Space Matrix Revisited

[Sonik Wiz]

As the end of April coming closer I am in sweet anticipation of receiving the new Surround Master v2 soon thereafter. It is bitter sweet because having the SM v2 will certainly become my primary go to decoder for stereo & encoded material. It will reduce almost to zero any urgency to fix my Tate 101A or my Sansui QSD-1000. I expect the SM v2 to excel in performance over these units.

So this has got me on a nostalgic binge of digging out various old decoders & appreciating them all one more time. My Electro Voice EVX-4 (which sounds a lot better than one would expect) and the Heathkit AD-2002 ( the new improved EV decoder that doesn’t sound as good on stereo) and the Fosgate Model 5 that just sounds so bland.

Also in my Closet of Quad is a Fosgate Audionics DSM 3606. DSM of course stands for Digital Space Matrix & also there is a Pro Plus tacked on at the end. If they used smaller type on the front panel maybe they could work in some more modifying superlatives such as Platinum or Mega. You can take a look at an ad for the 3602 here from Disclord’s archive:

DSM 3602

And while you’re looking at that page, who can find the embarrassing typo?

There were a total of 6 DSM models: 3600, 3601, 3602, 3604 3608, 3610. All had different combinations of wired/wireless remote, with/without rear channel power amps, with/without Dolby approved time delay on the back channels. There was also the Gavotte original, Mark II, and the RFQ 5000 that sported the DSM label. These were all intended for use in cars & not otherwise pertinent to this note. There are plenty of other notes on these models just search the forum.
Elsewhere Wagonmaster_91 made a good post about his DSM 3608 decoder & worth a read:

DSM 3608

You can see some good shots of the insides that is very close to mine before I started modifying it. Also in that post he made an astute observation:

The 3608 suffers the "middle child" syndrome in the 3600 series of 360 Space Matrix surround processors.

I’d carry that a step further & say that the entire DSM line suffers from middle child syndrome. It came right after the Fosgate Tate 101A that everybody compared it to & was somewhat disappointed. It came before the Model 3, 4, 5 etc that dropped the DSM label for digital servo logic surround processor. After that the Fosgate Audionics name disappeared, absorbed into the Land of Harman Kardon where the Shadows lie.

At any rate when I saw an affordable DSM 3606 pop up on EBay at least 10 years ago I snapped it up. I mean a collaborative design between Fosgate & Scheiber? Man, how great can that be?

Before I auditioned it I took the top lid off to inspect inside. Yes this was a used product so no worries about warranties. Actually that’s the first thing I do with new equipment as well (here’s looking at you SM v2). I void factory warranties faster than I can void my bladder.
I was disappointed when I found 2 large modules containing the directional control voltage generator and the actual matrix circuitry. I studied the PCB & circuitry a bit, put the cover back on & gave it a listen.

It just was the weirdest sounding decoder I ever heard. Pushing the mode buttons had an audible effect but nothing that really sounded clear, or what you expect from the function names. I tried stereo, SQ, QS. It certainly gave outputs from all speakers, center front was pretty much up front and ambience in the rear but everything else seemed wrong. I pulled it out of my system & went back to tinkering on it.

Eventually I removed the bottom panel & saw those modules had something like 12 pins on two sides soldered in. So what the heck, I pulled out my solder wick & started removing them. Successfully done I was again disappointed to see they were potted in RTV type silicone rubber! I couldn’t turn back now. I did some research and found a Loctite product called Chisel that would remove cured silicone rubber. It doesn’t really dissolve it away but it makes it goopy so you can clean it out with a Q tip or compressed air. Lather, rinse, repeat.

It took a while but studying the Control Voltage Generator module I realized it was almost exact as an example circuit in a Scheiber patent. Right down to individual component values. This was designed to look at front/back individually and left/right as a single varying polarity control signal. The matrix module I don’t remember so much about except there were some regular DIP Op Amps with the pins spread out & soldered so it was pseudo-surface mount. A hand full of transistors, a lot of resistors. I couldn’t match it to any circuitry I’d seen before.

