advertisement


PCM56 MSB adjustment

Mike P

Trade: Pickwell Audio
I'm afraid I've covered this topic in the past but I'm going to drag it up again....

I've just repaired an old Denon DCD-3300, which a lovely TOTL vintage player. It uses the Burr Brown PCM56 (one per channel) and has the facility to adjust the MSB error via a trimmer.

I understand the basic principle would be to play a test tone sine wave from a CD whilst observing the output on a distortion analyser or spectrum analyser and and then adjust the trimmer for lowest distortion (distortion meter) or lowest harmonic 'spikes', especially 3rd and 5th order harmonics (S.A.).

From reading around it seems most people use a 1KHz frequency. I don't know if there's any particular significance to this, I suppose it's in the critical audio spectrum and happens to be a convenient round number.

Again, reading around it seems most people use a very low level signal for this, typically -60dB. However, the service manuals for the Sony CDP-X7esd and CDP-557esd, which have similar adjustment facilities states 1KHz @ 0dB. The Sony service manual calls for a distortion analyser and not a spectrum analyser. I attempted the Sony procedure a few years ago and bought an old H.P. distortion analyser from Les at Avondale especially for the job but in the end it became apparent that the distortion analyser I'd bought wasn't up to the job and I abandoned the project.

I now have an old analogue CRT Iwatsu spectrum analyser and the Denon DCD-3300 has rekindled my interest in having a go at this again.

Unfortunately I no longer have any facility to burn a CDR with an appropriate test tone and so the first thing I need to do is find someone who'd be willing to do this for me. I think what I need is a long track of (say 10 mins at least) of a -60dB 1Khz sine wave test tone.

I'm unsure about whether or not this sine wave test tone needs to be dithered or not. Any advice on this would be greatly appreciated.
 
Mike

When i did such a thing on a friends pcm56-based Denon at least a decade ago I used a very low level tone, deliberately so, below -60dB (undithered) and set it up using a simple spectrum analysis software for lowest /smoothest roll-off of the obvious harmonics (only 48khz soundard iirc) My thought was - that low dig signal level boosts the obvious %thd / 'grass' visible on the plot, and dithering wasn't needed - it'd only add noise to the analogue output. Seemed to work.

I'd forward to reading your view after doing it by 'the book' :)

I still have / can write all sorts of arbitrary generated test-tones to CD - drop me a pm with a wishlist and postal address : )
 
Mike

When i did such a thing on a friends pcm56-based Denon a decade ago I used a very low level tone, deliberately so, around -90dB (undithered) and set it up using a simple spectrum analysis software for lowest /smoothest roll-off of the obvious harmonics. Seemed to work. I'd forward to reading your view after doing it by 'the book' :)

I still have / can write all sorts of arbitrary generated test-tones to CD - drop me a pm with a wishlist and postal address : )

Thanks Martin. I'll send you a PM.
 
Tweaking for best THD at full scale is just for the marketing headline figure. We cannot hear much less than 0.1% distortion. The DAC behaviour at 40 dB below full scale matters far more, real music does not go much quieter.
 
Has anyone heard anything from Martin recently? I sent him PM but he hasn't replied, which is unlike him. I hope he's OK.

Given that it's August I guess (hope) he's just on holiday.
 
Success! I've finally succeeded at doing the MSB adjustment on my recently repaired and restored Denon DCD-3300.

Martin very generously burned a CDR of test tones for me to use. He did four tracks for me as follows:

1) 1KHz @ 0dB
2) 1KHz @ -30dB
3) 1KHz @ -60dB
4) 1KHz @ -90dB

I used the -90dB test tone and left that track running on repeat whilst carrying out the adjustment.

The DCD-3300 has a headphone output and I found that using earbuds I could listen very effectively to a single channel and adjust the MSB trimmer for lowest noise. It's very easy to hear the noise fall/rise as you rotate the trimmer and quite easy to find the sweet spot of lowest noise.

And here's the proof as recorded on my vintage Iwatsu spectrum analyser.

Left channel deliberately misadjusted:

L CH before adjust by Michael Pickwell, on Flickr

Left channel after adjustment:
L CH after adjust by Michael Pickwell, on Flickr

And here's a pic of the internals of the wonderful DCD-3300.
DCD-3300 by Michael Pickwell, on Flickr
 
Last edited:
I've been playing with this again and have some interesting findings to share.

First, a bit of background...

In order to repair my DCD-3300 (above) I stole a DAC chips from my DCD-1700 (also an excellent machine). I've been wanting to get the DCD-1700 up and running again and so I started hunting around for a source of genuine PCM-56 chips, preferably the selected K grade.

