Modified Quad II pair – thoughts on what’s been done?

[Baracuts]

Hi all,

I recently acquired a pair of Quad II monoblocks from a hi-fi enthusiast. Both have been modified in essentially the same way, and I’d like to get a better understanding of what has been done, why it may have been done, and whether any of it follows established Quad II modification practice.

The original EF86/KT66 layout and transformers remain, but there are a number of changes:
- solid-state rectification in place of the GZ32
- 47µF / 500V F&T capacitor in the HT supply
- additional HT fuse
- reworked KT66 cathode resistor/bypass arrangement
- replacement coupling capacitors mounted under metal straps
- a 0.22µF / 100V capacitor on the main board
- IEC mains, RCA inputs and 7Ω speaker terminals

There are some hand-scratched service/workshop markings on the chassis. I’m unsure exactly how old the F&T capacitor is.
Both amplifiers are working normally.
I’d be very interested in any thoughts on the work: what the likely aims were, whether the implementation looks technically sound, and whether any of it resembles a known or published Quad II modification.

Photos in post below.

Cheers

[Barry]

Hello Julian,

To be able to offer some advice and comments we ideally need to see some clear photos of below the chassis.

Also can you say if the original GZ32 valve is missing?

Replacement of the original and now out-of-date socketry makes eminent sense, as does the HT fuse (as long as it is of a correct rating and safely mounted under the chassis.

As to the "reworked cathode resistor and bypass arrangement", again we would need to know the values in order to compare with those of the original circuit diagram. I will post the circuit diagram in my next post.

Cheers
Barry

[Baracuts]

These show the overall layout and some of the main modifications in more detail.

[Baracuts]

Thanks Barry.
Yes, the original GZ32 is completely absent – rectification is now handled by the four-diode board shown in the photos.
I haven’t been able to confirm the value of the reworked cathode resistor or bypass capacitor yet.
The markings are difficult to read without dismantling anything, which I haven’t done.
I can just about make out “18...” on the resistor, but I don’t want to assume the value without confirming it.

[Barry]

Thanks for the photos Julian.

In order to compare your modified and upgraded Quad II amps with unmodified version, here is the circuit diagram for the original design:

https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEjgeAkHyHvFqWNk47LMPjmk8eaNWcdSXNucb5tfzNrA8l CIibFHWEswGkQeWY8-TXcuAnrH0IZ_g4CUjeNh6_PpyvqBu0-QEYVvFtFKaBHx8yhSDk3-NIMi5I83Pizku8eSWCwYifs9TUF4/s1600/Quad_II_power_amplifier.gif

so we can compare component values.

The two most obvious changes are the replacement of the smoothing capacitors C4 and C6 (each having a capacitance of 16uF) with two separate and more modern versions, and the complete replacement of the GZ32 rectifier valve (V5) with a solid state bridge rectifier using four discrete semiconductor diodes. One of the smoothing capacitors is positioned where V5 used to be, with the second one mounted underneath the chassis. The terminals of these capacitors are at a potential of 330V (C4)and 340V (C6) respectively, so need to be well insulated so they cannot be touched; which they appear to be. I can't see the manufacturer's date on the capacitors, so have no idea if they too need to be renewed.
Again I cannot identify the four diodes used in the bridge rectifier, but as long as they can pass a current of say 200mA, and can withstand a reverse voltage of say 500V, they should be OK.

I have no idea why there are some high power resistors (one of which has heat sinking to the side of the chassis) connected to each end of the inductor of the CLC pi-network, or what their values are. None are used on the original design. The wiring and cursory insulation of them looks sloppy, and certainly needs to be tidied up to make it safe.

The addition of an HT fuse is a good idea (certainly it won't do any harm) as long as it has a rating of 250mA and is a 'slow blow' fuse (marked with a 'T' on one of the end caps).


Regarding the replacement of the original metal-clad DC blocking capacitors (C2 and C3) each having a capacitance of 1uF and 400V rating, with a pair of 470nF metallised PET capacitors of equal or higher voltage rating ought to be fine. (However there will be some small changes to the effective capacitance, since the original capacitors were metallised paper versions hermetically sealed into a metal housing which is clamped to the metal chassis and therefore earthed at 0V. Depending on which end of the axial capacitor is connected to the inner of the of the paper capacitor, and which end is connected to the outer part, there will be some very small additional capacitances, forming a delta network with the main 0.1uF value. The replacement capacitors (two 470nF in parallel are also secured by the same metal clamp, so there will be some parasitic capacitance, but this will probably be of a lower value. I can't comment if all this will make any difference, except to say that Quad kept some minor details of their designs to themselves.)

