Thank you, hopefully you can help me. I'd like to understand the spatial cues a little better. Using the same gears with similar acoustic loudness level, I noticed some recordings creates very wide, tall and deep image while others are rather flat. For example at the beginning of Spirit Within in Heart of the Rainforest album by Alexander Baker & Clair Marlo, I can hear the birds chirping seems to come from my neighbor's trees outside of my house. I'd like to know if there is a measurement we can make that show a recorded piece is really holographic, a bit compressed in depth, compressed in height or just plain outright flat.
Same principle I have done this on all ADSL drivers, etc. What I'm trying to understand how you measure THD into a load without output current. As stated Iout and Vout = 0, the horizontal scale on the plot makes no sense as well.
It's a low current model of a famous current dumping output stage. There is a huge increase in distortion at the point when the dumping transistors (or in this case, diodes) turn on.
The "THD vs level" looks innocuous, even the IMD looks "not so bad". But the circuit sucks. I've used this as an example of defect that does not show up "blatantly" on THD/IMD. It does show up, but it really doesn't look as bad as it should look. I mean, the output impedance varies in a non-linear way by a factor of 10 when output current moves a bit. It's terrible. Red lights should be flashing.
Anyway. I don't want to blow too much hot air, so when I build the setup and do actual measurements, I'll make a topic with detailed explanations of everything. For now, a quick summary:
This is a grown-up version of a setup I've already used and that worked very well, with the expected limitation of being a hacked together bunch of crap and clip leads.
The setup is designed to test an output stage in open loop (I'll call it the DUT), but it could also work on a full amp.
It sums two AC channels from a soundcard plus DC, and drives the output stage with that. Another DC DAC sets output stage bias current. And two other soundcard output channels, plus another DC, drive the output of this output stage through a stereo amp and resistors, to drive a current into the DUT output. The two power supplies of this output stage under test are also programmable. So it will set the input voltage, supply voltage, and output current of the DUT. I use 2 output channels summed to avoid intermodulation up to the DUT.
There are a bunch of FETs to apply transient load current to heat the transistors.
Meanwhile, a bunch of instrumentation amps measure Ib, Vbe, Ie of output devices (also for FETs) for acquisition.
The expected test signals are one high level, low frequency sine (LF) and one low level high frequency sine (HF), like 50Hz 3Amps and 24kHz 1mA. As said above, that controls either input voltage, or output current, or both. But they could be any other signal.
In order to do non-static tests I will not be using any FFT. Since distortion is signal dependent change in gain, I've had excellent results by simply filtering out the LF sine and detecting the amplitude of the HF sine in software. This gives one data point per HF period.
Once the small signal gain is known (by detecting the small HF signal amplitude) for each amplitude of the LF high current wave, then the transfer function can be reconstructed in software, and small signal THD computed at each operating point. To answer the above question, I know it sounds weird to have "current but no current", but really, small signal THD happens at zero amplitude, it is an approximation of the nonlinearity of the circuit at the operating point being considered. It's another way to look at the derivatives series of the transfer function. I might not even not compute it, since the transfer function has all the information in it anyway.
One application is to trace the gm of the output stage in real time. It shows the devices heating up over a couple tens of ms, bias changing, etc. But I'm also interested in changes in hFe vs temperature, and the previous setup didn't measure that. For example after a half-cycle of positive high current, the top device should be hot, which increases its hFe. When it crosses over to the other device which is cold, will there be a sharp change in current gain for the whole output stage?... Since the output stage is driven from a non-zero impedance, that could cause distortion that doesn't show up in static power tests. Also nonlinear change in AC output impedance vs output current, etc.
etc. To be continued.
Last edited:
So I don't know of anyone using current dumping these days and that circuit as an output stage used open loop would of course behave poorly. They make fuzz pedals with similar configurations.
BTW the diodes are the current dumping transistors in the WW 1975 article, using the same amplifier for the low and high current totally defeats the purpose and just makes a very bad amplifier as well as no sense.
BTW the diodes are the current dumping transistors in the WW 1975 article, using the same amplifier for the low and high current totally defeats the purpose and just makes a very bad amplifier as well as no sense.
Last edited:
> So I don't know of anyone using current dumping these days ....
DOGC-H by Dr. Jagodic
And the faithful followers insist that it sounds wonderful.
Patrick
DOGC-H by Dr. Jagodic
And the faithful followers insist that it sounds wonderful.
Patrick
The Fiio E5 sounds quite bad, in the sense that it will worsen the sound quality if the phone or DAP or PC sound-card behind it is high quality.
For example if someone has a Vivo X3S or Vivo X5 Max smartphone (they sound good), then you connect the Fiio E5 amp... it sounds bad, since the Fiio E5 made the sound dirty.
Hmm & the NE5532 op-amp sounds bad compared to the modern & high-end chips, like what Texas Instruments is selling now the OPA1611, I doubt it requires training / practice like people need to know what to listen for, it's like if you show someone a Blu-Ray for the first time versus 720P DVD it doesn't require training or practice to know what the difference is. Possibly the auditory stuff in the brain is just traditionally based on what is useful, which is hunting (/warfare), music & conversations (language)... there is not much in the brain designed for more than those three. Some people which were in a car accident & hit their head, became amusical after the injury (I learnt it in university & they showed the pictures), so then scientists started to believe that the brain has sectors dedicated 100% to music.
From a quick examination of an online board shot, it seems to have an abundance of X7R (or worse) coupling caps in the signal path
and they didn't read the bit in the datasheet that explains how to connect the decoupling caps to keep the charge pump noise away from the analog bits...
I will change my signature since I'm the aberrant force, then.
