JLH 10 Watt class A amplifier

I would test the meter before going any further. Make sure it is set for DC current and not AC.

If you have no suitable resistors to use for a test then try connecting an incandescent mains voltage bulb across the DC supply via the meter. That will pull enough current to get a reading of some sort. A linear halogen bulb would be ideal.

Thanks Mooly. You were right. My Fluke meter is the problem. I dug out another meter and I have amperage output to adjust
 
pDfAs4a.jpg


At No 4192 I gave a concept idea. It works very well except needs low resistor values. I will play with it as having gain is not to be sniffed at. If you look at the LM317 graphs the drop out voltage is better than it would be at full power. I think this circuit has one advantage in that the 2N3055 is not in the feedback loop of the LM317. As it costs very little it is as makers diagram including 2 x 1N4007 diodes.

Here is my banker idea. It's not great yet. It's getting there. I have a LD1084 which now I've thought it through should do nicely. Ripple is about -55 dB lower at the 2N3055 output. Hiss is a tiny bit better than my bench PSU. Top graph is ripple due to a 1 amp load before the regulator.

Some very good news for me. There was only a 2 mm stump of the broken transformer primary ( less than 2 mm I think ). I took a big risk and pulled on it and got 15 mm, I lost about 1 mm at the last pull which was OK. The Nuvotem RS0500 P1-2-025 with triple heatshrink and new insulation outer looks as good as new. I think it " is " as good as new now. It's quiet also.
 
pYbBjwm.jpg


Here is an idea I saw in the 1980's from a Creek amplifier, thanks Michael. He ran two rectifiers from one transformer. He also it for a quiter circuit, the preamp. 1 x WO2 1 amp rectictifier plus 1 x 2200 uF standard grade ( not high ripple ). I get 15 dB reduction in hum. There are a few harmonics that better layout will solve, that is mostly a better 0 V point. 87 dBV = 45 uV hum at 100( 120 ) Hz at 1.2 amps. Otherwize it is the same PSU as before. The ripple is 230 mV BIG capacitor and 15 mV 2200 uF to LM317. That 23 dB better than before of which we get 15 dB. The 2N3055 working very well. If someone tries I think we can work out the current the LM317 is using. 16 mA at a guess. That might only be 11 mA to drive this 2N3055 to 1.2A. That's rather good. 5 mA is used by the 240R/5K6 of the LM317.

I now can run the variac transformer down to 220 ( 110 ) VAC for only a mild change in hum ( 3 dB ), that's 10 VAC better. If you look the LM317 now has 1.3 VDC more to play with.

The beauty of this PSU is a novice should be able to make it. If a Super Beta transistor ( also fast ) was used I would expect slightly better hum. If the small cap was made 4700 uF and the big cap 22000uF 2N3055 will do fine for 2.5 A stereo. I have built better PSU's than this. None as easy to make. Thick wire mostly. I will sort out fusing later. I think 35 VDC at 40V spec is OK. This PSU is using this series of cap from my recycle box. Remember this 2N3055 must go on the main heatsink, LM317 should be off the heatsink as it is easier. Here are typical parts.

http://cpc.farnell.com/multicomp/lpr50v478m30x25/capacitor-4700uf-50v/dp/CA04834
http://cpc.farnell.com/unbranded/am152/bridge-rectifier-1-5a-200v/dp/SC07473
http://cpc.farnell.com/unbranded/cm3502/bridge-rectifier-35a-200v/dp/SC07541
http://cpc.farnell.com/unbranded/cm3502/bridge-rectifier-35a-200v/dp/SC07541

I did try a Darlington ( MJ3000/1 ). I didn't get it to do anything better. Also a BD135+ 2N3055. Same story.

Someone asked why Douglas Self insists AB needs only a simple power supply. The 2N3055 here is working in the same way as Self's amplifiers. At the collector 230,000 uV of ripple, at the emitter only 44 uV. That's nearly 75 dB reduction if the 2N3055 base is fed a clean referece voltage ( or music ). The LM317 reference is - 93 dBV ( - 87 dBV 2N3055 ) hum and hiss the same as the 2N3055. That means the 2N3055 which isn't in a loop of feedback is doing a great job, it has a form of feedback in itself ( link if you like below ). I could imagine this PSU will add a bit of second harmonic distortion, not the one to worry about, I will measure it and if not good enough will say. The output 1000 uF + 0R33 is to mimic the amplifier decoupling which I found to be important. With class A we have the luxury of having a small resistance to protect the 2N3055 during start up. I can return to a unity gain NPN+PNP pair if distortion is an issue. At unity gain it should equal or better this. A 2N3055+ 2N2955 would be fun. That way the Vce of 2N2955 should be excellent and no chance of blowing up the driver. It might even have better hum.

http://www.edaboard.com/thread324683.html

If you do the maths this PSU wastes least energy whilst being linear. If it is adjsted to 27V it should work with any mains voltage. Note I get 26.6 VAC when the rating is 25V at 230 V. This is because I am not using the full 500VA, a useful thing. If you look at a lower cost transformer of lets say 160 VA I would imagine it to be about the same more or less. As a rough guide the 2N3055 needs about >3V minimum in to out to get a good result. The LM317 at this light load should be fine and is getting 1.3V extra due to the rectifier working less hard. I will look at fusing later.
 
