Sound Quality Vs. Measurements

[tvrgeek]

Yea, they keep coming up with stuff too. Can't let those left coast guys get too far ahead The other news item that blew me away was IBM got some time on BlueJean and came up with an electrolyte for air-lithium batteries that won't break down. Design-a-chemical program. If their report pans out, it is a world changer. We are talking about ten times the energy density. A cell phone that lasts a month, laptop a week, a car 300 miles. There are a couple more game changers coming in processors and memory that I am not at liberty to share. Lets just say the ride ain't over yet.

Simetrex, uses SPICE models. Superb. Got both theirs and LTSpice so I can go read the manuals. If I can learn enough to repeat Wahab's sim, then I'll sim the little Rotel she likes. I bet it is not only very generic, but just plain well executed.

If my friends did not already know I was crazy, this experience has proved it to them. I'm having fun.

 

[ThorstenL]

Hi,

Thanks Jan, not I did not know of that distro.

Seems you are mainly an Open Source / 'nux geek and less of a tvrgeek...

LTSpice is a spice simulator from Linear Tech that is not a "distro" but proprietary software designed by Linear Tech for internal use (it evolved from "Switchercad", and made available as freeware to any who care to use it.

Linear Technology - Design Simulation and Device Models

It runs on Micro$oft Windoze. As it is one of the few freely available and unlimited Spice Simulators it has become the de-facto standard for many DIY'ers and you often see LTSpice Files posted with threads about designs here.

It is greatly beloved by the many "Spice Jokeys" here on DIYA who routinely crow about ridiculously low THD from their Sim's that are highly suspicious to anyone who ever builds and tests real circuits with real devices.

I found the learning curve for using it rather steep when I tried to start using it a few years ago. I routinely need custom models (including tubes, a wide range of transformers etc) and excitations for my work and back then this was hard to do at the time, so I stuck what I had and which already had all the customisation done...

There are many resources for LTSpice in this community and others, here are some that came to my note (by no means exhaustive) and which I bookmarked "for a rainy day":

LTSpice and What the New User Should Know - diyAudio

LTspice Tutorial

Undocumented LTspice - LTwiki

http://www.diyaudio.com/forums/tubes-valves/129950-how-simulate-opt-ltspice.html

http://www.diyaudio.com/forums/software-tools/199487-ltspice-how-measure-thd-n.html

http://www.diyaudio.com/forums/software-tools/202522-ltspice-thd-analyzer.html

http://www.diyaudio.com/forums/software-tools/199462-whats-best-way-show-output-impedance-ltspice.html

http://www.diyaudio.com/forums/solid-state/195607-jitter-phase-noise-ltspice.html

http://www.diyaudio.com/forums/tubes-valves/180144-zener-ltspice-model.html

A thread I was interested in that features extensive and at time gratuitous use of Spice Simulation with downloadable files of whole amplifiers including models etc. is here:

http://www.diyaudio.com/forums/solid-state/175289-goldmund-mods-improvements-stability.html

Ciao T

 

[abraxalito]

It runs on Micro$oft Windoze.

For the open source geeks it also runs passably under Wine on my Ubuntu Linux.

 

[ThorstenL]

Hi,

On his later designs, he eliminated R11 completely.

Yes, this "loading VAS and shunt compensation" has fallen out of favour. However, I would suggest you first try the Amp with the original design and let your wife listen. Only then try for example the compensation Wahab has shown and see if your wife still likes the Amp. It would be an excellent experiment.

On the multiple outputs, it would not be difficult to drill the heat sink to use two outputs.

If you have not received your replacement Fet's yet see if you can cancel the order and switch to double die TO-3 parts. They are available from a number of vendors, including Exicon:

ECF20N20

Maybe it's just me bing lazy, but i hate doing much mechanical work on gear I'm modifying.

I hate taking current through the mounting hardware on TO-3 packages. Learned that back in failure analysis.

Then don't...

Ciao T

 

[Wavebourn]

I am pretty sure his wife will like Wahab's compensation.

 

[Dark Fears]

I really don't think we should throw away measurement, but there is some truth to what he says about numbers not meaning everything. I have experienced this a number of times. Honest specifications are a really good tool to understand characteristics though.

