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 8586878889>
Author
Fatmangolf View Drop Down
Admin Group
Admin Group
Avatar

Joined: 23 Dec 2009
Location: Middlesbrough
Status: Offline
Points: 5267
Post Options Post Options   Thanks (0) Thanks(0)   Quote Fatmangolf Quote  Post ReplyReply Direct Link To This Post Posted: 14 Oct 2019 at 10:34pm
Thank you for explaining something pretty complicated Graham and the coal derrivatives tree above is great.
Jon

Open mind and ears, whilst owning GSP Genera, Accession, Elevator EXP, Solo ULDE, Proprius amps, Cusat50 cables, Lautus digital cable, Spatia cables and links, and a Majestic DAC.
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: 15 Oct 2019 at 9:40am
40WPC (0.5W class-A) Power Amplifier

Maximum output 1% 1kHz THD+N both channels driven

8 ohms: 40 watts (input 903mV) rms
4 ohms: 58 watts (input 783mV) rms

Frequency response at 32 watts rms into 8 ohms
continuous, both channels driven

20Hz (-1dB) to 20kHz (-2dB)

THD+N at 32 watts rms into 8 ohms
continuous, both channels driven

 20Hz: 0.052%
100Hz: 0.035%
315Hz: 0.035%
1kHz: 0.037%
3.15kHz: 0.044%
10kHz: 0.078%
20kHz: 0.137%

THD+N at 1 watt rms into 8 ohms
continuous, both channels driven

 20Hz: 0.023%
100Hz: 0.02%
315Hz: 0.02%
1kHz: 0.02%
3.15kHz: 0.02%
10kHz: 0.025%
20kHz: 0.026%

IMD at 32 watts rms into 8 ohms
continuous, both channels driven

0.136%

IMD at 1 watt rms into 8 ohms
continuous, both channels driven

0.045%

S/N ratio, 20Hz - 20kHz CCIR: 73dB

Output noise, 20Hz - 20kHz CCIR: -99dB

Crosstalk: -71dB

Idling temperature (heatsinks): 43°C (ambient 23°C)

Maximum temperature (heatsinks): 52°C (ambient 24°C) @ 58 watts rms into 4 ohms continuous, both channels driven

(temperatures taken with case lid removed)

Idling current (output transistors): 250mA

Test supply voltage: 240V ac (rating as per transformer primary)

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: 16 Oct 2019 at 9:47am
The trouble with class-A, or any other state of bias is the accuracy of its control.

A long time ago I made temperature controllers. They were used in industry to regulate the temperatures of chemical solutions used in the manufacture of various parts from Rolls Royce bumper bars to electricity meters.

Some academic had decided that a 1N4148 diode would make a good cheap sensor, and an op-amp had such good power supply rejection ratio (PSSR) that it would make the perfect comparator - it would only need a basic unregulated power supply - and therefore could all fit in a valve base module case.

Two major UK plating line manufacturers had based their systems on it, and had installed plating lines up and down the country. The lines were working but with problems.

It was around this time I was asked to make the temperature controllers because I was young, "cheap" and still "wet behind the ears"; and they wanted to save money. I was to find out that they were wasting money on factory visits where they'd spend hours head-scratching, but never solving the problem.

All I'd done was to clone what they gave me, and so my temperature controllers were just as bad! But with nowhere else to turn, they pressed me for the answer.

The academic's mistake was he never saw the diode sensor didn't have any power supply rejection - he'd just gone on op-amp data. It would seem to be a common problem judging by this: https://www.analog.com/en/analog-dialogue/articles/avoiding-op-amp-instability-problems.html

The diode when biased into conduction by a current is supposed to give a temperature dependent output, which the op-amp compares to a DC voltage set by a pot, and "switches" its output.

The trouble eventually identified was the change in supply voltage during switching - or simply a fluctuation in the mains supply - changed the value of bias current, which changed the voltage drop across the diode. It had been thought that the diode only changed with temperature, which isn't the case.

By the time this was recognised the industry had mostly moved to China and there wasn't any need for me to make any more temperature controllers...

