Will check out Aliexpress, have never used them before though. Was looking for a complete solution, ready to solder together. 🙂
Yeah, I have been thinking about the risk of getting counterfeit semiconductors too, hence the reason for posting here. 🙂
Thank you for the extensive reply.
The comment regarding baking the PCB is a very good one that I hadn't considered. I am expecting counterfeit transistors as well... I think I will purchase a kit, it seems to cheap not to try. I am on a budget, but have an old heatsink I think will suffice as well as 200W DELL 19,5VDC laptop brick I have laying around.
Will consider your comments and will post an update when I do purchase this kit.
🙂
I already have a Hiraga Le Monstre kit, also off ebay. Have no symmetrical PSU, hence that one has to wait a little.
Project 83 - MOSFET Power Follower
Here is a design that could to some ears sound better than the JLH. It has no loop negative feedback. The distortion at about 0.4 watts is about 0.065% stated and 0.43% 10 watts. It needs the full output of an op amp to get about 8 watts. One could reduce the voltage to perhaps 33VDC so as not to waste power, current could come down to 1.4 Amps I feel, Whilst a cheap MOSFET isn't the most linear device the distortion is far lower than most speakers. KEF LS50 about the same, they are considered low distortion types. Quad ESL are lower, the speaker JLH used as a reference.
I guess ZD3 could be 2 x Red LED ( Circa 1.65 V x 2 ). That should be OK if using 33V and 2K2 at 14 mA as bias. The output current is set using ohms law and value of resistor. 2.2V/1.4A = 1R5 approximately. If the devices will centre as neatly when 33V it is hard to say. In the 40 V version it is very good ( I wonder ).
The reason the distortion is reasonable is the Source Follower has unity ( or less ) voltage gain. The ESP storyline gives detail on this.
A single Darlington current source should be fine as the need for speed is minimal. Unlike the JLH how the current source works is pure ohms law and Vbe of the Darlington. I like the LED's as they show it's working. Most are 30 mA max so should be inside limits even when 40 V used. I suspect R5 could be 3K3.
One advantage of this amplifier is to work the preamp harder to get it's maximum output voltage. This means where critical stability issues arise the voltage gain section is not connected to the speakers which is a great stability bonus. The designer says the input capacitance is quite high. Most op amps should cope. I agree with Rod Elliot that a 12V gate Zener would be better. One could make the preamp feedback loop include this amplifier. The distortion would be very low. I would expect trouble doing that. Keep it simple if tempted to try this design.
Thought I would show this as it helps put the JLH in context.
Here is a design that could to some ears sound better than the JLH. It has no loop negative feedback. The distortion at about 0.4 watts is about 0.065% stated and 0.43% 10 watts. It needs the full output of an op amp to get about 8 watts. One could reduce the voltage to perhaps 33VDC so as not to waste power, current could come down to 1.4 Amps I feel, Whilst a cheap MOSFET isn't the most linear device the distortion is far lower than most speakers. KEF LS50 about the same, they are considered low distortion types. Quad ESL are lower, the speaker JLH used as a reference.
I guess ZD3 could be 2 x Red LED ( Circa 1.65 V x 2 ). That should be OK if using 33V and 2K2 at 14 mA as bias. The output current is set using ohms law and value of resistor. 2.2V/1.4A = 1R5 approximately. If the devices will centre as neatly when 33V it is hard to say. In the 40 V version it is very good ( I wonder ).
The reason the distortion is reasonable is the Source Follower has unity ( or less ) voltage gain. The ESP storyline gives detail on this.
A single Darlington current source should be fine as the need for speed is minimal. Unlike the JLH how the current source works is pure ohms law and Vbe of the Darlington. I like the LED's as they show it's working. Most are 30 mA max so should be inside limits even when 40 V used. I suspect R5 could be 3K3.
One advantage of this amplifier is to work the preamp harder to get it's maximum output voltage. This means where critical stability issues arise the voltage gain section is not connected to the speakers which is a great stability bonus. The designer says the input capacitance is quite high. Most op amps should cope. I agree with Rod Elliot that a 12V gate Zener would be better. One could make the preamp feedback loop include this amplifier. The distortion would be very low. I would expect trouble doing that. Keep it simple if tempted to try this design.