If the ultimate goal of audio for most of the people in this thread is to have a listening room which is identical to a live performance, so you just hit play & it sounds exactly (or nearby) to how a band sounds playing live (acoustic or with amplifiers), then you will need a soundproof listening room & some anechoic wood designs on the walls & some highly resolving speakers, either two or four... (I am pretty sure four sounds better i.e. stereo recordings x 2... especially in a rectangle or square room, unless it's a truly anechoic room or a sphere shaped room, then maybe two...)... then after that, you can start to argue about which amplifier or which DAC reproduces the live recording best. Well, wow, if someone has an anechoic listening room with expensive floorstanding speakers & everything is set up well, I don't think it's very important if the DAC is a Texas Instruments or an Analog Devices =)...... The overwhelming sound will make the DAC & amplifier stuff less important.. although if people then play around & compare 10 DAC's with the same recording, sure, that's what DAC developers do. National in the U.S. had a dedicated listening room where they would compare all their little chips all day.
~
Yes, Titanium & Aluminium have a sound signature. If someone designs a Titanium speaker (or Titanium dynamic driver in an IEM), Titanium always retains a sound signature, even if they spend several years on getting the THD down to 0.01% & the frequency response extended from 20 Hz to 20 kHz. Again a lot of perceived sound & the aesthetic expression is within the time domain, so people place too much spotlight on the FR.
Also if the time domain correlates to FR sometimes, well... if you play around with an equalizer it will change the IR, so there is some correlation to the IR & decay, although measuring a middle C on a Titanium diaphragm will have faster rise & decay, then if you magically change the diaphragm to paper, the rise & decay is different, so this has nothing to do with FR.
This is not the QUAD current dumping stage. It misses the balanced bridge feedback which was the whole point and makes all the difference!
Jan
I think the Benchmark AHB2 power amp is a kind of current dumper crossed with Class H variable rails but I wouldn't bet my life on it, just remember reading it somewhere.
From a quick examination of an online board shot, it seems to have an abundance of X7R (or worse) coupling caps in the signal path
Lel, thank u for the technical analysis of possibly why it sounds bad.
On the other hand the Fiio X5 gen 1 (PCM1792A into OPA1611, with a ton of Tantalum capacitors on the PCB......), is an ok sounding product, especially the Line-out, so I'm trying to buy one from eBay now to upgrade my PC or just enjoy the PCM1792A sound again, it's such a cool DAC sound actually, in my view.
I am surrounded by cheap ES9018 designs now for years, including in my LG smartphone, it's not musically interesting when it's the 'El Cheapo version. The Vivo phones made it sound ok (maybe Xiaomi as well, I never checked), although compared to the proper 9018... for example used in sound studios, or a properly designed DAC, then the smartphones don't sound the same.. someone can blindfold test it, it's just people are too not interested to blindfold test stuff.. humans just buy stuff, humans don't blindfold try everything they bought, it's not programmed as anything necessary.
From a quick examination of an online board shot, it seems to have an abundance of X7R (or worse) coupling caps in the signal path
Haha ye I will avoid the designs with "150 X7R in the signal path" in the future
In the land of the Lord of the measurements it's the best SINAD performer.
So it will sound spectacular.
We all want good sq whether it's a perfect repro or not. The closer the better is the challenge. After all, this is a hobby. As far as arguing is concerned, I defer to Chris, a man after my own heart.
We all want good sq whether it's a perfect repro or not. The closer the better is the challenge. After all, this is a hobby. As far as arguing is concerned, I defer to Chris, a man after my own heart.
Yes, everything is an enthusiastic interest only, until you start retailing or producing & selling something, then it becomes a profession - unless they're doing it on the side, or have an underling / manager, although then it's still a side-profession or they are lazily dictating to their underling what kind of results they are looking for.
The retailers and people selling products have income goals, so a lot of them argue that brand & the total design are so important... yeah, well.. a motorbike enthusiast knows every part of the motorbike, what is underneath, they see 100 pieces moving, not just ' a bike '... it's normal. Any enthusiast should be encouraged to investigate what is underneath.
------
Hm, I will also refer to NwAvGuy again as one of the insecure people which wants to pick a fight with everyone. He said that the SanDisk Clip+ already met the threshholds of human hearing, it's a $20 device from K-mart or Amazon, I already bought two of them, his claim that I need to provide evidence to the scientific community that the SanDisk Clip+ sounds different than the flagship DAC's from Analog Devices & TI is just.. evidence-myopia.
Any university which says the SanDisk Clip+ is the limit of human hearing didn't test people well enough, or the 20 students had 'tin ears'.
Secondly, after NwAvGuy decided to change from enthusiast to professional, then he joined some major company listed on the stock market & abandoned blogspot & all internet forums... so we went incognito & now makes money from audio, instead.
Hi indra1,
What you are describing occurs in the brain entirely. What them to poke wires into your brain to measure it?
If I had to guess, discordant sound probably kills the experience for you. Flat sound stage. If something has very low distortion, it would probably allow you to enjoy and process the sound and you could then imagine yourself in that sound field rather than hearing from the outside.
We can easily measure the signal, what your mind does with it is up to you and will vary upon how you feel.
-Chris
What you are describing occurs in the brain entirely. What them to poke wires into your brain to measure it?
If I had to guess, discordant sound probably kills the experience for you. Flat sound stage. If something has very low distortion, it would probably allow you to enjoy and process the sound and you could then imagine yourself in that sound field rather than hearing from the outside.
We can easily measure the signal, what your mind does with it is up to you and will vary upon how you feel.
-Chris
- Status
- Not open for further replies.