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 142143144145146 345>
Author
  Topic Search Topic Search  Topic Options Topic Options
BAK View Drop Down
Senior Member
Senior Member
Avatar

Joined: 14 Mar 2010
Location: Kentucky, USA
Status: Offline
Points: 1744 		Post Options Post Options   Thanks (0) Thanks(0)   Quote BAK Quote  Post ReplyReply Direct Link To This Post Posted: 03 Jul 2021 at 4:04pm
This is me in another design scenario 25 years ago, "it has to work this way!".
I never had that argument with Graham.

 Graham's designs are all his own doing. I may have helped with corrections in calculations or with reminding him of practicalities that he thought of before. But I have only endeavored to support his engineering ethic for the best sound quality that can be achieved.

 Along the way, I have gleaned some pearls of Graham's wisdom.

 I bow to Graham's engineering knowledge.
Bruce
AT-14SA, Pickering XV-15, Hana EL, Technics SL-1600MK2, Lautus, Majestic DAC, Technics SH-8055 spectrum analyzer, Eminence Beta8A custom cabs; Proprius & Reflex M or C, Enjoy Life your way!
Back to Top
Sylvain View Drop Down
Senior Member
Senior Member


Joined: 18 Jan 2010
Status: Offline
Points: 481 		Post Options Post Options   Thanks (0) Thanks(0)   Quote Sylvain Quote  Post ReplyReply Direct Link To This Post Posted: 04 Jul 2021 at 7:05pm
Reassurance not necessary, Proprius and Gerera and Headphones and power supply on my desk is evidence of passion and devotion and skills and a '' resource'' of gold.

But i have read through the last 25 years, other designers stating a change of capacitor and use of a copper piece here and there and a resitor change at some point of the circuit or '' better'' more expensive capacitors or you replicate the '' RIAA'' equalisation to effect a more analogue sound or '' tube like '' warmth as '' balonee'''

I feel for the head scratching hours and dilemma but in GOOD Respect for the authenticity, originality of GSP. 
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: 10 Jul 2021 at 11:15pm
An amplifier must be able to do justice to all genres, so it is extremely important for the designer to have a good ear for what sounds right, and the widest range of musical styles to listen to. Hopefully some of it will match the customer's tastes.

Orchestral and easy listening are easy, but jazz, gospel, massed choral voices, R&B, especially blue eyed soul, hard rock, progressive rock, psychedelic rock, and sometimes a mixture of it all, can be difficult to reproduce in a way that keeps the listener's attention.

So I decided to look at the harmonic distribution of the distortions, discarding all thoughts of gm multiples, and concentrated on minimising odd harmonics. Using CCIR 1k weighting the upper mids are boosted so you're able to get a good picture on the bar graph visualisation of the FFT (spectrum analyser). Adjusting the quiescent output stage voltage using a 1kHz fundamental, the third, fifth and seventh could be tucked down below -80dB, making the second harmonic dominant, with the fourth in second place, followed by the sixth - these being greater in amplitude than the odd harmonics.

Odd harmonics are the stuff of square waves and very objectionable, but even harmonics are more triangular, and closer to musical waveforms, so much so that high levels of second harmonics can go unnoticed. I guess the sixth is a cross between odd and even, but it was adjusted as low as it would go.

I was surprised to find Vq was near to 14 - 15mV, which is far from the onset of gm-doubling (circa 22mV), but on listening to some of the "awkward" genres - a mix of hard rock, jazz and blue eyed soul - I was pretty much glued to the music. It was less shouty with greater detail and more believable.

Incidentally, the distortion meter read higher and was nearer 0.05% THD1 at half a watt (real sort of domestic listening level).

I shall do some more work on it tomorrow.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Fatmangolf View Drop Down
Moderator Group
Moderator Group
Avatar

Joined: 23 Dec 2009
Location: Middlesbrough
Status: Offline
Points: 8989 		Post Options Post Options   Thanks (0) Thanks(0)   Quote Fatmangolf Quote  Post ReplyReply Direct Link To This Post Posted: 11 Jul 2021 at 10:28am
That makes a lot of sense and suggestssomeone else focussing on a simple THD % could mean a less 'musical' design due to combination of odd distortion harmonics which you have eliminated.
Jon

Open mind and ears whilst owning GSP Genera, Accession M, Accession MC, Elevator EXP, Solo ULDE, Proprius amps, Cusat50 cables, Lautus digital cable, Spatia cables and links, and a Majestic DAC.
Back to Top
Sylvain View Drop Down
Senior Member
Senior Member


Joined: 18 Jan 2010
Status: Offline
Points: 481 		Post Options Post Options   Thanks (0) Thanks(0)   Quote Sylvain Quote  Post ReplyReply Direct Link To This Post Posted: 11 Jul 2021 at 10:41am
I recall a particular audio quality brand (N)being designated to '' jazz and classical '' and another (M)designated ""Rock and Blues "" and other for ballads and BUT a ''One for all genres'' is a '' task'' indeed for designers. 

