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 6061626364 67>
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: 8484
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 20 May 2019 at 1:48pm
The bathtub curve:



Elcap, Jens Both [CC BY-SA 3.0 (http://creativecommons.org/licenses/by-sa/3.0/)]

The curve for 2200uF can be imagined to be just left of the 1000uF curve.

A 2200uF electrolytic used to reduce the emitter resistance of T2 to much below its intrinsic value (circa 4 ohms) introduces some non-linearities after around 1kHz.

The resistance is only small at less than 0.1 ohms where it transitions from a straight line, but it is 2.5% of the 4 ohms intrinsic resistance. We will lose 2.5% of the required open loop gain, and because of the change from capacitive towards inductive, it will hold implications for phase.

To maintain the 4 ohms intrinsic resistance as an impedance, such that it is 4 ohms at all frequencies - or at least those which give us the sibilance trouble - a bypass capacitor should be calculated to be better than 0.1 ohms at the harmonics we wish to control.

If that were 9kHz we find we need nearly 180uF, which would also be an electrolytic, so isn't going to help.

But if we think of it as paralleling resistors, then a smaller value might help. Trying 4.7uF in simulation showed a problem of paralleling which results in negative impedance due to the formation of a tank circuit with stray inductances.
emitter bypass capacitors


We find that 1uF works well in simulation, and its resonant frequency (dip) is around 2MHz, which we don't really want, but a value of 100nF simulates too high in frequency. The resonance might however be damped by the 1k (total) emitter resistance and the parallel 2200uF capacitor's remaining impedance.

So we'll give 1uF a try.


Edited by Graham Slee - 20 May 2019 at 1:49pm
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: 8484
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 20 May 2019 at 3:51pm
There is another way to make the VAS a bit more linear and that is by placing a small resistor in its emitter, making say (big) Re 10 ohms in addition to (little) re 4 ohms.

This however, causes peaking using the 1uF from above. We can use 100n instead which doesn't work as good but is better than nothing.

Overall this might result in a more relaxed sibilance band sound? Possibly, but burn-in is full of little surprises...


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: 8484
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 22 May 2019 at 7:19pm
Just read: "...power transformers do not have a sound. They are not in the signal path!"

The law of conservation of energy states that the total energy of an isolated system remains constant; it is said to be conserved over time. This law means that energy can neither be created nor destroyed; rather, it can only be transformed or transferred from one form to another.

Current is the flow of charges around a circuit.

Kirchhoff's Current Law says that the sum of all currents flowing into a node equals the sum of currents flowing out of the node.

So the sum of all currents flowing out of a node must have flowed into it.

Signal current is supplied by the power supply; therefore signal current must flow back to it. Current flow is from positive to negative (electron flow is opposite).

Then what? Does it just stop there? How can it? It would constitute the destruction of energy, which cannot happen. Perhaps it's dissipated as heat? Sorry, no!

Signal current flows back up the power supply!

Unless the reservoir capacitors are absolutely huge in a power amp, there is only one route back for low frequencies, and that is via the rectifiers and transformer secondary('s). It will contain crossover distortion of the bridge rectifier, but any harmonics high enough in frequency will "reflow" via the reservoir capacitors.

And this is why if there is sufficient ripple because the reservoir capacitors are too small, you get hum.

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: 8484
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 23 May 2019 at 10:57am
Q: Is isolating the transformer and associated power supply in a separate case 'all that it is cracked up to be'? Is the effect of the power supply 'components' only connected by the so-called "umbilical cord" to the amplifier going to alleviate distortion/magnetic fields IN the amplifier signal itself?

A: The thing about circulating currents is something I always check, but we only need to go back to schoolboy physics and the battery, bulb, switch and some of that cotton covered wire, to get the picture.

Current flows from battery to bulb (with switch closed) and then back to battery. Current flows in battery and if we make the bulb a heavier load, battery gets warm proving the point.

So simply adding more wire just adds resistance, or in ac terms impedance. There is also inductance and capacitance in the umbilical, but not a substantial amount, but then again it just depends on how sensitive circuitry is to the not substantial amount. Here I'm thinking voltage regulators.

