I felt that something was wrong with latest incarnation of the FE PS so I've tried to go back to BZX85 zeners, much better, more dynamic and entertaining.
The 1N53xx sounds in the FE PS too much warm and not as dynamic as the BZX85.
...
Thus it seems that the 1N53xx needs more current to work at its best...
Now I've to decide if it's better to increase the Zener current (and thus the LM317 CCS current) increasing also heat generation or simply use the BZX85s
I've tried to increase total current to 46mA and zener current to 19.6mA for 1N53xx.
A bit better but, apart that transistors and regulators starts to need heatsinking, it seems that the basic sound signature of both diodes is always evident...
So current seems the wrong parameter to consider.
The other thing that changes between the two diodes is dynamic impedance, according to datasheets there is an order of magnitude between them (1N53xx has the much lower one).
To the experts:
- could this be the parameter that give this signature?
- it could be that the fact 1N53xx are temperature compensated makes the difference?
BTW both diodes sounds good, only different timbre.
Last edited:
Yes, the dynamic impedance will affect the regulator performance.
I suspect you are getting a different spectrum of harmonics at the output of the regulator. That different spectrum is being "heard" as a different colouration.
That's why way back I advised against guessing at component changes.
I suspect you are getting a different spectrum of harmonics at the output of the regulator. That different spectrum is being "heard" as a different colouration.
That's why way back I advised against guessing at component changes.
Andrew, you have done the least explanation in this discussion, perhaps you could elaborate on each of my points as to how the rCLC PSU filters perform better in each?
1. The case of ripple, which the source is upstream of the filter, and we want to keep downstream voltage more constant.
2. The case of varying load, which the source is downstream of the filter, and we still want to keep the downstream voltage more constant.
1. The case of ripple, which the source is upstream of the filter, and we want to keep downstream voltage more constant.
2. The case of varying load, which the source is downstream of the filter, and we still want to keep the downstream voltage more constant.
This is a point where you will start going in circles if you rely only on listening tests. A problem I ran into a few years ago.
The harmonics caused by the interconnects and the interface is not small, in most cases just less than speakers and a lot more than good electronics in general. The main reason I had to dig and solve the interface problem. But if the MyRef has the high frequency filter like that, even my interconnect will not solve the problem.
As I go through various mods, the sound will generally get quieter, the timbre gets finer and more detail is revealed. This is also why I only modify both channels to the same when I think the difference reaches a good milestone,
Last edited:
Fine.
Can you elaborate a bit?
In particular do you think is better to have a lower impedance zener in this particular case?
Thanks in advance.
I do agree, it's not easy to understand which signature is better BTW I'm settling on the warmer presentation of 1N53xx but this could change...
What do you mean with 'timbre gets finer', warmer? clear? whatever?
Dario,
Originally Posted by ClaveFremen
In the meanwhile I've read again this great W. Jung's article and I'm tempted to try the LM317+MOSFET cascode CCS (image 13c).
In Jung's article he writes:
"If complementary source and sink circuits are needed for an application, it is actually better performance-wise to use a pair of LM317s ... than it would be to use an LM317 and an LM337."
Have you tried replacing the LM337 with an LM317 current-sink? Does it make a difference?
In post #297 one of LinxGuru's simulations was the equivalent of just replacing R1/R4 with a CCS (no regulator transistors). Adding LM317 CCSs would be a fairly trivial mod to the TP boards. Does anyone have a view on whether this would work or be worthwhile?
Sorry if that's a stupid question - just point me to a reference if it is.
Thanks
Geoff
Originally Posted by ClaveFremen
In the meanwhile I've read again this great W. Jung's article and I'm tempted to try the LM317+MOSFET cascode CCS (image 13c).
In Jung's article he writes:
"If complementary source and sink circuits are needed for an application, it is actually better performance-wise to use a pair of LM317s ... than it would be to use an LM317 and an LM337."
Have you tried replacing the LM337 with an LM317 current-sink? Does it make a difference?
In post #297 one of LinxGuru's simulations was the equivalent of just replacing R1/R4 with a CCS (no regulator transistors). Adding LM317 CCSs would be a fairly trivial mod to the TP boards. Does anyone have a view on whether this would work or be worthwhile?
Sorry if that's a stupid question - just point me to a reference if it is.
Thanks
Geoff
For example,
An externally hosted image should be here but it was not working when we last tested it.
This album can sound very dry on many systems. As the mods progress, the timbre begins to give a nice delicate impression with the timbre having a thin trail, not so long, but gives you an impression the the violin is right in the room. There is very little room reflections in this recording, but as the mods progress, the slight room reflection appear very faintly.
An externally hosted image should be here but it was not working when we last tested it.
This album sounds harsh in many systems. I remember not liking it so much, but once I was testing a new interconnect design at a friend's home, he had the Focal Utopia Be speakers and intentionally put this album in his CD player. To our surprise, the harshness was much less than we expected. As I go through the mods here, this album again gives a more delicate timbre, that makes the music very beautiful.
In both of these, one can gradually appreciate the playing skills the performer uses. It seems that you can feel the performer is using different strength and emotion in various portions of the performance.
Last edited:
Hi Geoff,
I've read too that passage but I've yet to try it, I'll report about it.
When I've tried the first version of the FE reg initially there were no CCS and it was a clear improvement.
The CCS was another not small step forward.
So I expect it to be a nice improvement even without the transistor.
I'll try it and report.
I was able to find only Salvatore Accardo's one as a digital download from Linn.
In the next days I'll search better but Classic is not the kind of music I listen to regularly, we'll see
Just tried, with OnSemi LM3x7 and 51R resistors, they're a nice step-up from resistors.
You can expect (much) tighter bass, refinement, slightly warmer sound and blacker background.
Different type of music reveal different things, I have not tried to see if other Salvatore Accardo albums have the same effect, because that one was the only one I had at the time, and it sounded the most terrible. Perhaps you could also recommend some albums you use, because I am not familiar with the more recent popular music. All my rock collections are pretty old artists. The strangest one I have is probably the Verve Remixed collection which I bought in Australia.I was able to find only Salvatore Accardo's one as a digital download from Linn.
In the next days I'll search better but Classic is not the kind of music I listen to regularly, we'll see
There's one pop disk I use for the same reason:
Paul McCartney - Pipes of peace
This CD master has a very high dynamic range and highs are very hot.
That is also an old recording. I have a few Beatles albums. but none of either after they broke up.
The problem with most popular music is that the recording is too heavily mixed, thus the spacial cues are lost or confusing.
A relative of mine recorded with a band "The Thirstbusters". He said it was recorded in a professional studio, but it was terribly done according in comparison with others.
TAS list are not so bad, although there are better ones out there.
The problem with most popular music is that the recording is too heavily mixed, thus the spacial cues are lost or confusing.
A relative of mine recorded with a band "The Thirstbusters". He said it was recorded in a professional studio, but it was terribly done according in comparison with others.
TAS list are not so bad, although there are better ones out there.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.