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 Club

 

Open to all owners plus those contemplating the purchase of a Graham Slee audio product wishing to use our loaner program: join here
Subscribe to our newsletter here (Rules on posting can be found here)


1970s Design Indulgence

 Post Reply Post Reply Page  <1 86878889>
Author
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8842
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 31 Oct 2019 at 6:06am
We may now be reaching the final tuning stage, or at least I'm hoping so.

Yesterday's fuse placement swap resulted in a better defined stereo image, benefiting the bass frequencies.

Putting a quick blow fuse of the correct value on the mains side to open during a fault, now rules out the use of a toroidal transformer: it would blow on switch-on due to the inrush current.

Mains transformer buzz (through the case) takes some hearing - it's quiet in comparison to most, and hum at the speakers can only be heard when you're closer than 30cm.

The highs however, were giving me a problem: Fripp's Lizard (King Crimson) being an acid test, as if I can't follow it I fall asleep, and that's no good.

The flute didn't sound like a flute - just a high pitched sound; Haskell's laughter sounded etched instead of fluid.

The inputs were wired using some RS shielded cable, so I replaced it with what's used for our single sided interconnects (CuSat50 and Lautus).

Straight away the flute sounded like a flute, and Haskell's laughter more lifelike; but still not as fluid as I'd like. This might be due to the type of input filter capacitor or me using a carbon film across the input (which charges the input coupling capacitor).

I think the inputs will have to be separate RCA jacks either end of the rear panel, so they can be close to the input pads on each board, allowing me to use short simple twisted pairs, which should have least effect on the sound. The input grounds will then go direct to case via their shells. I will report on this shortly.


Edited by Graham Slee - 31 Oct 2019 at 6:24am
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8842
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 01 Nov 2019 at 6:08pm
Negative feedback needs to be applied linearly otherwise it has to "try" to cure its own non-linearity, which it will never do.

Everything looks fine on the analytical test gear, and nothing untoward shows on the simulator using the NFB divider shown at "a.", but it can be seen that during the negative half cycle, it crosses 0V, but not on the positive half cycle.

At some point during the negative half cycle no current will flow.

The NFB divider shown at "b." biases the voltage seen across the divider resistor (R13 in parallel with R12 by virtue of C9) so that NFB current does not cross zero at any point.


Erratum: image shows transistor marked as T2. It should be T1


Edited by Graham Slee - 04 Nov 2019 at 5:41pm
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8842
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 04 Nov 2019 at 12:06pm
SID (slew-rate induced distortion)

Otala's target is difficult to achieve: 20kHz -3dB open-loop.

This can only be done by local NFB in the voltage stage.

Otala's riddance of Cdom is also problematic, but surmounted by a capacitor in the local NFB.

Pushing the value smaller, to 150pF, reduces phase margin to less than 90 degrees, but by only 7 degrees. This risks ringing on the high notes.

But with a 100k local NFB resistor, the -3dB break frequency is 10kHz.

Otala's overshoot combined with the possibility of ringing is not good for the sound.

But to reduce the overshoot means reducing open-loop gain further, which increases THD.

A value of 51k instead of 100k takes the open-loop break frequency to 21kHz, but reduces global NFB to 16dB.

With the input filter being a simple RC, source impedances within IEC 1938 (0 to 2.2k) will have break frequencies between 20.4kHz (2.2k) and 28kHz (0). As most sources have impedance we could assume a 24kHz average.

Overshoot would therefore be limited to just a 3kHz range of high musical harmonics, but with CD quality at 22kHz there's hardly any difference.

All that would remain is the 7 degrees of ringing.

The only way to reduce that ringing is to increase compensation capacitance, but that increases the risk of overshoot.

Another way to reduce ringing would be to remove the ferrite beads from the output transistor base leads, but that risks parasitic oscillation in the 40MHz region.

So, it looks like the final tuning option is to use the 51k local NFB resistance and hope for the best.
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8842
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 12 Nov 2019 at 7:14am
An impasse has been reached...

Without the lid screwed on we have an almost silent power amp. Screw the lid on, and we have a hum box.

Taming laminated transformers is not for the faint hearted, and requires a lot of help from the transformer manufacturer. The unfortunate thing for me is my transformer manufacturer is too busy.

My only option is to go backwards and fit the US made toroidal transformer which was "almost there", in the hopes that the mods done since give the positive result I'm looking for.

Let me explain what I'm trying to achieve: any old amplifier can make noises in the loudspeakers, but high fidelity stereo is about projecting the realism of the music into the room.

