I've been recording the audio output from my PSR3000. Whilst I've seen (and reported here) waveform clipping in the past due to excessive mix level in midi files I've now got a much stranger problem and I wonder if anyone else has seen this.

Basically, if you look at the audio waveform it is clearly not symmetrical. The +ve side of the waveform looks fine but the negative side looks compressed and low-pass filtered!

I get this on midi file, styles or live playing. I've used two different power supplies. I've changed the EQ. I've reset the keyboard. No change.

I'm capturing from the Phono sockets, I haven't bother capturing from the jack sockets are these are downstream from the phono so I'd expect the problem to be seen there as well.

It still sounds OK which may just confirm that my hearing is getting worse....

My only thought at present is take the lid off and waggle everything! Are there any
chips in sockets that might have crept a little?

To summarise: HELP!!!

Is there a way we can see what the you a talking about?

How can a level in a MIDI file be excessive?

I'd record just at least a snippet from the main outs, just to make sure...

Quite a lot of real sounds from real instruments exhibit non-symmetrical waveforms, only synths are perfect!

But I would get someone else with a 3k to confirm it for you. You MIGHT have had a component failure on one of the op-amps. I'd also double check your recording signal path. Have you eliminated other sources from the same problem? Your A/D card might be damaged, your mixer's op amps might be faulty, there's a whole signal path to troubleshoot (sorry if you already have)...

Try a recording of just ONE sound, say a synth pure tone recording (sawtooth or square wave) and see if it is those too, or just the result of non symmetrical acoustic waves...

Hpe this helps.

I seem to remember distortion problems with the PSR3000..the culprit was a damaged headphone circuit board..apparently all the sound is routed through this board...

Mac,

Post a small sample of your recording so we can try to analyze the problem.

Gary

You can also record from the headphone out, to see if it happens there too.
DonM

Thanks for that. I don't have any way to post images so Im about to send some of you a small email with a screen shot attached. I was lucky enough to find a recording from December last year of the same midi file so the screenshot shows current waveform to the left and last years waveform to the right.

I agree that real instruments are non-symmetrical but I think I'm well outside that amount of variation!

I'll have a chance to investigate further on Sunday (UK).

Re "How can a level in a MIDI file be excessive" you can get signal clipping caused by the way the channels of a file are mixed digitally within a keyboard.

The digital mixing takes the instantaneous value of each voice waveform, scales it by the mix level for the track, adds up all the resulting values and arrives at a total instantaneous value for the waveform. If this value is larger than the maximum possible value for the summed waveform then the value will effectively get clipped at that maximum.

This is not a fault as such but a consequence of the way the maths is done, in some ways it's the computing equivalent of gain structures.

As a result of this, on my 3k, I rarely set individual track levels above 70, especially drum or bass tracks as these are the ones that are most likely to overload the system.

[This message has been edited by MacAllcock (edited 01-21-2010).]

[This message has been edited by MacAllcock (edited 01-22-2010).]

[This message has been edited by MacAllcock (edited 01-22-2010).]

Hello John,

Got your email with your waveform images.
Definite offset of some kind.
DC offset maybe?
Still you've said it sounds ok.
Normally I work on the premise of; if it sounds ok, then it is ok.

I've never had issues with MIDI files and levels, but I don't use them that much, I've only a few bit's of kit that handle GM.
Have a Yammy XG1000 PCI card lying around somewhere, not currently installed in a PC at the moment.

I wondered about DC offset, but to me the difference in the +ve and -ve cycles looks too pronounced.

I may just have to live with "it sounds OK" until I've saved up for a 910!

I've just noticed a new recent firmware release for the 3K http://music.yamaha.com/products/downloads.html;jsessionid=05DEBA11BCA961B65CC8C75AE41697B8?productId=106898

I don't really see what difference it would make; I don't need the new functionality but is this worth a go just to see?

Sorry, Mac. I replied to your email before I saw this post. If that is the waveform display of the 3k before and afterwards, then of course, yes, you have a problem...

But, for it to pinpoint the 3k as the culprit, you need to make a recording of something else through your soundcard, to eliminate that as the cause.

"you need to make a recording of something else through your soundcard, to eliminate that as the cause."

Thats a very good point. I'll let you know what happens!

It strikes me that, if all the 3k's signal path goes through the headphone circuitry, you might need to be careful to ground yourself before plugging in a set of headphones that might already be on your head or in your hands, and possibly a small static charge built up. Touch the metallic barrel of a jack plug, or the metal on your mixer, or something like that, otherwise it is possible (if however unlikely) that the discharge get into the headphone circuitry.

