On a more serious note, what is the difference, if any, in using GR and BL grades of the JFETs ?
The bypass current source can compensate for the difference in bias current, not ?
In theory yes.
There are two points to be aware of.
With a 50R source resistor, a GR grade with typically 4mA Idss will have a bias current of ~1.8mA.
With a target bias of 1.2mA, we need to bypass 0.6mA.
You can use a 0.6mA current source, or a ~10k resistor.
A BL grade with typically 8mA Idss will have a bias current of ~3.2mA under the same condition.
You now have to use a 2mA current source for the bypass.
As things are in real life, a CCS of 0.6mA using degenerated 2SK208 is much more stable than one at 2mA.
Apart from that, the Yfs of the JFETs changes with actual bias current.
Yfs at 3.2mA is ~1.5x that at 1.8mA.
So the open loop gain also changes.
This is not to say that it will not work with BL grade.
It is a matter of how close you want to get to the original.
Obviously the lower the Idss, the better in that sense.
Patrick
The bypass current source can compensate for the difference in bias current, not ?
In theory yes.
There are two points to be aware of.
With a 50R source resistor, a GR grade with typically 4mA Idss will have a bias current of ~1.8mA.
With a target bias of 1.2mA, we need to bypass 0.6mA.
You can use a 0.6mA current source, or a ~10k resistor.
A BL grade with typically 8mA Idss will have a bias current of ~3.2mA under the same condition.
You now have to use a 2mA current source for the bypass.
As things are in real life, a CCS of 0.6mA using degenerated 2SK208 is much more stable than one at 2mA.
Apart from that, the Yfs of the JFETs changes with actual bias current.
Yfs at 3.2mA is ~1.5x that at 1.8mA.
So the open loop gain also changes.
This is not to say that it will not work with BL grade.
It is a matter of how close you want to get to the original.
Obviously the lower the Idss, the better in that sense.
Patrick
So then how about 2SK246Y / 2SJ103Y ?
They are easier to get, especially the SMD equivalent.
But they have way lower transconductance.
So you loose about 6~10dB in open loop gain, even taking the source resistor into account.
And of course they're also quite a bit noiser.
Patrick
They are easier to get, especially the SMD equivalent.
But they have way lower transconductance.
So you loose about 6~10dB in open loop gain, even taking the source resistor into account.
And of course they're also quite a bit noiser.
Patrick
What about temperature dependence? With the BL-grade the drain current will increase more with temperature than the GR-grade. Can this be compensated by the CCS?
According to Borbely, the zero tempco point of the JFET is given by :
Vgs = Vp + 0.63V
https://audioxpress.com/article/JFETs-The-New-Frontier-Part-1
That means not only is the zero tempco point different for different Idss grades.
It also means that e.g. a 2SK208GR has totally different tempco as a 2SK170GR.
Ideally one should use the JFETs at their zero tempco point.
Practically this is seldom possible given the limited availability of JFETs.
So one would just have to allow for the bias change between cold and warm.
In our examples, the steady state bias is about 20% higher than cold.
And large heatsinks always help.
Patrick
Vgs = Vp + 0.63V
https://audioxpress.com/article/JFETs-The-New-Frontier-Part-1
That means not only is the zero tempco point different for different Idss grades.
It also means that e.g. a 2SK208GR has totally different tempco as a 2SK170GR.
Ideally one should use the JFETs at their zero tempco point.
Practically this is seldom possible given the limited availability of JFETs.
So one would just have to allow for the bias change between cold and warm.
In our examples, the steady state bias is about 20% higher than cold.
And large heatsinks always help.
Patrick
Regarding testing:
We compare , also we contrast. What you drink when is just as important as what you drink.
And also important: inform the testers about what to expect, i.e. say that they can get served the same wine;-)
With amplifiers, for example, the regression can be heard more clearly, in my experience.
And as a recommendation: I like to use some small box, with almost nothing in it, the wires sticking out, criss-crossing, an electrolytic capacitor bumming around - something like this. I then put it up against a fat and expensive, what do I know, with a champagne front, weighing 20 kilos or more, open-ended price, really "high-end"-) The sound result is then described as "epochal", "my gosh is that cool" or "that's impossible"-)
😉
We compare , also we contrast. What you drink when is just as important as what you drink.
And also important: inform the testers about what to expect, i.e. say that they can get served the same wine;-)
With amplifiers, for example, the regression can be heard more clearly, in my experience.
And as a recommendation: I like to use some small box, with almost nothing in it, the wires sticking out, criss-crossing, an electrolytic capacitor bumming around - something like this. I then put it up against a fat and expensive, what do I know, with a champagne front, weighing 20 kilos or more, open-ended price, really "high-end"-) The sound result is then described as "epochal", "my gosh is that cool" or "that's impossible"-)
😉
I don't know if the question has been raised but do you intend to sell some? PCB or finished product.
For PCBs, you can use the Gerber in post #12, not ?
https://www.diyaudio.com/community/threads/hiraga-le-monstre-2024.413301/post-7696593
For JFETs, I was told that someone in Slovakia might have 100's of Y-grade.
So no need to ask us, as we only have GR grade.
And you can get GR grade on eBay.
For other transistors, depending on test results, either they can be bought from Mouser, or from eBay (for the originals).
But if you want to use what we have, when we have finished all the testing and listening, and if there is sufficient interest, we'll make a package available, like the F5Pi.
