That is a matter of operator skill. A skilled 'scope user can detect distortion in t 1-3% range.
Or, whatever fits the actual distortion spectrum that results.
No references, not accepted generally, probably that 40% of non-science.
More of that 40%, I fear.
Some cars can go real fast if you drop them off a cliff.
<shaking head> My tastes in science fiction are better satisfied by reading Analog magazine.
That is your point of view. I don't listen to music by slices... Or long enough for it makes sens and gives-you the time to get-in.
. Instant switch in the middle of a tune is a pure non-sens to me, you are not comparing the same thing (the same part).
"Sighted" ? I need to know at least a letter or a number to know which of two devices I compare. If you replace this number by a real name, who cares, unless you are impressed by a brand's name or its price ?
it seem obvious we are not living in the same planet, you and me. And the difference is, If I indicate what works better for me, I don't try to impose my point of view like a law for others.
Last edited:
No, it is a readily demonstrable fact.
I never suggested that anybody do so. From this statement, one might conclude that you don't understand the statement that your post responded to.
Please let me try to be more clear.
By switching time I mean the dead time between listening to alternatives.
Statements like that make it hard to believe that you have ever experienced a truly instant, clean switch over because it would be as close as you have ever come to comparing the same thing. That's one of the problems with your slow switching or cable swapping: You are not comparing the same thing. You are comparing a segment of music with whatever follows it, delayed by the time it takes to perform the switch over.
Obviously you have no clue about blind testing.
Please at least minimally educate yourself in the matter before you start pontificating about it. This may be a good starting point: ABX - Hydrogenaudio Knowledgebase
The one thing that a blind test provides for sure is a letter or number that identifies the pair of devices being compared.
Sighted refers to knowledge of what you are listening to at any particular instant. This can and must be concealed, but the general identity of the pair of devices being compared must be known so that a score can be kept.
Most guys understand that only a sordid ******** or a pitiful fool conduct discussions on technical matters using that kind of language. We know that type from playground at school and have in our working lives seen them under-achieve everywhere around us.
Thank you for lording it over me. You obviously have a superior understanding of the subject as compared to me.
Interesting how you attempt to correct a relatively mild factual statement offered with a relevant technical explanation and lacking in profanity with several fairly vicious personal attacks laced with foul language.
I really do not understand why they use this simple, low-quality recorded music for testing. Why do not they take good recording of Beethoven's no. 9, for example. The simple music like Tracy reveals nothing about high fidelity sound reproduction.
Do you know that from actual scientific experimentation with the recordings and attempting to hear differences that you know to be there by other means, or are you making your comments on general grounds?
There is a technology related to finding specific passages in specific recordings that make certain technical flaws most reliably detected.
It has been going on since the early 1980s and many of its findings have surprised many people.
The names of recordings and timings of segments of recordings, and compendiums of those musical passages burnt onto CDs, are circulated more-or-less privately, and have been for decades.
That's an interesting set of religious beliefs... ;-)
I believe in DBTs that produce relevant results.
I think it is wise to distinguish between lossless compression, lossy compression, and dynamics compression which the above statement seems to fail to do.
Unless the goal of the DBT is evaluation of a lossy compressed format, they are generally done with commercial recordings in format that is not lossy-compressed.
The conceit of that comment takes some beating when your continual hectoring is such that any sensible and reasonable people will just not bother to post when all they can look forward to by response is your cat's paw dissection of every comment and the ridicule which all too often is insinuated in your posts
[I for one would suggest that you follow the advice given you by Scott Wurcer a short time ago in another thread.]
Sorry to say the really high distortion graphs were never kept as 1% was my goal. Some look very similar at 5%. The 8 watts is just to say some useful output above 5 watts. What is interesting is if a 100 watt power amplifier of near zero distortion is run into the final valve g1. 5% is typical. That is well worth seeing as it makes some of what valve people say pure nonsense. Transistor hatred comes in. That makes my blood boil.
