Distortion problems

[Radio Wrangler]

One quirk of that Panasonic circuit is that it imposes resistance on the output of both opamps and that the resistance is within the overall feedback loop.

The circuit might do nice things for distortion reduction, but it's a known recipe for capacitive loads causing high frequency instability. The output resistors working into, say, cable capacitance creates an added pole within the feedback loop. As the resistors are appreciably larger than Zout of bare opamps (emitter follower output stages, usually) the frequency of this trash pole is brought down considerably. The task of designing a loop to cover the wanted frequency range and to be stable at all frequencies is made much more difficult, and will impose limits on the range of reactance component of the load which can be imposed.

I wonder if the input impedance of that Aktiv box isn't just purely resistive (rezistiv? ) and makes the output stage of the preamp kick off?

I wouldn't be running into expensive speakers while trying things out.

The Aktiv splitter's low input impedance isn't in the range most preamps are designed around, but something with a headphone output having a volume control might be a suitable driver to try.

David

[Diabolical Artificer]

I'd lay odd's on it's the Technic's amp, though could be wrong. These SU amps use those big amp bricks
rather than all discrete components and as Mark says are a weird design. Try reproducing the fault with only one source, then sub for a different amp. From your description it sounds like it distorts when something is overloading, try feeding it with a toneburst or something similar.

Could be completely wrong but.

Andy.

[Linnovice]

Thank you all so much for your suggestions and for taking the time to research them.
I confess most of the electronic ‘stuff’ has gone right over my head but some has stuck.
It may well be a problem with my hearing but up until four weeks ago when I had the M50 it was fine. It may be a coincidence that my hearing has dipped at the same time as the change. I will arrange for a check up.
I cannot just add a couple of power amps into the system. The way the Linn the system works is by removing the crossovers completely in the speakers themselves. These are replaced by dedicated mono cards designed to cover the three ranges. Bass, mid and treble. These cards are inserted in pairs actually into the circuitry of three individual stereo power amps (in my case LK100’s). This has the effect of converting them individually into three active power amps, one for each pair of driver units in the speakers.
I like the idea of connecting direct to the power amps and so bypassing the Technics. I have a couple of Revox A77’s that have variable output. I can try that.
I can also remove all the sources (including five R2R decks) and try it with just a CD player.
The overall impression I have is that the Technics is much clearer/brighter/incisive than the Tascam. I sat here last evening listening to a wide range of music, from Horowitz to Pink Floyd and most of it was superb.
I will report back after experimenting as recommended.
Thanks again.

[Hartley118]

Both Technics and Tascam are very reputable audio brands. If both are working properly and set to a flat response with no equalisation, they should sound exactly the same. Any audible difference between them suggests some kind of interfacing problem.

Martin

[mhennessy]

Originally Posted by Linnovice

Thank you all so much for your suggestions and for taking the time to research them.
I confess most of the electronic ‘stuff’ has gone right over my head but some has stuck.

Hi Linnovice.

I apologise for getting too deep with this - it often happens

Can I suggest that before changing anything, you try these two simple things:

1. Experiment with the "Direct" switch on the Technics. Is the distortion present at both settings of that switch?

2. Temporarily unplug whatever is plugged into the tape outputs of the Technics. Does this change the distortion?

Come back to us with the results of those tests, and we'll take it from there.

All the best,

Mark

[Linnovice]

Hi Mark,

Right.
With regards to the Direct button. Makes no difference whether it’s in or out as far as the noise is concerned. Obviously there is a change in the reproduction sound quality. When Direct is engaged it’s sharper than when it’s not. Presumably that’s because the signal is being sent around an eq stage.

With ALL inputs removed apart from the CD noise is still there.

Feeding the power amps direct from a tape deck (Teac A3440). Sound good with no noise. There’s a bonus with that one as it proves the ears are ok.

