Some Hfe meters only use a low current to test transistors which means it isn't a true representative of normal current flow Hfe.
Nigel "Any improvements by using different transistors would be minimal."
I am convinced that changing the passive components, the variations are rather marginal, but my experience with the tube suggests to me the opposite: the tube replacement involves significant changes.
I do not know if this happens even with the solid state, I have never tried and and the reason for my question.
Thank you.
I am convinced that changing the passive components, the variations are rather marginal, but my experience with the tube suggests to me the opposite: the tube replacement involves significant changes.
I do not know if this happens even with the solid state, I have never tried and and the reason for my question.
Thank you.
The sound quality that is unique to old and simple designs is often is due to the semis used. In the various JLH class A amplifier versions, there has a been an evolution from the original MJ480 output type to 2N3055, MJ15003 and now to plastic types as the favoured TO3 types become obsolete. See transistor substitutions here: The Class-A Amplifier Site - JLH Class-A Update
I like the idea of lower distortion but I'm not so happy that it means less of the sweet distortion that I built the thing for in the first place. Well, JLH advised that MJ15003 was evaluated by friends as slightly better sounding than 2N3055 and that's where I would stop, rather than step up to really good low frequency types like MJL21194 that come into their own in Class AB.
In other words, be certain of what is implied by improvements and whether you really want them before assuming that any or all will improve sound quality, for example.
I like the idea of lower distortion but I'm not so happy that it means less of the sweet distortion that I built the thing for in the first place. Well, JLH advised that MJ15003 was evaluated by friends as slightly better sounding than 2N3055 and that's where I would stop, rather than step up to really good low frequency types like MJL21194 that come into their own in Class AB.
In other words, be certain of what is implied by improvements and whether you really want them before assuming that any or all will improve sound quality, for example.
Hi Ian,
Thanks for the suggestion link, I must say that I already knew it.
My JLH has originated precisely from that written and from those schemes.
In fact I mounted the suggested final transistors (mj15003), and I like them as well, however I would like to know the difference in sound that the MJ21194 can give in practice. These are perhaps the highest expression of T03 technology for this kind of high thermal dissipation applications, in fact the much higher ft, the control on distortion even if accompanied by a lower Beta, on paper should offer superior performance. My doubt remains, not having experience in the matter and how they deviate sonically from each other to equal conditions: bass, mid, high, sound stage, dynamics, cleanliness, etc.
All this to understand if the substitution is good and improvements in some sound parameter, or if in the face of this quality may produce defects and therefore not worth it.
Surely if someone has tried physically can report their conclusions.
Thanks, greetings to everyone.
Mleod
Thanks for the suggestion link, I must say that I already knew it.
My JLH has originated precisely from that written and from those schemes.
In fact I mounted the suggested final transistors (mj15003), and I like them as well, however I would like to know the difference in sound that the MJ21194 can give in practice. These are perhaps the highest expression of T03 technology for this kind of high thermal dissipation applications, in fact the much higher ft, the control on distortion even if accompanied by a lower Beta, on paper should offer superior performance. My doubt remains, not having experience in the matter and how they deviate sonically from each other to equal conditions: bass, mid, high, sound stage, dynamics, cleanliness, etc.
All this to understand if the substitution is good and improvements in some sound parameter, or if in the face of this quality may produce defects and therefore not worth it.
Surely if someone has tried physically can report their conclusions.
Thanks, greetings to everyone.
Mleod
Newer BJTs
Just remember that the new BJTs will have to be left for at least 30 to 50 hours to "settle in" before one can make an evaluation of the change in sound.
Just remember that the new BJTs will have to be left for at least 30 to 50 hours to "settle in" before one can make an evaluation of the change in sound.
My prediction is the MJ21194 will bring better result. As an individual transistor, the MJ21194 is a better transistor. Whether it suits the existing design/circuit is another matter, something you may want to try.
1. fT AND SUITABILITY WITH EXISTING DESIGN.
MJ15003 and MJ21194 are both not the original transistor required by the design. But MJ21194 is closer to what JLH wanted in term of bandwidth (fT), which is at least 4M (MJ15003 is 2M) like the MJ480. Transistors with higher fT (like the modern Sankens) tend to give smoother high frequencies. But big change to fT requires change also to the amplifier design because it relates with stability and compensation.
2. CURRENT CAPABILITY.
Later after JLH 1969, there were more transistor options than 2N3055 or MJ480. It is the main secret sauces of Naim amplifiers to use high current transistors (VAS, driver, output). The BUX/BUV series are high current (switching) transistors that became rare today. Still exist from Motorola is MJ15003 and 2N3773 (but not at a level similar to original Naim transistor).
Current is responsible for dynamic and dynamic is responsible for musicality and enjoyment. Listen for the old Naim if you want to know what I'm talking about. So in this case, MJ15003 might have a slight edge over MJ21194 but I doubt it. The JLH was not designed for dynamics after all.
3. CAPACITANCE (Cob)
If you don't need more power, there is no need to use high Cob device (such as MJL3281 for high power and MJL0281 for smaller power). The MJ21194 has half the capacitance of MJ15003. That's why I'm sure the MJ21194 will sound better. I will not describe how this low Cob sounds like but if you are familiar with the sound of Toshiba TTC5200 (lower Cob) as opposed no mainstream Sanken (MT200) then you know what I'm talking about. From Sanken I favor 2SC4466 (and probably 2SC2837).
4. LINEARITY
They are both equal, even the 2N3773, for no more than 1A bias. Those high current transistors used in old Naim designs are almost all non-linear. Non-linear transistors tend to have higher distortion, but even if you manage to design a circuit with low distortion, I don't like non-linear transistors.
