Linn LK2 power amplifier

[Kat Manton]

Hi folks,

I recently acquired a Linn LK2 power amplifier (in exchange for overhauling a rather nice classic synth) and finally had a chance to try it out t'other day.

First things first; I made sure it wasn't going to fry anything. A meter on DC volts showed a few mV on the speaker outputs; nothing to worry about. Switching to AC volts similarly showed almost nothing. Sticking a finger on the input pin got me about a volt. So neither channel had gone hard over to to a supply rail, it didn't appear to be oscillating and it did seem to be amplifying. Now I could connect it to something.

Linn have used XLR connectors for the inputs but, defying convention, they used male XLR connectors. Just to add to the fun, the inputs to each channel are unbalanced and wired differently. Pin 1 is ground, as one would expect. The left input is on pin 2 but the right input is on pin 3. The pin-outs are shown on a label on the back, but it doesn't take much searching to find a few people have been caught out by this, probably buying off-the-shelf RCA-XLR cables only to find one channel doesn't work. It's worth bearing in mind if you're faced with a Linn LK2 (LK275, LK280) with a "one channel dead" fault!

Doubtless this wouldn't have been a problem originally if one had bought an LK1 pre-amplifier at the same time as it would come with the correct cables.

I hadn't got a pre-amplifier, but I did have a Cloud CX133 Zone Mixer. Well, it's got inputs, outputs and a volume knob...

Side Note: CX133 Zone Mixer?!

It'd been given to me a while ago, essentially as 'scrap' as it'd had something spilled inside it. Whatever that was, it was highly corrosive and had eaten PCB tracks and the leads of a number of components. I took this as a challenge and repaired it. The two 'Zone' outputs have VCAs, allowing remote volume adjustment. Both 'THAT 2150' VCAs for Zone 1 had been eaten, I hadn't got any, but I did have a few 'dbx 2151' VCAs; the earlier incarnation of the higher-specification (lower distortion) version of the same device. They're used in a few studio mixing consoles which have VCA faders and/or dynamics processing on every channel, so I reckoned they'd do.

Whilst inside, I reverse-engineered parts of the circuit then built inverting unity-gain buffers on strip-board so I could bypass the rear-panel bass/treble controls; I figured it'd be more use to me without tone controls. I also swapped Zone 1's output buffer (NE5532) for an OP275 and used another one in the 'tone bypass' amp for Zone 1 (Zone 2 uses a TL072) more out of curiosity than anything. (There are claims that different op-amps "sound different"; I wondered if I could hear and/or measure a difference.)

This departure from the main programme is brought to you in case anyone's wondering what on earth possessed me to used a zone mixer usually employed in pubs etc. as a pre-amplifier for a classic power amplifier. After my tweaks, it's a lot more 'pro/hi-fi' than 'pub', exceeding the original specification by a fair margin. It's 0.5 dB down at 10 Hz and 55 kHz, flat as a flat thing in-between and THD measures < 0.02% (on my ST1700B with the dodgy oscillator); I think it'll do...

Having made up a couple of short cables with a female XLR on both ends and one cable wired 'wrong', I pondered whether to measure it first then listen to it or vice versa...

I tried the LK2 out on headphones first. Stax SR-X Mark 3 (which are good) via SRD-6 passive adapter (which... isn't.) Source material was FLAC via Edirol FA-101.

Impressive! I've driven these headphones from all manner of amplifiers and never had them sound as neutral as this before. There's usually a sort-of mid-range 'shout' which I've tended to put down to the amplifier interacting with what's probably a somewhat reactive load (the passive adapter uses step-up transformers) but I've never really investigated. Driven by the LK2, I was now enjoying a more neutral tonal balance along with plenty of detail. Promising; certainly music was engaging enough for me to stay up half the night listening to everything from surf instrumentals to psychedelic dance music (leaving me with ears like two pieces of raw steak, I must get/make new ear-pads for the things!)

Now... speakers. I connected it up to the (modified/upgraded) Celestion Ditton 44s which lurk in the living room; source material this time was FLAC and the odd high bit-rate MP3 via M-Audio Delta 44.