In an amplifier it doesn't matter if the bias spreader is done with diodes or a transistor - the base-emitter junction has the same tempco - but with an unregulated supply the voltage drop is going to change along with changes in supply voltage.

The rate of change is mitigated to some extent by the addition of a resistor in the transistor bias spreader. Unfortunately, the transistor has gain and because of its own parasitics, wants to oscillate, and replacing it with a diode chain has improved the sound of this amplifier.

So now, I'm back to figuring out how to prevent supply voltage variations altering the degree of pre-set bias, which will obviously change the degree of class-A-ness.
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: 16 Oct 2019 at 8:17pm
By drawing more standing current through the output transistors we are making the difference between no signal consumption and maximum output consumption narrower.

In effect, it adds some power supply regulation similar to shunt regulation, but rather than power going to waste, it is being used to reduce output crossover distortion and the result is that the 1 watt THD is reduced.

The limiting factor is how much heat this generates and if it's acceptable. Without making the case ridiculously large to carry monster heatsinks; accepting the existing case as being aesthetically OK, then the limit is dictated by its thermal capacity which is roughly 1°C (or K for Kelvin to be scientifically correct) and how much temperature rise we can allow.

The commercial limit to protect the user from burns is about 55°C, and we'd have to fit thermal trips rated at 55°C to switch the amplifier off.

My earlier ramblings on the temperature rise, although correct, did not allow for the change in conductance in the sensing diodes due to mains voltage fluctuations; and so on 250 volts the case got hotter than on the 240 volts test voltage (hence my last rambling).

The only practical way around this is to reduce the standing current such that on 250 volts the case doesn't get too hot, and by measurement I found that a 10 volt rise in mains supply results in 30mA increased standing current. Therefore, reducing the standing current by 30mA on the test voltage will lead to the predicted temperature rise when powered by a 250 volts supply.

The downside in reducing standing current to 220mA at 240 volts supply results in "class-A" up to 0.387 watts, which doesn't look as good in print as 0.5 watts.

It should also be understood that this amplifier must be used at ambient temperatures of 35°C or less. It won't be much use for mid-summer outside entertainment!

Set as described here, the 1 watt THD is now 0.03% instead of 0.02%.

Another thing I did, which the circuit diagram shows, is remove all the snap-in capacitors and pack in as many 1000uF 100V radial capacitors as I could. Even so, I now have less than half the original reservoir capacitance.

The main difference between capacitor types is their mechanical construction: the radials have a rubber bung seal which also supports the bottom of the foil windings.

On swept frequency tests near full output - 32 watts to be precise - and going from zero to 32 watts and then sweeping over a period of a few seconds from 20Hz to 20kHz, an initial mechanical rustling could be heard during the low frequencies whilst the snap-in capacitors were in use. It is less obvious to non-existent using the radials.

If the foil windings of a capacitor can be upset by the requirement for power, then I guess it could have a bearing on the sound?

Currently the amplifier sounds more like the Proprius I'm used to. I just hope it stays like this.

Edited by Graham Slee - 16 Oct 2019 at 8:20pm
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: 20 Oct 2019 at 2:20am
If you think electronic components don't mess up the sound then solder a PESD24VL2BT TVS diode across an RIAA phono stage input.

It is a specially developed bidirectional diode designed to fulfil the requirements of ESD (electrostatic discharge) protection, whilst having minimal influence on the source.

Its capacitance is hardly anything at 12pF, and its non-linearity with voltage change is hardly anything either. It will shift the MM "bump" up and down by only a few Hz in proportion with signal amplitude. It should be perfectly transparent.

Unfortunately it isn't. The sound is veiled and removal of the device demonstrates how much so. Golden ears are not required.



But because the EU conductive finger test says so, I thought I might get away with putting one of the devices on the output. Still no joy.



The lack of capacitive stability certainly messes up the sound, so what else can cause this?

Read the next thrilling instalment...
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 (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 24 Oct 2019 at 3:49am
From space invaders to Facebook - tech has come a long way - but not in high fidelity audio.

