I think I recognise the heatsinks Hood used, If so he ran them at ~80C when it wasn't playing anything. 😉 Whoops - plus ambient.
I looked around my archives and found the original PCB layout of the black JLH shown above. It uses quite a hefty heat sink. The layout is a mirrored stereo amp on a single PCB using all 1/2 watt resistors. I can post pictures and Gerber should anybody be interested and is lazy to do your own layout. It uses the MJE4281A output transistors. It is hot but stable in every respect.
You are welcome. To anyone other, this is not: "I think it works amplifier, many was built and working perfectly."Please don't tell me I should have done this or that. Or the layout is, whatever you disagree with or don't like or feel or assume. There are nothing audiophile or special components or golden this or that in the layout. It was designed and tested several times. To those disagreeables, I won't even reply to such comments. If some data is missing, ask, and I will search my archives for it and provide it to you. HAVE FUN AND ENJOY YOUR JLH
Last edited:
Sorry Guys the PDFs was an updated one with a few changes on it from the Gerbers. Thus the PDFs are updates from the Gerber files. I do not have Gerbers of the PDFs on my current hard drives but the PDF should be 1:1 should you prefer working from the later version. I had no problem driving the amp with a HP 27V 3.5 amp SMPS.
Last edited:
The Miller capacitor is virtually certain to control oscillation as it generates the ideal 6dB/octave roll off at all frequencies without the unity gain effect of the phase lead capacitor. It has the effect of slew rate limiting the response to a fast input which you may note John Linsley Hood avoided using. With fast transistors I would recommend using the smallest value to give stability. A Tian simulation indicated that with 33pF the phase margin was 87 degrees and the gain margin over 40dB. You would probably need less and I suspect Nico's 22pF is adequate. Certainly the phase lead has less margin but as JLH said the use of an RC network at the output helps. The point he did not say is that it provides a resistive load at high frequencies which keeps the open loop gain under control, since an inductive load could provide a higher impedance and with it a higher OLG which is harder to stabilise. My proposal was to include the LR network as well.
Nice work Nico. Do you think the standard two potentiometers might be added- one in the bootstrap and the other in the bias network to set the Iq and centre rail voltage? It would be a hard decision because the sizes and shapes of potentiometers seem to get smaller and different each time I search, and you would need a 1W type for the bootstrap. Or possibly provide a position for parallelling resistors, so you start with the nominal top value expected for the highest gain transistors, measure the current you get and add a parallel transistor to one or both resistors in the chain as evaluation dictates.
With two large capacitors per channel is it worth placing a wire link between them in the power tracks so that if someone wants to put a resistor or a choke between them they could?
With two large capacitors per channel is it worth placing a wire link between them in the power tracks so that if someone wants to put a resistor or a choke between them they could?
Hello John,
Regarding the two pots, I found this video from Steve Wagner and wondered the same
I then found that I had the exact same kit and had put it on one side thinking it to be useless with no adjustment.
It's the one I use now £10 for the pair as I recall at the time, they've shot up to £14 now 🙂
Cheers
Regarding the two pots, I found this video from Steve Wagner and wondered the same
It's the one I use now £10 for the pair as I recall at the time, they've shot up to £14 now 🙂
Cheers
John, yes a bias pot would be okay, but I set my bias and offset with the power supply I suggested manually. With this amp, I always found it sounding a little rougher with the offset at zero than slightly above. For what reason I do not know, but when the output voltage is slightly above V/2 it sounded better. I think, looking at the design, John always insisted the higher gain output transistor was to be at the top. Interesting but true.
John, BTW I set the bias once with a pot and replaced it with fixed resistors, It made no sense to me having bias set to precisely 1.3 amp, if it was 1.27 amp or 1.32 amp who cared, it ran a little cooler or warmer. However, it must be noted that as soon as you exceed the set bias current when playing music it does not distort a little more like a typical AB amp, the sound just falls apart completely.
Edit: The bootstrap is just an old/cheap way to realize a CCS, in fact, personally a CCS sounds a little more "refined" but for some psychoacoustic reason the bootstrap sounded louder, do ask why, it is just so. Mind you swinging the extra volt or so closer to the rail on a 10 watt amp may sound louder????
Edit: The bootstrap is just an old/cheap way to realize a CCS, in fact, personally a CCS sounds a little more "refined" but for some psychoacoustic reason the bootstrap sounded louder, do ask why, it is just so. Mind you swinging the extra volt or so closer to the rail on a 10 watt amp may sound louder????
Last edited:
Now, probably the bad news. I liked the sound of JLH 1969 amp a lot, I had to spend a lot of money and very complex circuitry to surpass it sound signature. I have built JLH for many a friend and used other output devices and I must admit they all sounded the same. The only sound signature that appealed to me more was a single-ended design. But in the single ended designs, it was more noticeable (not day and night) what sound signature a specific output device offered. For single ended, a lateral mosfet sounded the sweetest to me, regardless of the music you listen to.
The highest gain should be at the bottom and output correct to 0.25v. A simple answer is bias high on purpose and fit that resistor in a way you can solder another in parallel with it, 110 or ?? 120k should do it. I would deliberately fit that one high,
Quick edit. His distortion tests indicate that the gain match needs to be as close as possible.
Quick edit. His distortion tests indicate that the gain match needs to be as close as possible.
Last edited:
I have all his articles in an archived HDD if anyone is interested. As mentioned, I will share what I remembered of the 40+ built with who needs it. I cannot imagine JLH direct words after 50 odd years, nor am I interested. Build, some yourself and then jump in, comment and help others do the same.
As not mentioned, the beast above is one channel of an 1080 watt/ch class A. It sounded spectacular. Only to be used in a ducted air-conditioned listening hall. Each power supply consisted of a 4kVA transformer. Besides the owner, only a few other people in the world were lucky enough to hear its performance. It took a design team more than a year to produce, and it was auditioned with two pairs of B&W 800s at the time we heard it.
As not mentioned, the beast above is one channel of an 1080 watt/ch class A. It sounded spectacular. Only to be used in a ducted air-conditioned listening hall. Each power supply consisted of a 4kVA transformer. Besides the owner, only a few other people in the world were lucky enough to hear its performance. It took a design team more than a year to produce, and it was auditioned with two pairs of B&W 800s at the time we heard it.
Last edited:
The last I used successfully was that of a Hewlett Packard laptop SMPS. I think it was 27V at 3.5 amp works fine and does not mind the high capacitive load as it has some switch on delay. I do not think I will ever divert back to ordinary transformers again.
