I don't care about the technical values of an amp, I mean, those that are measurable.
what matters to me is listening and what i feel.
a simulation can always give a good idea on the viability of a circuit and even on its "type of sonor signature", but nothing replaces real life.
for nothing in the world, I will not exchange a jlh 69 (and many other amps for that matter) for a big Accuphase or something like that and I speak with full knowledge of the facts because I have had a few.
and then one day, I had the opportunity to listen to an original Le Monstre 8w and my life changed completely.
what matters to me is listening and what i feel.
a simulation can always give a good idea on the viability of a circuit and even on its "type of sonor signature", but nothing replaces real life.
for nothing in the world, I will not exchange a jlh 69 (and many other amps for that matter) for a big Accuphase or something like that and I speak with full knowledge of the facts because I have had a few.
and then one day, I had the opportunity to listen to an original Le Monstre 8w and my life changed completely.
Member
Joined 2009
Paid Member
2N6338G ?
Anybody considered these 40MHz 200W transistors for the output instead of 2N3055 ?
“...designed for industrial-military power amplifier...”
https://www.onsemi.com/pub/Collateral/2N6338-D.PDF
Anybody considered these 40MHz 200W transistors for the output instead of 2N3055 ?
“...designed for industrial-military power amplifier...”
https://www.onsemi.com/pub/Collateral/2N6338-D.PDF
Member
Joined 2009
Paid Member
Member
Joined 2009
Paid Member
Hello Bigun,
I thought slower transistors are better suited for the design. In the JLH 2003 article, its mentioned that MJL3281 are not suitable , better ones are MJ21194 and MJ15003 which are slower.
The Class-A Amplifier Site - JLH Class-A Update
regards
prasi
I thought slower transistors are better suited for the design. In the JLH 2003 article, its mentioned that MJL3281 are not suitable , better ones are MJ21194 and MJ15003 which are slower.
The Class-A Amplifier Site - JLH Class-A Update
regards
prasi
Member
Joined 2009
Paid Member
Hi Prasi, I’ve read that too, but some folk have recently been very happy with fast transistors, such as Rallyfinnen. I’m interested to know what huggygood has tried too.
My version, TGM9, is super slow; it uses 8 parallel output pairs of 2N3055H which means very high total total Cob. I’ve not used the amp since installing my TGM8 but I feel interested to go back to my JLH and make improvements, possibly a new build so that I can compare. I have lots of modern, fast, power devices in my junk box but none are TO-3 which seems to be the traditional package to use here.
My version, TGM9, is super slow; it uses 8 parallel output pairs of 2N3055H which means very high total total Cob. I’ve not used the amp since installing my TGM8 but I feel interested to go back to my JLH and make improvements, possibly a new build so that I can compare. I have lots of modern, fast, power devices in my junk box but none are TO-3 which seems to be the traditional package to use here.
Last edited:
JLH seemed to think medium speed transistors worked best. He couldn't say exactly why. My feeling is gain and capacitance are the questions. The amplifier needs only full power to 5 kHz which should be 50 kHz for safety. That's not to say real music bandwidth should be restricted. The typical 250 kHz bandwidth of a JLH is actually very good. What high gain transistors can offer is lower distortion and more importantly a nicer balance of distortion. Sadly a super transistor often needs restricting to avoid oscillation. This is because eventually the negative feedback this amplifier requires becomes positive feedback. If the frequency range is restricted the feedback reverts to purely negative. This sounds like a disaster. It isn't because to restrict a moderately high speed transistor works well. Douglas Self says the internal capacitance of a lower grade transistor will assist stability. However it is the lowest grade of capacitance. Far better to add cog/NPO ceramic capacitors to the better quality devices. We should say we need high gain device which most likely will be fast types. When small transistors often for video fast wasn't always high gain. The designs were looking for different things. BF range by coincidence are good audio devices often used for the VAS. BD139 of the original type similar.
Don't forget to mention linear gain vs current. I still like the PNP version from Zero Zone best, and I think it is because the outputs are very linear (2SA1216), not because of their speed.
I have tried other fast outputs, and that sounded and measured slightly better than the MJ15003, Linearity has been pretty good on those too, but not 'flat'. I think it's also about having the output transistors working in their most linear region. Some outputs improved distortion at higher currents, but then we would only have to dissipate a lot of extra heat (that would probably kill the output transistor), or run on really low voltages and get low voltage to the speaker, so they don't really fit the application.
IMO the Z-Z PNP has something more to offer than the others I tried, in distortion measurements as well as in subjective sound. My belief is that this is because the outputs match the application very well.
I think the most interesting JLH 'evolution' for me would be to try variations on the PNP setup. Double outputs, ccs etc, but for now I'm actually quite happy with the sound it's making, so now I'm playing around with speakers a bit.
