Mine was a capacitance multiplier. A simple series resistor protection was the key item. 500V/ 20A. 24 R. It and the FET lost 5V. A zener diode to protect the gate. Until the 24'R fitted to the FET drain I lost the FET after some days. Afterwards it looked so obvious I should expect that. Obviously very high currents on switch on. An output filter of only 1uF was the only likely cause. Once the small problems overcome it gave remarkable performance. All it did was follow a capacitor. The FET having enough gain. JLH a Darlington. The amplifier hum reduced from -66 dB to -88 dB reference 1 watt. Not bad for £4. The FET TO220FP insulated device. 5 V 60 mA wasn't hot enough to need a heatsink. It was easy to do so was fixed to the chassis.
I never built one for the JLH. I am certain it would work.
The Naim nap250 was basically a nap160 with regulated power supply. It sounded very different.More open, detailed and less coloured. Sometimes I felt the 250 was slightly boring. I suspect the PSU wasn't perfection.
I never built one for the JLH. I am certain it would work.
The Naim nap250 was basically a nap160 with regulated power supply. It sounded very different.More open, detailed and less coloured. Sometimes I felt the 250 was slightly boring. I suspect the PSU wasn't perfection.
As I posted before, using 2 highpass filters in series (220R/470uF for each one) on the 'Capacitor' of the cap multiplier can bring the sawtooth waveform down even further. I think It was Rod Elliot who came up with that idea.
Regards
Mike
Regards
Mike
You might have seen putting a similar RC filter on s switch mode power supply was mentioned. If a class AB it might not be a good idea. When class A it should be ok due to constant current requirements. JLH is near enough that. I used 1 ohm 10 000 uF. It would convert a sawtooth to near sinusoida if non smps. A delay circuit was used to allow the surge current. This was very typical of valve radios. A twin capacitor with red and yellow terminals plus black common. The red was high ripple and the yellow less. A resistor or choke from red to yellow. Very economical and not bad considering the cost. When class AB the wire between the power supply and power amplifier can be the resistor. The local decoupling capacitor can be made large. 4700 uF minimum. The main capacitor can be industrial types. That makes the amplifier more stable and has ideal use of money.
High pass. That's intriguing. I was just assuming low pass. I sometimes use subsonic filtering.
I notice that no one looks at capacitance choices in the JLH. Not quality. Just quantity. Also gain. Then the bootstrap. If a bottle of wine it's the knowledge of similar factors that separate good from great. My guess greater than transistor types after a certain point.
I notice that no one looks at capacitance choices in the JLH. Not quality. Just quantity. Also gain. Then the bootstrap. If a bottle of wine it's the knowledge of similar factors that separate good from great. My guess greater than transistor types after a certain point.
This is a simplified power supply for my JLH69.
The regulator has 2SC1383 / SA683 as drivers and KSC5242 / KSA1962 as output transistors. I found 2SC1383 / SA683 in my stock and they have great features, the only bad is that they have a small Vce, that's why they are left unused.
The regulator has 2SC1383 / SA683 as drivers and KSC5242 / KSA1962 as output transistors. I found 2SC1383 / SA683 in my stock and they have great features, the only bad is that they have a small Vce, that's why they are left unused.
Attachments
Yes you are correct, it should read low pass to get rid of the sawtooth. 🙂
Happy listening
Mike
Hello.
I want to build the jlh amplifier, I have 2sc2608 2sa1117, 16 of each, would they cut it, specs look promising, what do you think?.
I want to build the jlh amplifier, I have 2sc2608 2sa1117, 16 of each, would they cut it, specs look promising, what do you think?.
Nearly any modern transistor will work. Even using very ordinary transistors like BC327/337 2N3055 already exceed JLH original design. The thing to remember is this is a class A amplifier so most of the folk law of class AB doesn't apply. What does matter is current gain and the modern super transistors have better gain. BC327/337 come in ultra high gain 40 code. The transistor is 0.6'watts so not ideal above 27 Volts or driving lower gain output transistors.
The more important thing to say is an oscilloscope is vital to see the amplifier is stable and a signal generator for testing.
The more important thing to say is an oscilloscope is vital to see the amplifier is stable and a signal generator for testing.
Good explanation I will go ahead then, and read through this long thread to pick up some knowledge, never build a class a before.
Have a great day..
Have a great day..
Very good luck. You are very nearly guaranteed success. Don't worry about six watts or twenty watts. It mostly sounds similar. Six watts is less difficult. John wanted fifteen watts like his Williamson valve amplifier. John was trying to get an identical sound. The Williamson is a legend. My friends Majestic Transformers still make the output transformers for them 73 years on. They were making Quad valve transformers when I visited recently.
Actually, I used some TDA2009 for 4 watts per channel at my house (singleton input made sources work). Also, after circa2020 winter security revisions, most JLH reference materials are back on line and available.
re #5810
I'd agree that the original design can work acceptably with many types of transistors rated for the duties. Some folk may even prefer the sound of quite the wrong types of semis too, at least as far as their judgement by simple listening tests goes. However, if you also want technically good, repeatable and reliable performance, at least the driver/phase splitter Q3 still needs to be a good quality transistor and it's surprising how few suitable types are still available.
