New Posts New Posts RSS Feed - 1970s Design Indulgence
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Welcome to the Graham Slee Audio Products Owners Forum

 

Open to all owners plus those contemplating the purchase of a Graham Slee HiFi System Components audio product and wishing to use this forum's loaner program: join here (Rules on posting can be found here)

This website along with trade marks Graham Slee and HiFi System Components are owned by Cadman Enterprises Ltd


1970s Design Indulgence

 Post Reply Post Reply Page  <1 6566676869 345>
Author
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 24 Jun 2019 at 11:13am
The tail wagging the dog

The design so far works well with some loudspeakers, but not so well with others. It tends to work well with older loudspeaker designs and not so well with newer loudspeaker designs.

The power supply's stiffness helps, and if regulated, the design copes extremely well, but here I want to use an unregulated simple transformer supply like used in other 70's amps (and still used in today's amps).

So, what else could help it work well with all loudspeakers?

Many of today's passive crossovers are very complex (as I tried to communicate earlier), and whether or not it is known to the crossover designer that parasitic spikes caused by series inductance (which exist) "reflect" or otherwise get back to the amplifier output, and thence into the negative feedback loop, is a moot point.

However, I will say that any HF spikes having an equivalent frequency to that beyond where sufficient negative feedback exists will not do wonders for the amplifier's handling of high harmonics (musical ones).

It was not until the passing of the decade that answers became forthcoming to the audible effect of "lntermodulation at the amplifier-loudspeaker interface", or should I say, that publication of studies into it didn't appear until late in 1980.

From Matti Otala and Jorma Lammasneimi; Wireless World Nov/Dec 1980:

"lntermodulation occurs between an amplfied signal and a delayed version returned from a loudspeaker through a feedback loop, when open-loop output impedance is high compared to speaker impedance. Part one of this article analyses this and a second part describes a measurement method with results of tests on different types of amplifier circuit and suggestions for avoiding the effect."

In part 2 they go on to say:

"The effect described is but one of the numerous phenomena affecting the quality of low-frequency sound reproduction. It does not seem probable that its distortion could be dramatically higher than the measured SMPTE intermodulation distortion of the amplifier, unless protection circuitry malfunctions. However, the theory presented may explain some of the subtle differences in the sound quality between different circuit topologies having otherwise equal standard measurement data."

And conclude:

"The output should provide a low open-loop output impedance to adequately attenuate the loud-speaker reaction signal so that the need for a feedback-generated damping is minimized"

The full text can be read at the reader’s leisure at:

https://www.americanradiohistory.com/Archive-Wireless-World/80s/Wireless-World-1980-11.pdf (page 45)

https://www.americanradiohistory.com/Archive-Wireless-World/80s/Wireless-World-1980-12.pdf (page 42)

I will continue on this subject in my next post...


Edited by Graham Slee - 24 Jun 2019 at 8:19pm
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 24 Jun 2019 at 7:48pm
While the above articles are valuable in figuring out the sources of intermodulation at the amplifier-loudspeaker interface, it isn't very helpful to misname the amplifiers in Fig 13 and Fig 14 - Fig 14 being quasi-complimentary, not Fig 13!

Assuming publisher error the diagrams look to have swapped places. The Jan 1981 Wireless World makes no apology or correction.

Again we need some evidence as a 'sanity check' so as not to accuse falsely, and here I cite the data sheet for the AD823 opamp (https://www.analog.com/media/en/technical-documentation/data-sheets/AD823.pdf)

We read in its description: "The complementary common emitter design of the output stage..." and compare that description with Fig 14: "Grounded-emitter complementary output circuit" - 'common' and 'ground' being used interchangeably - and we realise the article is in error.

However, let's have more proof:

The AD823 OUTPUT IMPEDANCE - The low frequency open-loop output impedance of the common-emitter output stage used in this design is approximately 30 kΩ. Although this is significantly higher than a typical emitter follower output stage, when it is connected with feedback, the output impedance is reduced by the open-loop gain of the op amp.

This further confirms the Wireless World error.

And something which is quasi is "apparently but not really; seemingly" which when coupled with the word complimentary, means apparently but not really complimentary, which describes the diagram in Fig 14, but definitely doesn't describe the diagram in Fig 13.

Therefore, we find the open-loop output impedance for a quasi-complimentary as measured in the article is 2.7 ohms and not 60 ohms!

Admittedly 2.7 ohms is poorer than the 1.2 ohms for the complimentary output stage correctly labelled in Fig 12.

What I wanted to know is the exact effect of the negative going signal on output impedance, but the article didn't exactly help me here.

More on this shortly...
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 25 Jun 2019 at 6:39am
The problem with wanting a flat frequency response (with gain!) way into the 100's of kHz is that you have to design into the short-wave and VHF radio frequencies - at least.

To get 50 watts out you have to put 1 volt in if you have a gain of 20.

If you want 100kHz as your -3dB (half power) turnover frequency, then to ensure high frequency stability its crossing frequency is 2MHz, and it should continue at -6dB/octave to at least 20MHz, and not "flick" back up, or the EMC bogey man will get you.

The problem in testing up to those frequencies at rated power output is that something will blow, so nobody ever does. After all, the amplifier will never be called upon to reproduce those frequencies, but you have to know it could do it which ensures it will not become a radio frequency oscillator and will not go unstable.

Before Simulation Program with Integrated Circuit Emphasis (SPICE), transistor parameters were used to calculate what the circuit would do. With SPICE those parameters are in the device models or subcircuits.

