Gotcha!! As they say,
Found the main problem causing loss of transparancey and muddied detail in my apparently low distortion amp. Despite the good THD measurments it turned out to be bad grounding techniques. To be fair the amp is still on the test bench on a lash up rig that i made for testing amp prototypes. It has its own PSU, res caps, volume controll and RCA inputs. The problem i had was due to most of my grounds going back to a central point a bit like a star ground. Only thing is, the resevoir caps also terminate at this point along with mains ground. There always was a faint hum when using this jig but i paid no attention to it as i knew it was the jig causing it and it didnt seem to bother the measurements i was making. Also i have been using a preamp/tone stack in front of the amp also with its ground return going to this central star point. Today i thought i had nothing to loose by sorting out the grounding of the prototype to get rid of the hum. I moved the preamps ground to the veroboards ground entry point, moved the volume pots ground there and the RCA connectors ground there too. Upon power up the faint hum had vanished. I did some scope tests for speed and stability and everthing seemed the same as it did before. No change i thought.
After the measurements i swaped the dummy load for some speakers and reconnected the CD player. My what a difference a few inches of wire make!! It sounded totally different. I put on a CD i love, know and use as a reference and one that is particulary difficult on the highs. 'Amongst The Stones' by Davey Spillane.
WOW! It was instant heaven, all the instruments jumped out at me, no muddying, no listening fatigue and the hair on my skin started to stand up. That was the effect i've been looking for!! So natural, crystal clear and already i'm sure i'm picking up background noises i havent heard before ( unless i've forgotten about them as its been a while since i've listened to a decent amp LOL )
So there we are, it seems that bad grounding can cause odd distortions that really offended my ears even though computer says YES. I wonder what these components are?? Slightly muddy base? i think not as the bass sounded fine before although it now seems more reinforced. Scratchy highs? Well again they seemed fine before, just didnt sound very natural. It was definatly doing somthing to ruin the music, my ears know but they are keeping it a secret from me. Bizzare, i now feel i need to go on a mission now to find out what kind of distortion was SOO subtle yet was indeed capable of destroying the music even in its undetectable levels. Hope this helps others out whome are suffering similar oddities.
Leigh
Found the main problem causing loss of transparancey and muddied detail in my apparently low distortion amp. Despite the good THD measurments it turned out to be bad grounding techniques. To be fair the amp is still on the test bench on a lash up rig that i made for testing amp prototypes. It has its own PSU, res caps, volume controll and RCA inputs. The problem i had was due to most of my grounds going back to a central point a bit like a star ground. Only thing is, the resevoir caps also terminate at this point along with mains ground. There always was a faint hum when using this jig but i paid no attention to it as i knew it was the jig causing it and it didnt seem to bother the measurements i was making. Also i have been using a preamp/tone stack in front of the amp also with its ground return going to this central star point. Today i thought i had nothing to loose by sorting out the grounding of the prototype to get rid of the hum. I moved the preamps ground to the veroboards ground entry point, moved the volume pots ground there and the RCA connectors ground there too. Upon power up the faint hum had vanished. I did some scope tests for speed and stability and everthing seemed the same as it did before. No change i thought.
After the measurements i swaped the dummy load for some speakers and reconnected the CD player. My what a difference a few inches of wire make!! It sounded totally different. I put on a CD i love, know and use as a reference and one that is particulary difficult on the highs. 'Amongst The Stones' by Davey Spillane.
WOW! It was instant heaven, all the instruments jumped out at me, no muddying, no listening fatigue and the hair on my skin started to stand up. That was the effect i've been looking for!! So natural, crystal clear and already i'm sure i'm picking up background noises i havent heard before ( unless i've forgotten about them as its been a while since i've listened to a decent amp LOL )
So there we are, it seems that bad grounding can cause odd distortions that really offended my ears even though computer says YES. I wonder what these components are?? Slightly muddy base? i think not as the bass sounded fine before although it now seems more reinforced. Scratchy highs? Well again they seemed fine before, just didnt sound very natural. It was definatly doing somthing to ruin the music, my ears know but they are keeping it a secret from me. Bizzare, i now feel i need to go on a mission now to find out what kind of distortion was SOO subtle yet was indeed capable of destroying the music even in its undetectable levels. Hope this helps others out whome are suffering similar oddities.
Leigh
Andy L. Francis said:
Andy, although the term Rise Time from a first approach it expresses the speed of response of the amplifier in steep signals injected in its input (thus sinusoidal signals are not valuable for this measurement) such as square waves or sawtooth with its rising edge from the left, in the real life it has relation with the bandwidth product (or frequency response). There is a coarse way to estimate directly the frequency response of a DUT if we divide the number 300 by the measured rise time. For your instance and if the measurement it is correct, then the frequency response of your amplifier at 60Vpp it is approximatelly up to 300/6,6=45,5KHz.
