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 9596979899 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 (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 15 Jan 2020 at 8:34pm
I was amazed, after a bedding-in period, to find the same sound from the trans-diodes as I'd got from the transistor bias spreader, and not a good sound at that - measurably fine but audibly sibilant.

The difference between a base-collector connected transistor-diode and a normal diode is the maximum reverse voltage, which is about 5V for a transistor, and about 100V for a 1N4148 diode.

The only thing I can imagine it being is that the reverse biased transistor zener behaviour generates noise - often taken advantage of in synthesisers to produce a particular quality of noise - and somehow, even though forward biased, there must be some charge at high frequencies due to the dynamics of the music. This might sound like me clutching at straws, but at this point I'm lost for any other explanation, so if you know better, please say!

A 1N4148 diode glued to each power transistor works fine, and doesn't seem to have this problem, and it might be just coincidence, but it doesn't act as a zener, and even if it were to, it would be back biased at 100 volts.

The trouble is the glue: I would much prefer to have the thermal detection device mechanically fastened, but a 1N4148 is such a finicky thing to bolt down without risking it breaking or shorting to a lead.

There are other ways of using a transistor as a diode, so I'm off to look at them.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
peterb View Drop Down
Senior Member
Senior Member
Avatar

Joined: 02 Feb 2017
Location: Cornwall
Status: Offline
Points: 332
Post Options Post Options   Thanks (0) Thanks(0)   Quote peterb Quote  Post ReplyReply Direct Link To This Post Posted: 15 Jan 2020 at 10:48pm
Would an SMD version of the 1N4148 stuck with a thermal conductive glue make it a bit easier?

Edited by peterb - 15 Jan 2020 at 10:49pm
Peter
---------------------------------------
Dual 505-1, Cyrus CD T, DIY 80W MosFet amp and PreAmp, 2xKEF 103.2
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 (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 17 Jan 2020 at 7:16pm
There are imagined, measured, and observed factors in amplifier design. I try to avoid the imagined, try not to rely on the measured too-much, but trust in observation.

People say if it sounds good it's right. I once made a simple phasing device which worked its magic whilst turning the control knob. It sounded good, but as a method of reproducing music accurately, it wasn't good.

I have listened to unstable amplifiers in audiophile systems and heard the occasional sound appear as if it is somewhere beyond the listening room, but other sounds remain within the walls. It can be a wow moment, but it isn't right.

I have heard Hendrix whilst playing walk round the back of the room, and that's either a clever or incidental studio panning trick, which in the "right" conditions works, but more often than not, doesn't.

Tuning by ear in such ways might "improve" some music, but often at the expense of other music. I think if you observe sufficient performances of music you learn the sound of instruments, and how they sound in venues, and how sound reinforcement, and its tricks, makes it really sound.

Then when you listen to similarly recorded music reproduced by the electronics you're working on, you have a good enough idea whether it's doing its job correctly or not.

Live music can be a full orchestra in a concert hall, or a nephew playing his electric guitar on his practice amp in your back room. It can be the organist and drummer at a social club, or a rock concert at an arena - it's all live - and it all has its artistic persuasion.

The organist and drummer demonstrate accidental musical interactions, where the low organ notes rattle the snare drum-skins, but in a studio recording where instruments are separated this obviously doesn't take place. If you have a good idea how well-produced music sounds, and how its tonal balance sounds, you can get a good idea if an amplifier is reproducing sounds correctly.

You don't need all the octaves to hear if the tonal balance is correct, so a good studio monitor will work, and I choose the LS3/5a for my monitoring. The room acoustics must be good, and sound absorption must be similar to the end-users abode.

Then it's just a case of listening to as much music as you can - quite a rich catalogue of genres - and all of them must provide some satisfaction, and not glower with glare at you, unless recorded particularly badly.

The other thing I do is keep the amplifier circuit switched on over long periods of time because the sound changes if something isn't right in a design. I believe most designers who might be employed 9 - 5 will not have the "luxury" and might therefore be blinded to a particular design discipline. This is quite dangerous to their reputations in my opinion.

