If one want to use the core and dare to use network no. 4 with defined T4, you can fall back on EXakt MK2. The correction consists of R15 & C9, it now moves from the input to the output, so it resides exactly between Supra24 and Ruby3 ... will that work 😉 (?).
Now, of course, we do without C206, or 100pF to 470pf does no harm. We now set Ruby's load (R203) to 5k6 as standard. The series resistor R107 can now also be set back to 47 to 51 ohms, depending on the required output resistance. We know 50 ohms, 150 ohms and 600 ohms, 75 ohms are also not unknown but uncommon in the LF sector.
Enjoy your journey through the pitfalls of analog audio signal processing,
HBt.
Now, of course, we do without C206, or 100pF to 470pf does no harm. We now set Ruby's load (R203) to 5k6 as standard. The series resistor R107 can now also be set back to 47 to 51 ohms, depending on the required output resistance. We know 50 ohms, 150 ohms and 600 ohms, 75 ohms are also not unknown but uncommon in the LF sector.
Enjoy your journey through the pitfalls of analog audio signal processing,
HBt.
Last edited:
Your pace is extraordinary. I have a lot to read up on in this thread beside the intriguing 23 page SLN-Thread You mentioned. Matching the amplifier interfaces is certainly an overlooked problem in hifi. I guess this is a reason people buy and justify insanely expensive interconnects without understanding the physics around them. Not that I understand, it's just a hypothetical cause to why people can hear the difference.
This is precisely the crux of the matter. Let's take the new core (component dimensions /values !) of the Supra and look at its dynamic output resistance. It is frequency-dependent because the negative feedback is frequency-dependent; it falls with the frequency. Initially it is huge - i.e. supra24 is not loadable and certainly not frequency-dependent loadable. My quick idea #161 of using the Lego bricks from another circuit is therefore very questionable. Everything we connect between the core and the buffer loads the core - and that's exactly what we don't want. This concept only works with the ruby3 output driver.
So if you see kit suppliers trying to sell you a simple Supra-Lego box, beware. It will lack the relief, the buffer at the output.The Elektror Supra original does not necessarily need this, due to the relatively high quiescent current and the lower internal resistance.
This side project was intended to provide practical proof that a statically precisely balanced core is a fine thing and that nothing really disturbs the sound /no noise. However, there is one very underestimated point and that is the input, if possible we should always try to design the input pair as jFet.
But that would be a new journey. Let's be clear: Rx and Cx must be variable, i.e. adaptable. The MM pickup system is our biggest technical problem.
HBt.
supra24 works as it is, so let's enjoy it.
So if you see kit suppliers trying to sell you a simple Supra-Lego box, beware. It will lack the relief, the buffer at the output.The Elektror Supra original does not necessarily need this, due to the relatively high quiescent current and the lower internal resistance.
This side project was intended to provide practical proof that a statically precisely balanced core is a fine thing and that nothing really disturbs the sound /no noise. However, there is one very underestimated point and that is the input, if possible we should always try to design the input pair as jFet.
But that would be a new journey. Let's be clear: Rx and Cx must be variable, i.e. adaptable. The MM pickup system is our biggest technical problem.
HBt.
supra24 works as it is, so let's enjoy it.
I have to apologize and go back to bed, influenza with a high fever for three days. It's extremely exhausting.
Be careful with You health. Rest to get well. We can wait. Best wishes
Thank you very much for your wishes 🙂
It is getting better, no more fever and all secondary infections are mild.
At the moment I don't know exactly how this thread can develop any further. I even think it was a mistake to give the core with its jFet buffer (from my posting #1) the name SUPRA2024.
The association with the original 1980s MC/ MM equalizer SUPRA was carelessly thrown into the room by the user adason, which Marcel and Nick then gladly took up to philosophize about the phenomenon of weighted noise ... You can read all about it in the relevant threads (not only in the OREAD thread). But the noise number is now so old that it just tires me out even more - and the same narrative style over and over again.
To keep this short:
oread uses a special bjt operational amplifier approach and supra24 uses /another one. With the latter, however, the core is not loadable. Actually, we don't want anything like that. We need our own, resilient output stage, basically the ideal impedance converter. This little illustrious appendix is ruby3.
If we look closely, we will see that the basic topology is the same, but we will also recognize that the dimensioning is fundamentally different. Unfortunately, we are now also using the network of the Elektor Supra as a frequency-dependent negative feedback, which of course reinforces the apparent similarity to the Elektor Supra.
So not only would a completely different choice of name have been necessary for a clear differentiation, but also a different network - for differentiation.
The present circuit is quite unique!
And with this (Morten's PCB) layout, it can be set up without complaint.
greetings,
HBt.
It is getting better, no more fever and all secondary infections are mild.
At the moment I don't know exactly how this thread can develop any further. I even think it was a mistake to give the core with its jFet buffer (from my posting #1) the name SUPRA2024.
