Quad 405-2 limiter circuit

Status
Not open for further replies.
Jump to Latest
#21
I've just looked at the 707 circuit, and although the output stage isn't biased as far into class-C as the old 405 originally was, there's no standing current flowing and I wouldn't call the output class-B (there's one diode drop between the bases of the output push-pull emitter follower to provide the standing bias).

Even with current-dumping it makes sense to design the circuit to ease the requirements on the dumpers as much as possible, while still keeping the output stage in cut-off under quiescent conditions.
 
#22
RogerGustavsson said:
janneman!

The answer to your question is yes. Keith Snook's frontend……..

Roger I hope you mean yes it does keep the current dumping principle 🙂

What I was trying to show is that the current dumping theory was well ahead of commercial component technology at the time ( Tr3 Tr4 also Tr2 ? ) and that when C8 can be correctly placed in the bridge circuit the theory works in practice and works well.

The "frontend" (IC1) and even my modification to it, is NOT part of the current dumping stage apart from the fact that it sets the d.c. offset at the speakers through the CD stage.

I have had a lot of feed back (no feed forward or negative) about the 405 modification which eventually got me to removed the current limiting after fitting "bigger" transistors for Tr7-Tr10 and applying Tiberiu Vicols current source for Tr1. I have kept the crowbar and fuses but hope not to use either.

There was a slight improvement (even lower distortion!) but not a significant sonic change like that of moving C8. I have only lived with the amp in this condition for about a year now so it’s too soon to comment, however I may add it to the 405 mod article in the near future.

Davidsrb (and all)

Have you considered that if the Dumpers were biased in class B or even nearer to class A then they would dominate the contribution of the Tr1-Tr7 class A stage and would introduce distortion at crossover which would be very noticeable at low level unless excessive feedback were applied.

Transistors (power or otherwise) have much lower than their published gains at low collector currents and the traditional idea of biasing to Class B only slightly improves the gain whereas class A……

Now consider the L2, C8, R38 and R20//R21 "bridge" that combines the 405 dumper stage and class A stage, At low level, when the dumpers are not conducting and not contributing distortion, the amplifier is predominantly class A with output impedance lower than R38 due to traditional feedback.

The dumpers switch in at a higher level but do not switch at the cross-over point - they switch either side of cross-over helping the class A stage deliver (dump) the higher current required for the higher output voltage across the load.

The fact that L2 has a low impedance at audio frequencies and helps reduce switching spikes and C8 is the dominant pole for the class A stage while both balance the bridge at all frequencies is just clever design.

It is possible that in a current dumping amplifier stage like the QUAD 405 that L2 and C8 are the only reactive components required, C7 (and R17) can go if Tr2 is replaced, use the Tiberiu Vicol current source for Tr1 and similar higher power source for R30/R31/C10 and L1 and L3 ? OK leave them.

So does your QUAD 405 need a raft of new capacitors fitted or just modifing to it's full potential, My current (present) 405 makes Elkie Brooks and Tom Waits sound like they are chewing rocks in the same room as me and recordings of my children same but substitute bubble gum for rocks.

regards

Keith (Snook)
Click to expand...
 
#23
dcd said:
[snip]Have you considered that if the Dumpers were biased in class B or even nearer to class A then they would dominate the contribution of the Tr1-Tr7 class A stage and would introduce distortion at crossover which would be very noticeable at low level unless excessive feedback were applied.[snip]Keith (Snook)
Click to expand...


Keith,

I dont agree to this. The principle of the CD is that whatever the contribution of the dumper stage is, the class A stage automagically adjusts itself to provide whatever is required for (ideally) undistorted output. Biasing the CD stage in AB or A wouldn't make any difference here, but it would also not do any good IN THEORY, except maybe that it would ease the load on the class A stage and thefore IN PRACTISE might give slightly better performance.

Jan Didden
 
#24
Jan

I am glad you say you do not agree with me as this may get others considering - the situation -

What is nice, is that you then go on to enforce my argument (as in mathematics, not fighting) by stating that….

