I would start with the speakers. The leaving of the requirement of 3-tones-amplification:
No multiway, no passive crossovers: circuit-salad. Too many characters, toooo different, homogenity, clearness NEVER.
No floorstand: too much surface that resonate and reflect.
No litz-wire, un-laqueured. Less diameters.
A better damping of the casings.
A better build-in of the drivers. A very complex modification of the driver - I do not know a good driver.
And so on;-)
No multiway, no passive crossovers: circuit-salad. Too many characters, toooo different, homogenity, clearness NEVER.
No floorstand: too much surface that resonate and reflect.
No litz-wire, un-laqueured. Less diameters.
A better damping of the casings.
A better build-in of the drivers. A very complex modification of the driver - I do not know a good driver.
And so on;-)
I use a 4 foot x 2 foot baffles in 50/50 MDF and plywood = 1 inch. I have been listening too loud, the JLH would take the volume down which would be a good idea, my friend measured 115 dB which seems fine and sounds like 100 dB, The Beatles seem just right with them and opera, 78's also. The bass unit is a 15 inch type of 5.7 ohms and Qts of 1.2. I use massive bass EQ to get 40 Hz flat and 30 Hz - 12 dB. I use a choke to roll it off starting at 250 Hz, that's a first order filter. Next a Eminance 12 Lta full range unit Which gives 100 Hz to 8 kHz. It has a few tweaks. Most surprising as they are very cheap tweeters DT74/8. These worked with the maths and were better than nothing. Very much better and actually nice. Because they cross at 6 kHz ( also first order ) they have lower distortion than typical. The big surprise is they work with the polar responce of the 12 Lta against any book of theory to give wider dispersion. Happy accident I would say.
Alas they are packed away for now. Being baffles 90% of the colouration of room and box is absent. The 1950's problems of the 12 Lta are minimal compared with the usual box colourations. I also have some Magnepan. It's hard to say which are best. Something to note. I put a blanket over the baffle backs. In theory it should sound good. It didn't and had box colouration. I think the reflected sound from the rear room walls is a bonus. I am told baffles sound much better in modern houses where there are hard reflective surfaces. It is the dipole beaming prevent indirect sound. The DT74 made it uniform which beats the Magnepans, about a good 4 foot central stage which is still good over 12 feet ( it's far less without the tweeter, I suspect the additions from 4 kHz up help, pure luck, there will be a hump about there, great for 78's ). I could add a DT74 to the Maggies, there's a thought. I would cross at 10 kHz if I did. Not to get an extra 4 kHz above the 15 kHz they have, just to widen the polar range.
Alas they are packed away for now. Being baffles 90% of the colouration of room and box is absent. The 1950's problems of the 12 Lta are minimal compared with the usual box colourations. I also have some Magnepan. It's hard to say which are best. Something to note. I put a blanket over the baffle backs. In theory it should sound good. It didn't and had box colouration. I think the reflected sound from the rear room walls is a bonus. I am told baffles sound much better in modern houses where there are hard reflective surfaces. It is the dipole beaming prevent indirect sound. The DT74 made it uniform which beats the Magnepans, about a good 4 foot central stage which is still good over 12 feet ( it's far less without the tweeter, I suspect the additions from 4 kHz up help, pure luck, there will be a hump about there, great for 78's ). I could add a DT74 to the Maggies, there's a thought. I would cross at 10 kHz if I did. Not to get an extra 4 kHz above the 15 kHz they have, just to widen the polar range.
![IMG_20171224_145128[2].jpg](/community/data/attachments/600/600049-65a02dcaca9d5244e6d07e9787970188.jpg)
@cumbb - I agree with Nigel that it is hard to see how the BD236 can sound better than an MJ2955, because they are both epitaxial devices from about the same technology generation with Ft's of 3MHz. Did you check that the quiescent current is set up correctly for each device type? JLH showed that it was necessary to adjust that, and if your MJ2955's have lower gain that would explain the distortion, if that is what you are hearing.