Interesting... Plenty of detail; it was possible to pick out and follow absolutely any instrument/sound in a track no matter how low down in the mix it was. Vocals, male and female, sounded natural as did acoustic instruments. Notes started and stopped when they should without 'hanging on' and... I'm in danger of sounding like a hi-fi magazine if I carry on.

So it's good, then? Yes, but... It was very laid-back to the extent that the lower registers dominated. Far from balanced and neutral, I found it difficult to listen to anything without my attention being drawn towards the bass. I've found the LK2 described elsewhere as 'laid-back' and 'dark' but this was... unsubtle.

(This wasn't the loose, uncontrolled bass of a valve amplifier failing to control a 12" bass driver on the far side of a fourth-order LPF; I know what that sounds like. This was tight, detailed, well-defined, controlled... there just seemed to be a bit too much of it. It's not often I found myself wishing there were tone controls somewhere!)

This is obviously a very good amplifier, as one would expect. But what's going on?

• I just need to get used to it? Perhaps the amp I was using before is a bit on the 'thin' side?
• There's something wrong with it? It is, after all, about 25 years old.
• It's interacting in some strange way with my speakers? There's a lot of L and C in those Ditton crossovers.

I think this needs investigating... It's time to dig out a diagram (the service manual can be downloaded here) and throw it at some test equipment!

(I also think this post is long enough and it's an appropriate point for me to grab a coffee before writing the next bit.)

Kat

[Kat Manton]

Hi again,

I've grabbed both a coffee and the near-illegible notes of a few measurements:

Left Channel:

• Frequency response, -3 dB points: < 10 Hz and 37 kHz
• Frequency response, -1 dB points: 18 Hz and 17.8 kHz
• Frequency response, -0.5 dB points: 25.0 Hz and 12.0 kHz
• Distortion: 0.012%
• 100 Hz spikes visible on output with no signal

Right Channel:

• Frequency response, -3 dB points: < 10 Hz and 38 kHz
• Frequency response, -1 dB points: 17.5 Hz and 19.0 kHz
• Frequency response, -0.5 dB points: 24.0 Hz and 13.0 kHz
• Distortion: 0.04%
• 100 Hz spikes visible on output with no signal

Okay, so the frequency response is gently rolling off towards high frequencies. I've measured other amplifiers which are 0.5 dB down at 30 kHz or higher and are practically flat from 20 Hz to 20 kHz, so this is unexpected but might explain something.

I found this 'sterophile' Linn LK1 preamplifier & LK280 power amplifier review. Wading through it, one eventually finds:

At a 1V output level, the '280 was ostensibly flat throughout most of the audio band, though there was a slight droop at the edges, the response being 0.75dB down at 21Hz and 20kHz.

This isn't a '280, but it's interesting to see how the later version has a flatter response.

There's also this:

As it transpired, however, I was less than impressed with the LK2, finding, as did Alvin Gold back in Vol.9 No.2, that while it had a somewhat laid-back balance, it also suffered a pervasive "gray" coloration, which dried out recorded ambience and obscured fine detail.

Could that be the frequency response and just possibly the effects of the 100 Hz rectifier switching spikes (which would probably be worse when the PSU is under some load)?

Finally, there's this:

Alvin Gold had reported on an inherent instability in the amp's predecessor, the LK2

Hmm... "instability"..?

Checking the serial number against this list of revisions, this one was made between October 1985 and April 1986.

The service manual mentions "reliability modifications", none of which appear to have been carried out. So this is a very early one which hasn't received any dealer upgrades. There's also what looks like a Zobel network visible in the photographs in the manual which doesn't seem to be present(!)

The diagram in the service manual is for the LK280 but hasn't got any component values on it. I tracked down a circuit for the LK2/LK275, but it looks reverse-engineered, I think I've spotted at least one mistake and it isn't laid-out either how I would've drawn it or like the one in the manual, making comparisons more difficult.

I identified one difference before giving up; the LK280 has 0.1 uF capacitors in parallel with the 100 uF electrolytics on the voltage regulator outputs; the LK2 doesn't. I'm thinking that may help with the 100 Hz spikes.

One of the first things to do is to identify why the distortion is higher on the right channel. Looking at the distortion residual on the 'scope, it looks like crossover distortion. My initial suspects are the resistors which set the biasing, one/more may have changed value.