I know I'm getting ahead of myself, here, just blueskying possible causes...

Do you recall any particular incidents (perhaps the power going out on you while the 3k was switched on) between now and when it WAS recording alright? Strictly speaking, with an external power supply, it should be much harder for power fluctuations to damage your unit, but it IS another possibility...

It's not the sound card or the cabling. Oh dear!

Well, bottom line is, if you can't HEAR any difference...

You say you have the identical file recorded before and after the malfunction? If you set up their levels carefully to be as close as possible (maybe your software has an RMS output meter?), you ought to be able to A/B the two in a good listening situation (maybe good headphones) and see if there IS anything to get worked up about.

Work around available although problem still there!

I've recorded the same midi file using the output from the main jacks, and from the headphones socket, and both look fine.

The output from the phono sockets is still strange (I thought I'd better double check).
I can hear a difference between good and bad, the bad one isn't as crisp and seems to lack a shade of punch. It's not a huge difference though, there's a case to be made that I'm persuading myself there's an audible difference because I know there's a waveform difference there in the first place!

Anyhow, I'm still going to take the lid off, disturb and reseat all connectors, and maybe even put a bit of contact cleaner into the main volume control (it's getting a bit crackly) but that's as far as it goes. If this doesn't resolve the issue with the phono output I'll record off the main jacks.

Thank you all for your assistance, it's made sorting this out much less of a shot it the dark!

John

So, have been inside, had a look at circuit board, nothing obvious, removed dust, squirted contact cleaner into volume pot, re-assembled.

Now have zero crackles volume control, phono output still strange. Which is what I expected to be honest but you have to try these things!

Internally it's pretty neat. All cable runs are taped down or clipped into specific slots in the case moulding. There are ferrite cores on a small number of links, and these are individually supported so they won't move about under their own weight during transit. The major audio-related electronics are inside a Faraday cage and there's also a shield under the main input/output board (which I removed to examine the PCB by the phono's).

Mentioning a Faraday cage reminded me of this video: http://www.topgear.com/uk/videos/car-lightning

Are you sure that the earlier recording you used for comparison was from the phono (RCA) outs? Maybe you had used the headphone or main outs before, and the RCA's were always a bit naff? This one is really a headscratcher. I can't think of why the RCA out would be the only one with the problem. I wouldn't have thought any additional circuitry should be involved. It's just another line out.

And why specifically the negative part of the waveform is distorted, but NOT the positive..? Strange. I'd still like to see another 3k do the same test.

You never know...

I'm stumped to get any logical answer to this as well. I've never used anything other that the phono's for recording so it's definately something that's turned up recently.

Looking at the PCB there seem to be a couple of surface mount transistors and capacitors in the circuit to the phono's but I can't see how a single-point component fault would affect both channels, and even then I don't see what sort of fault would give such an odd waveform. Hairline crack in PCB earth plane somewhere? I looked hard but couldn't see anything remotely suspicious. Oh well.

Well, look on the bright side... two out of three good outputs is still a majority!

BTW, you ever done any quiescent noise recordings from the main and headphone outs, see if maybe the headphone amplification circuitry is introducing a bit more noise than the main outs? I wouldn't be surprised if it did...

Bipolar voltage running the opamps.

The negative side isn't up to snuff.

Suspect a bad opamp here, since the other outputs are still healthy.

This is a job for a competent electronics service technician, as there are other things that could cause this, associated components to the output opamp for that output such as resistors, possibly capacitor, but the first thing to suspect is the output opamp, likely has a bad internal junction on the negative-going side.


--Mac

Would it have a separate op amp to the main amp, Mac? That is what is so puzzling. No bipolarity disorder on the main outs

Shouldn't the RCA outs simply mirror the mains?

I'd figured the mains would mirror the phono's as well!

I couldn't see an op-amp anywhere near the phono output, although there are some on the output board. Dodgy op-amp is the idea that makes most sense in all this.

Visibly the signal recorded from the mains doesnt look any noiser that that recorded from the phono, I'll check that toniqht,

Not necessarily.

If the designer did the breakout *before* the opamps that are used to drive each different output, a likely scenario, then there could be just one opamp driving the one set of outputs that is bad.

Schematic would be the place to find out, if one is available.


--Mac