The association with the original 1980s MC/ MM equalizer SUPRA was carelessly thrown into the room by the user adason, which Marcel and Nick then gladly took up to philosophize about the phenomenon of weighted noise ... You can read all about it in the relevant threads (not only in the OREAD thread). But the noise number is now so old that it just tires me out even more - and the same narrative style over and over again.
To keep this short:
oread uses a special bjt operational amplifier approach and supra24 uses /another one. With the latter, however, the core is not loadable. Actually, we don't want anything like that. We need our own, resilient output stage, basically the ideal impedance converter. This little illustrious appendix is ruby3.
If we look closely, we will see that the basic topology is the same, but we will also recognize that the dimensioning is fundamentally different. Unfortunately, we are now also using the network of the Elektor Supra as a frequency-dependent negative feedback, which of course reinforces the apparent similarity to the Elektor Supra.
So not only would a completely different choice of name have been necessary for a clear differentiation, but also a different network - for differentiation.
The present circuit is quite unique!
And with this (Morten's PCB) layout, it can be set up without complaint.
greetings,
HBt.
Could a different name for the by-product (a kind of booger) from the original small craft project have made it more popular? I now very much doubt it - and yet this discreet DIY project is also a worthwhile one.
Are there any questions, what can I (also) do?
Are there any questions, what can I (also) do?
Let's use MC12 and check how good the actual network is in a matter of seconds.
Just to play with this tool a bit.
Perhaps someone dares to use the mysterious network no. 3? In terms of sound, it will do absolutely nothing - the best network is number 4! But one could differentiate ourself from the competition 😉
Just to play with this tool a bit.
Perhaps someone dares to use the mysterious network no. 3? In terms of sound, it will do absolutely nothing - the best network is number 4! But one could differentiate ourself from the competition 😉
A comment on the sound:
the low noise is striking (this is exactly what the core should prove, no more and no less), but also the tendency to reconstruct shattering glass with all its hardness. supra24 and ruby3 are tough reproducers of hard, short strokes - no whitewashers. Absolutely the opposite of soft, they are extremely fast. Boom and done.
the low noise is striking (this is exactly what the core should prove, no more and no less), but also the tendency to reconstruct shattering glass with all its hardness. supra24 and ruby3 are tough reproducers of hard, short strokes - no whitewashers. Absolutely the opposite of soft, they are extremely fast. Boom and done.
RIAA deviation (theoretical), if we calculate the components of Network 2 exactly and use them in the same way.
150kOhm
12.09385kOhm
19.85nF
6.715nF
Now the sounding can begin ..!
150kOhm
12.09385kOhm
19.85nF
6.715nF
Now the sounding can begin ..!
Playground
Transfer function, the load represents the network itself ...
The dynamic input resistance of the negative feedback core ...
128k98 - 20kHz
131k23 - 10kHz
Transfer function, the load represents the network itself ...
The dynamic input resistance of the negative feedback core ...
128k98 - 20kHz
131k23 - 10kHz
Any news from the kit scene?
How does the direct comparison between OREAD and this branch the Supra24 plus Ruby3 compare?
How does the direct comparison between OREAD and this branch the Supra24 plus Ruby3 compare?
"Hausmittel"
Can the transfer function of this amplifier be determined directly and brutally? Yes, you can - using MLS and cross-correlation.
The left channel is completely congruent with the right channel and the relative deviation from the RIAA specification remains completely within limits, the measurements confirm the MC12 simulation result ..!
Can the transfer function of this amplifier be determined directly and brutally? Yes, you can - using MLS and cross-correlation.
The left channel is completely congruent with the right channel and the relative deviation from the RIAA specification remains completely within limits, the measurements confirm the MC12 simulation result ..!
Attachments
How can you convince the external USB module “Creative Sound Blaster SB1240” and "Adacity" to actually work with 24Bit/192kHz? fh is always exactly at 22k05Hz. Does win11 need a special driver or does the internal digital filter always take care of this limit?
It would be really nice if we practically didn't need any real measurement technology anymore, just a notebook, a bit of software and this external ADC / DAC even with an analog phono equalizer on board.
Does anyone know anything about this blaster?
It would be really nice if we practically didn't need any real measurement technology anymore, just a notebook, a bit of software and this external ADC / DAC even with an analog phono equalizer on board.
Does anyone know anything about this blaster?
Hardcore measurements with the already mentioned TACET test record ... this is my right channel ! Once again, the famous KEITHLEY 2015 THD Multimeter had to serve as the measuring device.
Values marked in red are definitely reading or measuring errors.
Interesting is the fact that we do not determine the absolute deviation of our EQ, but relative to the deviation of our EQ the deviation of the RIAA preemphasis EQ of the Neumann VMS80 cutting machine used for cutting the LP template ..!
Furthermore, we don't know whether the fh is >14kHz and falls at 6dB per octave or somewhere around 20kHz and falls at 12dB per octave.
TPC