If using a toroidal transformer amongst a sensitive circuit and the toroidal transformer is made lossy enough (less efficient), it won't vibrate so much due to saturation forces, and won't put out much of a field to be picked up by circuit wiring/PCB tracks.

But using a frame transformer (such as an EI) the radiated field is stronger, and even though I/you/we can obtain a specially made low magnetising-current transformer, there will be a signal level where it's bound to be picked up. Even a PCB track is inductive enough.

In that scenario it is best to make the supply remote because radiated energy decays with distance. For example I use a plastic cased PSU1 and tell people to mount it off the rack, which gives distance and also isolates mechanical mains frequency vibration from getting to the turntable arm and cartridge.

But, the current has to flow from, and therefore back to, the power supply, and where's it going to go inside it?

If the load current is small enough and we make the reservoir capacitors large enough, the entire signal current will flow in the reservoir capacitors because at the signal current the capacitor impedance will be smaller than the transformer secondary impedance. Caveat: the high frequencies might not flow too well in a large electrolytic, requiring some film capacitor bypass.

Even if the PSU output is a voltage regulator (a series type such as a 78 series, or an adjustable), it has to flow back to the starting point, which is the positive end of the caps. It cannot flow back up the voltage regulator sense pin. So, wherever we put it, the power supply is part of the amplifier.

Crap in the secondary either becomes part of the signal current, or debilitates the reservoir capacitors in some way if they don't work well with the type of crap. Either way they are now part of the signal.

Measuring it can be frustrating. FFT (spectrum analyser) displays on the test gear flicker all over the place because of LF noise anyway. The 1 kHz distortion test always looks the same. It would be nice to know a measurement which would reveal it. Measurements only get put on analysers by popular demand...

A 50/60 Hz transformer has to work on 60Hz, and therefore on 50Hz - a lower frequency - you will get greater saturation than at 60Hz. The trouble with toroids is they're more "tangential" at this increase.

If you're only drawing a little signal current the saturation is at its greatest, and the crap is at its greatest. Solid-state always sounds great loud, but might not be so good in the first watt. In home listening we rarely escape that first watt.

Then, if there is some household appliance going, it puts its rubbish on the supply, even if it's next door (like my workshop is off a separate supply to my house). The voltage offset (some wrongly call it DC) puts more voltage on one side of the ac swing than the other. The transformer asymmetrically saturates.

Music also contains asymmetric waveforms. The result is distortion at low or first watt, listening levels.

An EI transformer generally doesn't saturate as much, and when it does it is not as "tangential", reducing the chance of placing it in the signal current, and reducing its chance of debilitating the reservoir capacitors. The problem is then how to prevent hum being induced in adjacent circuitry.
Not simple enough for Google-Bot to understand...
Back to Top
Richardl60 View Drop Down
Senior Member
Senior Member


Joined: 04 Nov 2014
Location: Yorkshire
Status: Offline
Points: 1383
Post Options Post Options   Thanks (1) Thanks(1)   Quote Richardl60 Quote  Post ReplyReply Direct Link To This Post Posted: 23 May 2019 at 12:40pm
[QUOTE=Graham Slee]Q:
'So simply adding more wire just adds resistance, or in ac terms impedance. There is also inductance and capacitance in the umbilical, but not a substantial amount, but then again it just depends on how sensitive circuitry is to the not substantial amount'.
This is very true and whilst I am unsure how much 'measured' performance of the umbilical would show up there will be audible differences between different cable materials, constructions etc (and other connectors), so another variable in the system....

The Source turntable, Audiomods V micrometer, Dynavector XX2-2, Accession M, Elevator, Leema Antila 2Seco CD player, TDL studio 1.
Back to Top
DogBox View Drop Down
Regular
Regular


Joined: 10 Feb 2019
Status: Offline
Points: 57
Post Options Post Options   Thanks (0) Thanks(0)   Quote DogBox Quote  Post ReplyReply Direct Link To This Post Posted: 26 May 2019 at 3:45am
Now I am just putting forward what someone else did to control "their" Toroid giving off disturbing influence to it's surroundings eg. circuit board and other electronics... 
They got a thin (0.5mm) strip of copper about the height of the toroid and about 450mm long and wrapt it around the transformer in a "shorted turn" - like what was commonly done on E-I transformers, except they couldn't solder the join so a good polish was given on both sides and an overlap of 100mm given and all held firm with a large hose clamp with worm drive (like on radiator hoses on your car) but 90 - 100mm diameter. This gave an extra -11dB to lower the noise. 
I don't know what you think about this but it is mentioned with intention to help. 
DogBox 
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: 8484
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 26 May 2019 at 11:40am
Suggestion noted, and it saves having a specially made transformer, so top marks here for assisting DIYers!