 

[wahab]

Your version is easier for me to get my head around, that's for sure. As yo have done this sim, I will try to copy that to learn the tool and the main reason, to learn how these puppies work. As I have mentioned, reading textbooks is a great step one. Nothing beats being able to ask a question. I really appreciate your efforts and the patience of the many on this forum.

What would the sim say if you slowed down the outputs and changed R18 (23 on the mod) to 680 and R21 (R12 on themod) to 470? I am switching to the Exicon EC-10N16, P16 pair. They have half the gate cap. from the Hitachi. So the above gives about 1M for f3. Would I not risk some output stability with the 100 & 68? I am going to fly the gate resistors right off the socket.

150 on the CM emitters. OK, I was guessing 100. Actually, I happen to have some. How about a .1u across Q8?, or Q12 in the mod.

THose 2SC1775's sure do have a LOT more gain.

I gather the output network is a "typical" ?

You have moved the compensation to the output of the first and second stages into the feedback loop. No more big DC load on the VAS Do I look at them, C8/R13 and C5/R13 as if they were separate?

Simetrix. Found it. Simple is good as I don't intend to do this for a living. I am already on my third career. This is far more fun as a hobby.

The original P FET gate 470R resistor is too high and this produce
an overshoot on the positive going of the signal when checking
with a square signal.

The input differential current is 0.6mA/transistor , hence the 150R
to create a significant voltage drop of 90mV that will allow the CM
to work correctly despite components dispersion.

Shunting the VBE multiplier with a cap can be useful , i didnt check,
in case wich a 100nF cap should be enough.
Depending on the device used a 100/220pF can sometime be connected
between base and collector of the said transistor to prevent local instability.

The compensation i used is TMC , transitionnal miller compensation,
wich has been vastly popularized in this forum by Edmond Stuart.
The relevant threads can be easily found and will provide quite
an insight about this original way of compensating an amp.

As for the output network , well you re right , it s typical ,
albeit the original should be enough..

As mentionned by Thorsten you can use dual die cases
that will allow for better current capability without drilling
the heatspreader , not counting the shorter electric path
wich is instrumental with power fets.

Just be carefull to first set the iddle current to 0 after such a mod,
although the Exicon fets have higher thresholds , one never knows.......

 

[wahab]

Does Simetrix allow for Fourier analysis?

Of course , here two exemples.

 

[Speedskater]

While on the subject of Fourier. The Power Point and all scripts from "j.j." recent seminar on FFT are now at:

Scripts for Octave

 

[bcarso]

If you have not received your replacement Fet's yet see if you can cancel the order and switch to double die TO-3 parts. They are available from a number of vendors, including Exicon:

ECF20N20

Those look very interesting. The characteristic curves indicate that they have a negative tempco of drain current magnitude at any but very low levels, somewhat different than the Hitachi parts. And I like the availability in both packages --- I wonder which one has the lowest lead inductances?

Another question someone may know the answer for: are there built-in gate protection zeners? It would seem so from the specification on gate-source breakdown voltages.

Also, would anyone know what the capacitance of the body diode looks like, as opposed to the capacitance specifications given in the datasheet? As the body diode is rated for 16A of forward current, I would imagine that it is substantial.


Brad

 

[ThorstenL]

Hi,

Those look very interesting. The characteristic curves indicate that they have a negative tempco of drain current magnitude at any but very low levels, somewhat different than the Hitachi parts.

I think that is just the scale in which the graphs are printed.

Hitatchi parts are negative above around 100mA, that is the bottom fraction of a mm in the chart...

And I like the availability in both packages --- I wonder which one has the lowest lead inductances?

TO-3 of course. If you directly solder it to the PCB.

Another question someone may know the answer for: are there built-in gate protection zeners? It would seem so from the specification on gate-source breakdown voltages.

I may be mistaken here, but AFAIK laterals do not have a "gate protection zenner" build in as a seperate device, it is formed as part of the part works.

External clamps are usually recommended with any of these devices.

Also, would anyone know what the capacitance of the body diode looks like, as opposed to the capacitance specifications given in the datasheet?

Essentially, one of these parts is pretty exactly the same as two separate 8A Exicon Mosfets in parallel. These in turn are essentially Hitachi 2nd source parts with comparable (but usually slightly better) specifications as the original Hitachi Parts.

Ciao T

 

[mr_push_pull]

don't get this the wrong way, but maybe the title of the topic should be changed. the discussion, at least the latter part has absolutely nothing to do with the title. it's really misleading.