Then, I think it'd be interesting to play 'spot the difference' between this early LK2 and the LK280 to identify revisions which can be 'back-ported'.

The diagrams leave something to be desired; I'll start by re-drawing the LK280 diagram from the manual with the component values on it, then examine one of the LK2 boards and amend a copy of the diagram to reflect the differences.

Back... eventually; with decent circuit diagrams (with a bit of luck and more coffee.)

Has anyone else owned and/or repaired one of these? I'd be interested in knowing how it sounded. I have a feeling there's a particularly good amplifier inside this box, if I can fix the faults and tweak the frequency response without turning it into an oscillator...

Kat

[mhennessy]

Originally Posted by Kat Manton

It's time to dig out a diagram (the service manual can be downloaded here) and throw it at some test equipment!

The circuit diagram in that manual isn't too bad, though will be much better once you've added values. The next stage is to realise that the majority of transistors aren't really doing anything; many of them are simply cascoding other transistors, many of which are simple current sources or mirrors. The benefits of cascoding are questionable in many applications, although it's commonly done in any amplifier with hi-end pretensions...

To whit, and at the risk of stating the obvious, Q1/Q2 are the LTP, Q10 is the VAS, Q20/21/22 and Q23 is the positive OPS, in CFP configuration; likewise Q25/26/27 and Q24 is the negative half. Obviously, Q13 sets bias.

Q3/4 are the current mirror loading the input stage. Generally, a Good Idea.

Q5/6 forms the current sources for the LTP. R3 actually makes the LTP a LTP, if you see what I mean. Normally, one current source does the job, and R3 would be two separate resistors (assuming the designer chooses to degenerate the LTP, which is generally a Good Idea). It's common to use this "pi" configuration in something like an oscilloscope, where certain DC conditions can't be met be the simple "T" configuration, but in an audio amp, it's very rare indeed...

Q17 is supplying operating current to the collector of the VAS.

Looking at the top stuff, Q7 and Q8 form a decent "ring of two" current source. I like these; the extra transistor keeps the current more constant because the collector current variations in Q8 caused by Early Effect is corrected. Not that that matters here, as the output is cascoded by Q9 and Q11. The two diodes do the job of holding the bases constant, and also supply current to the lower input current sources via R11, obviously D3 volts that voltage still. The voltage to cascode the VAS comes from these diodes...

I think what happens here is that as the VAS brings the drive voltage on its collector up, more current flows through it. This means that less current flows through Q9, Q11 and Q14. This current is reflected by current mirror Q16 and Q17, which reduces the pull-down current seen by the VAS. As the VAS is turning off to reduce the voltage to drive the output down, the current supplied to the current mirror increases, pulling down further...

Normally, the output of a VAS is only loaded with a current source (or even bootstrapped resistors). But the trouble with this is that the VAS can source a lot more current that the current source can sink (assuming the VAS is at the positive rail, as here). This isn't a major problem in practice, but can lead to unequal slew rates, and if the output stage lacks current gain, unsymmetrical drive can increase even-order harmonics. In this arrangement, it looks like the push-pull currents available are equal (limited by the current source Q7/Q8). There are much easier ways to arrange this...

The block labelled U3 is very intriguing, to say the least. It's obviously regulating the PSU, with help from the LM317/337 regulators, and providing current limiting in lieu of current protection in the amplifier itself. Presumably U3 is a thick-film module? Hard to see in the photos in the manual...

This fits with my previous experience of Linn products, I'm afraid. Lots of unnecessary complexity for little apparent benefit - as an old colleague used to say, "designed by a graduate"! And the way it's put together looks to be a bit of a nightmare. There isn't a lot of space on the PCB for all those hot-running components, and I would expect to see several discoloured areas.

I would also advise caution when tweaking this one - things could get very expensive if it does turn into an oscillator. I'm not sure that 1dB down at ~18kHz is all that audible, but it would be safer to put a network ahead of the amplifier to boost the HF gently, rather than trying to increase the closed-loop bandwidth. Of course, take a close look at R2 and C1 - which should be forming an ultrasonic filter well above the audio band, but it will be dependant on source impedance... I will think harder about the problem once the component values are on the diagram though (damn this man-flu )

As it seems bass-heavy, what is your normal amp?

