To quote Joachim and as I promised him, I was going to post it to help other people with their build.
What happens is that gain varies big time with the minimum variation due to the OL nature of the design.
A little disclaimer is that I didn't have matched devices in the first place and ended up with 1dB difference within CH R and L.
So I took the channel with slighly less gain and I lowered the resistors from 31.6 down to 30.9 and the gain was matched down to 0.08dB.
This makes you think that it is not important to go crazy matching devices since offset is taken in care by the Servo, and matching can much easier being adjusted by emitter resistance.
By doing that the distortion difference between the two channel was under 0.001%.
Listening confirmed that both channel seem to sound pretty equal, so it's up to you guys if you want to take the extra time to match device.
Sure, having matched device might offer a tiny improvement on the sound, unfortunately I don't have time to try this out on this design, I will leave it up to you guys and if anybody tried this, please let us know if there was any significant improvement from matching.
Another thing that people have been concerned about is tolerance on the RIAA: should it be 0.1% or can it be more?
Well long story short, I have put 1% tolerance parts and the RIAA curve between the two channel is perfectly in track and among the entire audio band there is not even 0.1dB difference between the two channels
Let's now talk about RIAA's value.
I made precise/high-resolution RIAA measurements.
With 73.2K 9.91K 11nF and 32nF the RIAA curve is below 0.1dB on all the audio band.
If 73.2K could be made 73.18K or something like that precision would be even higher than 0.1dB.
Last subject I want to go through, but just as important as the points above, is: in order to make Paradise sound really nice and match RIAA to such a high level of precision you need to use no more than 4700uF on the boosting capacitor.
Forget about 6.8m or even wors 14mF which sound dull and velied.
That's it for now.
If you guys will follow the tips above and the other useful tips that the guys here have made available you will have a very fast, simple, smooth build with a very nice result.
Cheers!
I think we should hold on with the preamp and wait for the balance masterpiece to come alive and then build a balance linestage on the technology eventually learn from it that can also work SE of course with Paradise.
Just my two cents, then you guys can do it the other wat too
I first suggested 6800uF elcaps and MiiB raised them to double. When you lower the elcaps in value you get less deep extension in the very low bass, much under 20Hz. To lower the value of that caps can bring a welcome subsonic filter effect if you have a stability problem in your system.
As far as i know Stefano has rather big B&W speakers that have very high mass woofers.
Actually Stafano hears his complete system so i am not prepared yet to except the lower value as default.
As far as i know Stefano has rather big B&W speakers that have very high mass woofers.
Actually Stafano hears his complete system so i am not prepared yet to except the lower value as default.
Joachin,
thanks for your reply.
Both simulation and measurements on the bench show a +0.08dB at 20Hz.
A typical subsonic filter has -3dB at 20Hz,
Either way listening tests carried out today show a very nice bass extention.
Cello and double bass are reproduced with power and deeper notes have the right articulation.
In this configuration, final bass response will strongly depend on the rest of the equipment.
For instance, using a highly modified denon 103R brings a very nice soundstage and very clear highs and mids.
Unfortunately it's missing the high energy on the bass and body on the mids compared to the upper range ZYX 4D.
If I didn't know this difference on these 2 carts, I could wonder if the value were a tad low (although I could almost say the sound is blameless even with this conficuration and Denon on).
So, depending on what you put on the paradise, it will respond accordingly.
I can say based on measurements and listening test to try these values with confidence and then try also the one that Joachim suggested and then decide for your own.
People who will try different values, please post/share your opinion/experience here.
thanks for your reply.
Both simulation and measurements on the bench show a +0.08dB at 20Hz.
A typical subsonic filter has -3dB at 20Hz,
Either way listening tests carried out today show a very nice bass extention.
Cello and double bass are reproduced with power and deeper notes have the right articulation.
In this configuration, final bass response will strongly depend on the rest of the equipment.
For instance, using a highly modified denon 103R brings a very nice soundstage and very clear highs and mids.
Unfortunately it's missing the high energy on the bass and body on the mids compared to the upper range ZYX 4D.
If I didn't know this difference on these 2 carts, I could wonder if the value were a tad low (although I could almost say the sound is blameless even with this conficuration and Denon on).
So, depending on what you put on the paradise, it will respond accordingly.
I can say based on measurements and listening test to try these values with confidence and then try also the one that Joachim suggested and then decide for your own.
People who will try different values, please post/share your opinion/experience here.
BC327/337 HFE values and match questions
Hi. I started measuring the BC327/BC337, -40 grade HFE value to match them and got these preliminary results. Just checking if I'm on the right track:
BC327 (PNP), 125/500 tested so far:
Larger variation in the HFE values than for the BC337 (NPN) type,
values vary so far from 298 to 630. Many of them are around 370.
I should be able to get enough around 560-570 to match with the NPN
There is a even a possibility that I found enough 610 to 620 ones...
BC337 (NPN), 25 tested so far:
Less variation than PNP type, from 463 to 624, a lot are around 550-575.
Did you found the same type of variation from PNP to NPN type?
Gain and Matched transistor questions:
-I think the higher HFE, the better for the Gain Cell, right? Is 575 enough?