I've noticed that some better Chinese kit suppliers have recently made a point of authenticity🙄 by supplying the 1970 upgraded type 2N1711 transistors for Q3 even if theirs are fake copies. Here's what the class A amplifier site (Geoff Moss) had to say in 2003 about transistor substitutions:
I'd agree that the original design can work acceptably with many types of transistors rated for the duties. Some folk may even prefer the sound of quite the wrong types of semis too, at least as far as their judgement by simple listening tests goes. However, if you also want technically good, repeatable and reliable performance, at least the driver/phase splitter Q3 still needs to be a good quality transistor and it's surprising how few suitable types are still available.
I've noticed that some better Chinese kit suppliers have recently made a point of authenticity🙄 by supplying the 1970 upgraded type 2N1711 transistors for Q3 even if theirs are fake copies. Here's what the class A amplifier site (Geoff Moss) had to say in 2003 about transistor substitutions:
I suggest that we all read the available material before assuming that whatever sounds good at first take is really any better than an accidental effects box though. The Class-A Amplifier Site
Last edited:
I don't really believe transistors have a sound. What they have is an ability to do the job. I would liken the JLH to a three speed automatic gearbox. Considering what it is it's very good. If the bottom output device was exchanged for a BD139/2N2955 complimentary feedback pair it might well sound better despite being humble transistors. The reason is a big reduction in THD and IM distortion due to massive current gain. Oscillation is almost certain so an oscilloscope required. A similar driver might be possible. The complimentary feedback pair is highly linear. It's difficult to know if it could work as the pair may not offer the same working. If it isn't easy to do I feel it's no longer a JLH. Strictly speaking although topologically equivalent to a single NPN device I doubt it totally is. I have built a complimentary feedback pair vas or driver with excellent results. This was an I to V converter so very similar.
My problem with some ideas is they are hyper empirical without any understanding of why something worked. I believe using modern devices high current gain and IM distortion is the question. As IM is tricky to measure it most likely is reduced by high current gain if stability is ensured. I wouldn't call that empirical I would call that a theory with understanding of risk.
My problem with some ideas is they are hyper empirical without any understanding of why something worked. I believe using modern devices high current gain and IM distortion is the question. As IM is tricky to measure it most likely is reduced by high current gain if stability is ensured. I wouldn't call that empirical I would call that a theory with understanding of risk.
Sometimes it is diffucult to understand when people talk about an old version of JLH with an output cap or a newer - push pull.
I think it is a given that solid state electrical components don't have any "sound" of their own. However, audible distortion can be produced when inappropriate parts are fitted by guesswork, necessity or the actual design is poor, not optimised for the application etc. Perhaps though, a little even harmonic distortion will sound pleasant - subjectively better than a technically correct and lower distortion design, at least
. This last point is why certain amplifier designs - particularly the simple and cheap ones like the JLH '69, are evergreens and always popular with DIY builders. It's become a tinker's dream, is somewhat flexible with component types and suitable for almost anyone with an ear for music who can work safely 😎
The 2N1711 transistor made by Central Semiconductor and still listed by Mouser drives the output 2N3055 pair in split phase - pushing with variable pressure one way then another but not all the way into cutoff at either end.
I think it is a given that solid state electrical components don't have any "sound" of their own. However, audible distortion can be produced when inappropriate parts are fitted by guesswork, necessity or the actual design is poor, not optimised for the application etc. Perhaps though, a little even harmonic distortion will sound pleasant - subjectively better than a technically correct and lower distortion design, at least
This last point is why certain amplifier designs - particularly the simple and cheap ones like the JLH '69, are evergreens and always popular with DIY builders. It's become a tinker's dream, is somewhat flexible with component types and suitable for almost anyone with an ear for music who can work safely 😎
That's more or less what I think. I have moved into a new house which needs DIY. When more settled I am going to try the really nasty MJ3001 and smps with filter. Current 0.8 amps. I hope to prove current gain is the issue. BC337-40 should love that. JLH says it should drop in. If it works that slight transistor sound will go. Reducing current sometimes helps. Ripple is a colouration. From measurements I can get rid of smps noise. It won't be as good as a linear supply. Noise at -105 dB might cause IM we can perceive. A simple LM317 can do -130dB or better reference 27 VDC. I imagine it will sound a tad bright but not fatiguing.
I am putting together Hadcock GH 220 Garrard 401 Denon DL110 to go with my mildly modified NAD 3020. The NAD is my JLH preamp. I made a bespoke preamp for the Denon.I do the 75 uS at the cartridge. I am using some Mission 760i that I like very much. Celestion SL6 would be ideal. My Magnepans now given to my old boss. The Celestion although 82 dB per watt probably will be loud enough.
Phono stage. I found that a 470nF cap is reasonably close to 75 uS using the Denon DL 110 coil resistance. The inductance doesn't prevent it being reasonably accurate. More so than most LP's. The advantage of this is it removes the need in the electronics to remove the ultrasonic noises. After that a simple 3180/318 uS stage with high gain. I tried the same trick with other cartridges and failed. Sometimes a simple idea works. I used MC33078 and NE5532. MC in an intangible way was better. Sometimes I think that's fake and me preferring something because I feel it should be better. I have a Lyra Helikon and Ortofon MC25 calibration. With the bespoke phono stage the DL110 is really close. It's not really suited to MM inputs. The Hadcock GH 220 is a gamble. I liked the unipivot arm. I had a Naim Aro once and really liked that.