In my experience, amplifiers that sound a little "bright" have relatively high amounts of distortion, and "duller" amplifiers are actually much better in this regard. Occasionally though, an amplifier with high distortion can sound dull because it's been tweaked to roll off early. But in general, I've rarely found much use in steady-state frequency response checks, and a look at the distortion residual is much more valuable. Once it's down to below 0.01%, you should be OK

It's a nice box though. Depending on the size of the transformer, I'd be inclined to replace the works with a couple of LM3886 or similar. Obviously, I'm joking , but that's what I've been using for the last few years, despite having seen many other amps through here in that time. Obviously I'm biased though; everyone loves their babies

Looking forward to some pictures!

Cheers,

Mark

[PaulE27]

Hi Kat

If both channels of the Linn sound the same it is very unlikely there is a fault, so I reckon you are just noticing the different sound from your old amp. From your measurements the early hf roll-off might just be audible, but you are really hearing differences in the bass, aren't you?

As a matter of interest, what is the old amp?

Regards

Paul

[Kat Manton]

Hi,

I'll just answer this briefly as I think it's illustrative:

Originally Posted by PaulE27

As a matter of interest, what is the old amp?

It's not one, it's several; I have a few 'on rotation':

• Sansui AU-317 II
• Musical Fidelity B1
• Marantz 1150
• Pair of what used to be 100V line mixer amps (!)

None stand out as having any frequency response peculiarities; they're all broadly similar. I've also had other amps pass through and spend time between DACs and Dittons after being repaired. None have ever stood out by over/under-emphasising one part of the spectrum or another (unless they're faulty or just... not very good.)

That's what I find strange; the LK2 is either bass-heavy (or lacking treble, depending on how you look at it) and stands out from everything else because of it.

I've removed the right-channel board and found dry joints and components which look a bit... lightly roasted. It may just be down to components which have deteriorated in similar ways through long exposure to elevated temperatures rather than actually failing. I'll finish off the diagram then investigate

Kat

[Patrick Dixon]

I don't know about this amp in particular, but in general Linn stuff seems to sound fairly detailed but a bit uninvolving to me. The opposite to bass heavy, top light.

[Kat Manton]

Hi,

I think we may be coming up against that perennial communication problem, that of describing sound in written English. (How many times have we encountered faults described as a "whine" only to find out it's mains hum due to duff smoothing caps..?) Most, if not all words used have been borrowed from other senses and there's only a broad consensus on how they're applied.

To me, "bass heavy" and "top light" are almost the same, not opposites; I'd argue that the use of one or the other depends on how much the mid-range is affected. The opposite to both would be "bass light" though the word I usually use is "thin" (or "bright". Or "sharp". Or... I'll stop there..!)

If I ignore my ears and go with the measurements, the LK2 (or this LK2; I shouldn't judge them all from one example) is probably best described as "top light". I forgot to mention the 0 dB reference used for those measurements was at 1 kHz (1.0 V into a 10-ohm load, I need some more resistors!)

At it's "ostensibly flat" from 1 kHz down and 1 kHz is "mid-range", that makes "top light" probably a better description than "bass heavy".

So... I'm fairly sure I'm hearing the measured HF roll-off; it's the obvious explanation. (I've measured and listened to a lot of equipment over the years; I think, with experience, one can learn to correlate measurements with what one can hear. Incidentally, I can still hear past 21 kHz in my mid-forties; that used to be 24 kHz in my teens. Differences in HF response may be more noticeable to me as I'm possibly part bat...)

The circuit diagram isn't far off being finished. One 'spanner in the works' is that the BOM in the service manual is for the LK2/LK275 though the diagram is for the LK280. Also, the differences are fairly significant, I'm having to reverse-engineer the LK2 rather more than I expected.

Kat

[PJL]

Were the frequency response measurements with a load? Was it a dummy resistive load or a speaker? Being a believer in science, if you can hear it, you must be able to measure it. How about a side to side comparison on a scope running the same source through 2 different amps.

[Kat Manton]

IMO I don't need to look any further; I can hear a lack of treble and I've measured an HF roll-off.

I might measure one or more of the other amps for comparison; I'm fairly sure they aren't 1 dB down at 18 kHz though!