Employees of early tech companies took quite an interest, but since the 70's this pioneering interest seems to have waned.

The first and last authoritative work I know about on capacitors was undertaken by Walt Jung and Richard Marsh and was published a few weeks after the end of the 70's...

https://www.americanradiohistory.com/Archive-Audio/80s/Audio-1980-02.pdf
https://www.americanradiohistory.com/Archive-Audio/80s/Audio-1980-03.pdf

Jung and Marsh linked the tangent of loss angle with capacitor frequency stability in electrolytic capacitors, and solved my problems at a stroke.

Unfortunately I wasn't to find this information until the turn of the century, it being a USA publication.

I don't like giving too much away simply because I had to work and study hard to reach where I am, and my thought is you only really appreciate something when you've suffered to obtain it.

But a low tangent of loss angle equates to a low dissipation factor which equates to low dielectric absorption, and we use such as polypropylene and Teflon, which have low dielectric absorption, in audio circuits because they make a difference.

I chose the lowest tangent of loss angle reservoir capacitors I could find, but in my haste, missed where it said about derating it above 1000 uF.

Having replaced the 4700 uF capacitors with as many 1000 uF capacitors as I can pack in, the amplifier started sounding better. It has gone longer without the esses lacerating my eardrums.

There has also been the change to more standing current which has also helped in reducing low-level listening distortion.
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: 30 Oct 2019 at 1:03pm
It is now the 216th constantly on hour of the latest version of the power amplifier. There is still bass!

At 150 hours the mids and highs adopted a harsh sound which worsened up to the 168th hour, and I thought it was failure again.

However, I had accounts duties to do (3rd quarter VAT return) so just left it idling, or playing background music whilst doing the chore of entering-up figures.

Borrowing terms from a Monty Python sketch about words: "woody" and "tinny", are well placed to describe differences between the sound of valve and transistor technology. Why do we need confusing euphemisms?

It was quite obvious that 60s transistor radios sounded tinny, but the old "steam radio" sounded woody.

It was my choice to pursue solid-state rather than valves and too much water has passed under the bridge for me to change now.

It hasn't prevented me from trying to make solid-state sound more valve-like though. My red-line however, has been never to play around shaping the frequencies we can hear to achieve it.

Stepping back 50 years, it was apparent that when a transistor radio started distorting more than usual, the battery was going flat.

This is a rather crude way of explaining that the power supply has a considerable influence on how good things sound.

The one difference since 216 hours ago was ditching the 4700uF snap-in reservoir/decoupling capacitors for the 1000uF BC components radial electrolytics.

Even so, I still felt there was a power supply problem, and although decoupled by (now) 2 x 1,000uF capacitors there is a fuse rated at 4 amps in series with the power to the amplifier.

When requiring power to produce bass, the voltage drop across the fuse might only be a few millivolts, but it represents a modulation voltage, and any music in the mids will be modulated by it, which is distortion.

I had been threatening to link-out the fuses with heavy wire, but had to await some 1 amp quick-blow fuses to go in the mains side, if only to protect the workshop from fire.

I remembered I'd done a power amp around 40 years ago which only used a mains side fuse, but per channel, as it was a mono amp. Did it protect the amp? Well no, it didn't blow on short circuits, but neither did the amp. It simply burnt a lump out of the screwdriver I used to short the speaker terminals.

However, the output transistors were homotaxial 2N3055's and virtually bomb proof. Today's power transistors are epitaxial which means they will probably blow like an avalanche of the composite tiny bases they are.

Now, if this does result in the desired "sound quality" then I need to figure out exactly where the protection fuses need to go, and what rating I will be able to get away with.

I have a feeling that nowhere in the wired connection will be any good. In other words, nowhere in the secondary side, and so some considerable thought will have to go into the exact value.

I wonder how long the pieces of heavy wire which replace the existing fuses, will take to burn in? (I am joking - I hope).
Not simple enough for Google-Bot to understand...
Back to Top
 Post Reply Post Reply Page  <1 8586878889>
  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.