[SteveCG]

For what it is worth, I've spent some time chasing possible sources of intermodulation distortion in my hi-fi set-up. By far the most demanding musical source is choral music. However your challenge is to find a suitably pure source to aid debugging. My suggestion would be a live Radio 3 concert as distributed over its digital radio channel on the Freeview TV multiplex. One can record this on a PVR and replay the same section time and time again. Even this is not error free as the BBC uses "Optimod" sound level processing and you may doubt the PVR's hi-fi playback credentials. Other sources such as LP (Vinyl) and CD will almost certainly have been 'processed' (to the hilt sometimes!).

If this is not considered good enough as a test regime then do you have two good pure sine-wave oscillators so that you can set their frequencies a couple of kc/s apart in the 4 to 5 kcs range, add the signals together and listen out for curiosities?

I would suggest that repeatability of effect is to be sought after otherwise you can be chasing shadows.

[mhennessy]

Originally Posted by Linnovice

With regards to the Direct button. Makes no difference whether it’s in or out as far as the noise is concerned. Obviously there is a change in the reproduction sound quality. When Direct is engaged it’s sharper than when it’s not. Presumably that’s because the signal is being sent around an eq stage.

OK, so that probably rules out a problem with the pre-amp being able to drive the following Linn crossovers/amplifiers, though it would be nice to verify it by connecting a "vanilla" amplifier and speakers (or headphones) in place of your Linn active setup. You could even feed the output of the Technics preamp into a cassette or MiniDisc recorder that has a headphone socket.

The key thing is to listen to the output of the Technics when it's not being asked to drive what might be a rather peculiar input impedance. We don't have the circuit diagrams of the Linn modules, so we're basing this hunch on a potential breadcrumb in a spec sheet.

Originally Posted by Linnovice

With ALL inputs removed apart from the CD noise is still there.

At the risk of sounding pedantic, I meant that you needed to disconnect whatever was connected to the tape outputs of the Technics.

The fact that disconnecting the other source devices makes no difference is not a huge surprise.

But something that is picking up a signal from the Technics to make a recording might be loading the signal.

The odds of this are low, but still, it needs to be eliminated.

Originally Posted by Linnovice

Feeding the power amps direct from a tape deck (Teac A3440). Sound good with no noise. There’s a bonus with that one as it proves the ears are ok.

I've taken a quick look at the line output stage of the A3440, and don't see any particular reason why it should be any better than the Technics. It uses a less good op-amp and has a higher output impedance.

I note that you are saying "noise" rather than distortion. Can we just be clear that you are talking about a distortion that happens with high level, higher frequency signals, as described in your first post?

[mhennessy]

Originally Posted by Radio Wrangler

One quirk of that Panasonic circuit is that it imposes resistance on the output of both opamps and that the resistance is within the overall feedback loop.

That's quite common. It's a simple way to protect the op-amp from short-circuit (even though the internal protection should do that job). It's also commonly used in conjunction with a load resistor to act as a voltage limiter (see Quad 44, for example).

It's worth noting that the A3440 was OK, yet that does the same trick.

Some designs include a small cap between output and -ve input, and this improves stability into capacitive loads while maintaining a very low output impedance - potentially 0 ohms at low frequencies if you like. But all the designs looked at here include series resistance after the -ve feedback takeoff point, so should be inherently stable into capacitive loads.

Often the EMC caps at inputs and outputs are larger than the cable capacitance...

[mhennessy]

Originally Posted by SteveCG

For what it is worth, I've spent some time chasing possible sources of intermodulation distortion in my hi-fi set-up. By far the most demanding musical source is choral music. However your challenge is to find a suitably pure source to aid debugging. My suggestion would be a live Radio 3 concert as distributed over its digital radio channel on the Freeview TV multiplex. One can record this on a PVR and replay the same section time and time again. Even this is not error free as the BBC uses "Optimod" sound level processing and you may doubt the PVR's hi-fi playback credentials. Other sources such as LP (Vinyl) and CD will almost certainly have been 'processed' (to the hilt sometimes!).

If this is not considered good enough as a test regime then do you have two good pure sine-wave oscillators so that you can set their frequencies a couple of kc/s apart in the 4 to 5 kcs range, add the signals together and listen out for curiosities?