[mhennessy]

Originally Posted by Kat Manton

To me, "bass heavy" and "top light" are almost the same, not opposites; I'd argue that the use of one or the other depends on how much the mid-range is affected. The opposite to both would be "bass light" though the word I usually use is "thin" (or "bright". Or "sharp". Or... I'll stop there..!)

That's where I'd disagree; because to me, the spectrum is split between (at least) bass, midbass, lower mid, mid, upper mid, lower treble, upper treble. So I could legitimately describe something as being both bass heavy and sharp, to use your expressions.

It's very possible to have amplifiers that measure identically when performing frequency response plots into resistive loads, yet sound noticeably different when driving loudspeakers. You won't need me to explain how the impedance varies around resonance (e.g. from 4ohms to 100+ohms for a nominally 8 ohm loudspeaker), and of course there is a large reactive component to that. A lot of that stuff happens through the crossover region(s), and at HF too. The interface between amplifier and loudspeaker is still the matter of much research, and I wouldn't pretend to know all that much about it. I also mentioned harmonic content earlier, and this is important to the subjective impression too.

So... I'm fairly sure I'm hearing the measured HF roll-off; it's the obvious explanation. (I've measured and listened to a lot of equipment over the years; I think, with experience, one can learn to correlate measurements with what one can hear.

I would be interested to know what source material you are using that has significant energy at those frequencies. I've not come across much, apart from some electronica, although with that sort of music, it's interesting to judge differences, but not accuracy or neutrality.

All said, it's easy to prove or disprove the HF rolloff with the use of a simple filter to provide the appropriate HF boost ahead of the power amp, as I mentioned above. Would be 10 minutes work to build one and sort the values...

And another thing: you mention the other amplifiers, but were they auditioned using the same preamp?

Incidentally, I can still hear past 21 kHz in my mid-forties; that used to be 24 kHz in my teens. Differences in HF response may be more noticeable to me as I'm possibly part bat...)

I once convinced myself I could hear 23kHz (in my early 20s). Couldn't hear much at 22kHz or 24kHz, but definitely and distinctly 23kHz. Tried it with another tweeter, and could hear 21kHz, but nothing at 23kHz. Tried yet another, and found that it was reasonably up to about 19kHz, but nothing doing above...

Guess what - the first two were metal dome tweeters, the 3rd was a decent soft-dome unit. As you no-doubt know, metal-dome tweeters have hi-Q breakup nodes at these ultrasonic frequencies that produce audible tones. I avoid them at all costs these days, but some of my more "mature" colleagues like them (how's that for diplomacy )

I'm still the right side of 40, but can hear 19kHz still. But, if I'm completely honest with myself, I can't make any claims whatsoever because my measurements now and then were completely uncalibrated. I have no idea how many more dBs I'm having to feed into the same tweeter to hear the signal, but I'm sure it's some . Since my 20s, I've had 2 kids, and they make a lot of noise! Certainly, when I demonstrate things like this in class, I notice that I'm rather less bothered by high-level 15kHz than the younger people in the room. Line whistle still troubles me, sadly... Or reassuringly, perhaps

Needless to say (hopefully!), none of the above "confessional" is meant to challenge your experiences in any way

The circuit diagram isn't far off being finished. One 'spanner in the works' is that the BOM in the service manual is for the LK2/LK275 though the diagram is for the LK280. Also, the differences are fairly significant, I'm having to reverse-engineer the LK2 rather more than I expected.

That'll be good - it might be worth banging it into a simulator - perhaps to get a plot of open-loop gain so that we can work out how to safely flatten the response...

Cheers,

Mark

[mhennessy]

Originally Posted by Kat Manton

IMO I don't need to look any further; I can hear a lack of treble and I've measured an HF roll-off.

I might measure one or more of the other amps for comparison; I'm fairly sure they aren't 1 dB down at 18 kHz though!

It's accepted within the BBC at least that a SM can just hear a 1dB change at 10kHz, with a reference. But having met many of them, not all of them can still hear much at 18KHz. And, we've never gone above 15kHz anyway (think NICAM, which is how most analogue radio is still done, perhaps amazingly).

Some BBC loudspeakers use a 34mm Audax dome tweeter that starts rolling off above 16kHz, though there aren't too many of these left in the system these days...

