Reverse Engineering SSM2045?

[ryansupak]

Hi, I've been doing some research on SSM2045, which of course is hard to find. (Apparently they were extremely inconsistent even when they were manufactured -- as all SSM chips were.)

SSM2044, however, is plentiful, inexpensive, and still manufactured today

I making an assumption -- and I don't think too big of one -- that the 4-pole filter in the SSM2045 and the one in the SSM2044 are extremely similar in character. They have the same voltage ranges, same slopes, and same S/N ratio. Looking at the datasheets and reading the stats, I wouldn't be surprised if they're the exact same filter. (Based on first-hand anecdotes I have read about the inconsistency of SSM chips, I think it's devisable that the manufacturing variance in one could make it sound like the other, at any rate.)

The problem, of course, is that they have different feature sets and pinouts. However, by reading the EII schematic, I noticed that the EII is *only* using the 4-pole filter component of SSM2045, and *not* the VCA component or the 2-pole filter.

The inputs used for the SSM2045 implementation in the EII are as follows:
* 15V+ In
* Ground
* Signal In
* Frequency Control In
* Q Control In
* Pole 1 Control In
* Pole 2 Control In
* Pole 3 Control In
* Pole 4 Control In
* 4-Pole Filtered Signal Out

The SSM2044 has all of these inputs though it requires its inputs in a "balanced" configuration which is not a big problem. I suspect an SSM2044 on a breadboard with a few inverter opamps could replace an SSM2045 quite nicely.

Thoughts?
rs

[ryansupak]

Looking more at the schematic, a TL084 op-amp is feeding Signal In. (There is a 68K resistor in series with the TL084 positive output.) The TL084 negative output, which is apparently not needed in the case of an SSM2045, is still available for our use. This eliminates the need for an additional inverting op-amp.

The capacitors are wired a bit differently, but that would be quite easy to tweak.

I'd start with the assumption that the control values for Frequency, Q, and for the 4-Poles are identical between the SSM2045 and SSM2044, and adjust values to taste.

So, based on what I can tell, you could "clone" an SSM2045 with:

* a small breadboard
* a breakout for the SSM2045 IC socket
* an SSM2044 chip ($10 shipped)
* a 68k 5% resistor
* a very small amount of wire

Considering that I was quoted $95 USD shipped for a NOS SSM2045, this price ($25-$30 on the high side) sounds pretty good, and it also provides a "future proof" substitute for the most scarce part in the EII, now that the media problem has been solved by the HxC.

rs

[ryansupak]

As for the other "unobtainium" part of the EII output channels: It looks like S3528 is a digitally-controlled, 7th order (extremely steep) analog anti-aliasing filter. The more that a given voice item is transposed, the more that the cutoff point shifts up or down from the maximum sampling frequency. It's fair to say this filter is more of an engineering thing and not part of the "sound" or "warmth" of the unit. (It's probably there to prevent the "Sparkling Dust" effect that you get when you pitch down on an SP-1200). The S3528 can select one of 64 cutoff frequencies. I bet it uses 23 (center, 11 semitones down, 11 semitones up).

I bet that a circuit to replace this could be built for under $30 as well. It could be based on this (modern) LPF and some kind of simple microcontroller circuit to translate the digital word to clock: http://www.maxim-ic.com/datasheet/index.mvp/id/1443. The only other problem to solve would be to figure out the center frequency. I bet it's just above the sampling rate of the EII (27777 kHz). Once you know the center frequency, it's not a problem to calculate how much that frequency shifts per semitone. Then just program the microcontroller with a lookup table that translates digital words (input by EII main CPU) to clock frequency (which is what's used to control the antialiasing filter) and you're done.

The S3528 isn't one that I'll even attempt until I need it. The prospect of paying $80 for an obsolete S3528 antialiasing IC just doesn't sit well with me, is all.

But, the point is: I'm thinking the whole filter stack of an output channel could be built of common, modern components for $50-$60 as opposed to about $175 for two ancient, could-die-at-any-moment ICs. (The only ancient pieces of circuitry I'll pay a lot for are tubes, transformers, and tape-heads)

rs

[Futureman]

I like where you are going, but the issue for now will be that the filter replacement will have to sound very close to the original, else you'll have to upgrade all 8 voices, which will no longer be so cheap... But if they are 'close enough' then awesome.

FWIW, I don't think the E2's filter sounds like a 2044.. Maybe it's the way it's implemented, but it doesnt sound as good as my PolySix's filter.
I'd love the E2 to have a better filter.. Something that overdrives and has dirt.

[ryansupak]

From what I've read (which I know is not necessarily authoritative), most of the SSM filter "sound" comes from everything around the chip, not the chip itself.

I've still got some research to do but from what I can tell so far, the 2044 and 2045 4-Pole filter use exactly the same control ranges for everything - including the poles. Right now my working theory is that the 2045 is the exact same thing with the addition of (unused by EII) VCA sections that were designed for mixing boards, according to the datasheet.

I doubt that the 2045 has magic fairy dust in it, that is to say ... But of course, the final stage would be listening tests to get everything just right.

rs

[ryansupak]

I found the S3528 for $20 each at PCComponents.com. They have a few hundred in stock so reverse-engineering that part doesn't make sense for now. (I actually found S3528B but the pinout is identical.)

Coming up with a readily-available substitute for SSM2045 still makes sense though. I'll give that a shot in the medium-term (if/when the need arises for me).

rs

[ryansupak]

Here's a modular synth with an SSM2044-based filter:
<a href="http://www.youtube.com/v/CWfQRRoFmDU&rel=1" target="_blank">http://www.youtube.com/v/CWfQRRoFmDU&rel=1</a>

Here's a vintage synth with an SSM2045-based filter:
<a href="http://www.youtube.com/v/mb1lAvYgy94&rel=1" target="_blank">http://www.youtube.com/v/mb1lAvYgy94&rel=1</a>

Definitely an extremely similar sound.
rs

[ryansupak]

Don't let the existence of this thread stop you from doing what you want!
rs