I would suggest that repeatability of effect is to be sought after otherwise you can be chasing shadows.

I definitely agree that it would be nice to find a more objective and repeatable test - such as a sine wave test. Trouble is, that might be extremely tough on the speakers - but originally, the OP did say that it happens at all volumes, so if that's the case, then the volume control can be kept nice and low while testing.

The easiest way to do this is to use Audacity to make a test CD containing sine waves recorded at various levels up to 0dBFS, at a range of different frequencies. I'd start with 1kHz, 3kHz and 5kHz. Of course, an oscillator could be connected in place of the CD player, but as the problem occurs when playing CDs, it would be best to not change too much at once.

When it comes to audio, I've never been sold on intermodulation distortion measurements - it's complicated and doesn't tell you anything more than a regular THD+N measurement would - they are just different ways of observing non-linearity. Amongst others, Douglas Self taught me that, and so it would take quite a lot of work to convince me otherwise.

When it comes to Radio 3, the Optimods used for the digital services are set up to provide protection limiting only. The FM one only very gently compresses (despite what some people believe about it being changed on a timeswitch basis for drivetime, etc - I have spent weeks of my life debunking that, but the belief persists). Personally, I would be far more concerned about the effect of MP2 data-rate reduction than the Optimod here. On DAB, it's at least working in discrete stereo - it's joint stereo for TV platforms, and that distracts me far more than the other MP2 artefacts. The best way to get Radio 3 digitally is via the internet (320kb/s AAC), though that's obviously not as convenient for the OP, probably...


Linnovice, if you're following this, most of what I've said in this post can probably be ignored for now, but do think about the possibility of making a test CD. Or perhaps you already have one? No good CD collection is complete without a couple

[Linnovice]

Apologies for confusing things. It’s a distortion, not just a noise.

So that I am using a consistent source. I’ve now connected a Sony CD player with variable output directly to the power amps. The distortion is there, predomently on the right channel. This is on a Tallis Scholars CD of close harmony choral music.

On that basis it would appear it’s not the Technics playing up?

[Linnovice]

With just bass amp on the sound is clean.

With bass and treble there is a slight distortion.

With just the midrange on there is distinct distortion.

Same with both speakers (I’ve swapped them over).

[PJL]

I still favour this may not be distortion but a difference from your previous Tascam setup. Based on what you just posted, it is possible it was the Tascam that was poorly matched?

Do you use hearing aids? Hearing aids do distort the sound through multi-channel compression.

[Linnovice]

Hi PJL, yes I do wear hearing aids but the distortion is there when haven’t got them in. I’m not profoundly deaf. Just lost a bit of the top end.

[Hartley118]

Originally Posted by mhennessy

When it comes to audio, I've never been sold on intermodulation distortion measurements - it's complicated and doesn't tell you anything more than a regular THD+N measurement would - they are just different ways of observing non-linearity. Amongst others, Douglas Self taught me that, and so it would take quite a lot of work to convince me otherwise.

I agree that measurement of intermodulation and harmonic distortion are different but equivalent ways of measuring non-linearity in a wide-band system. However, if the system has restricted bandwidth, then the harmonics generated by non-linearity at high frequencies may well fall outside that bandwidth. For example, it can be difficult to measure high-frequency harmonics in an analogue tape recorder: all may appear to be well until it's presented with the high frequency energy of something like a jangled bunch of keys, when all sorts of unwanted low frequency intermodulation components typically cause very dirty reproduction.

AFAIK, intermodulation distortion measurements were first applied in the optical sound film industry in the 1930s where the restricted system bandwidth made meaningful harmonic distortion measurement very difficult, particularly at high frequencies. Engineers nevertheless knew that they were dealing with a pretty non-linear recording system and devised ways of measuring those intermod components.

It's arguable that, whilst we normally measure non-linearity by harmonic distortion, the objectionable sound of distortion that we actually perceive is actually due to the intermodulation components because they bear no harmonic relation to the original sound. The ear can cope with moderate low-order harmonics because they only change the tone character of a note. However, intermodulation components are immediately objectionable because something entirely different has entered the sound stage and it's not masked by the original sound.