DAB or DTV doesn't go about 20.25kHz, despite the 48kHz sampling, thanks to MUSICAM encoding.

We might consider using 96kHz or 192kHz for capture and post-production, but I don't believe there are any specific recommendations or requirements yet. Archiving might be using these - I'd have to check (at this stage, it's a rumour). I'm certainly open to higher sample rates - especially as storage is cheap - and I think we have a lot left to learn about all this, but I'm sure the issue is as much to do with the phase response of the filtering as anything else. Recently, Hi-Fi News started performing statistical analysis of hi-res downloads, because they found that some were simply up-sampled from 44.1/48kHz originals

Cheers,

Mark

[Kat Manton]

Hi,

Originally Posted by mhennessy

Looking forward to some pictures!

Edit: I've uploaded these at 800 x 600, you'll probably need to click them at least twice to view them full-size.

Overall view of RH amp PCB:

Originally Posted by mhennessy

And the way it's put together looks to be a bit of a nightmare. There isn't a lot of space on the PCB for all those hot-running components, and I would expect to see several discoloured areas.

Close-ups of the amplifier section:



I think the brown resistors in the middle were once the same shade of blue as the ones around the edge...

Oh, and doesn't polystyrene soften (if not melt) at a fairly low temperature? I think I'll remove those polystyrene caps and check them on the bridge, then try to space them (or replacements) a bit further from sources of heat. I definitely need to resolder a few joints and I think I'll measure the discoloured resistors in case any have changed value. That may deal with the crossover distortion on this channel, it could be varied by application of mechanical shock...

Originally Posted by mhennessy

The block labelled U3 is very intriguing, to say the least. It's obviously regulating the PSU, with help from the LM317/337 regulators, and providing current limiting in lieu of current protection in the amplifier itself. Presumably U3 is a thick-film module? Hard to see in the photos in the manual...

It's a thick-film module in the LK280, but discrete-component in the LK2/LK275. I'll trace that out and post the circuit later. I think it shuts the amp down if high current is detected on either supply rail and if a high DC offset is detected on the output.

I'd have just used fuses and a fairly simple circuit operating a relay, but what do I know? I'm self-taught and have no qualifications higher than a couple of rather average A-level grades. How I got R&D jobs in pro-audio I'll never know...

(For all that circuit complexity, it does go 'plop' when switched on. Maybe I'm being picky, but I prefer amps which do nothing when switched on, then go 'clunk' after a couple of seconds...)

Originally Posted by mhennessy

It's a nice box though. Depending on the size of the transformer, I'd be inclined to replace the works with a couple of LM3886 or similar. Obviously, I'm joking [...]

I think the transformer's something like 650VA and has two secondaries, one for each amp. The heatsinks look pretty good too...

Kat

[GrimJosef]

Originally Posted by mhennessy

I'm still the right side of 40 ...

Er ... which side is that ?

Cheers,

GJ

[Patrick Dixon]

Well I'm 52 and I can still hear 18KHz .... provided it's loud enough of course!

[Kat Manton]

Attached, a first attempt at a diagram, PSU included.

If there's anything which looks strange or just plain wrong, let me know and I'll check the diagram against reality

Kat

[Kat Manton]

And there's more...

The BOM in the service manual is dated 15/08/2000 and is described as being for the LK2(60)/LK275. Assuming it's probably for the last incarnation of the LK275, I've played 'spot the difference' between the BOM and what I found when I traced the circuit. Also included are some changes deduced from the BOM and the LK280 photos in the manual.

Differences, pre April 1986 LK2, BOM for LK275 (in service manual):

