Tonewise upgrading the SY77 to SY99 specs

[madtheory]

You're welcome. I meant it!

The filter makes total sense to me, it's the biggest part of the "sound" of a converter IMO.

Again, nice work

[Mooger5]

Yeah, here´s what I found during my research: DACs can have a "sound", especially the older R2R, or resistor ladder, Multibit ones. It´s expensive and very difficult to manufacture a set of resistors where each one is exactly the double or half in value to the next one, on-chip. With one resistor per bit (16 bit = 16 resistors). Not only the values aren´t perfectly matched, they can vary according to temperature. This generates non-linearities that the LPF will try to supress, otherwise the IMD will smear the audioband. As technology improves, they´ve become easier and relatively cheaper to make.
The steeper the filters are, the less flat their response becomes, however. But as smooth as a 12dB can be, it won´t be as effective as a 24dB one. The latter seems like the best compromise between filtering out aliases and imparting its own colour to the sound. I´ve ran the SY99 filter on some Java simulator and verified that it does rise the amplitude up a little ( though less than 1dB) from 6KHz to around 20k, with a 3dB point at the corner frequency at 24KHz.

So the better manufactured a DAC chip is, the most linear it will be, meaning less strain for the filter to battle against.
Burr-Brown is said to have launched the last great R2R DACs, the PCM1702, before being acquired by TI. They featured laser-trimmed resistors and were used in high-end domestic audio. I wonder if any synthesizer or pro audio piece ever used one as well. Think that by that time, mid-90s, every company had started to use Bitstream, or 1-bit DACs, more linear and cheaper to make, but also much more noisier.

That time was, I think, a turning point for digital synths. They all sound terrific, but a tad samey...

[Mooger5]

Oh and another thing: 1-bit DACs due to their nature are much more susceptible to jitter than R2Rs. Maybe you could say despite their apparent weaknesses with temperature dpendancy Multibit DACs still sound more exciting, with an organic-like character. Trying to create an internet myth to push the value of these glorious dinosaurs up a BIT heh

[Mooger5]

This might be worth investigating: in addition to the steeper filters and faster opamps, the SY99 also has MSB adjustment for the DAC. Which the 77s (both the SY and the TG) do not: their DAC chip(s) pins are floating, unconnected.
The PCM56 datasheet says it´s optional. You use it if you need the extra resolution.
There are two ways to adjust it using outboard instrumentation but I´ll just trust my own ears whilst twiddling the pot: if it won´t bring an audible improvement then there won´t be any need for it.

[Mooger5]

Hmm nope. It´s only a trimmer for zero DC on the DAC output adjustment in the 99, just like in the TG, but with better adjusting resolution. In the SY77 the trimmer is absent, obviously because it´s near impossible to make adjustments to a board that´s buried inside the synth. The boards inside the TG and the 99 are easily reachable while in operation. But guess what: Tweaking the trimmer in the TG77 doesn´t seem to do any change to the sound. So it´s not really something significant. The trimmer was readjusted for 0VDC afterwards, just in case.

Think there are no more mods to do, so job done

[Mooger5]

This may be relevant information: I purchased a pair of PCM56s advertised as "K" graded, the best and most linear version, but unfortunately they seem to be just regular albeit in newer packages.
The curious thing is the chip that came with the TG77 is also regular grade but sounds fuller and bassier than the new ones. I´d speculate the new chips were manufactured after the takeover by TI.
It may just be a one-off case but even so, based on this rather limited experience I´d still personally recommend, when searching for replacements, to get a "vintage" chip. Hardly a surprise, but that´s the way it is.

Old chip on top. Does anyone have a clue about those codes?

[Mooger5]

Bah this is cork sniffing Forget it. Bye

[pba]

For the PCM56P-K You will see in the right bottom corner of IC housing big K,
without it, it is just PCM56P
They have simply measure the linearity of the DAC and the better ones become J or K on housing.
But it possible to have PCM56P with spec near J or K - because the demand for standar version was the biggest,
there is also PCM56P-L - low cost witch is worst.
Other Burr-Browns dac uses also small painted dot - one for J and two for K - it was withe or golden.

