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 3839404142 345>
Author
BackinBlack View Drop Down
Senior Member
Senior Member
Avatar

Joined: 05 Feb 2012
Location: Hinton, N'hants
Status: Offline
Points: 2020
Post Options Post Options   Thanks (0) Thanks(0)   Quote BackinBlack Quote  Post ReplyReply Direct Link To This Post Posted: 28 Dec 2018 at 7:44pm
Or perhaps it is, all that current being drawn holds the voltage down. Wink
Just listen, if it sounds good to you, enjoy it.
Back to Top
Richardl60 View Drop Down
Senior Member
Senior Member


Joined: 04 Nov 2014
Location: Yorkshire
Status: Offline
Points: 1468
Post Options Post Options   Thanks (0) Thanks(0)   Quote Richardl60 Quote  Post ReplyReply Direct Link To This Post Posted: 28 Dec 2018 at 8:25pm
well i am on 237.7-238.6v assuming my Maplin device is reliable!
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: 31 Dec 2018 at 8:30am
Output Fuse Inside Negative Feedback Loop: BS?

The reason I put the output fuse inside the negative feedback was because I remember Bob Stuart (of Boothroyd-Stuart/Meridian) reportedly saying it improved the distortion it caused, and the sound.

Maybe that's not the case with this amplifier? So far I've not been fully satisfied with the sound, and I've jumped through many hoops trying to coax it to do what I expect. But I'd not considered the output fuse.

If you think about it, the thin piece of fuse wire is heating and cooling in sympathy with the current flowing in it, which is in-turn in sympathy with the signal due to the load.

As it heats and cools its resistance will change and so it forms a changing potential divider for the negative feedback, so the amount of negative feedback is constantly changing with signal "intensity". On loud sections there is less negative feedback, and on quieter sections there is more.

Thinking about it, the fuse resistance will not change as rapidly as to cause the imagined distortion mechanism, and also it might not be much of a difference, but it is a difference, and so the "character" of the sound might change? And therefore rather than leading to an improvement, maybe it only leads to a difference?

And so I decided to eliminate this possible source of dissatisfaction by replacing the fuse with a substantial piece of wire.
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: 03 Jan 2019 at 1:39am
Circuits design themselves. Another part

A circuit's going to do what a circuit wants to do.

And if we don't help them they're going to protest. They'll not settle until we understand and see to their needs.

When they settle they really perform.

Now, the title of this topic is 1970s Design Indulgence, and like anything you can over indulge and suffer the consequences. You have to know what's good for the circuit and what isn't.

You can be forgiven for thinking the original circuit designers knew what they were doing. They probably thought that too, but we, and things, develop over time.

I keep on saying that they didn't have the powerful simulators we have today, and that is true. They might have had SPICE (Simulation Program with Integrated Circuit Emphasis), but my guess is they didn't, and if they had, what would it run on? Windows 95 was a thing of the future back then.

You only need look at the botch-up by today's standards of the Philips TDA1024 (NE5534). A good audio op-amp back then, but easily beaten by a TL071 (and similar) if you know what you're doing and the situation suits.

So what I did was take it as shown and made a power amplifier using up to date transistors, and tweaked it as far as I could. The trouble was the circuit had faults.

Now, the fuse in the NFB loop wasn't their fault, it was a suggestion by Bob Stuart as previously mentioned. I did include it in a commercial design around 1992 and I cannot remember if it made it sound bad, but it does cause a problem with this circuit, and after linking it out the sound improved considerably.

But there were a couple of other things:

1. Stability and the compensation capacitor: the thinking here was that placing it in the main feedback route that the VAS transistor is not slew rate limited. Actually they'd taken it from VAS collector to IPS emitter, but a latter circuit by JLH put it in the overall NFB.

So, the VAS transistor can slew as fast as it can go, and it is loaded by the input capacitance of the output stage. At some frequency it crashes into a brick wall! That should be obvious. So the NFB loop compensation then sorts out the panel beating? Wrong! The damage has been done and cannot be fixed.

The slide rule calculations coupled with limited measuring abilities might well give rise to a nice looking Nyquist stability plot, but in real operation if the sound goes etched, flat, brittle or lifeless, there is instability.

The stability regime I use sensibly tackles VAS miller capacitance by shunting it with a higher value (was 47pf and is 47pf again), and additionally slows its rise and fall times to a value the output stage can accept. Modern silicon transistors must be lower capacitance than the old designs I've been indulging in, so you'd expect they understood the loading effect, or maybe it was at a time where "NFB cures everything"? It does not. A much older chap, Marcus Graham Scroggie, was trying to tell us about that during all that time.

