The last effort of the ARP engineering team, this synth adds MIDI to the previous "Chroma" interface, and uses albeit lo-res interface, real sliders to control the data input. (Eg. the resonance is an 8 position switch. Others are better fortunately.) The Polaris has an ADR envelope for VCA but has a 5 slider envelope for VCF making this machine a little more interesting than typical ADSR x2 synths. The fantastic 'assignalbe' knob has two halves on many of the parameters including pitch bend which allows you to on the upper half select the standard bend amount. On the lower half though it's bend amount ONLY for the notes currently held. So you can use the pedal and hold a chord with it then solo and bend just the solo note for example. You can also set a 'glide' amount and when you hit a chord and hit the damper pedal the next notes will be monophonic stack of remaining oscillators with the glide! Fantastic performance synth with these features.
At it's core is the 186 cpu which was used in some industrial applications. The 286 is slightly different and is the first "AT" processor in the PC world. At 6Mhz clock at the low end in both cases. Typical of the day 74LS and HC and 4000 series chips are used to interface the data and address busses to peripherals. A timer circuit makes assessments on pitches during tuning function in a fairly convoluted way explained in the manual.
This machine uses an actual ADC instead of the comparator stepping used in many other analogs like Siel DK600 for example. Values are multiplexed in from the controls to be converted to digital values which the processor then directly compares to the value stored in memory. (From the initial knob reads I guess when patch is loaded so that if anything is moved then that value is immediately slapped into patch buffer which then is read to the DAC during the s+H cell recharge cycle for that particular parameter).
As with the Chroma before it, a "Fast" and "Slow" option exists for updating the S+H cells so that a more gradual/smooth tapering of levels for instance is possible beyond the resolution of the 14 bit HS3140 DAC. The code is written to determine which is most appropriate in various circumstances. But the process involves two 4051 analog MUX chips. One has resistors in series to limit the rate of charge on the S+H capacitors. The other just charges them as quickly as the buffer amp will charge them.
The control of the Slow and Fast S+H MUX chips works like this. 74HC138 chip pair is used. One for slow and one for fast enables. They have common address and primary enable lines. The *fast* enable one though has an enable called SH LAG. This and the general enable and other things are controlled by latch Z30. If that line goes high then only the *slow* side will operate and hence an RC time constant of about 7MS will be introduced into the response time of changes to those S+H cells. So say were that line to somehow go high during auto-tune, the computer would be fooled into thinking it must drive the pitch higher than it actually has to and you will wind up with an oscillator that is detuned high when the capacitor finally gets fully charged. This phenomena has confused many a tech as you see in the notes below. But it's worth it to hear those smoother transitions I suppose.
Anyway 6 voice 2 oscillator per voice from the 6 dual vco CEM3374's. Ring modulator is preset full depth but really nice and can be used with detune and osc2 envelope amounts to create some fantastic effects. Oscillator Sync also. Pulse widths can be swept out of bounds which creates a rather gnashy sound of unpleasant nature so watch for that. One might think something is wrong with the machine :-).
Another interesting thing that is undocumented in the manual near as I can see, is that when you are in SAW mode on an oscillator, the pw slider becomes a SUB AMOUNT. If it's in the middle you get your root waveform only. Either up or down..it gives a sub octave amount! Gee...that's a pretty significant feature to go undocumented. :-) What is REALLY happening here is this. PW control works by mixing the SAW through 100K to the comparator input WITH a DC voltage offset from the PW amount S+H buffer. The RAW SAW signal is able to bypass the comparator circuit when an analog switch is turned on (when SAW rather than pulse is selected for that osc). So when pulse is selected you purely have a comparator output giving a pulse wave. The input signal to comparator has troughs at about .95V and peak at 3.95V with PW at min. At max trough is 4V and peak is 7V. The reference leg of the comparator is held at 4V. So when the PW control is at miniumum there is no change in the comparator output at all. (ref leg is the non-invert side of comparator so voltage is in HIGH state at this point). As the control is turned up quickly the narrow pulse will emerge as the voltage varies from 1.01 to 4.01 V etc. Actually due to the digital resolution limits I'm not sure what the first point will be but as the slider rises to the middle of travel we wind up with an approximate square wave and then it tapes the other direction back to narrow pulse then nothing.
Ok so when SAW is selection there is no switch to turn off the pulse. So to get a pure saw wave one must turn the pulse width control to one extreme or the other. There is a 120K resistor on the comparator output. There is 100K plus the impedence of the analog switch in series with the saw signal coming around the comparator and mixing with that comparator output almost 50/50. So in the middle you wind up with a square wave that is low at the peak of the saw wave. Half way through the falling ramp, the comparator switches high for the duration of the ramp's descent! The result mix of the two then simply creates a DOUBLE FREQUENCY HALF HEIGHT SAW WAVE. So that is what you are hearing as you move these PW controls, giving a wider range of useful timbres.