Thought I would show this as it helps put the JLH in context.
it works well with the B1 buffer as preamp after potentiometer.
I built it a few years ago, I found it a bit too "bright"
and I did not try to make it evolve
I built it a few years ago, I found it a bit too "bright"
and I did not try to make it evolve
Pavels Follower sounds very well. The "brightness" is, in my mind, a result of "cleanness", in the "bass" too;-) Endless breath and deep and fine. The "supply-circuit" for the gate;-) The elves sing;-)
A slimmed JLH - not a TO-3 version - sounds a little bit more lively and corporeal.
Another interesting link:
Single-Ended Amplifiers and Buffers
A slimmed JLH - not a TO-3 version - sounds a little bit more lively and corporeal.
Another interesting link:
Single-Ended Amplifiers and Buffers
Gents, my schematic shows no value for C9, which is the input coupling cap. It shows it as an electro, but no value. Someone elsewhere has suggested 0.5uF as a suitable value, I have 470nF polyesters to hand. Is this a good choice?
Which version are you discussing? C9 does not tell me anything.
0.47 uF sounds a bit small. It will give a-3dB point at 33Hz with a 10k load. If the amplifier has 100k input that would work, so if that is the case, then OK, as that is 3.3Hz.
Check what impedance your amp input is and see if it will work.
-3dB point =1/(2 x 3.14159 x R x 0.47E-6)
0.47 uF sounds a bit small. It will give a-3dB point at 33Hz with a 10k load. If the amplifier has 100k input that would work, so if that is the case, then OK, as that is 3.3Hz.
Check what impedance your amp input is and see if it will work.
-3dB point =1/(2 x 3.14159 x R x 0.47E-6)
Thanks. I have the schematic on a pdf but I don't knoiw how to post it here. Having had a look it is very similar to the Fig 13 shown on this site:
EARLY SOLID STATE AUDIO AMPLIFIER DESIGNS
"My" version uses 2N3055 outputs of course.
The cap I am talking about is C4 on this schema. Every version of the JLH I have seen uses 0.5uF here or 0.47, so I thought that a 470nF poly ought to be ideal. If I'm mistaken then I'm more than happy to be educated on this. Thanks again.
How do I estimate the input impedance of the amp?
EARLY SOLID STATE AUDIO AMPLIFIER DESIGNS
"My" version uses 2N3055 outputs of course.
The cap I am talking about is C4 on this schema. Every version of the JLH I have seen uses 0.5uF here or 0.47, so I thought that a 470nF poly ought to be ideal. If I'm mistaken then I'm more than happy to be educated on this. Thanks again.
How do I estimate the input impedance of the amp?
Last edited:
You should visit the Class A Audio Site, archived at ESP website or the Magazine archive which I've posted links to here recently - JLH discusses the parts and decisions for the several versions and articles there. The C4 input capacitor isn't critical, its a high-pass filter for DC, which will exist as a half supply voltage difference to the signal ground voltage there, so you need to block that DC with a cap. This only needs to be good grade of dielectric. 470nF=0.47uF so that's fine.
Polypropylene (MKP film type) capacitors are newer and better than the specified electrolytic capacitor and their voltage rating will be more than enough for the duty (half of say, 30-40VDC). 1uF is commonly fitted and getting large there, so look at images of the clones offered on Ebay or builds on the thread here and you'll get the idea of the type and size that might be suitable and is commonly supplied.
It won't need any changes but the impedance and other specifications are in the article and other references on TCAAS archive.
It's a good choice. If we take the input inpedance as circa 100K+100K in paralell we get Fo = 1/2Pi RC We can say Rin>50 000 ohms. Rearanging we get 10^9/6.284x50000x470 = <6.8 Hz ( as Rin >50K ). If you went to 1uF it would be 3.2Hz. The point is 250 V polyester caps sound good and are cheap. The risk of below 100V types is a less good sound, 6.8Hz is about right. You could try a 1000uF Panasonic FC on the side to the 220R gain setting resistor. You could if using the 1969 design use 1000uF of the same type to the bootstrap. If you look on JLH's distrortion gragh at low bass it seems to say these other caps matter most. I think the output cap could be 3300uF or even 4700uF. Too large output caps sound boomy if the power supply is not ideal. The suggested one will work well. It is very typical of 1960's designs. Armstrong 600 series being typical and a great sounding class AB amp.