"the class A stage automagically adjusts itself to provide whatever is required for (ideally) undistorted output"

As I said - when the dumpers are off at low output levels** they contribute nothing and so the class A stage is dominant. However if the output stage is biased on too much the emitter followers will then provide a low output impedance for low level signals which will dominate the output because the impedance of the class A is much higher and it’s effect is swamped.

So as I see it increasing the bias of the dumpers makes the amplifier more like a traditional design but prevents the class A stage from being effective at low output levels** where it normally does most of it’s work "filling in" for the dumpers.

That’s the only point I was making regarding the proposed benefit of increasing the bias of the dumpers. There appears to be general agreement that the bridge provides feedback obtained from both the class A and dumper stage but rather than keep quoting old articles let’s think differently about this.

If you really want something to disagree about…..

I think that when correctly implemented (my 405 mod, 306, 606, 707…) the bridge provides much less correction signal than normal feedback would and it is this that makes the design unique ….. The majority of the output current is not derived from a previously fed back signal because only deviations from the class A reproduction are corrected for and once in action the dumpers are simply followers.

**I should perhaps confirm that when I say low levels I refer to small output voltage deviations either side of zero, they could be part of a larger excursion not just low level signals.


Automagically – automatic and magic that’s just what the QUAD 405 amplifier is.

Keith
 
#25
When the 405 was released there was a lot of discussion about whether it worked as described and I think ultimately that it does. I think there is still a lot of misunderstanding of this elegant design and I think that it has a lot to offer. The 405 was designed without the aid of any simulation. I was wondering if anyone had simulated the 405 or later current dumping amplifiers. I have also often wondered if further de-biasing the output stage and increasing the power delivered by the class A amp would be worth investigating

Stuart
 
#26
dcd said:
Jan

I am glad you say you do not agree with me as this may get others considering - the situation -

What is nice, is that you then go on to enforce my argument (as in mathematics, not fighting) by stating that….

"the class A stage automagically adjusts itself to provide whatever is required for (ideally) undistorted output"

As I said - when the dumpers are off at low output levels** they contribute nothing and so the class A stage is dominant. However if the output stage is biased on too much the emitter followers will then provide a low output impedance for low level signals which will dominate the output because the impedance of the class A is much higher and it’s effect is swamped.

So as I see it increasing the bias of the dumpers makes the amplifier more like a traditional design but prevents the class A stage from being effective at low output levels** where it normally does most of it’s work "filling in" for the dumpers.

That’s the only point I was making regarding the proposed benefit of increasing the bias of the dumpers. There appears to be general agreement that the bridge provides feedback obtained from both the class A and dumper stage but rather than keep quoting old articles let’s think differently about this.

If you really want something to disagree about…..

I think that when correctly implemented (my 405 mod, 306, 606, 707…) the bridge provides much less correction signal than normal feedback would and it is this that makes the design unique ….. The majority of the output current is not derived from a previously fed back signal because only deviations from the class A reproduction are corrected for and once in action the dumpers are simply followers.

**I should perhaps confirm that when I say low levels I refer to small output voltage deviations either side of zero, they could be part of a larger excursion not just low level signals.


Automagically – automatic and magic that’s just what the QUAD 405 amplifier is.

Keith
Click to expand...


Sorry, I still don't agree. I think your arguments are muddled. The basic premisse of CD is that it is NOT feedback in the normal sense in that it trades gain for correction. With traditional feedback you can only decrease the non-linearityby the inverse of the gain you trow away (the loop gain).

With CD you precisely meter the amount of corrective feedback so that it JUST compensates for the non-linearity of the dumpers. Once you have that working, it does not matter at all what the dumpers do. If they are totally off, the corrective feedback is JUST enough to let the class A amp EXACTLY fill in what is missing - and in this case (dumpers off) that's the whole signal.