Someone said of switch mode PSU's " At least they don't have 50/60 Hz noise" I have some very bad news. They do. If you look in your switch mode power supply ( SMPS ) if lucky you might have a 330 uF 400 V capacitor. Lets pretend the SMPS is 100% efficient. When 25V rms ( AC ) I had 230 mV ripple when 35 VDC. That implies for laziness about 23 mV at 246 Vrms as my house. That was with 10 000 uF. If we now do 10000/330 ( cap sizes ) it's 30:1. That implies ripple of 23 x 30 = 690 mV or 0.7 V. That suggest the SMPS must show some 1 to 1000 Hz noise.

Here are the results of a Meanwell RS 50- 24 at 28.1V into 0R33 + 1000 uF at 1.2A. I will bracket linear PSU.

27.8 V ( 29.3 ) 1.2A

Ripple - 50 dBV 75Hz !!!! - 65 dBV 150 Hz. ( - 87 dBV )
Hiss 1 Hz to 4 kHz - 72 dBV ( - 105 dBV )
Hiss 4kHz to 10 kHz - 83 dBV ( -105 dBV )
Hiss 10 kHz to 20 kHz -65 dBV ( -105 dBV )
Hiss 20 kHz to 16 MHz - 88 dBV ( - 105 to - 107 dBV )

The ripple is a very odd mix of 50 + 100 Hz ( beat ). The load is 50 % max so very fair. If we take - 50 and - 87 that's 47 dB difference. That's 225 : 1 or 10 000 uV : ( 44 uV ). From what I remember of the Meanwell spec that's well inside it. It's respectable, it's not the levels JLH suggested. We use dB's as they are logarithmic. That means vastly big numbers come down to smaller values. We can simply subtract or add. 1 = 10 ( 20 dB ) 2 = 100 ( 40 dB ) 3 = 1000 ( 60 dB ). We use 20 log that number. Thus log 2 = 0.30103 20 log = 6.0206. A handy number in digital work. Log base 10 is used. These are voltage ratios as usually used.

Lets look at the spare heat problem. The Meanwell has none we need think od. My Linear has 2.5 amps at 6V = 15 watts, The amplifier 30 x 2.5 amps = 75 watts. To only have 20% extra heat is a nice happy accident of a 25 Vrms transformer fitting the needs. I will go on to try to get a little more lower voltage usability. As said dropping to 27 V will solve that anyway. One can use Schotkey diodes, I have doubts about them. The rectifier is heat wee need to be careful of. I like the big cheap rectifiers ( 35 or 50 A ) bolted to the chassis.

To sum up SMPS. Big ripple noise of - 50 dBV ( -87 linear ) . Nasty 15 dB hump 10 kHz to 20 kHz ( switching ? ). Comendable > 20 kHz range ( digital interfacing ? ) although nearly 20 dB worse than the Linear ( x 10 ) . I would suspect the a much bigger SMPS " could " reduce this hum if lightly loaded. Ripple " should " also be lower.

The best ever PSU I made was - 103 dBV hum ( 6 uV ? ) and - 111dBV ( 2.5 uV ?) hiss to 16 MHz at 1.5 amps. -87dBV and - 105 dBV are not bad. I seem to remember a standard LM317 claims 20uV hiss. This might get to 500 uV if being lazy. The 5K6 + 47 uF deals with most of the hiss and a lot of ripple. It's 10 pence you must spend. In data sheets it say 10 uF. 47 uF is slightly better. LM317 is vastly better than people think. It is not so good on big complex loads where it can go unstable.The single 2N3055 emitter follower should be much better. A class A amplifier must be a LM317's best possible big load. You can use one per amplifier ( just ) LD1084/85 measure the same when I tested them and have low drop out and higher current. For now I like the LM317 + 2N3055 as it seems logically a nicer solution.

I have graphs if needed.
 
Hello all, I've decided I want to build a JLH Class A from a kit, using a case and heat sink that I have around and 2x 19V SMPS. It's going to be my first DIY amp and I haven't done much electronics since my sixth form days. Nevertheless I can read a schematic and at a push I could make this on Veroboard. However a kit would be easier, I can't find the Hart kits any more, can anyone recommend a decent one? The Chinese Ebay specials seem OK but I'd like to avoid the dodgy 2N3055s of unknown origin if possible and get one with better devices. If this still means a 2N3055 then so be it, but I understand that there are better devices these days, both in TO3 case and whatever they call the flat black ones that are possibly easier to engineer by way of flying leads.

So, does anyone have any recommendations please? There could be a whole £10 at stake!
 
A update on my progress. First, I have not yet electrocuted myself which is a good thing. I still have adjusting to do. I had problems getting voltage down to 50%. I swapped out the transformer from a 400va 24 volt secondaries unit to a 300va 18 volt unit. Same deal. I can only get to 14 volts, not 11 as desired. The channel with the closest adjustment has terrible DC offset at the outputs. Starts at 14 volts at startup and takes two minutes to settle to 0.01. I am waiting for the fuses to arrive for my second meter. Having two setup, one on amperage and the other on voltage should help. Thanks again for the help and knowledge.
 