Also, as i reffered in a previous post, related with the original subject of this thread of our friend Nitrate about the THD, i pointed out that with the high quality parts offered today in the market, the only way to make a device with poor THD it is the gross mistakes during the practical implementation of the theoretical circuit. As you can see, already Nitrate discovered errors in ground nodes connection.
As for me, i am not so much occupied with THD, IMD etc. I am interested mainly about the dynamic behaviour of the pre and the power amplifier such the steep bass and the creaky high (i like verry much to hear brass wind instruments, the slapy kick drum sound and the electric bass low notes) so i preffer bookshelf type loudspeakers.
Fotios
fotios said:
Is this correct? I believe it is supposed to be 0.35/risetime, so the -3dB bandwidth is 0.35/6E-6 = 58.3KHz.
Or is yours the -1dB bandwidth?
nitrate said:
Leigh,
Can you go the other way? Could you return the system to the original configuration? Would the fatigue and lifeless sound faithfully return? Then re-apply the changes one-at-a-time to see which is the 'critical' flaw. It seems to me like a really worthwhile exercise.
Congratulations on your progress,
Edward
Boris_The_Blade said:
As i said, this is a coarse estimation so the difference between mine and your calculation it is 13,5KHz. So it does not matters this difference because it is measured with reference a 10KHz square wave. In the classic audio, when the early decade scale used instead the log scale, a decade was any ten-part interval such as the decade resistance boxes which of controls calibrated from 1 to 9 on each knob (ref. D.Davis & C.Davis Sound System Engineering book). Then the low limit was 30Hz and the first decade extended from 30Hz to 300Hz, the second decade from 300Hz to 3000Hz, the third from 3KHz to 30KHz and so on. Periodically expressed the same it is: 33,33ms - 3,333ms - 0,333ms - 33,33ìs and so on. The scale used today it is from 20 to 20000Hz which represents 3 linear decades. So the nr. 300 derives from the linear scale of the first decade used early. Also, it does not matters if we measure the rise time in a 100Hz or in a 1KHz or in a 10KHz signal. In a DSO, if we activate the X10 mag. time base button, for the same amplitude of the signal the calculation of rise time gives the same result in any of the three frequencies.
As for the reference level of measurement, if you see in the plot of my previous post link, it is taken automatically from a Hameg DSO and from the cursors position calculated as: 20log48,4Vpp/60Vpp=-1,86dB
I don't know if there is any mistake in my description, because it is written with hasty, then any correction it is well accepted.
Fotios
Edward,
just for you i reversed all the good work and moved things one by one. Seems the offending item was the volume pot/RCA input connector assembly. They are wired as one 'blob' and secured to a wooden frame connected to one end of a peice of coax about 10inch long. The other end connects to the amp input and the sheild is used as the ground. When this ground is connected to the jigs ground that includes the main resevoir caps return there is a faint ( barley noticable ) hum and the sound is distorted in a way that caused all the trouble mentioned earlier in the thread. When the coax ground is moved to a point were the ground enters onto the proto veroboard the faint hum vanishes and the musical perfermance returns to expected levels. All the hard to trace distortion vanishes and the amp becomes transparent and detailed again. Strange eh??
Leigh
just for you i reversed all the good work and moved things one by one. Seems the offending item was the volume pot/RCA input connector assembly. They are wired as one 'blob' and secured to a wooden frame connected to one end of a peice of coax about 10inch long. The other end connects to the amp input and the sheild is used as the ground. When this ground is connected to the jigs ground that includes the main resevoir caps return there is a faint ( barley noticable ) hum and the sound is distorted in a way that caused all the trouble mentioned earlier in the thread. When the coax ground is moved to a point were the ground enters onto the proto veroboard the faint hum vanishes and the musical perfermance returns to expected levels. All the hard to trace distortion vanishes and the amp becomes transparent and detailed again. Strange eh??
Leigh
nitrate said:
Hi Leigh,
Thanks for sharing this with us. These are very good points.
It's too late now, but with the benefit of hindsight, I'd ask you about some distortion tests on the amplifier before you repaired the grounds. For example, you stated that the amplifier had good THD, but I'm wondering if you measured low-frequency THD, like 50 Hz THD, as a function of power, might you have seen a problem? I would expect the bad effects of a bad grounding topology to reveal themselves as higher levels of low-frequency THD, either as harmonics themselves or as mains harmonics in the THD residual. Unfortunately, to do this kind of measurement really properly, one needs to look at the 50 Hz THD residual with a spectrum analyzer.
Any thoughts?