The cathode resistor and bypass capacitor (R12 and C5) of the original design have values 180R (3W arting) and 25uF (50V rating). Unless the resistor is showing signs of overheating (e.g.scorch marks), it can be left as it is. The capacitor would be best replaced with a modern electrolytic of 22uF or 25uF capacitance with a voltage rating of 63V or higher.

Another change is the capacitor C1, which originally was a 0.1uF, 150V rating metallised paper capacitor, which should be replaced with a metallised polyester or polypropylene equivalent. Why it has been replaced with a 0.22uF 100V device (as clearly shown in the fifth photograph), I have no idea; but I am suspicious.

All the other resistors are Erie 8 or Erie 9 types. These are quite old now and some of the high resistance value types (R1, R2 and R3) could well have drifted out of tolerance (+/- 10%), and may need replacing.

That's about it. I can't quite read the serial number of the amplifier as shown in the second photograph (it looks like it is 82,625), but given that about 90,000 Quad IIs were made, and production stopped around 1970 - 1971, yours are late samples.

[Patrick Dixon]

I have no idea why there are some high power resistors (one of which has heat sinking to the side of the chassis) connected to each end of the inductor of the CLC pi-network, or what their values are. None are used on the original design.

Possibly to reduce the HT voltage since the valve rectifier will drop more volts than the silicon replacement.

[Barry]

Yes - you could well be right; it makes sense. Thanks!

[Barry]

Hello Julian,

Have our comments and observations of your modified Quad II power amplifiers been helpful?

[Baracuts]

Yes, very helpful, thank you. This is exactly the sort of info I was hoping for.

Patrick’s suggestion that the added power resistors may be there to compensate for the lower voltage drop of the silicon rectifier also makes a lot of sense. C1 is very interesting, since the reason for changing it from the original value isn’t obvious.

I’ll also have a closer look at some of the later soldering and wiring around the modifications. A few joints look pretty rough, so it will be worth checking and tidying those when I go through the amps properly.

I mainly want to understand the reasoning behind the modifications before changing anything.

Thanks again!

[Barry]

I have been doing a bit more research on the mods performed on your Quad II amps. I'll come to the change in value of C1 later, but first some general observations regarding the power supply.

The first is that with the original GZ32 rectifier valve fitted, the reservior capacitors were charged up slowly, because the GZ32 had to warm up along with all the other valves. The new solid-state rectifier will charge up the capacitors near instantly. Now it could be that at the moment of switch on, the mains AC cycle could be either at its maximum voltage of 325V, or above the RMS value of 230V (which it is for 50% of the time). If the mains supply reaches its maximum RMS value of 253V (which is allowed by EU regulations), then the peak voltage becomes 358V. Thus to be on the safe side, make sure the voltage rating of the electrolytic capacitors is at least 450V.

Now to the compensation for the reduced voltage drop of the solid state rectifiers. The voltage drop across the GZ32 will be about 25V, whereas across the diodes it is 0.7 - 0.75V.. This means the voltage of the power supply rail will be 340 + 24.3 = 364.3 V. Now the simplest way to drop the voltage would be to fit a series resistor of 24.3/0.15 = 162 Ohm. However this will only work for a near-constant current draw of 150mA, whereas the current drawn by the Quad II will depend on the volume and will be changing all the time. So the modifier has opted to fit two current 'bleed' resistors instead. The calculation of the value of these bleed resistors is a lot more complicated, and requires the calculation of the source resistance of the power supply.

Starting with the power transformer, the resistances of the primary and secondary windings have been measured to be 15 Ohm (for the 240V tapping) and 125 Ohm respectively. (The inter-winding capacitance has also been measured to be 500 - 600pF, but the parallel reactance at 50Hz is so high it can be ignored.) The mains supply source impedance in the EU is 0.4 + j0.25 Ohm. The reactance corresponds to a series inductance of 0.8mH, which again we can ignore. The turns ratio of the amplifier's mains transformer is 620/240 = 2.58, so the overall source impedance is:


(0.4 +15).(2.58)2 + 125 = 227.8 Ohm.

The bleed resistor has to draw current so as to cause the voltage to drop from 365V to 340V. Therefore the bleed current is the difference between 364.3/227.77 = 1.599 A and 340/227.77 = 1.493 A, or 0.106A. Thus the required resistance is 340/0.106 = 3,207 Ohm, or 3.0 kOhm and 200 Ohm in series. The larger resistor will dissipate (0.106)2.(3.103) = 33.7W, and the smaller resistor will dissipate (0.106)2(200) = 2.47W, or a total of 36.2W. That would explain the use of two resistors, the larger of the two being mounted on a heat sink attached to the chassis.

These calculated figures look alarming, but I cannot find any fault in my calculations (neither can Google when asked directly, or when alternative suggested EE sites are examined). It would be interesting to know what value resistors have been fitted to your Quads.