To say why someone who loves transistors would build this amp I better say. I have always thought as few devices as possible would be best. Valves offer the better solution if wanting to take that to the ultimate limit. The quality one hears is layers in the music and space. Also a feeling of minute detail that isn't really the music and might not even please the musical producer.
The SE valve amp also overcomes a major defect of transformer coupled design. The DC bias in the transformer is a good thing. It make linearity without feedback better. It could be likened to tape head biasing of the nasty type. An SE amp is the purest class A there is. Crossover distortion can not exist nor can BH curve dead spot distortion. This amp is OK at 20 Hz ( - 1 dB and 2 % THD ) and gives something at 15 Hz ( - 3 dB 5 % THD ). 62 kHz is not too bad. Hypex is - 3 dB 47 kHz if I remember . Hypex is a very good amp.
This amplifer has two valves and a 620 mV sensetivity. There is no loop feedback and no cathode bootstrapping to valve 1. There are some transistor amp ideas used. The design was proven using input valves with failed emission ( maybe 20 tried and many makes ). The maximum distortion was 1.5% and identical spectra. Only one silicon device, a MOSFET multiplier.
How can this work ? Simple. I read the 1940's notes on distributed load amplifers before and around the time of Hafler. The Blumlein didn't give me much except to say he knew what he was saying. The inspiration was a gentleman who said approximately that if you draw pentode and triode curves your " eye " tells you there should be a very linear curve between them. Somehow he made me see it as he had. His best statement was UL was being talked about and he was as yet to see the patent. He was inferring how it might work. The real curves do have little kinks. His were linear with a small dog leg.
I worked out that in a single pentode this needed a leap of imagination to think how you could dial up this curve from one device. One should think cascodes to get a clearer idea.
My thinking was if we call standard ultra linear North South Ultra Linear we might also have East West UL, in to out. To point out, even two triodes will by inverting the stages distortion cancel. It's never is as good as one would hope for.
I have usually disliked UL amps. However some are very good. This suggests that UL makes life harder. A typical problem is a nasel sound which is highly unpleasent. Often the simple distortion tests are OK, but not the sound. I noticed the 1972 GEC book of KT88 amps have Zobel type circuits to the UL tap. Doubtless this is the trouble when loop feedback added.
On my amp I have by chance spare taps at 28 and 82 % ( 100 % being triode ). Very interestingly the 82% tap is very good. It is about the same as pure triode except offering a noticable gain improvement. The graphs are exactly the same as far as it was easy to tell. The 82% graph usually is rejected as too much like a triode. If right the gain was + 30 %. The 28% near pentode didn't float my boat.
This amplifer is repeatable and much more sensetive than the popular RH34. In my tests RH34 had worse distortion and hopeless sensetivity. The parts are cheap for either.
To test ideas about 20 silicon parts were used as CCS, CCSink and PSU multiplier. The multiplier MOSFET was retained ( one silicon device that could be an expensive choke if a purist ). The MOSFET does add a tiny bit of second harmonic. Hum is about - 85 dB refered to 1 watt ( - 89 dB 100 Hz ).
Notice how close to a pure sine wave this is. This is because the added harmonics balloon the wave. Interestingly what looks OK sounds OK. Equally lower distortion waves can look more distorted. And they sound more distorted. The ear measures realitive distortion.
The important thing to grasp is that test tools are test tools. If I make another amp it must be the same and work with the bucket of 20 used valves. The fact it is 1% THD ( not in this set of graphs, I did get there, it was 0.2% 1 watt ) is not the question. It is they are all the same and have the ordered harmonics. It has been argued this is a 0% distortion amp. I doubt if many will have a clue what I mean.
BTW. Some have an audio memory. It means they can carry a sound in their heads for years. As I type I remember a fiddle and triangle played outside in the freezing cold one New Years day at the pub. I remember thinking how like a transistor amp the triangle sounded and how valve like the fiddle ( the music wasn't violin stuff ). It was an 4 things in one. It was like two amplifers in use, it was stupid to play a violin at -2C. It was loud in an open space (!!!!!) and it was good.