I replaced the modules & re-soldered them. I also made some circuit mods such as improving a few capacitors & upgrading the numerous TL074 op amps to AD 0110. The rear channel output transistors ran hot even with no loads hooked up so I removed them. When I was done it looked like this:

Yes some of those ugly caps you see are my mods but sometimes electricity doesn’t care how pretty it looks. So I hooked it all up again,& again was underwhelmed by the decoding. The sound was crisp & clean, no artifacts but it sure didn’t fill my needs. I pulled it out my system & it found a long term home in my Closet of Quad. In fairness to the unit I did not do any alignment or tune up since I didn’t have instructions to do so. A few weeks ago I said what the heck what harm can I do? Why not try? So I did.

More to follow.

Any further info or corrections are most welcome.

 

[Sonik Wiz]

I started by preparing a test disc with sound positions based on RM/QS. More specifically RM which doesn’t require any 90 deg phase shifters, just proper amplitude/polarity levels. The signals were 1kHz sine wave two minutes long made on Adobe Audition. I also had some left to right, right to left pans made straight and also with out of phase blending. Keep in mind this unit is not marketed as a quadraphonic decoder but rather processor for playing video via original Dolby Surround and as a stereo to surround convertor. Using something like RM/QS coordinates gives known values that can be very likely to exist in stereo recordings.
I pulled my Panasonic Quadscope out of my system & on to my workbench. The DSM 3606 sat on top and the playback source was a Toshiba SD-9200. I adjusted the input balance using the balance null switch on the Fosgate. Visually on the quadscope it’s very easy to see when it’s balanced spot on.

When I ran through a few quick checks I guess I shouldn’t be surprised that the scope display looked as wonky as the unit sounded. I should have taken a few pics it was bizarre but it just didn’t occur to me.

On the board there are 3 adjustment pots. The one on the left is labeled CB F. This adjust front to back crosstalk reduction. In other words apply a center front signal & adjust until center back signal is at minimum. The middle pot is labeled CF F. This controls leakage from back to front. Apply a center back signal & adjust for minimum leakage in the front. The 3rd pot on the right is marked P RL. This controls left/right sensing. This one really perplexed me. When things are quite tuned up & looking good adjusting this pot a bit off factory setting seems to do nothing. However if it is not just right, going back & trying to tweak front and back is impossible. I settled on a setting just barely off factory setting & the same can be said of the other two adjustments. I won’t call these adjustments jumpy but a very little change makes a big difference in the decoded output. In practice it is more interactive than I make it sound. It took me a few hours over 2 days to get it adjusted as best I can. There are various TP connections on the PC board. I’m sure the factory used these not my method.

The results can be seen in the below photos. All were done on the WIDE setting as that is the only one that is designated for music. I will say that the other modes supply a mono rear channel & moves the left only / right only progressively up front. The surround output level was set 12:00 straight up.

The squiggle in the center front scope trace is a known artifact of the scope; it appears no matter where the center front input comes from. And by center front I mean L=R 2 speaker no center channel used. All the positions closely relate to what you should expect, much better than when I started out. Observations is that center left and center right are noticeably lower in level than any corner location or center F/B. That gives the expectation that this is decoder that will have very strong corner decoding as well as F/B but lacking on equal power for Center L/R. Look at the last picture taken from an actual stereo CD: Thomas Dolby; Aliens Ate My Buick.

You can see strong activity in the rear, distinct left front/ right front activity with quite a bit random stuff going on in between. The center L/R directions are pretty much missing. Since this is just a still pic I can tell you there was a lot of bouncing around activity between the rear channels, and more back & forth between front /rear than this would show. For a playback like this I would keep it on the WIDE setting & probably reduce the rear channel level & bring up the sound level overall. Or any other way to get the front /rear in balance.

Also of course there is not perfect crosstalk elimination and output levels “around the clock” are not perfectly matching. Still I think this is quite good for a decoder as old as it is. Some may ask if I tested it with SQ & the answer is no, there’s no need to. Performance like this puts it much closer to a QS decoder than SQ. Predictably playing SQ through it would get the front 3 positions right, and center back. But with any rear L/R activity the 90 deg phase shifts turns into an incoherent mess, just like my long gone one time girlfriend Jann on a Sunday morning.

Any corrections and further info is most welcome!

More to follow.