One thing about the bootstrap. How it works is circa 90% positive feedback to the far end of the driver transistor collector resistor makes it seem the resistor is vastly higher in value. It's like the collector is using a > 100V supply rail. Usually it is a great way of getting crossover distortion into the collector of your BD139 or whatever. When class A there is virtually no down side as the crossover distortion is not there. Don't let anyone tell you bootstrapping is a failed 1960's idea. It is super fast and is a mathamatical certainty. I have used as little at 50% positive feedback with spectacular results. It was at very high voltage so I wanted to keep it simple. It avoided custom built parts.
If you use 101% positive feedback you will be in trouble. The only well known circuit that did was the PW Texan. Even then it was controlled by negative feedback. That circuit is not as bad as it looks. A voltage gain of 3 suits it best. To get > 97% gain from a current amp is near impossible. 97% is usually where the trouble starts. Again we find the JLH very sophisticated whilst pretending to be simple. To be honest John put many to shame.
If you used no cap at all it could be OK. The output cap will stop speaker damage. If the DC mid point changes greatly stop. If using a volume pot it will not work as it will change the DC mid point at the output cap. If you use a preamp with an output cap it is very possible no input cap will be required. If so optmise the preamp output cap to perhaps 1uF polyester. I am happy to be told this is wrong. It's a good discussion regardless. It needs a little care. The mid point tells you what you need to know. If it is mid point +/-0.5V it's OK.
I think that one problem in using too small a capacitor is noise. Low frequency noise is a problem in all devices, and source impedance is a factor in controlling it. If the input capacitor is too small the impedance will be high(ish) and that might stop noise being suppressed. (This is voltage noise - there is also current noise which increases with lower impedances, but should be low with low currents). While I agree that 470nF is suitable according to JLH's original (1969 vintage) circuit I would tend to use larger, though as the circuit is not particularly noisy a larger capacitor might not be obviously better.
If we are talking about the original, which seems to be the case, I'd say you definitely need a capacitor or the mid rail will be thrown off balance.
But I will repeat my concern that the JLH bootstrap can only respond as fast as the output transistor - because that is driving it. So a faster transistor will give a faster response. The current 2N3055 being epi should be "fast enough" for most listeners, but I have got good results using MJ3281A's or 2sC5200's. I've not looked into the high input capacitance effects - this may counter the fast response somewhat when driven by a limited current VAS (that is, without driver transistors).
If we are talking about the original, which seems to be the case, I'd say you definitely need a capacitor or the mid rail will be thrown off balance.
But I will repeat my concern that the JLH bootstrap can only respond as fast as the output transistor - because that is driving it. So a faster transistor will give a faster response. The current 2N3055 being epi should be "fast enough" for most listeners, but I have got good results using MJ3281A's or 2sC5200's. I've not looked into the high input capacitance effects - this may counter the fast response somewhat when driven by a limited current VAS (that is, without driver transistors).
John, I did measure noise on my birds nest JLH with BC327-40, to be honest it wasn't fantastic. BC327 is a very low noise device not unlike BC560C whilst being more available ( for now that is ). The Rbb- is very low which shows it to be a mini power device. BC327/337 are not unlike 2N4401/4403 that are said to be switching types.
The 55K input resistance doesn't help. My assumption was the low impedance of the preamp would reward me. It is possible a 10uF electrolytic as is typical has hidden virtue. However polorizing electrolytics isn't good. They seem to work best below 0.4V DC and with less than 250 mV rms signal.
Cyril Bateman's Capacitor Sound articles | Linear Audio NL
The 55K input resistance doesn't help. My assumption was the low impedance of the preamp would reward me. It is possible a 10uF electrolytic as is typical has hidden virtue. However polorizing electrolytics isn't good. They seem to work best below 0.4V DC and with less than 250 mV rms signal.