But this is not CD, there will be definnetly some improvement, but You will havemore improvment tweeking other parts of the circuit.
More important is power supply line - please replace the electrolitic capacitors for power supply lines of opamps with sanyo oscon - You can incerase the capacity, and remove the ceramic capacitors from the power supply decoupling line. Yhe same for headphone opamp, which i have not replace.

Replace also the capacitors around PCM56P power supply.

Remember that it is synergy between the power supply line behavior and the music.

Every opamp has his own signature - i have use here mix of OPA2604 and OPA2134

The opamp will burn-in during first 20-50hours and it will sound on the begining too agressive.

I have also change the bipolar electroitycs for Nichicon Muse ES - but the Elna BP is fine, i have measure it and it is still in spec after all this years.

Remember that only one DAC is used for 4 chanels - multiplexed by the YM3028 chip which can be the bigest problem of the signal path (here You can also replace the caps on power supply for Sanyo Oscon).

BTW.Sanyo Oscon is now Panasonic Oscon since few years, it is polymer capacitor with very good spec, and it simply sounds good but it needs some time to burn-in. But it is only for power suply decupling, not for the signal path.

You can also change the I/V converter of the PCM56P - You can use some extetnal opamp i/V converter on some fast opamp at least OPA604, but better will be some OPA627(sounds dry) or AD797 or some AD8xx - sorry i don't remember the numbers now - but the slew rate is important, and silver mica capacitors in the feedback loop does very good job. Other option is to try simple resitor i/V conversion - it is not so linear, but it is witout the (timing)errors of feedback loop , and sometimes it simply sounds better - You can read a lot on diyaudio.com formum about PCM56P I/V conversion. And check the .pdf for PCM56 - Analog Devices replecements also sounds a bit different AD1856.
Using higher bit dac will give You no improvement, and the signal level will be lower - with worse signal to noise ratio.

[Mooger5]

Thank you for the info on the pcm56.

I also got a few PCM61P-K , one reserved for the JD800, which I have tried on the TG77 but the results were like you said, very low output of the music signal, but with loud digital grunge ontop.

Regarding the use of newer faster opamps, I guess you missed a few messages in this topic. I don´t blame you, it´s been a long thread.

Here:

What about using higher spec opamps like the MC33078 or the LME49720, do you think that there will be any obvious improvement ?

Regarding the new opamps, in this specific context that is upgrading the 77 to 99 specs, I personally don´t see a point in using something audio related that was manufactured after the synth you´re trying to improve went out of production. If you want a synth from a certain vintage, it´s because there´s something about the sound that inspires you, and for whatever reason. So use components that were contemporary, I say. Now to mod it with modern parts... then why not get a modern synth in the first place?

So, the goal was to replicate the tone of the SY99, staying loyal to the "vintage integrity", not to update the 77 to modern standards. I think the goal has been met.

I´ve been a member of DIYaudio . com for15 years. It´s a great forum. I don´t post for ages, but I lurk from time to time

[Mooger5]

BTW and a bit off topic: I built a NonOS DAC once, based on the TDA1543 and a CS8412, and wasn´t much impressed. I think the 1543 is a really bad dac chip, even when paralleled for more linearity and so many people used to rave about it just because one was found inside a 47 Labs product.

This viewtopic.php?f=1&t=105282&start=30#p807523 was a true revelation. Listen to the brass section at the 00.20 mark.

So I´ve been toying with the idea of building a voltage controlled reconstruction filter just to see if there´s an advantage in fine tuning the corner frequency to taste, even down to around 18KHz or lower if it sounds "better".

[philip]

I wish u we're living in my area. Btw I like sy77 grainy tone a bit more than the crystal clear 99

[philip]

I wish u could upgrade my new modx to sound as grainy as the sy77 ))) Is it really possible?

[Mooger5]

Haha. That would be so crazy