 

[Sonik Wiz]

After tuning up the DSM 3606 to the best of my ability I hooked it up in my system again. First impression was I couldn't believe this was the same decoder I last listened several years ago. The sound was not as spacious as the best PC decoding can do but at least it sounded correct. The Thomas Dolby track I used in testing played back different than expected based on the scope traces. The rear levels were actually quite low & I boosted them up about 5 dB for better balance. Very sharp corner placement & the separation front to back & vice versa (turning off the front power amps & going back & forth) was total, going only be my ears. What also was cool is that listening to only the front or back there's no audible pumping or artifacts. Just smooth sound. Punching in the center front speaker produced a level so loud it needed to be reduced by 8 dB to sound correct. The front L/R speakers pick up even more apparent separation doing this. Adjusting the bass EQ from zero only made things worse.

The decoder is not as detailed & crisp as I am used to listening to. As J. Gordon Holt would say "there are only two forms of audio distortion, one of commission & one of omission." Or as I like to say additive & subtractive respectively. Additive is when noise or audible THD, clipping distortion or frequency response is boosted at certain frequencies. Subtractive is when detail is lost, cymbals might sound dull, bass might be diminished. This decoder definitely falls into the subtractive group. And it's an uneven thing. On some pieces of music it's like I've never heard the reed buzz so clear on a saxophone. But most of the time I get the subjective feeling that somethings missing.

Much to my surprise the 3606 is much more sensitive to input level than I expected. I did see some of this on the test bench..... the input level needed to be set just right or the scope trace was distorted no matter where the trimpots were adjusted at. My oppo plugs directly into the 3606 & turning the level up past mid-point results in severe distortion. Also this unit needs a few minutes to warm up to stabilize. Both the front & back LED's light up slightly when first turned on & that goes away after a few minutes.

Lastly the internal output level trimpots were not adjusted when I did the testing & listening. Overall the output level is much lower than through DPL II on my Anthem. I may pull it one more time & tweak those a bit.

A fun adventure but still looking forward to the Involve SM v2.

 

[Soundfield]

Me too!

Which one is the evil twin?

 

[Sonik Wiz]

Which one is the evil twin?

The other one.

 

[Soundfield]

The other one.

I knew it, I never trusted him.

 

[Sonik Wiz]

I knew it, I never trusted him.

Me either.

 

[Sonik Wiz]

What?! No one saw the typo on Disclord's archive for the Fosgate?

 

[elguapo511]

Good read. What is this thing good for among its contemporaries? On eBay the prices are all over the map.

 

[Sonik Wiz]

Good read. What is this thing good for among its contemporaries? On eBay the prices are all over the map.

Just like the Proton SD-1000 it is good for playing back movies in original Dolby Surround & creating a quadraphonic sounfield from stereo music. Good as an QS decoder when aligned properly.

 

[quadtrade]

I know someone with a surround master first version and a recapped tate and the tate wins in almost all aspects of that duel.

 

[Sonik Wiz]

I know someone with a surround master first version and a recapped tate and the tate wins in almost all aspects of that dual.

Don't troll me. What duel ( or dual ^) are you referring to? Can you substantiate? I would like to learn more. At any rate my reference was to the SM v2, not the first one.

Except for my opening and closing remarks this note is about the DSM 3606, not Tate or Surround Master.

 

[quadtrade]

Who is trolling anyone. Hope that was in jest. Been doing this for 45 years. Know lot of people with lots of equipment. He Had 3 different units set up. Tate had better separation. SM was a little cleaner sound.

 

[par4ken]

A friend of mine had one of these Digital Space Matrix units, I forget which model. I added an input level indicator and a relay to bypass the unit when powered off, at his request. Despite my attempts to explain to him that the inputs need to be carefully matched he insisted on running his equalizer with different settings on each channel before the decoder and then he wondered why it didn't seem to do anything. As I recall it worked OK, I was still using my S&IC at the time and so nothing else seemed nearly as good. He pawned it shortly after, all my hard work for nothing!
A couple of years ago I purchased a Gavotte on eBay, ('m currently using a circle surround decoder in the car, which I like as the rear effect is adjustable), the original Gavotte lacks this. I tested the Gavotte it in the house, it just needs 12 Volts to work, although it could benefit from a pre-synth circuit to throw more sound to the rear, it still sounded good.