JLH Class-A help needed! [newbie alert]

[w00t]

Hi all,

I'm currently stuck between the 1996 JLH and the 2000 JLH, not knowing the difference in quality. I have absolutely no experience in building amps using discrete components (have built amps with tubes and opamps though) and hence would like to seek your valuable advice.

JLH 2000:
http://www.tcaas.btinternet.co.uk/jlh2000.pdf

It shows the schematic on the last page, which requires +36V, unlike the other designs I have seen utilising +22V or so. Are there any 36V supplies to recommend?

There is also the 'Set V.dc out' and 'Set Iq' which I do not know how to go about doing it. Is there a way where I can solder in an ammeter/voltmeter to set v.dc and iq on the go?

Regarding the 2N3055 used, I remember reading somewhere that MJ21194 could be better sonically. Would I just swap the 2N3055 (Q1) with the MJ21194? Thanks!

JLH1996:
http://www.tcaas.btinternet.co.uk/jlh1996.pdf

More or less the same questions as the JLH2000, the power supply transformers have no rating on them.

Thanks, and cheers!

[eLarson]

Welcome to the forum. I see you already found Geoff's great Class A Amplifier Site. Read over Tim Andrew's comments on building the modified JLH. He gives good descriptions of the differences that he heard as he modified his JLH. Note, too, that he ran into some low-level noise while using MJ21194 output devices. He had better luck with the MJ15003, finding similar- sounding improvements as with the MJ21194.

I highly recommend reading through this thread: <a href='http://www.diyaudio.com/forums/showthread.php?s=&threadid=3075&highlight='>JLH 10 Watt class A amplifier</a>

[Geoff]

Hi w00t

The schematic in the jlh2000.pdf was a slight rehash of the original 1969 design and was not intended to supercede the 1996 version.

Before going any further, as Eric suggests, check out the 'JLH Update' page on my website which offers some potential improvements to the 1996 version. The use of a ccs in the output dc offset control circuitry removes the possibility of noise generated by the 7815 from being injected into the feedback node.

Also, the 1996 quiescent current control circuit does not operate as well as it should because the voltage reference applied to the base of Tr5 varies according to the gains of Tr1 and Tr2 (which change with junction temperature as the amp warms up), due to the volt drops through RV2 and R7 varying with Tr1 and Tr2 base current.

The MJ21194 has too low a gain for my liking and my preferred output transistor at the moment is the MJ15003.

Please feel free to email me with any specific queries you may have. Usually I try to reply the same day, but just at the moment there may be short delay. My current pc is on its last legs (well, it is 5 years old!) and has been giving me some dire warnings during the past couple of days so I am frantically trying to transfer all the accumulated programs, files and data onto a new pc that I bought today. Hopefully, normal service will be resumed before too long.

Geoff

[hlyytine]

Hello w00t,

It shows the schematic on the last page, which requires +36V, unlike the other designs I have seen utilising +22V or so. Are there any 36V supplies to recommend?

Those 22 V and 36 V figures assume a specific load impedance. The higher the targeted load impedance, the higher rail voltage you will need. Don't worry about supply voltage, I'm currently running my ever evolving prototype (note to other readers: I've currently plugged a 2SC945 to drive the phase splitter, the collector of 2SC945 connected to ground. One day I tried CFP input stage but couldn't get rid of oscillations, now I think I've found the cure) at +/- 14 volts and a quiescent current of 3.5 A (max. power at about 2.5 Ohms). If you indent to use those 2N3055's, you will be safe even at +/- 25 V (or 50 volts in single-rail versions). Just build one, you will be impressed anyway

There is also the 'Set V.dc out' and 'Set Iq' which I do not know how to go about doing it. Is there a way where I can solder in an ammeter/voltmeter to set v.dc and iq on the go?

I'm referring to figure 2 at http://www.tcaas.btinternet.co.uk/jlhupdate.htm. You could measure the voltage drop across R10 to determine the quiescent current. Since the majority of the current drawn from the positive rail consists of the quiescent current through Q2 and Q1 (only few tens of mAs goes to the rest), I just plugged the ammeter to the positive rail. Plug a voltmeter across the output or even better, a scope if you have one (you'll see if it oscillates or not).