So to summarise: check the reservoir capacitors have a rating of at least 450 V, and if you can manage it, read off the values of the two bleeder resistors. And try to tidy up the wiring of the latter: there is about 350V around them.

[Barry]

Regarding the change of value of the capacitor C1, I have done a little reading around; simply because I couldn't see why it was needed in the first place.

The Quad II uses an unusual and elegant circuit topology, called a cathode-coupled paraphase phase splitter.


Pentodes require a stable screen grid voltage relative to their cathodes to maintain constant gain.
Because the driver stage lacks a standard bypass capacitor to ground on the screen grids, Peter Walker used C! to AC-couple the screen grids of of both EF86 valves together.
Since the two valves handle signals that are 180o out of phase, the signal variations at the screen grids are equal and opposite. C1 forces them to cancel each other out establishing a virtual AC signal ground right between the two screen grids without actually tying them to the physical chassis ground.


Thus C1 is an AC coupling capacitor connected between the screen grids of V1 and V2, to ensure the input to each EF86 pentode is always in phase. V1 and V2 form what is known as a paraphase driver. C1 forces an AC connection between them to restore AC balance and counteract any dynamic asymmetries between the two valves.

(Apologies if I'm stating what you already know, but it is all new to me, despite having owned two sets of Quad II amps in the past!)

Increasing the value from 0.1uF to 0.22uF is not a good idea for the following reasons:

1 Alteration of the phase splitter's AC balance

Increasing the value to 0.22uF extends this balancing action to a lower frequency. But because 0.1uF is already fully effective across the entire audible spectrum, in reality the performance gain in AC balance due to increasing the value of C1 is negligible.


2 Doubles the time constant with slower recovery

Doubling the capacitance doubles the RC time constant of the screen grid circuit. If the amplifier experiences a heavy low-frequency transient, severe overload or clipping, the larger 0.22uF capacitor will take twice as long to charge and discharge, which can result in prolonged transient blocking or momentary bias shifts after a heavy signal peak.


3 Impact on low-frequency phase margin
The Quad II uses a substantial amount of global negative feedback (around 18 - 20dB). Modifying reactive component values in the driver stage alters the low-frequency phase shifts within the feedback loop. A larger time constant due to increasing the value of C1, shifts the phase characteristic at subsonic frequencies which can reduce the amplifier's stability margin, and at worse can make it prone to low frequency instability or "motorboating" (so-called because the sound is like that of an outboard motor)

Increasing the value of C1 is usually made by hobbyists, who misunderstand the purpose of C1, thinking it to be a standard inter-stage audio coupling capacitor (such as C2 and C3). Increasing the value of a coupling capacitor is often used to increase the usable audio bandwidth.

It is recommended that the original Hunts capacitor, which were notoriously unstable, and became 'leaky' to DC, is replaced with a modern 0.1uF polypropylene film capacitor (with a rating of 160V to 250V.).

[Baracuts]

Thanks Barry,

Your explanation of the added power resistors makes sense, and it’s very useful to have a possible reason for why they were added. I’ll check their actual values, the remaining HT capacitor, and compare both monoblocks. I’ll also take a closer look at the wiring and soldering in that area, especially given the voltages involved.

The explanation of C1 is also very useful. I did'nt know exactly how it functions in the circuit, so your explanation makes the change to 0.22µF much easier to understand. I’ll confirm that the 0.22µF component is indeed C1, and if so, changing it back to the original 0.1µF value with a suitable modern capacitor seems right.

Thanks again for taking the time to go through the circuit and work all of this out in such detail!

Cheers,
Julian

[Barry]

Hello Julian,

I've been thinking about your modified Quad II amps. First of all the idea that there are two bleed resistors effectively connected in parallel about the choke L1 is completely wrong. There can only be one resistor, which in your case appears to be made up of two resistors in series. I have corrected my original post.

My mistake was caused by my misinterpreting the photograph showing two resistors connected to the choke. I don't believe the two are attached to each terminal of the choke. If you could carefully examine the wiring in the area around the choke, and try to describe it I would be greatful.

Also I notice the new capacitors have a capacitance nearly three times that of their replacements. This means the inrush current during charging will be much higher, but if the rectifier diodes are 1N4007 with a 1A forward current rating that should be OK.

I'm out of the country at the moment, so can only do more calculations of the ripple current when I get back in about a week's time.

One last thing: the shunt bleeder resistors form a CR network with one of the capacitors (the one upstream from the choke). The time constant of this CR combination is (45. 10-6).3200 = 145.10-3s, or 145ms. This is a sufficiently long discharge time given the charge will be topped up every 10ms (that of a rectified 50Hz sine wave).