Cheers,
Bob
Bob, where I see my grounding mistakes is in sidebands. If I'm doing a 1kHz distortion measurement and the grounding isn't right, I'll get blips at 880, 940, 1060, and 1120. That can't be good for the sound. Admittedly, it takes a long sample length and careful windowing to pick all that out, but it's there and it's often more prominent than the harmonics.
Hi Bob,
The problem is easy to reproduce. I'm currently putting my test computer back together with a half decent sound card that is capable of sub 100db measurements. The old one was a soundblaster16 and it wasnt capable of detecting really low distortion. As soon as i do that i will run ARTA on the amp and do some low freq measurments. It may show somthing previously missed on the old SB16 setup running RMMA as it has better spectral analysis functions. If it does i'll do some screen captures and post them in here. I must say the old setup used to show a slight harmonic at around 50Hz but i just ignored it and put it down to the ground hum i already knew about. I suspect that maybe very low freq distortion or harmonics can play havock with my hearing, causing listner fatigue to trick my brain into masking details 'in my head'???
Hmm, hopefully things will reveal themselves in the most unexpected of ways.
Leigh
The problem is easy to reproduce. I'm currently putting my test computer back together with a half decent sound card that is capable of sub 100db measurements. The old one was a soundblaster16 and it wasnt capable of detecting really low distortion. As soon as i do that i will run ARTA on the amp and do some low freq measurments. It may show somthing previously missed on the old SB16 setup running RMMA as it has better spectral analysis functions. If it does i'll do some screen captures and post them in here. I must say the old setup used to show a slight harmonic at around 50Hz but i just ignored it and put it down to the ground hum i already knew about. I suspect that maybe very low freq distortion or harmonics can play havock with my hearing, causing listner fatigue to trick my brain into masking details 'in my head'???
Hmm, hopefully things will reveal themselves in the most unexpected of ways.
Leigh
Janneman,
Your graphs are the first objective argument I've seen i favor tube amplifiers. Did you run those graphs yourself?
Your graphs are the first objective argument I've seen i favor tube amplifiers. Did you run those graphs yourself?
SY said:
Sy, that's a very good point, and makes sense. How far down are those sidebands that you see, usually?
If you run a lower fundamental, like 50, 100 or 150 Hz, that is not synchronous with the mains frequency, those sidebands might be easier to see and more separated from the fundamental.
The sidebands you point out at, like, 880 and 1120, may not show up very well with a THD analyzer as they are fairly close to the fundamental notch filter frequency.
Thanks,
Bob
nitrate said:Hi Bob,
The problem is easy to reproduce. I'm currently putting my test computer back together with a half decent sound card that is capable of sub 100db measurements. The old one was a soundblaster16 and it wasnt capable of detecting really low distortion. As soon as i do that i will run ARTA on the amp and do some low freq measurments. It may show somthing previously missed on the old SB16 setup running RMMA as it has better spectral analysis functions. If it does i'll do some screen captures and post them in here. I must say the old setup used to show a slight harmonic at around 50Hz but i just ignored it and put it down to the ground hum i already knew about. I suspect that maybe very low freq distortion or harmonics can play havock with my hearing, causing listner fatigue to trick my brain into masking details 'in my head'???
Hmm, hopefully things will reveal themselves in the most unexpected of ways.
Leigh
Hi Leigh,
Good. Keep us posted.
Along those lines of looking more closely at low frequency distortions, I have been toying with the idea of a new test for power amplifiers. It is a variation on CCIF IM at 19 & 20 kHz. It is instead two-tone IM at 19 Hz and 20 Hz, resulting in a 1 Hz beat. Looking at the results of this test with a spectrum analyzer in the frequency ranges around 20 Hz, 40 Hz and 60 Hz could prove very interesting.
A variation of this test offsetting the frequencies from multiples of the mains frequency might even be better. Such a combination might be 18 Hz and 19 Hz.
One can also look for any evidence of even-order distortion products at 1 Hz with fairly inexpensive filtering.
The 1 Hz variation in output power under this test might also expose some thermal distortion mechanisms.
Yet another variation of such a test could be a low-frequency version of my MIM test, with three equal tones at 9 Hz, 10.05 Hz and 20 Hz. Who knows what we might find.
Cheers,
Bob
sam9 said:
No, they were shown by Mr. de Lima at his lecture for ETF2007. They used to be on his website but I can't find them anymore.
I did an interview with him which will be published in AX but I don't know when.
But Gerrit Boers has done some independent measurements and found also some cases where a particular combination of (SE) tube gear and speakers led to unusually low system distortion.
Jan Didden
Bob, I've picked them out at -60dB or so, depending on how bad the grounding goof. Measurement resolution is everything- lots of samples.
I'll see if I can dig up a specimen of my own stupidity for all to marvel at.
I'll see if I can dig up a specimen of my own stupidity for all to marvel at.
- Status
- Not open for further replies.