• C7, C8, was 22/10 Tant., changed to 33/10 Tant.
• C9, was connected to 0V via R29, changed to direct connection
• C16, C17, was 6800/63 Al. Elect., changed to 10000/64 Al. Elect.
• C21, addition, 330nF, 250V, Evox/Arcolectric (above board, in series with R29 across output)
• C22, C23, addition, 100/63 Al. Elect. (above board, adjacent to Q20, Q24)
• D12, D13, was N.F., changed to MR752 (above board, directly across R25, R26)
• Q20, Q27, was BD250C, replaced with 2SA1386 (but TIP36C fitted)
• Q24, Q31, was BD249C, replaced with 2SC3519 (but TIP35C fitted)
• R4, was 150R 0.25W 1% MF, replaced with 4k3 0.25W 1% MF (significant difference, increases NFB, error?)
• R11, R50, was 3k3 0.25W 1% MF, replaced with 4k3 0.25W 1% MF (reduce curr. D1, D5)
• R24, was 4R7 0.25W 1% MF, no change listed in BOM (error? should match R27)
• R25, R26, was 0R47 2.5W 10% W21, replaced with 4R7 4W 5% w/w cer.
• R27, was 4R7 0.25W 1% MF, replaced with 12R 0.25W 1% MF (error? should match R24)
• R29, was 10R 0.25W, replaced with 0R47 2.5W 10% W21 (above board, in series with C21 across output)
• R33, was 20k 0.25W 1% MF, replaced with 20k 2.5W 1% MF ('reliability mod.')
• R49A, R49B, was 2 x 1k2 0.25W 1% MF, replaced with 1 x 2k4 2.5W 10% W21 ('reliability mod.')
• R54, was 27R 0.25W 1% MF, not listed in BOM (definite error)
• R101, R102, C101, C102 (my designations) not listed in any form (early mod. not fitted to later units?)

Kat

[dave cox]

Is there something missing from the 'net at the anode of D4 ?

dc

Ahhh, maybe R50, R11 and D1 are what i'm looking for ...

[mhennessy]

Hi Kat,

That's a little different to the diagram I was describing earlier, but much of what I said is still true. I've worked out the DC conditions as best as possible, along with a few other things. In lieu of a scanner, here's what I reckon:

The supply rails are just over 40 volts. I assumed 40 for simplicity subsequently...

No wonder R33 was changed from 0.25 watt. It dissipates more than half a watt, depending on the value of the unregulated supplies...

The current limit is close to 6 amps. An overload on either channel brings down the other, thanks to R39 or R48. The arrangement is fairly neat - the IC works with the power transistor in the normal way (detailed on the 317 datasheet), but Q28 (or Q33) facilitates the shutdown - both are saturated normally thanks to R33. Q29 or Q33 will turn off the supplies, which Q30 or Q34 is able to do...

The input filter turns over at around 40kHz with zero output impedance from the preamp. At 1K, this falls to ~24kHz. It's not an "easy" input impedance, and not all preamps will drive this well - an NE5532 would be excellent here, but some more "trendy" alternatives might not fare as well.

The total LTP current is 10mA; Q5 and Q6 are 5mA current sources. Rather high...

The power dissipated in R50 and R11 is ~0.37 watts!

Q7 can supply up to 33mA.

The VAS output is loaded with R15/R17, to reduce its open loop gain, but it considerably increases VAS distortion. Generally not a Good Thing.

The VAS Miller compensation is unusually high in value - but is two pole. You might expect 100p or lower here. Suggests that they had problems elsewhere. I presume that the high LTP current and some of the complexity around the VAS is to increase the current available to charge/discharge Cdom. And to drive R15/17...

Pleased to see that they have improved the bias setting stage, and made it thermally track the drivers...

Based on what I see everywhere, I would again urge caution before attempting to increase the closed-loop bandwidth. Though playing with C1 is very tempting. When you measured it, what was the Zout of your oscillator? Just in case...

CFP output stages can be problematic, taking off at VHF frequencies. The added components (R101.102 and C101/102) would have been to tame this...

A lot of resistors are fairly marginal in rating - it's no wonder many have discoloured. Whoever designed this forgot that the 0.25W rating is at 25C! Still, it's lasted 20+ years, so I shouldn't be too judgemental. I do wonder how many components will have drifted, and I wonder if some of the transistors will be degraded as a result of the long-term heat?

Cheers,

Mark

[mhennessy]

Originally Posted by dave cox

Ahhh, maybe R50, R11 and D1 are what i'm looking for ...

Yes indeed. If the diagram gets revised, it would be worth making this clearer. Move R50 down the page, make the line from D4 horizontal... I'd be inclined to stretch the whole thing vertically so that the middle sections are clearer - no idea how easy that is as I use a 0.5mm HB for my CAD

I miss the circles from the transistors too

[dave cox]

The input filter R1,R2 and C1 looks to have quite a low turnover point or is my maths all wrong ?

dc