As noted in service manual, one of the biggest issues people have is using the tape interface. I keep forgetting how so I'm going to document one workable way. They recommend going to cassette (lower function "C") and turning off 3 and 4, which turns it from digital to audio (recommended) and from high to low level (not recommended). Leave light 4 on in other words if you are using a sound blaster live like me or probably most other sound cards out there. Turn your volumes all the way up on wave and output in the computer's audio controls. Get a cable from your speaker out to the 'in' sync of the polaris and you should have success after hitting 'upper functions 12' or 'load all'. After you get done loading programs you hit it again and run the file with sequences in it.
I've seen op amps fail in various places. And 4051 analog mux chips give drift to vco's. Power supplies have been problems so always verify rails of course. The membrane switch panels are notorious for cracking, and several switches may cease to function due to a break in the line somewhere. I'm an expert at fixing these problems and can almost guarantee success. It's not a trivial proceedure to teach someone and depending on the situation there are a few different methods which may need to be employed so just send it if you need it done. Shipping is cheap when you removet he panel. I can test it here without the whole synth. I just did my own unit and it took about 4 hours! Arrghh. That's by far the worst one so far. But they all work and will forever or until the plastic turns to dust now.
A service manual is very helpful on these units. If missing voices though, hey, try a 'tune all' :-). Sometimes they're SO far out of bounds that you can't hear them. ALSO the manual was weak on tape dump verify. #11 just flashes if all is going well. It sounded like you were supposed to have it tracking through the patch LED's but that only happens if you do an actual load.
I've never used the simple sequencer and in fact mine is fouled up or something. AFter working on it I get an error honk as though memory is full all the time though I've emptied it completely! When it honks the stop light quickly toggles to play I think or record and back. Strange. Doesn't affect the sound fortunately so I just turned the honk down and live with it for now.
NOTE: Regarding CEM3372 VCF's. We had a recent bit of experience trying to get these to work in Doug Terrebone's Prophet 600 and in one I had here. None of the stock I had would sound the same (not to mention that two I got from chipforbrains some time back were totally doa... bummer. Thought those were NOS!) They lacked the whispy overtones with Resonance all the way up in particular. THey were all "C" revision. The "B" rev's were in the P600's but Doug said he finally got in an AX80 that had Rev "C's" and those worked fine! Go figure! Anyway just beware that there is some variation in these chips.
I'd previously gotten some confused information about the reversal of designated trimmers. That doesn't seem to be true. Anyway I've had several recent repairs on these nearing the end of 2009. A couple of the strangest ever. Ryan Powers witnessed something that was quite interesting. This is why I love having people around when I work. We first did the membrane switch ribbon mod on all four cracked cables. We got the insulation off it appeared well. Three of the cables worked fine but the left side 7 connector one had 5 bad lines!! I pulled it after determining that it wasn't just dirty switches which I've never seen on these actually. And recleaned it a bit more. GREAT! NOW NONE OF THEM 7 WORK!! We kept recleaning it for an hour or something. Totally insane. None ever worked again. I Told Ryan "Sometimes..I have to pray for things. It's..just me...." Well I prayed and felt led to swab it lightly one more time and surface scrape it one more time and put it back in. ALL 7 WORKED AND NEVER FAILED AGAIN!
The second problem which I'd tried to address was that the ADR Release slider was acting like a channel volume! Now that function, as we know, resides on the assignables slider. And I thought..mmmayybe the 4051 is somehow..bridged there? I replaced it and it worked! Until..Ryan came back when it immediately made a fool of me and started acting exactly as it had been when he'd left for a bit. Now the assignables slider had a broken shaft but I checked the resistance measurements and it was actually working fine. But I prayed again and felt led to simply replace that slider and put things back together. It worked fine and never manifested that again. Does ANYONE have a theory as to what was causing that? Please...don't say 'supernatural beings'. I already have surmised that this is the most likely cause :-)
Tonight I worked on a Polaris II. It has a problem of 3374 number 2 having one VCO off tune by about a half step! All others would tune-all just fine. I replaced the vco chip..no effect. I replaced the comparator since the thing uses the square waves generated by it to tune I believe. Still fouled up. I'd cleaned the socket and I'm wondering if something didn't eventually help a connection or bad solder joint because I hit tune-all again and finally they all just tuned. Probably should redo the solder on that one. There was a little rust on a couple comparator pins near where they join with the plastic package. (see below, I bet it was the same issue. read it all)
Similarly a Polaris 300477 (which I'd already had back once I think and I resoldered the board and it behaved for a while) had some elusive dirty sockets that caused a leaky S+H effect on the volume control line for the vca on the first three channels. ALSO though this synth had a peculariar problem with voice 5 turning off during auto tune. I viewed the tune buffer and found that it's level was a db hotter than others. 4's was lower though. It tuned fine. Actually 5 tuned fine it just had to be turned on manually each time using lowerfunction E - 5. Anyway I swapped CEM3372 chips and the levels changed but now BOTH 4 and 5 wouldn't turn on! I took a different pair from a scrap unit and 4 was back but 5 was still off. THe levels ...well 4 was perfect now but 5 was right back where it was initially at a little high. I swapped the chips AGAIN and now all 6 tune and turn on fine! Go figure. I can't see any reason for this at all at the moment. I mean there are no doubt minor resistance variations and something was just on threshhold but I'm not sure who two different filters would fail in some other test that is carried out on 5 yet ONE of them work on 4 but not 5! quite peculiar!
