JLH 10 Watt class A amplifier

Well... just a thread to say that at last I finished my JLH 10Watter last weekend... I dicided to buy a commercial lab-powersupply.. this made things alot easier..

The results: Great!

Sound is a difinite upgrade from my Sony integrated amp .. more instruments are comming through even in loud part of the music.. high's are excelent .. well at least for me they are.. and supprisingly the bass somehow has more detail.. I experienced the new sound like I experinced my first Grado headphone listeing, but now with the stereophonic effect working..

The amp was easy to build, I even flipped a transistor around but the amp still worked.! just a little noisy.. Iq is set at [email protected] 27Volt, the 4 heasinks get hot but can be touched for more than 10 seconds.. well permanent actually higher voltages and Iq goes really hot!

10Khz square wave looks crystal clean but a 50KHz reveals the assymetric slewrate.. bandwidth goes from 10Hz to beyond 200KHz...

conclusion are that it is a very simple amp to build, no PCB used, no instability of any kind.. and sound is quite revealing.. highly recommended

but then again, my initials aren't H.H. ;)

I built the amp. Great amp
i built mine on +22,-22 volt. No cap at the output
You could change the 2n3055 to MJ15003 makes alot of difference but the cost would be twice of 2n3055 plus mj15003 has a bigger Safe Operating Area they can withstand 250 w instead of 115 w of 3055. I wouldn't say 3055 are bad but they almost antique stuff :) but still in production mj15003 would be the upgrade to it. Better bass, treble and midrange evident when change. If can get the 2n1711 that would be a bonus ( another exotic transistor from where i come from). Apart from that, caps and resistor exspecially the feedback caps are good to be change. Hey you try a choke and cap filter forming a LC filter. Looks good on simulation. I will get back to you when i build the LC filter. BUt first, enjoy the music. MOst important part in audio some people might actually forget its importantance
Thanks for the tip!.. I would like to try those output transistors soon .. just need to find them first :) ... I did use the 2N1711.. Hfe was something like 300.. the input transistor bc560 had a Hfe of 360, I also checked a bunch of BD139's: they all had a Hfe of 110 to 130 ..
the 2N3055 were matched at Hfe of 80/89 and 260(!)/340(!)... which brings me to a question?

I test my transistors with a simple multimeter: I that usefull, since the thansistor will be used at (much) higher currents than my multimeter uses..?
Not bad, my 2n1711 was only 130 tested from the multimeter too funny. Well For transistor testing try looking up sound.au.com. Rod design a simple circuit to testing transistors in generally. The current can be set for small signal like the bc560 to 2A for the 2n3055. Well for the 2n3055 of hfe of 260 and 340 is quite high. Could be defect, i am not sure though. Well for my mj15003 hfe was only mere 40. Also for good a sound try( I am saying this if money grows on trees in your area) matching them almost identical like both channels at around hfe 80 i guess you get almost identical characteristics. But to do this you need alot of transistors. Any upgrades of caps material of resistor will invoke the law of diminishing returns on you just for you to note. Well for the input cap of 470nf i used a film and foil polyprop. My has not particular brand. If you really want hi-fi try hovland music cap will cost a bomb. and for 470 uf try blackgate non-polar i think the model is NX or something else as usually a bomb but the voltage of the cap you choose can be at 6.3 volt. Hence using a non-polar at that high value is possible and not so costly. But test the voltage first mine was a dual supply. Well you could try to change the 220 ohm resistor connecting to the 470uf to maybe to something better. That's the feedback resistor that's what i heard. Experimenting must done to check this out cause i build mine using good resistor here. Hope this helps
Just thought I'd add my 10c, as I've finished-up the final version of my Rod Elliott DoZ amp (Project 36), a JLH variant, and just love this amp!

Initially I had tried some different makes and gains of the 2N3055 in a DoZ prototype and had just thought the glaring mids and too-bright top-end were a function of using surplus parts and paying no attention to gains. So in my final version of the amp I used hq parts and matched 2N3055's with Hfe's of 200 for the output trannies and 160 for the 2N3055 current sources (per the JLH site better to have the one lower than the other). However I was dissapointed, once again it seemed lacking - mids and highs dominating. So I replaced the 2N's with MJ15003's (with gains of 60 & 55 respectively for output & source) and there is an absolutely unequivocal significant improvement - so would highly recommend. (And to whomever said somewhere on this forum something about the DoZ amp lacking deep bass I can happily state - you're completely wrong! EBTG's Temperamental was shaking the floor last night.)

Other substitutions if interested: I've used NTE373's selected for high gain for the BD139 (JLH - 2n1711) drivers . (I also have just aquired some high gain MJE180's (hfe - 250) that I may try sometime, as they have a better Ic operating range than the 373's and worked fine in the prototype). Input transistors are matched Motorola 48-869412's (with Hfe's around 300), a BC559C or NTE234 equivalent.
Thanks every one... I know the 2n3055 are old and considered not suiteble for quality audio.. I also read that in D.Self's book.. so I would like to switch to the MJ15003's.. I just need to find them first:(

Dear P Lacombe

I don't want to be annoying but do you have any explanation why the 2n3055 are totaly unasable besides that they are:
-old designed (that doesn't make them bad)
-cheap transistors? (that is a good quality :) )

Thanks for the comments!

Again with great effort to avoid becoming anoying I must reply to your comments:

2N3055 = Slow:
They are fast enough for a near perfect 50KHz square wave without any overshoot and a bandwidth beyond 200KHz. Their Ft(=0.8MHz to 4MHz) prevents high frequentie oscilation. I consider it useless to use faster ouput transistors if that mean installing of an compensation cap which I now do not need. This design is inherent stable.