I have never tried them. I know they are mentioned as devices similar to triacs ( two base connections ).
One thing I have tried to do is use devices I can almost guarantee are not fakes.
Last edited:
One thing I do believe can be true is T03 devices because of the steel case may sound different to TIP versions. However, because the T03 case is very large the conduction will still be reasonably good. One way to get the best from T03 is to clamp tages to both fixing bolts on the hat top side with think wires to the next component, it avoids the bolt being a prime conductor. I did think if I could saw the hat top off and locate the collector solder point. Then take a wire from there. I would seal with high temperature silicon sealant or similar. I think I read the silicon should not be exposed to the air, any air in the can is quickly depleated of oxygen, not so if open to the room. The more I think about it the more TIP35 looks good.
I am seeing the house moving people this week. If I am to build this amplifier it needs to be soon.
I am seeing the house moving people this week. If I am to build this amplifier it needs to be soon.
it's no fake,just NOS to my very old stock
if you want,i can send you some for test
perfect for oscillator (relaxation oscillator)
Last edited:
Indeed, why not the best, most linear high gain transistors available? I think there is a good reason why we use the older generation epitaxial industrial type transistors instead of more linear audio types. It is about distortion, or rather not enough of it. The distortion of older power transistor types in this design proves to be more pleasant sounding so the technical obligation to try for lowest distortion is not necessarily a good idea - at least not for this design and usual purpose.
Many experienced builders including JLH, seem to agree that MJ15003 comes out on top of the few low-cost TO3 transistors. Unfortunately, plastic versions of the cheap and popular 2N3055 don't seem as good-sounding as their TO3 cousins but there is a lot of variation among the many sources of 2N3055, depending on the manufacturer or the actual chips in typical fakes supplied in the kits. YMMV
I think we build JLH amplifiers not just because they are simple and cheap, but because they sound so good with many of the atmospheric and solo vocalist music types that we often listen to for relaxation and inspiration. My experience with this 1969 version and similar small class A amplifiers, is that trying to "gild the lily"with refinements to improve performance is usually counter-productive. A very clean and quiet power supply is one of the few improvements that is useful with JLH'69 and probably most other basic amplifiers which generally have low PSRR.
If you prefer less warmth and sweetness, bigger, denser sound from synths, choruses or bands,orchestras etc. where there is more need for very low distortion, perhaps this isn't the amplifier for you. A more conventional and up-to-date design in either class A, AB or even D will probably be more suitable.
I believe you are right but for the wrong reasons.
I believe (and there's plenty of discussion several hundred pages ago to support this) that the "niceness" of the JLH '69 comes from the following factors:
1) it's class A. With a decent power supply this means extremely good detail and a "black background" at the low levels we spend 90% of time at.
2) it does not use an LTP diff amp stage. The diff amp is outside the feedback loop and will generate a "flavour". (That is, characteristic distortion products)
3) the asymmetry of the design generates mostly even order harmonics when it does distort on large signals. (Lush, as you put it)
4) When it does clip, it is relatively well behaved. This makes it "louder" than 10W "should" sound.
Most attempts at making the JKH 69 "better" break one or more of the above characteristics, messing with "the sound".
Where the JLH falls over is difficult loads demanding lots of current or voltage swing and/or music with sustained high levels: this is not a result of the architecture but limits of the 2N3055 SOAR curve.
Last edited:
....wrong reasons? You may consider that 2/4 of the reasons you listed also centre on distortion as a basis of JLH's amplifier popularity. There is also a clear difference when fitting more modern audio transistors, there being a nice compromise that meets SOA and sound quality requirements at about where JLH left it. However, I did try not to go into topology details or to analyse the causes, keeping it simple as just "distortion" since many here won't be interested in long-winded details. I half expected there would be others who did want to add some topology features they considered important. Fair enough.
As you say, a "black background" is a feature of a good power supply which I also included as an improvement, since most builds here will be from cheap kits completed and used as is, with only the suggested simple transformer/rectifier supply. I even have a commercially built version which is just as basic. The amplifier is not known for low noise or high fidelity in it's basic form but this is how most DIY builds will be auditioned and probably used.