PNP transistors often have better gain. The JLH isn't a pure current amplifier output as in class AB. Thus the linearity is not the current amplification curves. David Mate the designer of Solid State Logic said the linearity of the output stage is not very important. What is important is the ability to use negative feedback without oscillation. To him this solved the great sound poor linearity of FETs. David liked the hybrid bipolar fet compound pair as did JLH in his later amplifier. David was shown an amplifier called C Audio that made him come to this point of view a technician at SSL told me. David's Orchadis amplifier was extremely good. I had one for a while. It wasn't produced if wondering.
Gain linearity is usually more a problem with power transistors used in class AB, where extreme variations in collector current combined with low maximum gain, means distortion could easily become gross with increasing output power. What sounds nice at personal listening levels may be unpleasant at high power - sounding more like a cheap PA sound system.
Class A is not such a problem for old, industrial transistors if used well within their power limits and only at low power - just as we are doing here. However, the sound quality of JLH'69, at typically only 1.2A bias current and 8R loads, still demonstrates how good they can be at entertainment in small spaces. Their good sound quality though, is not necessarily related to low distortion. In fact, if you investigate, most people prefer the "nice" distribution of harmonics it has rather than having the ultimate in low THD for their personal entertainment. For this purpose, don't waste your time on tests and measurements for hunting down and eliminating all distortion with nothing to show for it after years of effort and expense.
See the attached hFE v Ic graphs to see just how bad the linearity of the standard transistor types is, particularly for 2955, even at low current and assuming the devices you have are genuine spec. if not genuine product. Perhaps there is a reason there for your preference of the PNP versions?
Class A is not such a problem for old, industrial transistors if used well within their power limits and only at low power - just as we are doing here. However, the sound quality of JLH'69, at typically only 1.2A bias current and 8R loads, still demonstrates how good they can be at entertainment in small spaces. Their good sound quality though, is not necessarily related to low distortion. In fact, if you investigate, most people prefer the "nice" distribution of harmonics it has rather than having the ultimate in low THD for their personal entertainment. For this purpose, don't waste your time on tests and measurements for hunting down and eliminating all distortion with nothing to show for it after years of effort and expense.
See the attached hFE v Ic graphs to see just how bad the linearity of the standard transistor types is, particularly for 2955, even at low current and assuming the devices you have are genuine spec. if not genuine product. Perhaps there is a reason there for your preference of the PNP versions?
Attachments
Last edited:
As much as it looks to be a quasi complimentary amplifier with bad VAS current the JLH isn't exactly that. It gets away with a gain deficiency by forced transconductance from input to vas base. Because this outputs never switch off the distortion is reasonable. The output stage distortion must be about 8%. open loop. The output impedance high. Only by negative feedback can it work. I dare say 3% with everything right. It's current gain throughout the amplifier that matters. There is enough voltage gain.
Bonjour vous êtes tous daccord pur dire que le jlh est un tres bon ampli alors pourquoi le modifier?que cherchez vous a ameliorer ? Parce que entre les traductions nulles ,les abreviations, les courbes,et tout ces chiffres je suis complètement perdu . Est ce qu un d entre vous a fait la version 2003 telle qu on la trouve sur le net et en est il satisfait ? MERCI impossible de poster le texte en anglais j arrive à le traduire mais pas à le coller je tente Durand même EXCUSES
I followed your talk about about the pnp version of jlh with great interest, I still have a bunch of 2sa1117 pnp, and I would like to try make one , and see if it behaves.
What alterations to make on my type pcb?
Change the input transistor to pnp type, and the current sources
Reverse capacitor polarity
Direction of diodes
Switch rails polarity
And of course output transistors.
Would that work or am I missing something?
What alterations to make on my type pcb?
Change the input transistor to pnp type, and the current sources
Reverse capacitor polarity
Direction of diodes
Switch rails polarity
And of course output transistors.
Would that work or am I missing something?
hello you all agree that the jlh is a very good amp so why change it? what are you looking to improve? because between the null translations, all these curves these abbreviations and all these figures I am completely lost. Did any of you mount the 2003 version as found on the net and are you satisfied with it? thank you
I guess because humans are curious and driven to invent and improve, I agree that the jlh standard sounds heavenly.
I'm not that much knowledgeable as others here, but I'm sure that it's way more difficult to build a really simple amp like the jlh, and keep components count to a minimum, that takes skills.
I'm not that much knowledgeable as others here, but I'm sure that it's way more difficult to build a really simple amp like the jlh, and keep components count to a minimum, that takes skills.
I think I understand that. You could say you proved your point. It is a near perfect design. John Elliot found like myself 0.03% distortion. Some changes because we have to use modern parts. I am dyslexic in English, I would be worse in French.