-I guest I have to match both channels for the Gain cell HFE values, right?
-How close is an acceptable match for the Gain Cell? For example 560 to 570 is acceptable for the Gain Cell?
-What kind of match and HFE values are acceptable for the current mirror and buffer sections?
-Finally, do I need special HFE values for the few 327/337 in the shunt regulator sections?
Thanks...
Hi. I started measuring the BC327/BC337, -40 grade HFE value to match them and got these preliminary results. Just checking if I'm on the right track:
BC327 (PNP), 125/500 tested so far:
Larger variation in the HFE values than for the BC337 (NPN) type,
values vary so far from 298 to 630. Many of them are around 370.
I should be able to get enough around 560-570 to match with the NPN
There is a even a possibility that I found enough 610 to 620 ones...
BC337 (NPN), 25 tested so far:
Less variation than PNP type, from 463 to 624, a lot are around 550-575.
Did you found the same type of variation from PNP to NPN type?
Gain and Matched transistor questions:
-I think the higher HFE, the better for the Gain Cell, right? Is 575 enough?
-I guest I have to match both channels for the Gain cell HFE values, right?
-How close is an acceptable match for the Gain Cell? For example 560 to 570 is acceptable for the Gain Cell?
-What kind of match and HFE values are acceptable for the current mirror and buffer sections?
-Finally, do I need special HFE values for the few 327/337 in the shunt regulator sections?
Thanks...
Stefano, again this thread is long. I have discussed the issue of how big the elcaps have to be with Ricardo here because he also asked what they do. I told him that they affect the response in the very deep bass. Actually they form a high pass with the emitter impedance of the input stage. Ricardo did not come back so far about what he found while experimenting with the elcaps. Maybe he got some answers. In general i have no problem when people experiment with the elcap values. I think it is audible how big they are and even with 4700uF the audible range is not effected much so in a certain system adjusting the elecaps will be the ticket.
Algar, i would use the higher Hfe ones in the input stage. 10% matching from PNP to NPN is fine and with 5% you are on the safe side. The higher Hfe produces less base current at the input ( so less DC into the cartridge ) and the matching makes sure that you end up with the same gain in both channels. If you run into a gain match problem try to adjust the emitter resistors. For absolute lowest DC offset at the input you can also adjust Ube. I found that the Ube difference in the transistors we use is very small anyway. They are very symmetric to begin with. That is one of the reasons i prefer them over other BJTs.
I gave that advice to Stefano and he came up with some numbers.
Hfe value and matching in the current mirrors is less critical. Go for better then 10% and Hfe from 400 up will be just fine. For the PSU i think Hfe is not critical at all, so better then 10% and hfe higher then 400 is more then ok in the PSU.
I gave that advice to Stefano and he came up with some numbers.
Hfe value and matching in the current mirrors is less critical. Go for better then 10% and Hfe from 400 up will be just fine. For the PSU i think Hfe is not critical at all, so better then 10% and hfe higher then 400 is more then ok in the PSU.
Last edited:
The ElCAP's value has a lower limit which below that the gain at the bottom will collapse.
The 4.7mF in conjunction with the RIAA's value I gave here, offer a precision of 0.1dB (absolute value or +/-0.05 relative) from 20Hz-20KHz so I don't understand what could be wrong with this value.
If you bump that up, you will have a bump between 20-40Hz, given same values for the RIAA, unless you re-correct them to get it back to that kind of precision, but this will result in the SAME frequency response, with MORE ELCAP on the signal Path.
IMHO the frequency response should be as flat as possible (I am not a big fan of purposly bumbing up the low bass freq response) and it is better to have the lowest ELCAP value your frequency response can afford.
I have experiemented with different values and I feel pretty confortable to relate this aspect of the sound.
Obviously my system is full range and doesn't need any push at the bass; on the flip side, if bass response was weak, you can notice it immediately.
I don't want to go against your recommendation, so I will keep it quite now and all I have left to say is just to try that out and let me know!
Last edited:
To keep phasehift out of the base area you should extend it as deep as possible not just to 20 Hz but preferably one or 2 octaves deeper than that. It does the same as extending the bandwidth upward. I think we all can agree on the need of circuits extending beyond our hearing limit of say 15-16 KHz...
I have no respect for your argument on keeping the cap value as low as possible...what should be the benefit of that..??
I have no respect for your argument on keeping the cap value as low as possible...what should be the benefit of that..??
well try for yourself!
If you don't respect my statement, that's ok, but at least try it first and listen to the difference and then you can disagree with it
with the 2x6800 it starts to roll off below app 2 HZ, think that's sufficient...
Stfanoo in your system band-with limitation and subsonic filtering may sound better, but you state is like it's true for all systems. For that you have no valid knowledge. What is good for you may not suit the rest.
I state here why the larger capacitor was chosen with solid arguments.
Stfanoo in your system band-with limitation and subsonic filtering may sound better, but you state is like it's true for all systems. For that you have no valid knowledge. What is good for you may not suit the rest.
I state here why the larger capacitor was chosen with solid arguments.
