Quad 303 : Upgrade output stage to be fully complementary

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi!

I've upgraded the caps and transistors in my 303.

Currently I'm using my 303 for the tweeters (Scan Speak D2905/9900) in my 2-way active system. Actually I think that the 303 has a softer more relaxed sound than my Quad 405,2. Above all I insist on soft sounding tweeters, but I must admit that the 303 is not as detailed as the 405,2.

I'm not and expert, but I know that the quad 303 uses old output technology. At:
http://www.net-audio.co.uk/quad303upgrade.html

I read:
Quote
Additional Modifications for the experienced.
The 303 can also be upgraded further by modifying the output stage to be fully complementary instead of the original quasi complementary design. Quasi complementary designs use two NPN output devices with one NPN and one PNP driver to produce a push pull output stage...
UnQuote

303 standard (left) and 303 Modified (right):
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.


Has anybody outthere tried this?

Or do you have comments?

Regards, Ask
 
Administrator
Joined 2004
Paid Member
Mods

Hi Ask,
I normally wouldn't do this unless I had a channel blown, but why not? You may hear more of an improvement with the current source mod. You don't need to use a constant current diode, it's just easier. You can construct your own constant current source as well.
I have not tried either of these mods but they look sound.
-Chris
 
Administrator
Joined 2004
Paid Member
amp sounds bad

Hi analog_sa,
I agree with you that the 303 isn't the best amp I've heard. It would seem an excellent candidate for improvement. The power supply is required for stability and noise reasons. Other brands have done this as well to good effect.
I would say that it's just an old design that could use updating. I think that Ask is on the right track and the 303 will sound much better at the end. Ask will gain some experience, and that's always worthwhile.
-Chris
 
Quad 303/405,2 -> D2905/9900

hmmmmm....

I think that my upgraded Quad 303/405,2 can drive my D2905/9900 fine!

* I agree that a Quad 44 with TL71 opamps has bad treble
* I agree that an upgraded Quad 33 has even worse treble
* I agree that a NON-upgraded 303/405 has very bad treble performance

My source is an Arcam Delta 270 with LClock 2 and AD8620 opamps - I tried to connect a Marantz 6000 KI Signature which has softer treble on my system. Since I can tell the detailed difference in CD-Players, this test leads to the conclusion that my 303 is not an overshadowing bottleneck for the tweeter performance. Besides this I've heard many other systems...

Maybe analog_sa never heard an fully upgraded trimed Quad 303+44?

Regards, Ask
 
I think it was Stan Curtis of Cambridge fame who
said complementarys are as similar as a husband
and wife having the same surname.

Your suggestions here of changing the output
to a complementary output. The triple Quad designed is
a valid design. even Doug Self hasnt ventured comment
on triples. Id take care changing any polaritys here
as you may then be just providing drive for the
final transistor and discounting Quads triple design.

Re sound of the 303 , I much prefer the 303 into
electrostatics than a 405. the output capacitor
much maligned is of huge benefit not only in
protecting against DC but allowing current to precede
voltage which helps the electrostatic be driven.

Cheers / Chris
 
Hello Chris ( again) I agree the 303 does well into QUADs. esp my old ones. But I also use it on my Lowther horns.

Don't know why they sound so bad in SA. But then, everyone to their own taste.

Lowthers with no crossovers may go well with the Maynard amp !
Cheers Tony
 
Complimentary or just consecutive numbers

I have to agree with Chris Daly (Hi Chris)

The fact that Doug (I designed the noisiest pre-amp with most electrolytics) Self has made no comment would suggest that it is a good design, but then I know it is... it just doesn't travel well but can be rejuvenated without expensive capacitors or changing too many active components.

The output triples give a very high gain "complimentary" pair with the bits that matter, the output devices being truely equal in performance, and the bits that cause the quiescent current to change, the first in each pair, run low current and kept away from the output devices (take note Doug).

Quad went to great lengths to ensure that the linearity of each half was the same although this is not reflected in the fact that the resistor values are the same each side. They also went to great lengths (100R emitter resistors) to prevent high frequency instability which may occur with higher hfe and Ft devices.

the real test would be to set the quiescent current (Iq) at the normal 20mA and apply a 1kHz triangle wave with the feedback removed, you should see no visible crossover distortion with full output swing in resistive load.

If the crossover is minimum at a lower Iq then the amp may be more sensitive to temperature changes, If crossover is minimum at a much higher Iq then you will be generating your own temperature changes.


Chris you know I can't let you get away with a comment like...

"huge benefit not only in protecting against DC but allowing current to precede voltage"

If the current was to lead (or lag) the voltage at the output then the frequency response (into a resistive load) could never be flat. With a 303 connected to an ESL57 you may well find that it isn't but for other reasons like the load is not resistive.