Now, because there is real voltage gain in the IPS transistor, it stands to reason that some overall frequency compensation is needed to satisfy it. And this is where the smaller (10pf) capacitor connects back from VAS collector to IPS emitter. Note that it does not do the entire journey via the output stage, but is kept local.

Imagine that it could be placed IPS collector to base. It then depends on there being a fixed amount of input resistance or impedance in the right frequency range, but it is the input and who can know the source impedance? Nobody.

By connecting it VAS collector to IPS emitter we are in fact connecting it IPS collector to base, adjusted for the gain of both transistors (if the gain were lower the capacitor value would be higher).

The problem however is it will not take the HF gain below unity, it can never be below 1 in this configuration, so we must rely on VAS collector to base compensation to do the "heavy lifting".

When combined as here the phase margin for the chosen values gives us better than 70 degrees and the gain margin is in excess of 20dB. But it is the action of output stage input capacitance (which causes them to have what's called Ft) which is also loading the compensated VAS which is helping here. This can be seen in the "kick back" I referred to somewhere else, where parasitic capacitive coupling happens inside the output stage transistors, trying to return the output toward unity gain as we pass into the GHz region, but not quite doing it, thank goodness!

Maybe it was considered stupid to consider such ultra-high frequencies in the day, but SPICE shows they happen.

2. The Colpitts oscillator strikes again!

I must take this opportunity of thanking Doug Self for his unique explanation about emitter followers and how a stage with less than unity gain can oscillate. Thank you Mr Self! The explanation can be found in his book: Small Signal Audio Design (well worth buying).

I had misgivings about placing the CFP (T5 and T7) emitter resistor in the emitter side of T7, and rightly so, but I did it anyway because that was what Tobey and Dinsdale had done. Mr Self correctly points out that T7 is simply slave compliment to T5 (but not in those words) for the purpose of current handling. Therefore the true emitter of the CFP is above, not below. And this is where I should have put it, and since have.

By placing it the wrong side I will have to conjecture about it making an accidental Colpitt's oscillator which must have made its oscillations across the emitter resistor. The value and voltage across it would make whatever oscillations it was doing quite small, but even so it would have been modulating the output.

This had a similar sonic effect to the fuse in the NFB.

I am at pains to calculate why, but all the FFT plots had a small spike at 20kHz. Things like this tend to get put down as measurement artefacts, but this turned out to be real.

I can see the input capacitance of T7 is about 7nF to 8 nF from the data sheet, and it would need a resistance of about 1k to make 20kHz, but I cannot see that value. I can only see 100 ohms and 0.33 ohms. However, it can be seen that the 100 ohm resistor is connected in bootstrap form to the junction of T7 emitter to the 0.33 ohm resistor. The bootstrap should be capable of lifting 100 ohms to in the region of 1k, and if so, that calculates as in the region of 20kHz.

Moving the 0.33 ohm resistor to the upper side completely got rid of the 20kHz spike. Distortion (THD+N) fell away to below 0.008% between 10 watts and 25 watts output power. Obviously the 20kHz spike was adding to the distortion sum.

It sounds better too!

If we try to understand what the circuit is doing we can help it do more for us, or we can fight it and end up with nothing.



Edited by Graham Slee - 03 Jan 2019 at 1:40am
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: 05 Jan 2019 at 3:44am
Dial-A-Distortion: See-Saw

or Circuits Design Themselves: Yet another part!

Stability (as the vid we saw) occurs where phase margin is 45° or greater (135° of phase shift or lower), and -180° is reached at -10dB or better (-11dB being better for example).

However, phase margins less than 90° are subject to ringing: certain types of percussion, fingers scraping on acoustic guitar strings, shrieking voices, strong esses etc, take on what some in hi-fi call more detail, but normal people hear as tiring false brightness.

The current gain in an amplifier is such that low current tries to drive high capacitance junctions, and in doing so results in phase shift, and that is sent back to the input stage as negative feedback.

Phase at low to upper mid frequencies doesn't change much, but high treble and above is subjected to progressive phase shift. At low to upper mid frequencies the negative feedback is -180°, but as frequency increases phase shift is added, so in the case of 135° phase shift, the result is -315°, and if phase shifts further it reaches a total of -360° degrees, which is the same as 0°, which is positive feedback, and we have an oscillator instead of an amplifier.

If we let that happen, and it will be at a higher frequency than we can hear, the transistors in the circuit will be destroyed.