The trouble is these parameters were measured artificially and do not reflect what the device does in every real world application. Luckily "we" have Bob Cordell who started out at Bell Labs in the 70s, and being an amplifier enthusiast, has been able to modify the parameters in SPICE models to better suit what the transistors really do.

Now, getting back to the discussion regarding output impedance of different topologies, we are unable to use simulation to discover the open-loop output impedance, because the models don't show it - simple!

So how do we figure-out if the complimentary darlington of our quasi complimentary output stage has high output impedance?

Otalla says his was 2.7 ohms compared to 1.2 ohms for an all emitter follower design. Did his test take the average? The average of what? The article didn't say.

Subtracting impedances (same as subtracting resistances) would reveal what the complimentary darlington output resistance is, if we knew where to start - but we don't! Are these the sum of parallel or series impedances? Or am I way off target? Such articles shed light but only dimly. It hasn't answered the question I'd hoped it would answer, and the publication error would have one going on a hiding to nowhere!

The average hi-fi buyer is blissfully unaware of all of this - doesn't know what happens under the hood - reads what the press says - buys from the simpleton information he's fed. Hasn't a clue what the output stage is or does.

To be continued...


Edited by Graham Slee - 25 Jun 2019 at 6:41am
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Dave Friday View Drop Down
Senior Member
Senior Member


Joined: 07 Apr 2011
Location: Spain
Status: Offline
Points: 173
Post Options Post Options   Thanks (0) Thanks(0)   Quote Dave Friday Quote  Post ReplyReply Direct Link To This Post Posted: 25 Jun 2019 at 7:43am
Graham good morning, I look forward to each update in much the same way as I did to "The hitchhikers guide to the galaxy ".
Do you think that "difficult to drive"speakers is a sort of cult?
Kr,Baz.
lp12,oc9mk3,ca610p,krimson40watt pa,kef105.4
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 25 Jun 2019 at 5:02pm
I think difficult to drive speakers are the result of designers trying to fulfil the impossible, which is driven by

1. Placing speakers 1m from the wall/room corner (no help for the bass)

2. The anti-tone control movement

Both insisted on by the hi-fi press. But this is just my opinion and I'll not go into details right now.

Anyway, continuing on this intriguing journey into 1970s high fidelity amplifier design, I decided to make a major circuit excursion...

The amplifier output stage is now all emitter-follower (EF) and running at Mr Self's preferred 100mA standing current.

It's an ostentatious little thing, and if it were to sound just as good in a couple of weeks of being always on, it would be worth further investigation.

For now I'm just relaxing with this "new toy" after a few hours of disorganised circuit surgery. How long the fun will last is anybody’s guess!
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 26 Jun 2019 at 1:25pm
Here we see the new output stage.



Just two components differ from the original:

The PNP transistor T4 replaces the NPN transistor, taking its base to T2 emitter instead of T2 collector.

R3 takes the place of the two driver emitter resistors.

As these things existed in the 70s they can be used here without fear of taking the topic out of its context.

It evolved (actually people did it...) to what H C Lin would have preferred, but had to use what was available, which was the complimentary feedback pair (CFP).

It is now symmetrical. That is to say, the upper and lower parts are complimentary instead of being quasi-complimentary.

The (lower) CFP could be seen as having a higher slew rate (faster) than the (upper) EF, and had the added advantage of working in the direction of the VAS, making the negative slew rate actively fast, whilst the positive slew rate relies on a constant current source - in this case a capacitor bootstrap.

The VAS and source still do the same, but now each output half is 'equal'.

The other thing which can now happen is the charges can be swept out of the output transistor bases due to the "push-pull" effect of R3. This otherwise being a high frequency problem, where residual charge allows one output transistor to remain partially conductive whilst it should only be the other one.

Thus, the HF draw on the power supply had been more demanding, and now should be less demanding.

This should help towards getting the power supply right.

Just a note about the now gone (lower) CFP: to reduce its slew rate a resistor was placed in its emitter (see several pages back). However, if that was made too large it would cross zero earlier than the (upper) EF, resulting in the wrong HF compensation. And too small, it would race the EF. Such a balancing act wasn't considered worth the effort.

More information on quasi-complimentary problems appear in the appendix of a Linear Technology application note (AN-48 page 22) here: https://www.analog.com/media/en/technical-documentation/application-notes/an48.pdf

Even with the inclusion of the "Shaw diode" the problem of asymmetry persists, and when driving reactive (capacitive) loads the audible results are quite obvious, making some op-amps a little picky about their applications. In addition to the opamps listed, add the 5534/5532/5533 family, and most likely the OP275!
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 30 Jun 2019 at 10:58pm
If parasitics can happen they will happen unless something stands in their way, and EFs comprise the constituent parts for parasitics, namely base-emitter capacitance, and load capacitance (the output transistor base as far as the driver is concerned).

Taming them is fun. Base stopper resistors are about as far as you can go, but that makes low pass filters so that when you add the effect of Cdom, the output rolls down steeper than you want and robs phase margin, which leads to ringing, which is just as bad as the parasitics (provided the parasitics are not destructive).

One trick is to place a zero in Cdom so its effect reduces as the base stoppers begin their roll-off. The zero needs to be very close to the unity gain crossing.


Edited by Graham Slee - 30 Jun 2019 at 10:58pm
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
 Post Reply Post Reply Page  <1 6566676869 345>
  Share Topic   

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.01
Copyright ©2001-2018 Web Wiz Ltd.

This page was generated in 0.219 seconds.