 

[BFNY]

While on the subject of Fourier. The Power Point and all scripts from "j.j." recent seminar on FFT are now at:

Scripts for Octave

I Know this guy is really good, but that explanation of the FFT is really pretty jumbled, IMHO. If you want a better one, try reading

App Note 1405-2
http://cp.literature.agilent.com/litweb/pdf/5988-6774EN.pdf
App Note 1405-1
http://cp.literature.agilent.com/litweb/pdf/5988-6765EN.pdf

App Note 243

http://cp.literature.agilent.com/litweb/pdf/5952-8898E.pdf

 

[Wavebourn]

don't get this the wrong way, but maybe the title of the topic should be changed. the discussion, at least the latter part has absolutely nothing to do with the title. it's really misleading.

No, it's not. For experienced people who participate in this thread is obvious that common overall measurements regardless of parts, topology, and regimes, is impossible when we speak of high-end quality of sound reproduction that can be affected by myriads of factors. The way to get better sound reproduction is to improve things step by step, peace by peace, to get still optimal (but not the only the best that is impossible) result. Unfortunately we don't have any "Bester-ometer" device to judge high-end sound reproduction, except our perception apparatus. All measurements are useful when we measure parameters that we are going to improve. As the example, we discussed one good amp's topology, and people proposed what and how to improve. It is the only valid way I know today. After implementing of proposed improvements we'll see if our guesses were valid, and how sound quality was improved, or not.

 

[a.wayne]

don't get this the wrong way, but maybe the title of the topic should be changed. the discussion, at least the latter part has absolutely nothing to do with the title. it's really misleading.

Yes TVR has taken us way of course and it seems no one wants to discuss measurements, answers are short and with limited discussion ...

 

[bcarso]

don't get this the wrong way, but maybe the title of the topic should be changed. the discussion, at least the latter part has absolutely nothing to do with the title. it's really misleading.

I can appreciate that point-of-view, but woven within these details, things emerge that do have relevance to the topic IMO. In particular, as Thorsten stated, tvr's possible modification and before-and-after listening evaluation of the amplifier that's serving as the focus recently can be a fascinating study.

To wit: the slew rate of the stock DH-120 is fairly low and asymmetrical. The slew rate of the alternate design from wahab is symmetrical and much higher. Both slew rates are compatible more-or-less with a full-rated-power 20kHz sinusoid, but wahab's amp has lots of slew rate to spare as it were.

So we have, possibly, the opportunity to explore questions about symmetry and slew rate's effect on audio quality.

Perhaps a structural feature of this site could be envisioned, in which technical details could be placed on sidebars or some such. I know that, for example, my question about the details of the lateral DMOS parts could have been the subject of a new thread. But right now we have a concentration of experts here who are following this. I wouldn't necessarily expect all of them to jump on a new thread. So at risk of bloating the flow I thought it acceptable to ask a few questions here.


Brad

 

[BFNY]

Also, while in the Agilent App Note store, grab this one

http://cp.literature.agilent.com/litweb/pdf/5953-5136E.pdf

Although the title may sound foreign, it is a great overview of feedback theory which may pertain to amplifiers that use it ("a closed loop control system")

All the good stuff - poles and zeros, gain and phase margins

 

[Wavebourn]

To wit: the slew rate of the stock DH-120 is fairly low and asymmetrical. The slew rate of the alternate design from wahab is symmetrical and much higher. Both slew rates are compatible more-or-less with a full-rated-power 20kHz sinusoid, but wahab's amp has lots of slew rate to spare as it were.

I think it is the major improvement still proposed that will be reflected on sound quality. It can in some degree compensate as well results of another proposal, to get more powerful output devices.

 

[mr_push_pull]

is obvious that common overall measurements regardless of parts, topology, and regimes, is impossible when we speak of high-end quality of sound reproduction

really...

 

[a.wayne]

Really.....!!!!!

Well based on graphs i have posted and the comments afterwards, measurement don't account for a can of beans , you like it or you don't ....


I guess.......


Ultimate Audo Amp would have j-fet inputs, tube Vas , later mosfets outputs or BGT ( thermal tracking errors) , not sure , hard to get discussion going , my Idea for TVR to mod the Hafler amplifier along these lines .


Go for it TVR ...........