Do you recommend still using the 2N3055 then? I'm discouraged by hearing that a lot of so-called 2n3055 transistors are just semiconductors that fail to meet any more demanding specs so they are labelled as 2N3055 and thrown out to an unsuspecting world.
Ask ONsemi for samples of MJ15003. They will charge you for postage, but you know you are getting genuine devices.
You could ask for samples of the 3055 as well.
 
Do you recommend still using the 2N3055 then? I'm discouraged by hearing that a lot of so-called 2n3055 transistors are just semiconductors that fail to meet any more demanding specs so they are labelled as 2N3055 and thrown out to an unsuspecting world.

Onsemi devices are genuine. You can get then directly from Onsemi or buy them from Mouser.

From everything I have read, 2N3055 may be the best. I'm not sure the design benefits from faster devices.
 
I am using Indian 2N3055's which by measuring alone seem excellent. The better choice in the UK from CPC is the On MJ15015 that seems to be in the 3055 family and very cheap. To be frank any T03 device of > 60 V with Ft > 1 MHz should sound very good. I have limited my version to 30V 1.2 amps ( +/- 15 if split rail version ). I have a massive heatsink that must be 0.3C/W. Even with one channel it gets very warm. I think in stereo it will be the limit. If we take Vmax to be 36V with my linear PSU that's 36 x 2.5 A = 90W into the heatsink. 90 x 0.3 = 27C. In a typical UK max room temp that's 54C if I can get a good air flow ( not the best look ). 54C is as far as I dare go. The PSU is a surplus 6V at 2.5A = 15W, If we take the T03 case and heat washer to be 2C/W and 18 W that's = 36C. So as best we can get it that's a 90C case temperature. I have a felling it will be higher. My 3055's were £1 for 5 from Rapid some time ago. Everything I can from the scrap bin.

There are two things to say here. Class A does not need fast transistors. However most fast ones are also high gain which is good. My Indian 3055's seem to have a gain of 100 ( not 20 as often thought ). I doubt that can be bettered when not a Darlington. Also 6 watts or 15 watts won't really be a big deal. That's where JLH's 1980 class AB needs thinking about.

Here are very cheap very OK 2N3055's ( < £1 ) from CPC. Look very carefully at the spec sheet. They are very good. Ft > 2.5 MHz , Gain > 100 @ 1.2A @ 150C. The 150C is realistic for the silicon wafer when class A. If you analyze carefully you won't better this. Remember Ft > 2.5 MHz is rather good. I am getting 160 kHz ( -3dB ) close loop gain at gain of 12.25. That implies 2 MHz. I had to limit the gain using 33 pF. It looks to me it bettered 2.5 MHz. Without the 33 pF the amplifier measured far worse, unstable due to the 3055's being fast. The BC337-40 I am using seems a 6 pF device so needs the 33 pF to mimic a slower device. 160 kHz implies 100 kHz dead flat which is excellent.

http://www.farnell.com/datasheets/1673726.pdf
 
I think that the 2N3055 "bad press" originated when epitaxial transistors became prevalent. The original device had an excellent safe operating area, but that is not necessary in the JLH and other amplifiers with lower Vcc's which do not push the device into the second breakdown limited area. Douglas Self's comments that the 2N3055 transistor might no longer be the best probably referred to the original, 0.8MHz fT device, which I would agree with. But the epi 2N3055 is comparable to the MJ15003 at around 2MHz, and indeed similar to the MJ802 which if I recall was Self's choice of output device in the Blameless. I still use 3055's (epi) in one of my 50W designs, which sound good.
And as mentioned, ON semi still manufacture these. ST have declared their 3055 obsolete, but still offer the TIP3055.
 
Fast way remembering;-)
JLH without the first transistor. And shortening of some parts. 10 minutes to do. Sounds ... wonderful;-)
You do not need a board! Sounds ugly! Unnecessary materials and horrible solder jounts and resonances and paths ...-(
Now TO-220-transistors.
May be, christmas I will finish.-))))

My experience:
The capacitance. The higher the cleaner and more powerful and "harder" and so on. But not (a) big cap (big ones do scratch very very vwell;-), better a lot of little ones. At ONE circuit;-! And NO little bridging caps!
Shottky-diodes too.
Active psu, of course.
... but all this does not counter TO-3 character enough;-)


Any TO-220 or TO-247 I would use.

LG

A little bit better;-)


do you have a schema of the JLH without the first transistor?
 
@ Badge:
Did you control the values of the resistors, trimmers?

For example:
JLH AMPLIFIER CLASS "A" VERSION 2000


I bought Chinese ready assembled boards. I have only used the two variable resistance controls to attempt adjustment. One is 2K and the other 100K. I hit a wall where I can't get the voltage any lower by adjusting. I understand each one has a effect on the adjustment and hope to have my second meter back working on amperage so I can have both setup at the same time. I built another amp a number of years back and don't remember the adjustment process to be this involved. Thanks for your help