Cyril Bateman's Capacitor Sound articles | Linear Audio NL
BTW. I think bootstraps can be slower than is intuitive as it is just the ends of a resistor being made to raise into the air so to speak. The time lag just reduces the ability to hide the resistor, simple wave addition. As said, even 50% positive feedback will do plenty. It equaled a MJE350 CCS at 300V. I derived that from a transformer that is easilly the worst way to do it. The driver stage makes 28 Vrms and I had 14 Vrms to drive the bootstrap. The reduction of overal distortion was about 300%. That amplifier had no negative feedback at all. Not even local. It had less than 0.2% THD at 1 watt which is an excellent result. It did this by using pre-distortion. Pentode with CCS and Triode out. The triode has internal feedback if being correct.
I think we need to clearly distinguish between experimental ideas, other versions of the JLH design and the thread Topic or others like steve67 will wind up confused or misled by what is being proposed. For one thing, the JLH '69 has a single power supply rail, so we can't reasonably ditch the input or output caps.
Let him build and try his JLH'69 project with recommended parts or better ones first before suggesting options or changes then see if he wants to tinker with the design so that he can make something different out of it.
MKP caps by the way, are not as expensive relative to polyester, as they once were. In fact, I find them more readily available than MKT and cheap up to a few uF, so long as your'e not expecting Solen or film/foil types which will cost, if only because of the limited market.
Let him build and try his JLH'69 project with recommended parts or better ones first before suggesting options or changes then see if he wants to tinker with the design so that he can make something different out of it.
MKP caps by the way, are not as expensive relative to polyester, as they once were. In fact, I find them more readily available than MKT and cheap up to a few uF, so long as your'e not expecting Solen or film/foil types which will cost, if only because of the limited market.
Thanks Ian,
What are MKP caps? I learnt my (elementary) electronics as a 6th former 30 years ago, back then there were electros, polyesters like multicoloured mint humbugs, polystyrene and ceramics. Tants were for the wealthy. I'm returning to it now, having had a passive interest for 30-odd years, and finding it changed but entertaining. Re-learning stuff that I used to do in my sleep is fun, like the parallel resistors calculations and resistor/cap colour codes. Either this stuff has got a lot smaller or my eyes were better then. Both, I'm sure.
What are MKP caps? I learnt my (elementary) electronics as a 6th former 30 years ago, back then there were electros, polyesters like multicoloured mint humbugs, polystyrene and ceramics. Tants were for the wealthy. I'm returning to it now, having had a passive interest for 30-odd years, and finding it changed but entertaining. Re-learning stuff that I used to do in my sleep is fun, like the parallel resistors calculations and resistor/cap colour codes. Either this stuff has got a lot smaller or my eyes were better then. Both, I'm sure.
I try to stick to through hole components for my designs.
Some of the SMD components are minute.
Heat sinking.
I have worked out how the TO3 package goes on the pcb and the connections are made. Easy, all i need are mounting washers, screws and heat paste. I have a nice big heat sink, one per side (channel), each about 20cm long by 6cm high, fins every cm and 3cm deep. (8" long, 2.5" high, 3/8 spacing and 1 1/4" deep for our USA cousins). I don't know its °C/W spec but it looks tasty to me. I'm hanging it outside the case, fins vertical for max convection and airflow. I now need a bracket, the boards will be horizontal and I need to attach them to a vertical heat sink. The ones I have seen are ally angle, can I buy these pre-drilled (or not) and just bolt them to the heat sink? Alternatively, can I just fabricate something in steel? I have angle iron enough to cobble dogs in the workshop and light engineering is easy. However if it won't work for reasons of conductivity and SHC then I'd rather find out before it fries a few 2N3055s. Is ally the only solution?
I have worked out how the TO3 package goes on the pcb and the connections are made. Easy, all i need are mounting washers, screws and heat paste. I have a nice big heat sink, one per side (channel), each about 20cm long by 6cm high, fins every cm and 3cm deep. (8" long, 2.5" high, 3/8 spacing and 1 1/4" deep for our USA cousins). I don't know its °C/W spec but it looks tasty to me. I'm hanging it outside the case, fins vertical for max convection and airflow. I now need a bracket, the boards will be horizontal and I need to attach them to a vertical heat sink. The ones I have seen are ally angle, can I buy these pre-drilled (or not) and just bolt them to the heat sink? Alternatively, can I just fabricate something in steel? I have angle iron enough to cobble dogs in the workshop and light engineering is easy. However if it won't work for reasons of conductivity and SHC then I'd rather find out before it fries a few 2N3055s. Is ally the only solution?