We decided to implement the 405-II current-limiters on request only. We have built the limiters with normal components. If you insist on having the limiters on your boards, ask for a quotation.
Why don’t we implement the limiter in our High-end boards?
It affects the sound quality, not only because of the topology of the limiter itself but also the way Quad implement it. In a 405, but still in the 909, the current trough the sensing resistor modulates the voltage of the positive power rail of the class A stage and the Opamp.
The Opamp is the least of our worries, because it has a high power supply ripple rejection ratio.
The modulation is done by a dirty half wave class C signal from the upper dumper transistor. In the original 1975 days this was not such a problem, noise, crosstalk and other distortions masked this problem. But in search for the highest quality, the limiter has to go. Also the current sensing resistors can now be short circuited.
The right way to implement a limiter with this topology: (This is a 405-2 example, the 405-1 is less complex)
The limiter in the negative part of the output circuit can stay as it is, it is part of the dumper section, the class A stage will compensate for errors via the current dumping principle!
Input limit is the role of the Opamp.
In the first production version of the 405 the power supply voltage of the Opamp was plus and minus 12 V. Also due to the DC stabilisation there was a DC offset at the output of the Opamp of 1.7 V. This means the theoretical input voltage of the Opamp stage is 486mV max, conclusion; with 500mV input no 100W in 8 ohm undistorted output power was possible!
The PSU voltage went up to 15V and the DC offset was lowered to 1.2V, still the headroom is only 2.3dB for undistorted output. So a popular en reasonable update/mod is raising the PSU voltage and selecting the proper Opamp. An OPA604 running on 24V gives a healthy 6dB plus headroom in the OpAmp stage.
The rest? Calculation shows that the maximum theoretical input voltage needed to let the dumper stage clip is only 1,9 dB above the nominal input, so the figure of 2,3dB implemented by Quad looks sufficient. So be careful with the input level, there is only a small overload margin.
The output can be limited to 20V by inserting a link in the OpAmp circuit; this was done to protect the ESL57 from overloading. We do not advice this option, this is not a soft clipping device, and feeding 20V square waves in to a fragile ESL57 is not a good thing to do. Buy a refurbished 303 or Quad II to drive the ESL57. Clipping of a 303 and an II is still possible, but only with high input levels.
DC detection and protection.
A low pass filter is placed in parallel with the output, when the DC voltage is about 6 volt, the Diac opens and the Triac will short circuit the output. Due to the characteristic of the Diac the output stays in short circuit mode, hence the name ”Clamp circuit”. It was the intention that the fuses would be blown, in practice also the output transistors will be thermically killed! But there was a problem anyway with the circuit. Low voltage Diac’s are hard to get in these days.
Dada Electronics' own DC protection circuit is relay based and has a switch on delay function built in to it. See Ebay-object 200243187533.