So what we try to do is get the largest margin possible with the grail being 90° at the crossing point (0dB), and also a gain margin of 10 (20dB).

But power amplifiers don't want to do that! We can try to include further stages of current amplification to reduce phase shift stage by stage, but it can lead to further (often unforeseen) problems.

This amplifier is uncannily stable. A conventional modern day differential input amplifier would have suffered damage by now, the way I've treated this one.

The phase margin is controlled by Cdom, a small capacitor connected from collector to base of T2 (the voltage amplifier stage known as VAS), plus a small capacitor (10pf) from T2 collector to T1 emitter which is used to "lengthen" the gain margin.

The only way in this design of achieving 90° is by making Cdom large (220pf). This reduces the amount of negative feedback at high frequencies which increases high frequency distortion.

It can be made small: 22pf reduces phase margin to less than 45° but the amplifier survives. 47pf is around the minimum which improves phase margin to the mid 60's in degrees.

These small values improve measured distortion but the sound is more distorted to the ear. We can boast 0.007% THD+N but it is no good if it doesn't sound right (this being the reason for somebody's dissatisfaction with one of my phono stages...).

Going to the opposite extreme (220pf) we are approaching 90°, the measured distortion increases, but should still be inaudible at less than 0.06%, but we have slowed the amplifier to around 9V/uS slew rate as well. It does sound distorted because the musical input signal can slew faster than the amplifier. This might be disputed by other engineers, but it is the only explanation I have left.

Trying 68pf the sound is better, around 70° phase margin, and the measured distortion is quite good indeed at less than 0.02%, but it does tend to be a little too bright for me.

After also trying 150pf, and not liking that either, I settled on 100pf (tried before). This allegedly gives 80° phase margin and 25dB gain margin, depending on which transistor simulation models you trust!

Because it is less than the 90° grail, it will ring to a certain extent, but not as bad as at 70° or 60°, and the slew rate will be about 23V/uS. What we need now is to slow the input such that slew induced distortion (SID) does not result. And in doing so the ringing should be reduced. This is a bit difficult to achieve as we cannot know the source impedance of the users source.

Here we have to look at standards: IEC 61938 aims to make source equipment output impedance (driving impedance, not what it drives into) no more than 2.2k. If our amplifier has its -3dB point at 100kHz (all indications point to this) then if we make an input filter with a turnover frequency of 50kHz, the requirement should be satisfied. We also have a 300 ohm resistor before T1 and this is where we connect the input capacitor to ground.

The total resistance is therefore 2.5k ohms, and we calculate (from 1/2piFR) 1.27nF. Problems arise when somebody uses it with a passive preamp or even a potentiometer (similar things) where the source resistance is even higher, because 50kHz then becomes a much lower frequency, rolling off the highs.

So let's risk it and make it match the 100kHz amplifier turnover, so it rolls off at the same rate. We need a capacitor value of 0.63nF. But what if the source resistance is lower than 2.2k? We'd need a larger capacitor.

OK then, I'll try it at 1nF.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
musicdude View Drop Down
Senior Member
Senior Member
Avatar

Joined: 24 Jul 2011
Location: Chicago
Status: Offline
Points: 115
Post Options Post Options   Thanks (0) Thanks(0)   Quote musicdude Quote  Post ReplyReply Direct Link To This Post Posted: 05 Jan 2019 at 4:31am
Hats off to your never ending search for perfection. I'm glad there are still people out there who do things right! Looking forward for an other exceptional piece of HiFi perfection.
BTW, Happy New Year!
Andy

ProJect Xtension 10 with Clearaudio Charisma V2, GA2 SE, Majestic DAC, CuSat50, Yamaha CD-S1000, Yamaha M-80, Revel F-52.
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: 06 Jan 2019 at 6:45am
A bit of nostalgia...

I used to work with these. Tandberg Series 15 reel-reel. Note the combination of silicon and germanium transistors in the power amp.



Preamp section: follow the PB and 7.1/2" settings. Not far from being an RIAA stage. Pin 28 connects to point T on the volume control next image...



Power amp section: similar circuit to the Mitsubishi-Teleton (and most other amps of the day). Note the tone controls in the power amp negative feedback!!! (this seems to have been the norm with many tape recorders)

The comment regarding the sound of a lot of these circuits was "sounds better than they have a right to". The germanium transistors and their designs were much more benign than todays silicon brutes, but without the power you can get today.



Edited by Graham Slee - 06 Jan 2019 at 7:47am
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 3839404142 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 0.109 seconds.