As any good chippy would tell you "measure twice and cut once"

Best regards

BLUE
 
Dear Blue, and others interested in the Quad 303

It is a fact the Quad uses an electrolytic capacitor
some 2200uf to couple the AC audio waveform
( er .. music to most people ) to the speaker.

This principally arises from approx 33 volts DC
on the output ,a Dc condition arising from
a single rail power supply - although one should
not fail to mention regulation to 67volts, involves the
diode bridge negative duration driving TR201 and TR3 ,
- the regulator is a great circuit in itself ,worthy of
discussion, the Quad 303's single rail cant be viewed
as a simple + and earth set up.

Im interested in your comment that frequency is
effected by a condition where current leads voltage
and I agree load resistance/ reactance may play a
part in that -if loaded excessively. However it is the
case the Quad 303 is an amp principally designed to
drive one type of speaker and does use a capacitor
to couple its audio , a feature which I think Quad
cleaverly use to drive the ESL57- its a unison act.

The ESL57 a subject in itself contains a somewhat
awkward load for most amps. In practical use the
303 and ESL57 is a very good combination. There
is from what I hear no lack of high frequency from
using a capacitor on its output ,if anything the 303
has an excellent high frequency capability into the
ESL57.


A product of an AC (audio) signal running into a capacitor
from what I understand is that current leads voltage by 90 degrees expressed as the phase relationship between
current and voltage. Ian Hickman's Analog Electronics
book 1990 (published by BH Newnes )explores this
relationship on Page 13 , providing explanation using
vector diagrams and shown graphically at Figure 1.7 on
Page 12 at diagram (e) of that book.

Quad Im convinced in using an output capacitor use this
asset to their advantage in driving the ESL57. Ask
yourself why the old Quad 11 was so good into ESL57s
-well once again its a coupling network but in the case of
the Quad 11 a transformer. Yes there are differences
between transformers and capacitors, but there are
also similaritys when viewed as AC coupling devices

The 303 was designed to drive the ESL57 , and from
what I can see is very good in that role. The 303
will also drive other speakers - its an amp remember.

Quads philosophy of amp and speaker aligned
was rather unique and quaint in comparison to these
days where (-other than a few boutique brands) -
our amp( x) will drive anything called a speaker (x)

Yes I agree triples do produce parasitcs on voltage rails,
but as has been pointed out ,if ft values are carefully
chosen they shouldnt detract from the amplifiers
performance. If this presents a problem in the Quad 303
try a 1N4004 diode to partition R104 R130 R112 and
R116 from the + voltage rail feeding the transistor triples
ie TR103 TR105 and TR1L , the negligible voltage drop
shouldnt upset the current values certainly no more
so- than the specified resistor tolerance.

A great amp is the little 303, just refresh the electrolytic
capacitors to renew its great sound.

Hope this is of interest

Cheers / Chris
 
Chris and anyone else following..

The point was perhaps not ideally expressed and I am not having a dig. I did however stipulate a resistive load and the current and voltage in an electrically short (length) resistor at any given moment will always be in phase.

In practice the large output capacitor of a QUAD 303, Armstrong, Leak etc. will have a VERY small impedance (reactance) compared to the design load, this is made further insignificant by the application of negative feedback. The current through "the" capacitor will therefore produce very little voltage across it and but what little voltage is produced across it will indeed lag the current by 90deg but the current in the load resistor will always be in phase with the voltage across it.

As the capacitor and load are in SERIES the CURRENT in the load will always be in phase with the CURRENT through the capacitor and as the capacitor simply bridges between the amplifier output devices and the load it effectively does not exist at all but the lowest frequencies. The VOLTAGE across the load will be virtually the VOLTAGE at the amplifier output devices at all but the lowest frequencies which is exactly what is required.

At low frequencies the VOLTAGE across the load drops as the reactance of the capacitor increases and some of the amplifier output VOLTAGE appears across the capacitor and not the load resistor, this follows all the way down to dc when all the VOLTAGE at the amplifier output appears across the capacitor and none across the load. At all times however the voltage and current in a resistive load will be in phase.

If you can accept the above then substituting a reactive load (R±JW where R is the bit producing the music) should make no difference although in practice the reactive or resistive element may be very low and would upset the amplifier even if it were not capacitor coupled. There is also an argument about which side of the capacitor to take the ac feedback from but I’ll keep out of that one as the deed is done.

Chris; I was rather stupidly pushing the point that the relationship between voltage and current at the output terminals of an amplifier like the QUAD 303 only depends on the nature of the load impedance, within the designed frequency range.