Last edited:
The Ft of MJ21194 is not substantially better than MJ15003 but linearity is and that means lower distortion, as said. That will likely be a problem if you built a JLH amplifier for its unique sound quality (and why else would you build such a basic, hot-running and inefficient amplifier?) Perhaps a visit to the Pass Labs Forum would enlighten us when we see how popular a single, high distortion and very hot power Mosfet can be as a power amplifier.
I won't be rushing to try the substitution and maybe the only way to satisfy curiosity will be to do it yourself, assuming you can find affordable TO3 versions if needed for your existing amplifier. You may find that the JLH class A design isn't all that special to you anyway and conventional class A or other classes suit you just as well. I don't think opinions will help with that sort of desision.
Regarding "settling in", I agree that there is a period of time needed for our hearing to normalise with audio system changes but not vice-versa for such a length of time. This is easy to demonstrate with 2 or more amplifier channels of the same type, one just completed and the other having being completed and listened to for some days. It's possible that power supply caps being in common with both channels would affect the result but this can be accommodated if you want to prove a point.
I won't be rushing to try the substitution and maybe the only way to satisfy curiosity will be to do it yourself, assuming you can find affordable TO3 versions if needed for your existing amplifier. You may find that the JLH class A design isn't all that special to you anyway and conventional class A or other classes suit you just as well. I don't think opinions will help with that sort of desision.
Regarding "settling in", I agree that there is a period of time needed for our hearing to normalise with audio system changes but not vice-versa for such a length of time. This is easy to demonstrate with 2 or more amplifier channels of the same type, one just completed and the other having being completed and listened to for some days. It's possible that power supply caps being in common with both channels would affect the result but this can be accommodated if you want to prove a point.
Yes, many people are deft but they worry too much about amplifier sound quality
Seriously, JLH is still the best amp at one certain criterion. I know only the Ranchu-AKSA that comes close.
I got this schematic: Quasi FET/BJT 2016 (Ranchu, AKSA) | minek123 | Flickr
There is info available via Google. It looks like a similar layout to the JLH - it appears to be Class A and has 2 transistors setting up a voltage divider arrangement on a single power supply rail, as the JLH. I'm not sure if this is correctly referred to as "single ended". The difference is that the R-A has a BJT in the lower half of the voltage divider pair and a FET for the upper.
There is info available via Google. It looks like a similar layout to the JLH - it appears to be Class A and has 2 transistors setting up a voltage divider arrangement on a single power supply rail, as the JLH. I'm not sure if this is correctly referred to as "single ended". The difference is that the R-A has a BJT in the lower half of the voltage divider pair and a FET for the upper.
Member
Joined 2009
Paid Member
I have found that modern silicon almost always means excellent sound - the issue is the rest of the circuit around it, whether it is harsh or soft depends a lot on the feedback loop and it's compensation. I haven't tried an output device upgrade with my JLH yet though (which I would like to do) - BUT from the articles he wrote there were significant differences in performance to be obtained by the differences in Hfe for the output parts, not only in absolute terms but also whether the 'top' or 'bottom' output device had the higher Hfe.
HI everyone!
Bigium
Even I, reading, I had understood the importance of the HF substrate, if I am not mistaken, even by JLH himself, but this value seems to have passed into the background.
Today more considered are other characteristics.
However, returning to the comparison between the two types of transistors here is the highest Hfe value of mj15003 on mj21194 Extremely opposite from what other features might suggest.
Here comes my biggest doubt, from which the request for impressions and suggestions, especially from those who have really done this test.
Thank you all.
Greetings.
Mleod
Bigium
Even I, reading, I had understood the importance of the HF substrate, if I am not mistaken, even by JLH himself, but this value seems to have passed into the background.
Today more considered are other characteristics.
However, returning to the comparison between the two types of transistors here is the highest Hfe value of mj15003 on mj21194 Extremely opposite from what other features might suggest.
Here comes my biggest doubt, from which the request for impressions and suggestions, especially from those who have really done this test.
Thank you all.
Greetings.
Mleod
Beat me to it.
AKSA is Hugh Dean, who has been making not-quite-your-grandfather's ClassAB amplifier for a number of decades as Aspen Amplifiers.
Indeed it's probably fair to describe him as not just one of the iconoclasts of conventional amplifier design but the one who managed to make a conventional amplifier actually sound good (per Crowhurst 1957).
I don't personally know Ranchu but he did the heavy lifting to make the amplifier happen.
And that's how it got its name.
It'll put out about 50W/8ohm before it starts clipping.
Just wanted to say to anyone who didn't know Andrew T as we knew him died on March 12th aged 67. I asked about the JLH current setting which was answerd by Andrew a few pages back. I thought at the time it was a slightly less detailed answer and hoped he might say more. For what it is worth I think the best explanation is that the JLH is a bit like a primitive class AB ( Leak ) where in that example a resistor and diode chain set bias. The JLH mid point sets a reference point with DC feedback to make it stable, it needs regulated voltage to be as good as it can be. If you like it's resembles one half of an overbiased class AB. That was the converstion I hoped for. Now it becomes clear why we didn't get a follow up. I will miss Andrew.
Here is a typical Andrew thread and very useful.
connecting audio ground to Safety Earth
http://www.diyaudio.com/forums/in-memoriam/320518-andrewt.html#post5381674
Here is a typical Andrew thread and very useful.
connecting audio ground to Safety Earth
http://www.diyaudio.com/forums/in-memoriam/320518-andrewt.html#post5381674