After a fair bit of fruitless searching I decided to take a gamble on buying a selection of PCM-56 chips from China via eBay. The first batch of 10 chips to arrive arrive looked nothing like the genuine chips installed in my Denon players. They were also obviously sanded and re-marked and the shape of the 'body' of the IC and the shape of the pins was different. I sent a message to the eBay seller and they immediately issued a refund, which I guess is a strong indication that they know they are selling counterfeit parts. I haven't risked trying any of the chips from this batch in a CDP.

The second batch of chips I received were much more promising. They'd again been re-marked and look to me to be 'pulls' from old boards but at least they did look to be PCM-56 chips. I decided to risk trying one in the DCD-1700 and it worked, so they were indeed real PCM-56 chips but what grade? I strongly suspect that the answer is that they're any old grade of PCM-56 pulled from scrap boards and then re-marked as K grade chips to be sold at a premium price.

In this pic you can see the markings rub off with acetone to reveal the sanded surface underneath (top). One of the re-marked supposedly K grade chips from China (middle) and a genuine K grade chip (bottom). It's pretty clear that the re-marked chip doesn't quite have the right typeface, the colour of the body of the IC is too black and the colour of the text is too bright and silver.

IMG_20210930_073723 by Michael Pickwell, on Flickr

BTW the easiest way to catch most counterfeit IC's is to rub the markings with acetone. The vast majority of genuine IC's will be totally unaffected but the acetone will quickly take off the print and blacking from the counterfeit re-marked parts. For an example see here.

Next I set about trying to assess the performance of each of the five chips from this second batch as follows:

1) I numbered each of the Chinese chips 1-5 and also took out the remaining genuine K grade chip from the other channel on DCD-1700 which I numbered as chip 6.
2) I installed some IC sockets into the DCD-1700.
3) I individually installed the chips into the player and then adjusted the MSB trimmer to the optimum setting as explained in my earlier posts.
4) After adjustment I looked at the spectrum on my spectrum analyser and took a photo of the display so that I could easily visually compare the results.

Here are the results:

Sorry about the crap photos, I hadn't pre-planned to share them public when I was doing this.

Chip 1
Chip 1 by Michael Pickwell, on Flickr

Chip 2 by Michael Pickwell, on Flickr

Chip 3 by Michael Pickwell, on Flickr

Chip 4 by Michael Pickwell, on Flickr

Chip 5 by Michael Pickwell, on Flickr

chip 6 by Michael Pickwell, on Flickr

As you can see there are very obvious differences between the chips.

Chip 1 is kinda average.
Chip 2 was dead/non-functional.
Chip 3 measures spectacularly well.
Chip 4 is awful.
Chip 5 and 6 are similarly good.

I installed chips 5 and 6 into the DCD-1700 as they seem to be a nice 'matched' pair with good performance and I'm delighted with the sound quality.

I've now gone and ordered a bunch more chips from the same eBay seller in the hope of finding some more that measure as well a chip 1.

Interesting isn't it!
 
Would be good to see what a real 'K' spec measures like and if No3 is an example.

Are the BB chips not laser marked? I'm surprised they've not made a home-spun laser marker for this dodgy dealing.
 
Perhaps I wasn't very clear. Number 6 is a genuine K grade out of my CD player.

5 looks to be just as good as the genuine K and 3 is even better.
 
Hi Mike,

Sorry about jacking a two-year old thread, but I have a dead PCM56P-K on my Kenwood DP-1100sg CD player, and I know you’re the CDP reference member here ;)
I pulled the good one from the other channel, and I know for certain it’s faulty, and no audio signal got out anyway apart from hash (distortion noise).

What are the failure modes on this chip? Why do they fail anyway? Will it happen again?

[I’ve ordered supposedly fake or counterfeit ones from good old China on eBay.
Possibly just rebadged standard PCM56P, because the casing really looks like genuine BB, with the vertical — in the circle stamping at left – or have the Chinese refined their art of deception this far?
I’ll obviously do the acetone test as soon as they arrive… not really holding my breath here anyway (pun intended). :p]

Cheers Mike.
 
Hi Chartz,

All the PCM56-K chips I bought from China were re-marked but they did all work.

I my case I was able to test the chips and find the good ones and so the grading wasn't important. Some chips performed much better than others.

There's also the AD1856 which is a direct drop-in replacement (same part from a different manufacturer). I have a small stash of these and they all seem to perform well when tested. The sound is clearer and more sharply focused than the PCM56, which is by comparison a little more warm.
 
I’ve received some Chinese fake PCM56P.
The markings go away with acetone and you can see the case – definitely not BB – was sanded.
It works really well, so I suspect the chip to be a real AD1856.
Why they do this is beyond me though.
I adjusted the MSB with a -40, then -50 and finally -60 dB piano recording, by ear. Surprisingly, the fake IC behaves perfectly at -60 dB after adjustment. Odd, unless the fake PCM56P is a real AD1860… :p
 
Last edited:


advertisement


Back
Top