A lot of well-informed discussion here seems to draw from a wider class A field than applies to to the JLH type. That isn't always clarified and results in a bit of confusion among DIYs in the process of completing and evaluating their own. It may also be a unique solid state design but that doesn't make it a paragon of textbook design virtue or performance either - we just need to play to its strengths.
Last edited:
Ah, that make sense and is clearer (to me!
): I misinterpreted where you were going with your statements. I'm very interested in understanding both what audible differences exist and the root causes of those differences.
Have found "my" TO-3;-)
Copper-plate - the only one, fT: 200 kHz, no piiiiing per knock - the steel-bodys do kliiing and piiing - it is audible. Germanium;-)
Cool sound. A bit overcast, but very disciplined and fine and ... "right" behind;-)
And the ASZ17 I would use, but with a mechanical damping: clamp or putty or so. Than the BD318 and the MJ15004.
The JLH sounds good, cause just a handful parts and few indifferent parts and quasi se and ... clean enough - my mind;-)
Copper-plate - the only one, fT: 200 kHz, no piiiiing per knock - the steel-bodys do kliiing and piiing - it is audible. Germanium;-)
Cool sound. A bit overcast, but very disciplined and fine and ... "right" behind;-)
And the ASZ17 I would use, but with a mechanical damping: clamp or putty or so. Than the BD318 and the MJ15004.
The JLH sounds good, cause just a handful parts and few indifferent parts and quasi se and ... clean enough - my mind;-)
Many of us seem to share that urge, perhaps because of the surprising performance from a handful of parts, the cult following or even the British heritage thing. A formal study could even lead to more insight of how we perceive the playback of music recordings. It occurred to me though, it would save a bit of time if I started searching by looking closer at what others have created along similar lines.
Rod Elliott comes very close with his "Death of Zen" design since it borrows many of JLH's design features. He was onto something there, at least more than some other attempts I've read about and heard: Death of Zen - A new Class-A power amp., My build of it was (I think) a notch above the 3 JLH amplifiers I kept and though Rod's article goes some way to explaining why, I was always going to get around to investigating it for myself.
I think the last bit is just how this type of class A works. It's a bit like a rev limiter. If 2SC5200 sounded better it most likely would be gain advantage. Bog standard Indian 3055 take us to 160 kHz. I looked at BC327-40 337-40 as available and perhaps better than the 1969 devices. If I looked in SMD types I might have found even better ones as far as watts is concerned. I haven't been able to kill the BC337 yet. It looks to me the constant current function of the JLH is enough to prevent excessive 3055 base current. I will use a LM317 + 3055 as PSU. Test show very low ripple. No feedback between 317 and 3055 which to me is a better idea. The fixed voltage fixes the current. It's the only weekness of the 1969 version which might also be it's strength. Do it right is the only option.
The distortion is closer to the Jean Hiraga ideal that many ideal SE valve designs can give. That is the harmonics are on an exponential decay cure. Hiraga argued that as the ear has this type of distortion it would seem to have less distortion. A bit like colour balance with TV.
The JLH doesn't soft clip as such. It just says it's clipping without drama. You can have your cake and eat it. For simplicity early in the thread I showed how 2 bipolar transistors and 2 Audio FET's ( BUZ900/905 perhasp ) can be set at 0.7A standing current in overbiased class AB+A. This was to meet a circa 5 watts 6R type spec. The 0.7A was what the heat sink could allow. The design gave 20 watts class AB 8R and far more into 3R. The amplifier was far from sounding romantic and was low distortion. As it had no long tail pair it preserved the harmonic structure, albeit at a level well below human ability to hear. It also equalled a Self complimentary feedback pair with OPA 604 driver. Hard to say which was better. Of note was the Self type idea could work without loop feedback and still measure very well ( only double local ). The FET was loop feedback although as simple as is possible.
I have to repeat what Ivor Tiefenbrum often says. If it sounds better it is better.