Tuesday, September 05, 2017

An exploration of the practical issues and benefits of two of the Quad 405-1 modifications suggested in Berndt Luwdig's paper on the Quad 405.

These modifications are not for the inexperienced or those without adequate test equipment, and are not under consideration for inclusion in any Dada Electronics Kit. The D13 modification is included in our HE 405 board. This article is provided for information only. Use at own risk.

This article is an exploration of the practical issues and benefits of two of the Quad 405-1 modifications suggested in Berndt Luwdig's paper on the Quad 405.

The first of these is the addition of D13, as found in the 405-2 and later Quad current dumping amplifiers: 606, 306, 707, 909. The Quad service manual refers to this as 'correcting the response at 20KHz'. For brevity I will call this the 'D13 modification'.

The second is an alteration to the current limiting circuit to bring it more into line with what the networks N1, N2 do in a 405-2 or later current dumping amplifier. It consists of the addition of two 36V zeners and some resistor value changes around Tr5 and Tr6. It raises the current limits closer to the 405-2's, which enhances high-power performance into 4-ohm (and
2-ohm?) loads. For brevity I will call this the '36VZ modification'.

1. D13 modification. This is very straightforward to implement, just requiring removal of D5, cutting the track to the base of Tr9, and replacing
D5 with a 1N4003 diode from its anode connection point directly to Tr9 base and adding another 1N4003 from there to the original cathode connection point for D5, both now located on the track side of the PCB, both oriented 'cathode down' in the circuit. D6 should also be replaced with 1N4003. All this is better described in the Ludwig
paper: interested readers should certainly refer to that, rather than taking this bald description as an adequate specification.

However I had considerable difficulty with this modification. After much time and trouble this was eventually tracked down to two problems:

1.1 A faulty triac in the clamp circuit. Somehow this was behaving like a shunt capacitor presenting very low impedance at 20KHz, so when I tested with a large enough signal I would blow both power transistors and rail fuses. I was testing into an instrument with 100k input impedance at this point, not an 8 ohm dummy speaker load, so the current limiting didn't save me. I eventually found this fault by powering off the 405 and injecting the test signal into the speaker terminals, and measuring it, which showed the low impedance and low-pass filter/shunt capacitance behaviour.

1.2 A major slew-rate problem from about 4KHz upwards. This was eventually tracked down to Tr5 and Tr6. The 36VZ current-limit modification had already been applied, so it is uncertain whether that modification brought out the behaviour in these two transistors, which were the original BC214s, or whether they were already faulty. In any case they seem to have had some kind of excess capacitance issue, which was resolved by replacing them both with BC560. (The 36VZ modification is therefore not to blame.) So I would recommend replacing both these transistors before attempting the D13 modification.

It would be difficult or maybe impossible to even spot this problem without a signal generator and oscilloscope and/or distortion analyzer, and unless it is absent or resolved there is no point whatsoever in the
D13 modification, as the slewing distortion will completely mask it.

When complete, the D13 modification improved the THD of the amplifer as

1KHz: 0.0225% became 0.0062%
10KHz: 0.0217% became 0.0095%
20KHz: 0.028% became 0.0095%

IMD (DIN method, 250HZ and 8KHz in 4::1 ratio, 4VRMS output level):
0.0125% became 0.0077%

SNR: Unchanged.

Measured with Tektronix SG505 and SG505/01 oscillators and AA501/01 distortion analyzer at 4VRMS output level into an 8 ohm dummy load, i.e.
2 watts. (I rarely test at full power.)

It should also be noted that with the D13 modification, the upper dumper
Tr9 now comes on a bit sooner than Tr10, as you can intuit by measuring the standing b/e voltages on both; and the signal at the base of Tr10 is no longer symmetrical. The former doesn't matter much, as it only imposes a tiny extra load on Tr9 at very low signal voltages around
0.5VRMS: the latter doesn't matter at all: you are going to see all kinds of strange signals inside any feedback amplifier anyway.

2. 36VZ modification. This is very straightforward, as described in Ludwig.
I adopted the solution of scaling up all the resistors as described there, rather than using 2W resistors. I had no trouble with this modification beyond getting the diodes the right way around. The lower zener/resistor combination is fitted near the fuses and should be insulated from contact with them.

Esmond Pitt
Dada Electronics Australia
Copyright (c) Esmond Pitt, 2017. All rights reserved.


Post a Comment

Links to this post:

Create a Link

<< Home