If the dumpers DO something, the corrective feedback is JUST enough to let the class A fill in, again, what is missing. In this set up it doesn't matter whether the dumpers are completely off (as in class B at zero-crossing of the signal) or not, as in class AB. The point is that the feedback ratio adjusts itself to the non-linearity, while of course in traditional negative feedback apps the feedback ratio is constant.

It is PRECISELY the crux of CD that the dumper action gets corrected 100% exact, in theory. That's why I said that in practise, biasing the dumpers in AB may make a difference; the class A amp is not ideal, and easing it's life a bit is bound to be advantageous. But, in theory, the dumper non-linearity is exactly compensated.

Jan Didden
 
#27
Stuart:

I have modelled and analysed the 405, in fact I believe the subject was mentioned some time ago when you called me to buy a transformer case…..

A few things become clear (to a muddled old me at least) when you can add and remove components quickly on a computer to see what happens, although I also implemented many changes to check theory in practice and confirm the model information.

I did not keep records of the analysis runs I made in 2001/2003 so I have re-done a few on the modified schematic which i cannot attach here because it's too large - 9 pages 700kB - file can be found at:

http://www.dc-daylight.ltd.uk/Valve-Audio-Interest/QUAD/QUAD-405 Modification/QUAD-405-mods.html

The voltage waveforms don’t give much away but the current waveforms at 1Vout I think are interesting. Note Tr10 comes into play a lot earlier than Tr9. sorry about the colours.


Jan

Perhaps I am muddled but the signals produced by my modified 405 are definitely not.

All

I have seen a few references to complimentary output for the 405 to "balance" the dumpers, note the Tr8/Tr10 combination or similar is required because R30+R31 cannot provide the base current required by a single PNP like MJ15004 in this position.


Keith
 
#29
Jan

the frustrating thing for me is that we are in general agreement about most points.

I am happy for us to agree to disagree especially if it provokes more input on the subject which of course we may both agree or disagree with ;-)

Best regards

Keith
 
#30
405man said:
When the 405 was released there was a lot of discussion about whether it worked as described and I think ultimately that it does. I think there is still a lot of misunderstanding of this elegant design and I think that it has a lot to offer. The 405 was designed without the aid of any simulation. I was wondering if anyone had simulated the 405 or later current dumping amplifiers. I have also often wondered if further de-biasing the output stage and increasing the power delivered by the class A amp would be worth investigating

Stuart
Click to expand...


Hello Stuart,

Currently I'm working to implement Nelson Pass Aleph current source into QUAD 405 class A stage. First results are very impressive - more power, dinamics ... 😎

I'll share schematic soon.

regards, tibi
 
#32
Hi everyone. I know this is an old thread but I have modded my Quad 405 and think I have a problem and I would very much appreciate any help given.

It does make music, but I am unsure of the quality as I daren't connect to decent speakers.

I have done the dc daylight mods, but now I think R31 is overheating. It gets very hot, very quickly and smells like it's going to give up! It is like this on both channels.

Please help, anyone.

Regards, Lee.
 
#33
Lee

R31 and R30 are in series and so should have exactly the same current flowing through them - they should both get hot as they are dissipating about 1.1W - my mods do not affect this area of the circuit.

If R31 is significantly hotter than R30 then you have a problem with C10 which is passing d.c. - is it connected correctly ?? please tell me you did NOT fit a bipolar cap for C10.

Regards

Keith

DCD
 
#34
dcd said:

If R31 is significantly hotter than R30 then you have a problem with C10 which is passing d.c. - is it connected correctly ?? please tell me you did NOT fit a bipolar cap for C10.

Regards

Keith

DCD
Click to expand...

Thanks for the reply.

Nope, no bipolar. It is a correctly fitted 100uF 100v Panasonic FC.

Should I also have upgraded tr1? Your M12368 to mod-3 schem shows 2sa1085 but this is not mentioned in your upgrade text.

I read correct voltages on both sides of r31 but insure of current.

Regards, Lee.
 
Status
Not open for further replies.
Top Bottom