Quad 303

[nutteronthebus]

Hi thanks to Andy (Diabolical Artificer ) I am now a proud owner of a Quad 303
this matches the 33 and FM3 that I upgraded in June to the DADA style,
new BC550 in place of BC109 and new caps they now drive a set of Celef speakers and sounds great . Has anyone else done the DADA upgrade to there Quad 33 preamp and how is it sounding

[Craig Sawyers]

Well, I have a DaDa kit for the 33. I've done the power supply, which increases the DC from 12V to 16V, and some resistors have to be changed to ensure transistor bias points are right.

But changing the transistors to BC550 has less clear benefit. The noise factor of a BC109 is 4dB for a 500 ohm source resistor at 100uA collector current. Since the resistance of a typical moving magnet cartridge is about 500 ohms, the critical input transistor is generating about the same noise as the cartridge, and is not really significant. Changing to the BC550 elsewhere has an entirely insignificant effect on noise.

Don't get me wrong - I'm a real fan of DaDa, and have bought QUAD boards and knobs from them.

[Craig Sawyers]

Tried to edit, but my 30 minute edit limit had expired.

Amended second paragraph:

But changing the transistors to BC550 has less clear benefit. The noise factor of a BC109 is, at 500 ohms source and 200uA collector current:

4dB at 120Hz
2dB at 1kHz
2dB at 10kHz

Since the resistance of a typical moving magnet cartridge is about 500 ohms, the critical input transistor is generating less than the noise as the cartridge, and hence is insignificant. Changing to the BC550 elsewhere has an even less than insignificant (!) effect on noise.

[Trevor]

i think you are over-simplifying the cartridge the input will see impedance not resistance and this will be a lot higher because of the inductive element of the aforementioned cartridge
Trev

[nutteronthebus]

I upgraded the power supply and changed the BC109 and caps + resistors for low noise ones the parts I used were from RS (I have 250 BC550 min order for £8.50 ) this cost a lot less than the DADA kit (I have 3 quad 33 in various rooms along with both my sons who have 33/405 combos ) Also a mate who has 33/303 combo. With the 16v power supply giving more headroom they sound great

Dave

[Tim]

I have done a couple of these(and 405's) and the faults have nearly always been caused by those red "Bakelite" electrolytic capacitors going low value/open.

[evingar]

Originally Posted by Tim

I have done a couple of these(and 405's) and the faults have nearly always been caused by those red "Bakelite" electrolytic capacitors going low value/open.

Yep, that was my experience as well.

I bought a tatty example off ebay and one channel was low and distorted. A recap sorted it out nicely - the large smoothers measured out fine ESR wise so they stayed in.
It has been working perfectly as a sound system on my PC for years.

[Craig Sawyers]

Originally Posted by Trevor

i think you are over-simplifying the cartridge the input will see impedance not resistance and this will be a lot higher because of the inductive element of the aforementioned cartridge
Trev

Well sure - if you are doing the complete analysis you have to take account of the 500mH inductance of the cartridge, which progressively isolates the 47k load resistor and leaves that is the dominant noise source.
Assuming current noise is low enough (you have to try quite hard for it to dominate), all these effects are common whatever the inherent noise of the amp is.

As a good example, the 5532 opamp has an input noise voltage of 5nV/rootHz, which is the noise of a 1500 ohm resistor. Even so, when loaded with an M75ED Shure cartridge, and after RIAA EQ and 30x gain at 1kHz, but not weighted, gives a measured -88dBu at the output, which is very quiet indeed.

Now the BC109 has an effective noise resistance of around 300 ohms - so 1/5 that of the 5532 - which is amply good enough.

[Radio Wrangler]

That may be right, but you also have to include the current noise contribution in the effective impedance presented to the opamp or transistor. Taking the voltage noise component alone doesn't always tell the full story. If voltage noise happens to be dominant, then the result will be good, but you have to check both to make sure the simplification is valid. Otherwise, you can find more noise than expected when you build something.

The NE5534 opamp family are fine for moving magnet RIAA stages and produce less noise than the cartridge and other components around them. I've used them for many years in a circuit published by Peter Baxendall in Wireless World.

Similarly, the BC109 is fine and I've used them in another RIAA stage from Wireless World, this time by Hugh Walker. Theoretically,the BC109 should be quieter in this range of impedances, but as neither active device is at all dominant, the end results show little difference.

Signetics are long-gone, as I suppose is the fab where Mullard once made BC109s so any current devices must be from other manufacturers. Noise behaviour is affected by the diffusion processing, not just the maskwork so there has to be some doubt over whether new NE5534/NE5532 and BC109 behave like the originals and whether current manufacturers check awkward specs like noise behaviour. I might be inclined to go for one of Analog Device's offerings, not out of audiophile quackery but just to have a bit more confidence that I knew what I was getting and that I knew who the specs were guaranteed by. Probably it wouldn't out-perform a 5534/5532 to any noticeable extent but the probability of a dodgy copy of the 5534/5532 would be avoided.

David

[Craig Sawyers]

Originally Posted by Radio Wrangler

That may be right, but you also have to include the current noise contribution in the effective impedance presented to the opamp or transistor. Taking the voltage noise component alone doesn't always tell the full story. If voltage noise happens to be dominant, then the result will be good, but you have to check both to make sure the simplification is valid. Otherwise, you can find more noise than expected when you build something.

Oh absolutely. The worst case noise resistance is 47k, which has a a Johnson noise of 27nV/rootHz, you would need 0.5pA/root Hz of noise current to equal that value. Doing the sums (I wrote a speadsheet to do this) the input current noise of a transistor with an rbb' of 600 ohms, a beta of 200 and a collector current of 0.2mA (so a BC109) gives an input noise current of 0.6pa/rootHz. So the Johnson noise is about the same as the current noise through the same resistor.

But only at high frequencies of course once the 500mH cartridge inductance has cut in big time - so around 10kHz. And that is in the region where there is a 20dB cut via RIAA.

The NE5534 opamp family are fine for moving magnet RIAA stages and produce less noise than the cartridge and other components around them. I've used them for many years in a circuit published by Peter Baxendall in Wireless World.

Similarly, the BC109 is fine and I've used them in another RIAA stage from Wireless World, this time by Hugh Walker. Theoretically,the BC109 should be quieter in this range of impedances, but as neither active device is at all dominant, the end results show little difference.

Signetics are long-gone, as I suppose is the fab where Mullard once made BC109s so any current devices must be from other manufacturers. Noise behaviour is affected by the diffusion processing, not just the maskwork so there has to be some doubt over whether new NE5534/NE5532 and BC109 behave like the originals and whether current manufacturers check awkward specs like noise behaviour. I might be inclined to go for one of Analog Device's offerings, not out of audiophile quackery but just to have a bit more confidence that I knew what I was getting and that I knew who the specs were guaranteed by. Probably it wouldn't out-perform a 5534/5532 to any noticeable extent but the probability of a dodgy copy of the 5534/5532 would be avoided.

I agree with all of that. If I were making a choice of an opamp to do the job perfectly (at a price) the AD797 would be the way to go.

Craig

[Radio Wrangler]

Noise issues get more complex when reactive source impedances get involved (achtung! bad pun alert!)

The RF world has had to handle this issue. Reactive impedances are free of thermal noise, but the noise behaviour of an active device is affected by them, so RF devices (well-documented ones at least!) have Smith charts of complex source impedance with contours of constant noise figure plotted on them.

David