Unfortunately, every bit of modern technology seems based on the television transmission idea of AM packet transmission. Therefore cell-phones and computer communications radiate RFI, and AM RF is detected by diode junctions, and power amplifiers have diode junctions aplenty.

RFI noise is coupled through every part of a power amplifier, but mainly via the power supply. Defensive capacitors when "run-in" are effective inductors (antenna).

High fidelity has a fight on its hands.
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8842
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 12 Nov 2019 at 4:20pm
As requested earlier in these pages, the toroidal transformer is now back: Antek AS-3450 300VA.

Amp adjusted on test for 0.05% THD at 1kHz and 1W output. It can go as low as 0.025% at 1W but for the fact it gets too hot with the standing current that requires.
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8842
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 13 Nov 2019 at 12:30pm
On day one of the transformer swap, the difference to the music was night and day, initially it felt more like the "dead" of night.

Vieled might explain how it sounded, but with it there was imagery, which might seem contradictive. It was simply the highs which were not as tingly.

The case temperature was also deemed to be too hot.

A further adjustment to standing current seemed to make the sound more acceptable and bring down case temperature.

So today, I decided to go fully Otala'd, throwing caution to the wind by using up a bit more phase margin. Changing the local NFB capacitor down from 150pF to 100pF pushes up open-loop frequency response to 30kHz, such that the amp is faster than its input filter.

It's a case of trading Otala overshoot with stability overshoot, and finding which has most influence over the sound.

It has to be accepted that a simple circuit such as this will never measure as good as today's amplifier technology, but if it sounds as good I will be happy.

Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8842
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: Yesterday at 5:21am
Dinsdale noted the distortion producing stereo loop when he added another channel to the 1961 Tobey & Dinsdale mono amplifier to convert it to stereo.

I tried his loop breaker some time ago but didn't think it helped, and after-all, manufacturers prefer to star the voltage amp ground off the 0V node of the power supply reservoir capacitors.

This prevents the speaker current running through the ground connection to chassis where it joins together at the input terminals... or does it?

I will attempt to answer that in a moment, but first I will add Dinsdale's comment that a more elegant solution would be the use of two separate power packs.

Now, having taken that on board a number of months ago, I had provided it with a transformer having two secondaries such that each channel had a separate power supply, but only using one transformer.

I then started to get difficulties once each incarnation had "bedded in" for a number of hours (72 hours plus, which just happens to be the soak period specified in IEC60950).

Each transformer was replaced by a better-specified specimen, but to no avail, and then I decided on a frame transformer, but had to use two separate ones to obtain the VA rating as a single larger transformer would not fit.

That, if you remember, improved matters greatly, and sent me off thinking about interference, which I'd reasoned wasn't as bad using the split bobbins of a frame transformer, but that might have been a "red-herring".

Case hum had proven to be a problem, which recently led to reinstalling the best of the discarded toroidal transformers.

But then the poor sound returned.

I had also fallen-out with the large 4700uF snap-in electrolytics, and replaced each with 2 x 1,000uF radial electrolytics. This improved matters to a degree.

With each improvement I dreamt up the measured performance also improved, and so I have been at a loss to explain why the bad sound?

It surely could not have been stereo loop distortion as described by Dinsdale because of the use of two separate power packs...

...but are they really separate?

Going back and trying three different values of "loop-breaker" resistor, I found an interesting improvement to the sound using a value of 1 ohm, this being one-tenth of Dinsdale's 10 ohm solution.

But for this to effect somewhat of a cure, then, there has to be a stereo-loop. Why?

The bad sound is, and I've convinced myself of this, second harmonic. If you have lashings of second harmonic distortion it is akin to a frequency doubler. Actually you still have the fundamental, but you have a strong harmonic leading the ear to hear brightness.

The lashings of second harmonics required would have to be more than 1% to my mind, but the analyser has been reading better than 0.1%, which is inaudible.

But read what Dinsdale wrote: "Connection of a commercial stereophonic pickup or microphone with a common earth line now causes each channel to become distorted by the even harmonic products of both channels, causing unpleasant distortion particularly when one channel has a transient such as a cymbal clash."

As soon as the grounds are joined, which they must ultimately be, and are in a stereo amplifier, even-harmonics (second harmonic being even) make the sound bad.

But I don't have a stereo loop do I? Analysis might show that I have...

(To be continued)
Not simple enough for Google-Bot to understand...
Back to Top
 Post Reply Post Reply Page  <1 86878889>
  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 0.078 seconds.