ALSO NOTE: 300477 at least had a mod which was probably a transition in revision changes. The Zx08 4053 chips had legs 12 and 13 pulle dout of the sockets and 13 was hard wired and 12 left floating. ALSO though the destinations of the pins on this chip were SWAPPED between the section on the schematic that switches Alt and main out...and the section that grounds signal for the SYNC function! BEWARE of this as you troubleshoot. THe schematic may have different pin numbers here and maybe other places I missed but since that's where the hack was this is likely the only issue. ALSO this rev has 33K instead of 15K resistors from the saw to the cap of the sync input and from that point to ground.
That keyboard keeps coming back. First case of one of my panel repairs not quite getting done right and developing a kink that broke the panel for good. But this thing of oscillators tuning then instantly being out of tune with one another on all but 2 and 3 channels. 1 and 6 being the worst. Obviously no common thing to tie these manifestations to. 6 again..seemed to improve when I cleaned around the resistor I previously soldered in the current reference circuit. Moving/tapping on things seems to have no effect. Something is causing the computer to think those channels are tuning but there is a false reading occurring with them somehow it seems.
HOLD THE PRESSES! I've found someone else with an identical sounding manifestation. It's always the pitch B controlled oscillators. Well what could cause *that*? Only the sync lines I would think. Could it be that some noise is causing a little bit of random 'sync' trigger? Shorting the sink line makes no difference so I'm sure it's not that it's getting noise on it from a bad ground etc. Rather we also hear a little portamento when hitting the notes at first it seems often. I'm thinking there must be some slower charging on those channels of the S+H caps because after all...that would create a lag such that the cpu *thinks* it needs to go further perhaps on those oscillators that have this issue to get it in tune. THen after it has time to settle in more it's sharp! Which is always the case because you can move 'detune' but it will only make it more out of tune since detune only goes up in pitch.
When I cleaned Z60 socket it got all better. I don't understand the 'channel (and B osc) specific results' at this time. ALl caps measure identical in the s+h circuit. As do resistances. (oh btw don't be confused by the shorted 1M resistors in the resistor packs (pin 5 to pin 10) for the slow 'volume'..ie. between the pin 4's of the two 4051's. They deliberately shorted the ones on channel 1, 2 and 6 it appears! They aren't in use anyway so ..wth why not confuse technicians! haha. For whatever reason they didn't short 3 also. As you notice in the diagram 4 and 5 channels appear to be in use but in reality only channel 5 is actually used. 4 was wired to the J17 header and my guess is it was in case they decided to implement stereo volume control. eg. put a chorus bbd circuit in or something and have stereo outputs. So there it sits doing nothing. Channel 5's op amp is the only one of that pin number on each channel to be used at this time and it controls the main volume of the machine.
As a final note refer to the circuit description above for troubleshooting help if you ever see this issue involving the S+H control circuitry. But after getting this one back together nearly it DIED (locked up) putting the cover on. I spent some hours tapping around trying to localize what was the issue. It seemed hitting the back near the jacks on the right or the bottom right would always produce a lockup and I'd have to just power cycle to get the synth back working. I could get NO distinct correlation to anything on the circuit boards however. yet sometimes tapping in any random place it might do it if I did it fairly hard. Not keyboard and not panel related for sure. Then I noticed I could barely tap the power input and that seemed to be it and i cleaned and it seemed good til...you guessed it..I got the panel screws in yet again. I worked over the ins and solder on input board for ac. It got worse after that and I finally figured out it was the GROUND WIRE under the front of the power transformer! Getting needle nose vise grips on the back of the screw head and tightening it with pliars on the front did the trick. Finally a solid unit!
I think I have pretty much every part electronically for these save the exact CPU. I was mistaken previously. You have to get a 80186 as there are differences. I was under the impression that it was just the early 286. Not true.
Contact Sound Doctorin'