sub-plate under chassis top. This would look very well, and look like no mod had been done. Then 6CG7 can be used
and these give excellent performance which is at least equal or better than good samples of 6SN7. But ECC99 can be used for
V2, V3, and maybe 12AU7 for V1, the 9pin tubes are more available.

In old Quad-II amps I have tried using 6BX6/EF80 instead of EF86. With careful setting up, the 6BX6 can have lower RLa and
much higher Iadc, and hence higher gm which puts their operation further into the linear region of class A signal pentodes.
And I have use the pair of pentodes in differential mode with common cathode taken to -350V rail Input is to one 6BX6 grid,
and GNFB is to the other grid, and this works with 1/2 the THD of the Quad arrangement. While this is acceptable, and better
than original Quad, the use of 2 triodes as I have them in Fig 3 is the very best which can be done, IMHO.
The anode load for V1 triode could be a CCS using MJE350, but the resulting THD reduction and gain increase is marginal.
Maybe better sounding though, so keen DIYers will use the extra CCS.

The V1, V2, V3 as I have the use much more anode current than original 6SH7. Notice Rg for KT88 at 120k, so bias is held down
without effects of positive grid current at idle. The original amps had Rg = 470k = too high. 
Notice my usual critical damping networks needed for unconditional stability, see R8 & C6,  R8 & C7, R28 & C15, R10 & C8.

Old Quad-II OPTs had quite high Rw and high winding losses, and needed critical damping networks. Every old

and recently made tube amplifier which found its way to my bench has needed adjustments of critical damping networks to make the
amps stable for whatever loads can be configured with L, C and R or with no load connected. OPTs in Quad-II-40 are better than
in Quad-II, with less Rw, but the networks I have are necessary for unconditional stability.
I always use more R&C stability networks in all old amps which are often designed by designers who believe shit does not happen.   

Fig 4.


Fig 4 shows where I have retained the pair of 82uF (C11, C12) originally used to make 41uF after 5U4 rectifier.
R18 & R19 make 41r between 5U4 cathode and top C11 to limit peak charge current in 5U4.
Vripple at C11 with 161mAdc total = 8.6Vrms. This is filtered down by L1 4H + C9 & C10 235uF,
so Vripple at OPT anode CT = 24mVrms, much lower than the original amp.
The rest of PSU needs no explanations about its integrity.

Fig 5.


Fig 5 tells most people very little because most have no idea how to interpret the graph curves.
The two solid dark line curves show levels of Po at 1dB below clipping at the two available outputs, com-to-4r0,
and com-to-8r0.
Look along the bottom axis for any speaker load value, say choose 8r0.
Then go vertically up from 8r0, and you intersect the 4r0 outlet curve at 26Watts and 8r0 outlet curve
at 32 Watts.
Notice that the maximum Po for 4r7 is 32 Watts for both curves. But the use of 4r0 outlet will give better
DF, less THD, and better tolerance of Z less than 5r0.

Nearly everyone with say 4ohm speakers will plug the speaker cables to the two terminals labelled "Com" and "4ohm."
With 4r0 and music, drum beats and short duration signal peaks will give maybe 38Watts maximum before
there is any chance of B+ sag and Ek rise which occurs in class AB amps with cathode biasing, due to long time
constants in B+ rail and cathode R&C networks.
There are 10Watts of pure class A available before the amp works in class AB1. All the ppl I know use less
than 1 Watt average from each channel for average SPL of 87dB total with two channels with busy orchestral
or rock or jazz. So each speaker makes 84dB SPL
If speakers are rated for a rather low 87dB/M/W, then the each amp need only make 0.5Watts, and total Po
from both amps = 1.0Watt average of 87dB SPL.

Po both channels, Watts 0.01
2.0 4.0 10.0

Both amps can generate 109dB SPL at 80 Watts total. It is more than enough headroom room for most music
for most people, most of the time, ( but maybe never enough for teenage sons trying to impress their friends ).
If this all seems a poor outcome, get a pair of 100W amps which will make 200Watts total and give 110dB SPL.
But you may damage your hearing, which cannot be fixed. Music is about what sounds pleasurable at SPL
between say 70dB and about 97dB.
In 1960, few ppl ever needed more than 12Watts from a pair of KT66/6L6/807 in triode mode but with
modern less sensitive speakers the 40Watts is needed.  

You can conclude that 4 ohm speakers used on 4ohm outlet can give excessive levels.

What happens if 8 ohm speakers are used at "com" to "8 ohms" outlet? The same SPLs and damping factor

and distortion occur as using 4r0 speakers at com-to-4r.

Most people will find using 8 ohm speakers plugged to com-to-4r will give adequate headroom.

The graph certainly indicates 26W max from each amp so total of 52W so you get about 102dB SPL total.
The amps work mainly in class A, the damping factor is doubled, THD halved and winding losses halved.

I would conclude that you only should use the com-to-4r terminals which should power most speakers between 3

ohms and 12 ohms, which these days means all modern speakers made after 1970. The 4 ohm outlet offers
the most pure class A power possible for most speakers which will please most owners.
The 8ohm outlet probably best suits old speakers of 16r0, and will power ESL63 and other later Quad ESL models.
ESL57 have stricter limits on applied voltage so 4ohm outlet may be best.

Speakers of 4r0 MUST NEVER be plugged in to com-to-8r0 outlets.

This can lead to your amp becoming seriously damaged.

The Quad-II-40 has OPT ratios :-

For com-to-8r0, TR = 22.7 : 1, ZR = 515 : 1, so for 8r0 load, the load for KT88 RLa-a = 4k1, for class AB.

For load of 16r0, RLa-a = 8k2, for mainly pure class A.

For com-to-4r0, TR = 32.0 : 1, so ZR = 1024 : 1, so for 4r0, the load for KT88 RLa-a = 4k1, for class AB.

For load of 8r0, RLa-a = 8k2, for mainly pure class A.

With 4r0 speakers, you cannot obtain more than 10.2 Watts of pure class A.

There is no way you can alter the OPT windings to give an outlet for Com-to-2r0

Please make sure YOU don't get confused by these figures.

When in doubt about the impedance of your speakers, ALWAYS ONLY use the com-to-4r0 amp terminals.

For driving loads of 2r0, you need a speaker matching transformer made by Paul Speltz at


I have been able to alter old Quad-II amps to make almost the same amount of low THD power as the more recent

Quad-II-Forty, see the page at  http://www.turneraudio.com.au/quad2powerampmods.htm

Happy listening.

Back to re-engineered amplifiers

Back to Education and DIY

Back to Index