But the problem here is not noise from the transformer. If it was it would be evident from switch-on or shortly afterwards. The deteriorating SQ here happens three days into a continuously-on audition.

Prior to the deterioration the SQ is very desirable. The problem is finding the cause of it going down. Putting it straight back on the audio analyser reveals no change, and therefore Audio Precision don't have any suitable test for whatever it is, out of its lengthy software list. And if they don't have a test, then it's hard to imagine anybody else has.

One could surmise that because a sine wave test is unable to pin-point the problem, then we need to test using something more dynamic, except none of the intermodulation distortion tests help.

The deterioration has not been as bad after replacing the RS transformer with the Trans-Tronic "standard range" toroidal transformer. It improved through adding a snubber network to each secondary. Bass loss was not as evident. Mid frequency distortion did not improve.



Thinking this could be a saturation problem I placed the variac in the supply and adjusted its output voltage, and slight differences in SQ were observed. Unfortunately as the supply voltage changes, obviously so does the HT (the amplifier DC supply), and so few conclusions could be drawn.

The only way to test the theory of saturation was to reduce the possibility of it occurring, and that would mean using a different type of transformer. Two E/I core dual-bobbin transformers were ordered and fitted.



An immediate improvement to bass smoothness was obvious, but after nearly three days it was apparent that the upper mids were accompanied by the previously noted harmonics. Others hearing the music from this amplifier had not commented on this, but their ears have not done the "mileage" with the amplifier mine have.

Adjusting Iq did not give rise to any audible difference to this distortion, so I adapted the outputs to drive my HD250 headphones with dummy loads instead of speakers, with no dropper resistors!

Trying to adjust and detect any distortion one ear at a time proved very difficult and odd. But results suggested that 25mA was optimum. Playing some 'electronic music' it was adjusted further to take out the 'added fuzz' because 'added fuzz' on real fuzz demonstrates distortion quite well in my opinion!

The two channels measurements didn't match! So this sent me looking for what could be the problem with the Vbe generator.

In the '70s amps' the Vbe generation was quite often done with a couple of signal diodes in series with a trimmer, and often bypassed with a 47uF capacitor. Things changed because it was difficult to attach these diodes to the heatsink to obtain thermal 'feedback', or should that be called 'foldback' (I don't really care - I know what I mean).

The Japanese made a package containing these two diodes equipped with a mounting tag. Us British would use araldite and glue the diodes to the cases of TO-3 devices...

The idea of using a tab mount transistor like an adjustable zener came about and most of us adopted that idea, and that is used in this design.

Eventually I was able to see an uncompensated collector base junction. Looking at other designs I could see the same, so perhaps this is of no concern?

Is the miller capacitance of a transistor in the main voltage gain "chain" of no concern?

Could its parasitics eventually after three days have any effect on the ability of reservoir capacitors across this circuit's HT? Perhaps it doesn't have any parasitics?

I stumbled on a circuit of a 400 watt amplifier and observed a 47nF capacitor between collector and base, but it also had one base to emitter. I could see the reason for the one between collector and base - to swamp miller capacitance - but all the base-emitter 47nF will do is put maximum current through the output stage on switch-on for a split second. I don't like that idea.

But the 47nF collector to base I understand, and I calculated a value of around 30nF for it to turnover around 1kHz, so decided to follow suit and fit 47nF.

I have not since readjusted Iq, but four days from fitting the E/I transformers, and two days from fitting the 47nF's, the SQ is much improved over what it was.
Not simple enough for Google-Bot to understand...
Back to Top
 Post Reply Post Reply Page  <1 6061626364 67>
  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 0.063 seconds.