How exactly it might look like, we don't know yet.
Depends on which version comes out as winner.
Maybe there's no clear winner afterall.
🙂
Patrick
https://www.diyaudio.com/community/threads/hiraga-le-monstre-2024.413301/post-7696593
For JFETs, I was told that someone in Slovakia might have 100's of Y-grade.
So no need to ask us, as we only have GR grade.
And you can get GR grade on eBay.
For other transistors, depending on test results, either they can be bought from Mouser, or from eBay (for the originals).
But if you want to use what we have, when we have finished all the testing and listening, and if there is sufficient interest, we'll make a package available, like the F5Pi.
How exactly it might look like, we don't know yet.
Depends on which version comes out as winner.
Maybe there's no clear winner afterall.
🙂
Patrick
Thanks for the response Patrick.
I must have all the transistors in stock including the k170.
I have very little personal time and I choose projects that I store for when I have more time.
I had an original monster on battery many years ago and like many people like us "I sold it for something better" but I always had excellent memories of it and I would like to have one again. the House.
I will follow and see how it develops.
I must have all the transistors in stock including the k170.
I have very little personal time and I choose projects that I store for when I have more time.
I had an original monster on battery many years ago and like many people like us "I sold it for something better" but I always had excellent memories of it and I would like to have one again. the House.
I will follow and see how it develops.
I also started with batteries permanent "in floating", unfortunately the final electricity bill at the end of the year was a sour (WAF) surprise....
But, music in heaven with my Octopus monster ...
Alles, salukes.
Karel
But, music in heaven with my Octopus monster ...
Alles, salukes.
Karel
@EUVL
Thanks for the interesting article and the distortion measurements.
Do you think the old 2SA872 and its complement are contributing to higher distortion, or is it a function of the older output devices?
(I was planning on using these old Hitachis in a project of mine because of their excellent datasheet specs, that suit me very well, especially high V_Early of the PNP and high beta at low collector currents).
Thanks,
Alex
Thanks for the interesting article and the distortion measurements.
Do you think the old 2SA872 and its complement are contributing to higher distortion, or is it a function of the older output devices?
(I was planning on using these old Hitachis in a project of mine because of their excellent datasheet specs, that suit me very well, especially high V_Early of the PNP and high beta at low collector currents).
Thanks,
Alex
I had a rummage. It's been ages, but it makes me want to use them again. It's also should be encouraging to build Hiraga's Le Monstre without a circuit board. You could compare what a setup without a PCB would sound like...
Here with a pair of SK246/SJ103. 2 pairs of BC550/BC560, a pair of TIP2955/TIP3055. In a few weeks I might have time to replace the TO-247 TIPs with any TO-126 types. Let's see what works with them. The sound was already very snappy, rich and funky;-)
However, I won't play around with them for long, because listening to music with complementary-parts-pp is no fulfillment if you have other ways and means)-;
If you use batteries, accus, you should listen without any caps;-)
PS: Regarding the wine test: We once had a fabulously good Sauvignon Blanc in our wine bar. A dream. The next delivery arrived, same label, same vintage, same producer: we had to tip this one away. Our guests thought so too. We called the winemaker and he confessed that he had to buy from another winemaker because his Sauvignon Blanc tasted so good and was selling like hot cakes;-)
Here with a pair of SK246/SJ103. 2 pairs of BC550/BC560, a pair of TIP2955/TIP3055. In a few weeks I might have time to replace the TO-247 TIPs with any TO-126 types. Let's see what works with them. The sound was already very snappy, rich and funky;-)
However, I won't play around with them for long, because listening to music with complementary-parts-pp is no fulfillment if you have other ways and means)-;
If you use batteries, accus, you should listen without any caps;-)
PS: Regarding the wine test: We once had a fabulously good Sauvignon Blanc in our wine bar. A dream. The next delivery arrived, same label, same vintage, same producer: we had to tip this one away. Our guests thought so too. We called the winemaker and he confessed that he had to buy from another winemaker because his Sauvignon Blanc tasted so good and was selling like hot cakes;-)
Attachments
![DSCN0463[1].JPG](/community/data/attachments/1228/1228311-51eac776b9818f9ebb9cb4c628f3c165.jpg?hash=UerHdrmBj5)
All the easily available transistors we used in the Le Monstre "Modern" have very good hfe vs Ie linearity.
So it is tempting to say that the devices chosen 40 years ago contributes to the high distortion.
But the measurements in post #40 confuses the picture, as the distortion is quite low, even with those NOS transistors.
https://www.diyaudio.com/community/threads/hiraga-le-monstre-2024.413301/post-7698384
It may well be that (some of the) transistors deteriorated after 40 years.
It may be something else we do not yet understand.
It may even be that people like some combinations of distortion to "sweeten" the sound.
Fran is on holiday.
And after that he still needs to prepare two complete heatsink enclosures and power supplies before we can do proper ABx listening.
So some weeks to go before we have proper listening impressions.
But he seemed to be loving it for the few hours he spent on the LM-Modern and the LM-Original.
🙂
Patrick
Yes, I got to spend several hours yesterday listening to both versions, but not directly one against the other. I was very impressed, and that was just running from a lab supply into large caps and then to each amp board. It certainly has given encouragement for the work that has yet to come in building the versions so that we can listen to one against another.