Swinik VS Quad -- where is your Bridge, gentlemen?

[Wavebourn]

Here is the Quad 405-2 schematics (thanks for the link!),

http://www.dc-daylight.ltd.uk/Valve-...-Evolution.pdf

here is the Swinik:



Main difference: they used 4 transistors in a voltage amplifier, I used only 2 as fastest as was possible that time. And mine worked with more current than their. Also, mine had 2 times less of a voltage amplification. The result is, I could use a global feedback on the whole frequency band, they could not, they had to fight against parasitics EVEN WITHOUT a global feedback, while I used only one RC network (not shown on the schematics). Also, I limited gain of the OpAmp by a third feedback on the whole band except lower end (and DC, corresspondently, for less DC offset on output), for better control of transcient processes. Also, I used better impedance match for less POWER amplification of the whole amplifier taking more CURRENT from the OpAmp. As the result, Swinik was faster and had LESS distortions in the hearable frequency band, despite I used Soviet transistors and OpAmp that only were available for me that time. Their quazi - complementary output with additional phase asymmetry on high end of spectrum when complementary pairs were already available...

And I did not discover famous "Bridge" that gentlemen criticizing Swinik mentioned... Just the same resistor (R1 in Swinik, R38 in Quad) for transfer function approximation that was obvious long before 1975... However, I don't know if they could patent an approximation instead of "Current Dumping", I believe...

[latala]

The swink is a very basic idea that has been preceeded to an even better level by such amplifiers as the crown dc 300/phase linear 400/700b as well as many other high power pa amplifiers
the clasiffication would be for the swink A/C where the crown would be AB/C
As for the quad if you visit quad world you will find the (bridge)
it may not be obvious but carefull study will reveal it !
I have tested the Quad and under steady state sine waves the distortion was so low as not to be measurable with my equipment "Radford/sugden" this not the same for the others i have mentioned I wish it was so easy we would all be making Swinks
regards Trev

[Wavebourn]

Thank you Latala,
the link on the Quad World is above,
will you be so kind to show me The Bridge that exists in Quad, but don't exist in Swinik?

[latala]

With respect
The quad bridge has been very well documented over the last 25 years
this is a Gm correcting device! I suggest that you look it up I am reluctant to do so for you if you dont study you wont learn. I have here amongst my papers the origional treatise on the 405 and the current dumping concept but it will take me to long to sort it out for you
If you look on the forum you will find even more comments re the amp
I am not a fan of the 405 myself I prefer more conventional amplifiers
I will if I can find the time send you a copy of the origional article
Regards Trev

[ilimzn]

Last schematic in your pdf link: R38, L2, C8, (R20 || R21). If you look carefully you will find the same components in every schematic.

[Wavebourn]

Originally posted by ilimzn
Last schematic in your pdf link: R38, L2, C8, (R20 || R21). If you look carefully you will find the same components in every schematic.

I also choosed R1 resistor and all feedbacks according to load impedance and don't see something extraordinary that may be patented. Just an understanding what and why do you do something...

[ilimzn]

Originally posted by Wavebourn


I also choosed R1 resistor and all feedbacks according to load impedance and don't see something extraordinary that may be patented. Just an understanding what and why do you do something...

The bridge components are not selected for load impedance. There is a long thread called 'Death of Gainclone' here, IIRC there is a lonk to the original documents and a longer treatise called 'error feedforward' in it. Use the Search function...

[Wavebourn]

Originally posted by ilimzn


The bridge components are not selected for load impedance. There is a long thread called 'Death of Gainclone' here, IIRC there is a lonk to the original documents and a longer treatise called 'error feedforward' in it. Use the Search function...

Thank you ilimzn. I understand what do you mean; transfer function of the C - class stage as if don't depend on a load impedance. Unfortunately it is not true.
If to use imagination it is possible to find a Gainclone jumping from the London bridge, as well as Anna Karenina jumping under the train of dumped current. But if to look from another side, approximation means several functions that continue each other as smoothly as possible. For example, A class amplifier with gain G1 and output resistance R1, C class symmetrical emitter follower with gain G2 and output impedance R2, and symmetrical source followers with output impedance R3. Despite of such smoothness additional global feedback helps to reduce nonlineariries, as well as faster amplifiers help to minimize their impact on audible frequencies.

--Anatoliy

[edl]

Dear Wavebourn,

This design looks me so exciting.
Have you got full schema? I would look into that in my simulator.

Regards,

[Wavebourn]

Originally posted by edl
Dear Wavebourn,

This design looks me so exciting.
Have you got full schema? I would look into that in my simulator.

Regards,

What do you see on the picture, IS the "full schema" of the amplifier as it was built back in 1977. However, there was also a RC-network to stop parasitics and power regulators for OpAmp.
In the original design A - class stage was loaded by couple of 390 Ohm/20W resistors in series, it's own gain was 5. Today I would use a CCS on a fast PNP transistor instead of resistors with a positive feedback through a capacitor that increases distortions. Input resistor was 10K, global feedback resistor was 300K, I don't remember values of resistors across OpAmp and OpAmp+Class A amp, I played with them using pots to obtain results I liked the best. I remember I used 1 MOhm pots with 0.1 MF caps in series. But I do not remember final values of resistors.
You are free to play with details to obtain your optimal results, they will be yours.

Or wait untill I decide to disclose the current 3-step approximation version that uses vacuum tubes instead of OpAmp and additional FETs on output.