 'Acoustic 'instruments musical tone against the digital electronic instruments of this century is mile apart in tone.  
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: 11 Jul 2021 at 12:40pm
The main problem is keeping constant the setting which results in the desired harmonic distribution. There needs to be a condition under which Vq can be set such that it always results in the same harmonic distribution. Once set, the thermal "feedback" must be able to maintain the setting. The thermal "feedback" of this design is very carefully thought out, to make any setting very stable. The voltage amplifier has multiple NFB paths to reduce drift to a minimum, but there will be drift because the amplifier will do whatever the laws of physics require. However, stability can be cyclic in nature, and provided that's the case, it will eventually converge, then it might diverge only to converge again - that is acceptable.

The great difficulty is that there is little or no latitude, and how can there be? There is only one correct answer. How can a different setting lead to the same harmonic distribution? It cannot.

This will be a problem in manufacture, either in-house, or for the DIY constructor. Setting by ear doesn't work unless perhaps if you're musically trained. There needs to be a precise setting which the user can set in the knowledge that his (her) duplicate will have the same harmonic distribution - hence the meters.

At least one manufacturer has comprehended the problem, and has computerised the amplifier to make it calibrate to the set point. Such technology is beyond my reach and that of the average DIY constructor. As long as we are capable of setting meters accurately and I'm capable of giving you the right value, the harmonics should cycle around the fulcrum of "best" distribution.

The limitation is threefold, 

1. the analyser can only do steady-state because it's impossible to correlate multiple readings within finite time, and the "electronic scales" must balance for the reading to be correct. Therefore an arbitrary value which reflects actual usage must be estimated, and here it corresponds with one watt out. Vq is then varied until 3rd and 5th harmonics become even in amplitude, which incidentally, is the lowest they will go. By cyclic, the 3rd and 5th vie for position, and Vq is set to give equal vying over a long time period that might be several minutes. The amplifier is left in that state and monitored over a period of say one hour. After some considerable time the cyclic pattern emerges, this being the thermal response. Unfortunately, room temperature is part of the equation, and it has to be assumed that the user's room temperature will be kept constant at any temperature that is humanly comfortable.

2. once the cyclic vying is balanced, the amplifier must adjust in itself to a steady no-signal state within that room temperature. The time this takes is considerable - the meters being read at intervals - perhaps 30 minutes or longer.

3. One has to hope the mains supply voltage doesn't change during all of this.

It is a good idea to test over again at different times, and perhaps in another day.

So far, the meters have read 13.5 to 14 mV (the meters are uA scaled such that 1uA = 1mV). within the limits of the needle width.

The FFT harmonic distribution is shown below (annotated by me)

harmonic distribution
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: 12 Jul 2021 at 9:03pm
Further study of the meaning of Oliver's crossover distortion analysis, I continue to concur with Self's gm-doubling assertion that Vq can be no greater than Vt. As per Oliver and Cordell, Vq is dropped by Rbase/beta plus real emitter resistance rather than just the latter.

Cordell's MJL21193 and MJL21184 (output transistors in this design) model gave me values that do not appear to work in practice. Using values for Rbase (RB) and beta (BF) from On-Semiconductor (the manufacturer) resulted in a much better match of 24.91mV and 24.225mV for upper and lower complements, based on the combined RB/BF + RE.

The DC NFB draws 7.3mA away from the lower complement, requiring its RE to be slightly larger than the upper complement, but is not exact enough. To close the gap, a bleed resistor from HT drops 1.8mA into the output junction. The additional current gave the 24.225mV (above).

Therefore, the real emitter resistors drop 14.36mV and 15.72mV, respectively. The total emitter to emitter voltage is, therefore, 30.08mV. The meters read the average as if one emitter resistor, and so the set voltage is 15mV.

When first measuring for the "best" harmonic distribution, I said the indicated setting was 14 to 15 mV. Later, I found it closer to 13.5mV. The correct answer, according to Oliver, is anything between 7.5mV and 15mV (adjusted for this ratio from his 13 - 26 mV).

Oliver considered the result academic because perfect temperature control wasn't an option at the time and suggested a different solution, which would require another topic to describe.

The age of the thermal tracking transistor was still to come (and has since gone), and Oliver's paper had suggested it as a solution. The knowledge gained as to the on-chip diode used in the NJL series transistors, and the access to the power transistor's collector lead at its point of exit, has allowed a close approximation of thermal track characteristics.

A large heatsink as part of an aluminium case and additional series base spreader diodes "sensing" the case internal temperature has led to a very thermally stable design.

As such, the setting of optimal Vq is no longer academic. The upper end of Oliver's range (26mV) is touted as correct by Self and Cordell, and there is little chance of thermal runaway; it can be fully exploited.

The DC operating point is maintained stable by the multi-path DC negative feedback input stage. The only two variables seem to be the quiescent HT voltage, directly related to mains voltage stability and room temperature. In my experience, these can be sufficiently constant.

With the aid of the meters and the user's Vq trimmer, the optimum 15mV (30mV E-E) can be readily set and monitored.
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 142143144145146 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.