If you compare too quickly the better sound is either lost or carries on in the brain. Mostly a good sound is one that stops you going to bed. Could it be some love music enough to want a good sound, but not enough to know good from very good? For all of people's pontificating we can only guess what another might hear. Like perfect pitch some have it and others don't. Somethings I read here are as mad as thinking we should all drink distilled water. The sceince of that is well known so all would laugh. We haven't quite convinced some at DIY Audio about similar things. Sad thing is they don't believe a word of what they say. They just know it is well reasoned and has value in that alone. You're the looser if you start to believe you own doctrines. Truth is I love people who do that. Sometimes they do tell a big truth. I am glad it's not me who has to be the door keeper to the truth. I just love music and don't mind what tricks I use to get a good sound. Usually my tricks are maths.
That's the point, Pavel. The reasoning is there's not much on this recording, so there's little to get wrong, so ours will sound as good as anybody's.
Have you noticed how demos tend to use relatively simple chamber music, often recorde at low levels? Why not use something really dynamic? Is it perhaps because the demo system doesn't do dynamics very well?
Apart from the haughty and peremptory tone of your response, just a question: If the length of the piece you listen too to compare is longer than 10S, how much time between the same part will arrive to your ears ?
Second question: If you have difficulties to feel the differences between two gears at 20 second of time's difference, I can see only 3 reasons.
1- You have a memory of a red fish, time to consult.
2- You are deaf or do not know what to listen.
3- There is no decisive differences between the two gears.
Whatever the reason(s), in this situation, you can conclude that, for you, there is no noticeable difference, and that you can chose the best looking one, or the less expensive ;-)
More important for me is that there is no "accident" between the two plays, that can hurt your brain. So, i ordinary create a loop "in the rhythm" when I prepare the samples I will use to evaluate the gears.
More important for me is to listen several time to those pieces of equipment, in various "moods". With and without focusing my attention, at various levels, and when it is possible, on several days.
When I was in the way to evaluate a new amp in my studio, I used to work on each sample during one hour each, on several days... Don't need to "compare" blind, and it was the best method I knew.
Can-you really think you can teach-me, after a life dedicated to this job, (recording and engineering) how to do things with your theories ? (Read my signature).
Don't you think a musician is able to evaluate a guitar, just by playing-it, with one day of delay with the last time he was playing his own one ?
Last edited:
I had a brief try at making a Pye Mozart type design using standard bits ( this was my bench mark for my other amp). This amplifier measures well ( I forget what file name it has if wanting graphes ). It will have some small advantges over the RH34 and my final design. Note how it gets the spare ECC81 to bring the sensetivity up to being about right for passive preamp use. Don't copy the PSU. It will fail as drawn. The ECC81 is not well liked. Daft as it measures very nicely here. The valve two is a V to I converter. This will not be obvious as the circuit looks standard. If an ECC82 is used the V to I function is made almost non existant. As one would guess the outcome is similar but no longer is using the shunt feedback to lower V3 Rp ( circa 900R ). The special bit is valve two and three form a super triode especially if pentode out used. It is claimed the resulting super triode has lower Rp than a KT88 triode strapped. As there is a gain reduction via the shunt feedback that is quite noticable that seems possible. Rp is proportional. I am using about 16 db feedback to cathode V2. The shunt feedback via anode V3 and V3 g1 that kisses the anode of V2. Some call it anode to anode feedback which it isn't. One hopes for a high Rp of V2 which ideally would be a pentode, ECC82 has a low Rp. The V1 is a low gain use of an ECC81, an RF valve. ECC81 like 7199 works nicely in audio. EF184 is a pentode that looks like ECC81 if triode strapped. Again a TV type.
I did try conventional split resistor feedback to cathode V2 and think it might have been better.
- Status
- Not open for further replies.