Come to think of it, Linnovoice might find a carefully made digital recording of a jangled bunch of keys a useful signal source for diagnosing the source of the distortion problem.

Martin

[mhennessy]

This is progress . Or is it?

It's interesting that there is a difference between channels.

As you've swapped the speakers, it points to the Linn amps. But it wasn't there with the Teac A3440?

I think at this point, I'd be reaching for the test CD


PS: Could you tell me what model the Sony CD player is? I'd like to take a quick look at the service manual, just to see what the variable output stage is like.

[Linnovice]

Hi Mark,

The Sony CD player is the CDP X303ES model. Quite a nice piece of kit.

I’ve been looking up the specs of a Linn Kairn pre amp. They quote 100ohms output impedance which I believe is substantially different to my Technics. Think I am going to bite the bullet and get a Kairn. That’s the only way I can prove it.

[mhennessy]

Hi Linnovice,

Output impedance might still be a red herring here.

First off, the Technics is 180 ohms, which is not all that far from 100 ohms.

Secondly, the Teac A3440 has a much higher output impedance - 1k IIRC. Yet that was OK.

Finally, looking at the manual for the Sony CD player, the output impedance of the variable output depends on where you put the volume control - it comes straight from a 10k pot via a 100 ohm resistor. Therefore, the output impedance could be anything up to ~2.6k. That high value appears at half output voltage (which won't be half-way mechanically, because it's a log pot). At the maximum setting, it's 200 ohms.

But the Sony exhibited the problem. I wonder if it came and went with the setting of the volume control?

Anyway, if we were considering just output impedance in isolation, then all of the above is illogical. I think there must be more to it.

Here's a thought: do you have an adaptor lead that will get you from the headphone socket of the Sony into your Linn amps? If so, you could take advantage of the headphone amplifier, which has an output impedance of 56 ohms. Note that the headphone amp has around 10dB of gain, so you'll need to keep the volume a bit lower than before. If you still get the problem under these conditions, then that debunks output impedance as the cause.

Finally, it's worth saying this: if you have convinced yourself that the output impedance really is the problem, then you could change the output impedance of the Technics pre for the price of 2 resistors. I see no problem making it 100 ohms. Certainly a lot cheaper than a Linn preamp, which I sincerely believe isn't necessary

Hope this helps,

Mark



PS: Could I have some details of the Linn crossover cards - I'd like to look them up, but I don't know the Linn range very well. I realise that I probably won't get a schematic (which is why I don't use Linn gear), but I'd like to find out what I can nonetheless.

[Radio Wrangler]

Originally Posted by mhennessy

PS: Could I have some details of the Linn crossover cards - I'd like to look them up, but I don't know the Linn range very well. I realise that I probably won't get a schematic (which is why I don't use Linn gear), but I'd like to find out what I can nonetheless.

That's also the reason why I have none of their equipment.

You can bet your bottom dollar that their competitors have traced them out for all their products, so they're not protecting some secret rocket fuel formula, but it does inconvenience the owners of their stuff.

There is always the possibility that something wierd was done in their preamp/poweramp setup maybe for marketing reasons, maybe to appeal to the cultists.

Remember that the designer of the Naim amps did not believe in Zobel isolators for RF stability on the outputs of his amps and lo the whole business of weird speaker cables burning out amplifiers came to pass.

If the setup with out the preamp still sounds bad, then the next thing is to see if you can isolate it to amps versus speakers and amp versus amp.

If you have a decent full-range speaker with crossover inside, you can still connect it in place of any of the sections of your Linn speakers and have a listen. You'll get bass only, midrange only or tweeter only sound, but you should be able to tell if its clean or not.

Then you can interchange the power amps to see if anything moves with one of them.

David

[Ted Kendall]

Just a thought here - I had trouble once with a Revox A77 connected to a Radford preamplifier. When powered all was fine, but when the A77 was switched off it loaded the unbuffered tape output and hence the main signal path. Took me a while to find...

If there are no buffers in the preamp, this might have something to do with it.