Now the most important: before you power on the whole thing, adjust VR2 to maximum resistance and VR1 to halfway. As you slowly crank up the quiescent current using VR2, keep the output at zero by adjusting VR1. Just a note from my experiences: the quiescent current should track VR2 adjustments without jumps. Blew a couple of MJ15003's and an electrolytic capacitor Sunday

Regarding the 2N3055 used, I remember reading somewhere that MJ21194 could be better sonically. Would I just swap the 2N3055 (Q1) with the MJ21194? Thanks!

I don't know about MJ21194 but at least with MJ15003 the improvement is definitely real. But since you're new to JLH, I'd suggest you build a prototype first using 2N3055's. They're cheaper and after you get it running straight, throw in the MJ15003's (but do readjust the quiescent current, at least my MJ15003's have higher gain than my 2N3055's).

More or less the same questions as the JLH2000, the power supply transformers have no rating on them.

They say that the transformer rating has to be five times the amplifier output power. The lesser will work definitely and you can upgrade your supply later.

[w00t]

Hi everyone,

Thanks for the inputs. I'm still digesting them...

So, what's the winding voltages and VA i need for the JLH1996 design assuming I am driving an 8ohm load?

Thanks!

[w00t]

I just noticed the 7815 on the 1996 JLH where the out of the 7815 is pointed to "the other channel".

From what I read, itone channel is receiving 22V and the other channel 15V? Please enlighten me. Thanks!

[Geoff]

Originally posted by w00t
So, what's the winding voltages and VA i need for the JLH1996 design assuming I am driving an 8ohm load?

This depends on the supply rail voltage and quiescent current that you intend to use and also on the type of power supply (simple capacitor or regulated).

For the standard 1996 version with +/-22V rails, an Iq of 2A and a regulated supply you will need a transformer with two secondaries each rated at 22Vrms. Ideally, the VA rating should be about twice the minimum requirement. The minimum will be 2*22V*2A per channel, so with a monoblock arrangement you should be looking for about 175VA per channel (a standard 160VA will do) or if you intend to use a single power supply for both channels then you will need a transformer rated at about 350VA ( the nearest standard rating of 300VA should be OK).

Geoff

[Geoff]

Originally posted by w00t
I just noticed the 7815 on the 1996 JLH where the out of the 7815 is pointed to "the other channel".

From what I read, itone channel is receiving 22V and the other channel 15V? Please enlighten me. Thanks!

Both channels require the +/-22V supply rails. JLH had a policy of not using more components than strictly necessary, so if certain ones could be made common to both channels then that is what he did. In the case of the 1996 version, he proposed the use of a single 7815 (and C3) feeding a separate RV1/R1 in each channel.

IMO, the use of a 7815 in this position is not ideal on two grounds. First off, the 7815 is not renown for being a particularly quiet device and I do not like the idea of injecting HF noise into the feedback node.

Secondly, the 7815 is only supplying about 500uA of current (or 1mA if a single 7815 is used for both channels) which is below the minimum required to ensure stable operation of the regulator. One solution to this problem is to fit a resistor in parallel with C3, sized to ensure that a minimum of, say, 5mA flows through the 7815.

Alternatively, the 7815 can be replaced with a TL431 shunt regulator, or a ccs (as shown on my website), which will solve the regulator instability problems and will reduced the amount of noise injected into the feedback node.

Geoff

[hlyytine]

w00t,

Geoff already answered your questions, but I recommend that you build the 2000 version (two constant current sources). That way you will not need any 7815's polluting your feedback node. Essentially the 2000 version is as simple as the 1996 version.

I must admit I haven't build the 1996 version but somehow I feel you really should try the 2000 version first. Please have a look what Geoff and Tim Andrews have found out while the 2000 version evolved from the 1996 version, the full story is at Geoff's website.

[w00t]

Thanks for all the inputs!

I guess I shall go with the 2000 version.

Between http://www.tcaas.btinternet.co.uk/jlhupdate.htm (Fig 2) and the 2000 JLH, Fig 2 seems have a slightly higher component count.

If I were to build the 2000 JLH, how should I go about modifying the power supply? I guess the tranformer windings would have to be rated at 25-0-25 now.

Thanks for answering my questions! Will probably drive the Corals Flat 8A or Foxtex FE103E with this amp.

Cheers!

edit: I see that there is a 2200uF capacitor near the LS output. Can this cap. be of the lowest voltage rating possible?

edit2: Regarding the heatsinks, what are the dimensions (aluminium finned) you would recommend for the JLH 2000? Thanks once again!