2N3055 = bad gain linearity:
Well considering that this design uses the 2N3055 at 1.2A Iq thus so only at a 0.1 to 2.2A current swing, the devices are used in their lineair regeon (well according to D. Self in his book). Add to that that I discoved I do not need mor then 0.5A current swing for high SPL levels I considerd it as not so important

2N3055 = not robust
Well they are use at an Iq of 1/7th of the rating and a power dissipation 1/10th of their rating. I do not have to expect any failers.

Seems to me that no argument against the 2N3055 can be made other than the most important:
Subjective Experience.

The NAD 3020 used the 2N3055 as did the Bedini 25/25. Both these amps have practically cult status. The John Curl JC-2 used 2N5884s and 2N5884s which were also 4 MHz devices. Many construction articles are designed for ease of build and availiblity of parts. My favorite surplus store has plenty of 2N3055s and no 2SC3281s. These transistors are fine for a beginer to cut his teeth on and are more tolerant of non-optimum layout and compensation than the new fast stuff. And they really don't sound too bad.

best of british.......

Hello tschrama and Harry, thanks I was going to mention the 3020, but left it out.

My recollection of these is that they achieved their cult status mainly due to the British hifi press extolling their virtues in the context of low cost, minimalistish, synergystic systems that worked well in a small (British) listening environment.
In a small room, with 6 1/2" or 8" 2 way, yes indeed they worked quite well.

They were never great shakes hifi and they were never intended to be - their strength was british designed circuit and layout, european semiconductors ( I reckon they sound different to Jap and US semis), minimal parts count, cheap construction and built in Taiwan, and hence relatively cheap and easily available.
They also had nice sounding tone control section in addition to reasonably pleasant overload characteristics - ie compression and soft clipping, that enabled overdrive without getting unduly nasty.
Overall this was quite a pleasant amplifier for its day but goes nowhere near the detail, resoloution and cleanness of more modern circuits and devices.
A penalty of 'slow' output devices is increased HF and IMD distortion, and this is sonically evident.
The resultant sound can be a bit like Musak, and that can be pleasant enough in it's own right too.

Lets hope you try substituting the 3055's and report your results,
Regards, Eric.
NAD 3020

Many of the short comings of the NAD were the usual short comings of many amps built to cost targets. I put Cadas input RCAs, Edison Price Music Post output connectors, Panasonic HFQ power supply caps, and HexFred diodes in the thing. No other changes. The amp sounded seriously good driving modified Magnapan MGIIIs. I did not change any transistors in the amp. It is really easy to attribute the sound of an amp to preconcieved notions. Attention to details, passive parts quality, power supply design, and stability can easily count for as much as semiconductor choice. I also find that better the topology, the more sensistive the design is to these factors. Simple topologies lend themselves to less parts and make using good parts more an option at a given cost. some of the old topologies are competetive with todays designs. I bet a Curl JC-3 wth Toshbia Jfets and Toshiba and Sanyo Bipolars would rock your world and that is a 25 year old design.

Harry, sounds like those parts cost as much as the amp itself.
Yes, although the NAD 3020 semis were european, the passive parts were most definitely the cheapest asian that they could find !.
I recall that although not wonderfully clean nor uncoloured, they were always musical, even erring on the slightly polite side, which is perfectly acceptable to me.

Harry, my comment on the subtle sound character/origin of transistors is from long sonic observations - do you agree ?

Regards, Eric.

Originally quoted by Sidney:

"So in my final version of the amp I used hq parts and matched 2N3055's with Hfe's of 200 for the output trannies and 160 for the 2N3055 current sources (per the JLH site better to have the one lower than the other)."

There is (or should be :) nowhere on my site that states that it is better to have one output device with a higher or lower gain than the other. The best performance for this circuit is achieved when the two output transistors are matched. If matching is not possible, then the lowest distortion is obtained if the higher gain device is used for Tr1. See Table 2 in the 1969 article.

A number of people have reported an audible improvement when changing from the 2N3055 to MJ15003. Care must be taken if other alternatives are considered since the JLH circuit does not have any fixed compensation and relies on component characteristics to achieve stability. Higher speed devices such as the 3281 may or may not work. I am aware of one constructor who has used the 2SC3281 in a 1969 version without any apparent problems whereas another tried the MJL3281A in his 1996 version and the result was oscillation.



I'm afraid I have to disagree with you that the sound of the NAD is due to its use of European transistors. If you look at the schematic you will find that there is a mixture of Japanese (2SC1400), European (various BC and BD devices) and American (2N6551, 2N6554, 2N3055 etc) devices.

I think it is far more likely that the performance is due to the circuit topology. The amplifier avoids the use of long tailed pairs (differentials) and constant current sources. If I have read the schematic correctly, the power amp section has a single input transistor (current feedback) followed by a VAS with boot-strapped current source and then a Darlington EF output stage. The phono stage uses a cascoded single transistor input stage and the pre-amp section also manages to work without the current, seemingly mandatory, differential input.

I agree with you about the quality of the passive devices. All bar four of the resistors in each channel were 1/4W or 1/2W carbon. The only non-carbons used were the series resistor in the phono input (1/4W metal film), the output Zobel resistor (1W metal oxide) and the output stage emitter resistors (2W cement). Non-electrolytic capacitors were either ceramic or Mylar.

It would be very interesting to hear this circuit built with quality components.