All amplifiers sound the same, The distortions they introduce is all that distinguishes them.

BLUE

The use of CAPITALS is for clarity and not intended to be patronising.
 
AX tech editor
Joined 2002
Paid Member
analog_sa said:
The 303 is one of the worst SS amps i've ever heard. Is it the regged PS starving it of current? Are the output caps too horrible?
No idea. Just the concept of improving something that bad seems strange.


Actually, it is the output V/I protection that is very limited in the 303, and will come in at relatively low output current and voltage combinations. The 405 has almost double the V/I area. 405's with serial numbers above 29,000 have a larger area yet, especially around zero output voltage they can source or sink up to 6 amps versus just 4 amps for the earlier versions. At high output levels the 405 can source 9 amps and sink almost 4 amps.
The 303 will do 4 amps regardless of the output voltage.

Depending on the type of speakers used, this may be a cause for the difference between the 303 and the 405.

Jan Didden
 
  • Like
Reactions: 1 user
Dear Blue and others interested in the Quad 303

Thanks for a great discussion , you are indeed correct
that the load ( speaker ) pushes frequency around,
( as a product- voltage and current in equal amounts )
however the trick is to use a sufficient amplifier to keep
it all relative to audio reproduction within the
design criteria. This would appear to be the case with
the 303 amp and the ESL57.

A comprehensive analysis of this combination of
amplifier and speaker is found at

http://www.st-and.demon.co.uk/AudioMisc/57and303/interact.html

It would suggest that 2200uf as a coupling capacitor is
too large in the 303 , and a decision based on economics
relative to the same 100v capacitor used for the power supply
was made by Quad. A check with a calculator to find the
corner frequency and if im not mistaken this is, 1 divided by
2pi x RxC would find 330uf as more than sufficient.

The resonance displayed at 13Hz in these graphs cant be
reproduced by the Quad ESL 57 , but may explain why some people find it a good amp for sub bass use. Albiet it really
hasnt sufficient current capability for that purpose.

It may help this forum discussion if people assessing and
commenting on the sound of this amp have firstly changed all electrolytic capacitors for new ones.

A hint here is to use 80v capacitors in series for the power
supply rather than trying exhaustively to source 100v caps
(Quad may still stock replacements, if you have to use 100v )
remembering you end up with half the value in uf of one
capacitor, but twice the voltage capability by connecting
capacitors this way. The wiring can easily be altered to accomodate power supply caps in series rather than parrallel.
Its interesting to note the physical size of capacitors has
reduced enormously since the 303 was made.

The Quad 303 remains a valid design for electrostatic speaker
use which is quite remarkable considering it was introduced in 1968 and remained in production in the 1980's.
The Quad 50e using similar case work was introduced
about 1966 and was taken up by the BBC, but apparently
had a slightly different circuit design - has anyone a
schematic ?

Cheers / Chris
 
http://www.st-and.demon.co.uk/Audio...3/interact.html

Chris:

I had a look at this site and the article has several fundamental mistakes.....

The model of the speaker is incorrect as is the model of the 303 output but even if they were correct the analysis simply shows the voltage across the "speaker" input and because this is not a pure resistance we get the response shown it is exactly what I previosly described although maybe not so well.

Assuming the model were correct, the only bits that would produce energy (hopefully music energy) are the resistors, it is hoped that these energies at different frequencies all combine to produce the sound required at the listeners ear, if this could be modelled and the "speaker" analyzed with different sources then I would pay more attention.

Have I missed anything, I did not notice the measured results from a real 303 and ESL57 pair, this correlation would appear to be so easy to do, but a record of air presure, temperature and humidity would also have to be made as well as the volume (space) around the ESL57 to enable others to make a comparison.

Has any one done this ??

BLUE.

The band is just fantastic that is really what we think...... PINK
 
I agree with you that the 303 isn't the best amp I've heard.

Comments: You need to compare it to the alternatives available when it first appeared. The small signal section, sinle rail + ground and output cap topology are certainly old fashioned viewed in 2005, but the output tripple appearently reduced crossover distortion below what was previouisly attainable and did not require thermal tracking. Even today a means to maintain stable bias without thermal tracking is not to be sniffed at. If you were starting from scratch using full complementary rather than quasi would make sense except that I would not be surprised if the benefits are too slight to emerge through the other topological limitations. If modded, it I think you need to do it in a way that preserve the Sziklai circuit, otherwise you may loose the freedom from thermal tracking that is a non-trivial benefit. However, both Sziklais and tripples have a reputaion for oscillations that are difficult to track down and suppress.

Using the Quad 303 output section (without the cap!) in a more current topology (Self's "Blameless") could be a very interesting undertaking.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.