I have listened to circuits which sound incredibly good, only to find they sound as distorted as hell after a few days of being switched on. Returning them to the test apparatus, nothing is amiss, but to the ear it obviously is.

Therefore observation is the most important of all, no matter what the know-it-all’s say or believe.

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: 17 Jan 2020 at 7:17pm
Originally posted by peterb peterb wrote:

Would an SMD version of the 1N4148 stuck with a thermal conductive glue make it a bit easier?


If you can attach two wires to it, all well and good. Personally I can't. Wink
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: 31 Jan 2020 at 2:15pm
All amplifiers are a collection of parasitics interfering with the music. All semiconductor devices exhibit negative impedance. The reason for semiconductors is for communications, to perform the oscillation required to transmit and receive. To obtain perfect amplification requires the 'taming' of oscillations, most of which are parasitic (unintended) and the job is to identify and compensate for them.

Here we see a real oscillator (left) and the possibility of parasitic oscillation (right). The real oscillator is optimised to oscillate. The transistor circuit is far from being optimised but shares features with the glow-tube.

The glow-tube was the valve version of a zener diode, using gas ionization instead of avalanche ionisation.

The transistor drives a capacitive load. The transistor is in-effect a zener diode in series with a forward biased diode.

The transistor zener is (or should be) nowhere near its avalanche operation point, but nothing is so straightforward in amplification. If a similarity exists, it often results in unintended behaviour.

By connecting the collector output of one transistor to the base of a subsequent transistor, which has parasitic capacitance, the zener is loaded by a capacitance, and oscillations might be established.

Simply put, a zener outside avalanche still has leakage prior to avalanche, and some noise will be present, and noise in semiconductors is ever present with any current flow to some degree.

A small value resistor - a stopper - inserted between collector and capacitive load limits the possibility of a parasitic oscillator forming.

That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
peterb View Drop Down
Senior Member
Senior Member
Avatar

Joined: 02 Feb 2017
Location: Cornwall
Status: Offline
Points: 332
Post Options Post Options   Thanks (0) Thanks(0)   Quote peterb Quote  Post ReplyReply Direct Link To This Post Posted: 31 Jan 2020 at 4:13pm
Originally posted by Graham Slee Graham Slee wrote:

Originally posted by peterb peterb wrote:

Would an SMD version of the 1N4148 stuck with a thermal conductive glue make it a bit easier?


If you can attach two wires to it, all well and good. Personally I can't. Wink

Me neither until I bought magnifying goggles and a 'third hand' ! Smile
Peter
---------------------------------------
Dual 505-1, Cyrus CD T, DIY 80W MosFet amp and PreAmp, 2xKEF 103.2
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: 12 Feb 2020 at 11:18am
It seems the closer the emitter resistor voltage (Vq) gets to the thermal voltage of the output transistor (Vt), the better this amplifier sounds.

There are other things which must also be contributing: the "excellence associated with Beyschlag, Philips, and Centralab" (according to Vishay) is audible. Component choice is, however, only one part of a complex mechanism, which any amplifier is.

This amplifier's reliance on the accuracy of adjustment gives it a valve flavour reminiscent of instrumentation in the past. It isn't modern dressed up as retro; it is retro brought into the modern.

Valves could never be "set and forget" - something accepted by our forefathers - the TV always had "on the fly" adjusters under its flap!

A significant reason for this was the lack of stabilization circuitry - there were no Zener diodes - no laser-trimmed bandgap reference ICs.

Voltages would, therefore, drift along with ambient temperature and variations in supply voltage, and it was always a matter of tweaking it "on-song."

Likewise, here, the basic original transistor amplifiers bore similarities with valve circuitry by not featuring temperature and supply voltage correction.

The problem with (BJT) transistors is Vt, with its reliance on the constants of current and temperature.

The "transistor sound" comes from crossover distortion, which can be "tuned out" but without the necessary constants, drifts with changing current and temperature: a matter of the right amplifier subjected to the wrong conditions.

Barnie Oliver discovered this "tuning out" of the "transistor sound" as being making Vq = Vt in 1971.

If the circuitry automatically compensates for this, the amplifier is no-longer retro; so if the goal is a retro amplifier made to sound good, it has to have "on the fly" adjustment, and the user must be made aware of what it does, and use it for that purpose.

Prior to European harmonization, continental and UK mains supply voltages differed by a nominal 20 volts. This is a significant factor in retro amplifier design but often ignored because we are led to believe the Europe-wide voltage is 230 volts. There is no such voltage!

The lack of circuit regulation means the HT will change in proportion with the mains supply. When the HT is above or below its intended voltage value, Vq will have shifted in value sufficiently that it no longer matches Vt.

The two primary voltages are 220 volts across most of Europe, and 240 volts in the UK. These are allowed a variance of 6%, meaning 220 volts can range from 207 to 233 volts, and 240 volts can range from 226 to 254 volts. With the diversity of power generation, the grid where you are might vary across the full allowable range, making it more challenging to arrive at the correct setting for Vq.

Supply voltage correction did exist in a rudimentary way, through having several transformer primary taps. Your local TV and radio shop would know to set a product to the local voltage using the equipment's supply voltage selector and lock it in place.

The discount "sheds" supplanted the local shops, and voltage selection became very much "Hobson's choice." Transistorized circuitry "evolved" to be less-fussy through the use of Zener and bandgap regulation, or a least a constant current source.

It is only now, where we require to operate this retro amplifier design, we realise its performance relies on having the right voltage, but a voltage we would be hard-pressed to know for sure.

Even if we had the availability of muti-tapped primary transformers, we might discover the mains supply had shifted out of range. Most often, however, we are stuck with 230 volt (or 115 volt) primary transformers.

The adjustment has to be placed somewhere, and all we have to hand is the bias spreader itself (used to set Vq), and so this is where the manual tweak control must be.

The user cannot be expected to be an engineer, so he must have some human interface device to know where to tweak to. It would be hypocritical to have a "go, no-go" indication, such as an LED because this in itself requires a voltage reference. It must be passive, and the only device I know which fits the requirement is an analogue meter fixed to the amplifier.

The meter can be conveniently fed by placing its terminals across the output transistor emitters and choosing a meter sensitivity suitable to read-off Vq.

At this point, we must understand that Vq with signal becomes dynamic Vq, which is different from the static Vq we need to match to Vt, which is also DC, and therefore a static quantity.

If the meter is to be permanently connected, it must be capable of handling dynamic Vq, as this will "grow" with the output stage delivering power to the loudspeakers.

In effect, the meter doubles as an output meter, albeit one with an offset, which is the static value of Vq.

Another problem that arises is that dynamic Vq increases beyond the meter's overload, which will burn-out its windings if the amplifier is operated at constant maximum output.

Therefore, it isn't at all a matter of just connecting the meter between emitters. It can be kept simple, though, and only requires three additional components: Firstly, some series resistance, and I chose two paralleled 1k resistors to make 500 ohms; Secondly, a clamping device. Ideally, this would be a germanium diode for its low forward voltage, but these are very rare today, and so a 1N5817 Schottky diode with approx. 300mV forward voltage at this small current is what I chose.

The meter is rated for 5x current overload, and the diode Vf of 300mV with the meter resistance of 1500 ohms limits it to 200uA.

As for the meter, it is a matter of scaling, and a 100uA type fits nicely. The 52mV Vq across the output stage emitters translates to a 26uA meter deflection.

This being easily read-off, it leaves a reasonable amount of movement to indicate a rudimentary output level.

The photo below shows the meter "tagged on" to test out the idea before incorporating it into the amplifier. Two meters will be required: one for each channel.




Edited by Graham Slee - 12 Feb 2020 at 3:49pm
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 9596979899 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.