Vintage Synth Explorer Forums

Have got a VFX SD who do not works. the Display is always in "SELF TEST MODE". I've got the service manual but no shematics, so following their troubleshooting guide i end on: "Replace the Main board, send it to Ensoniq". I'm not able to perform any of the software RESET or HARD RESET. I've never been able to get the boot/start screen.I've got some skills reparing old synth but never had a Ensoniq. I've scoped CPU looks ok, tested connexions they are ok. Power supply is OK too. Cleaned old flux too, without succes.

So my question is would you have any idea how to solve this synth or how to get the software running at least?
Is there a part to specially look at?
Maybe someone has got the shematics for the VFX SD to share ? As maybe it will help to repair it.
Thank you

If any of the chips is faulty (FX, generator, display driver etc) the system won't boot, as AFAIR it is based on serial communication with "ring" topology. So if CPU doesn't receive proper response, the boot sequence freezes. In my VFX I had to replace FX chip. The good news - those custom chips (generator, FX) are the same that were then used in much more common SQ1 and SQ2.

http://s000.tinyupload.com/index.php?fi ... 3756412852 - here are VFX schematics that I've downloaded 5 years ago, when I was repairing my VFX. I don't remember, where I found them.

I posted the service manual, schematics, and a service bulletin regarding this as well:

http://www.vintagesynth.com/forum/viewt ... 60#p781027

And the ROMs:

http://www.vintagesynth.com/forum/viewt ... =1&t=93160

Edit: Given that the Giebler EDE/EDM utilities are annoyingly buggy, obsolete, and a generally pointless nuisance these days, I've made the VFX SD OS 2.10 OS sequencer reinitialize disk available in universal raw (Omniflop) format. This will work on a whole bunch of newer machines where Giebler would choke and die (for multiple reasons, including Turbo Pascal bugs and floppy disk controller problems) and is also easily convertible with HxC software if you're running that as a floppy drive replacement. It should be treated as an Ensoniq VFX-SD 3.5" 800K DD disk, of course.

https://mega.nz/file/HN0EhYyR#qCkWs3RNe ... eKnyzHW4C0


You should also check that the RESET signal is getting to the CPU properly. There's a fail-safe that turns it off if anything wonky is happening. It's controlled by a 5-pin LM2925 regulator on the PSU.

I have an EPS16+ with the same self test mode problem. I have a rack mounted unit, as well. I plan to side by side them to see if I can track down the source of the problem, unless I can find a mainboard before then. diagnostics & repair must be more costly than to buy another unit...

Hyde wrote:I have an EPS16+ with the same self test mode problem. I have a rack mounted unit, as well. I plan to side by side them to see if I can track down the source of the problem, unless I can find a mainboard before then. diagnostics & repair must be more costly than to buy another unit...

Pretty much all the Ensoniq stuff of that era do the same (annoying) thing. The good thing about the VFX SD (or classic VFX with the 2.0+ ROMs), the keyboard can be pulled out of the loop and it'll still boot past the display self-test, albeit with a keyboard calibration error [Retry/Ingore]. The older VFXs will choke in the same situation.

Edit: Incidentally, if you have an older VFX and you want to rule out the keyboard, try unplugging the 20-pin key cable from the mainboard and then jumper (the now exposed header) pins 17 and 19 on the mainboard--the two left most on the top row of the keyboard header. I've tested this on a VFX SD and it auto-passes the keyboard calibration, so it should do the same with the original VFX. What it does is connects the display serial out directly to the UART serial in, thereby taking keyboard serial communications out of the loop.

As mentioned earlier in the thread, a good portion of the chips in Ensoniq stuff are common between various models, so you can steal/borrow/swap in various things from stuff you might own or have access to. For example, you can take the CPU out of an EPS series board and drop it in the VFX to test a dead CPU. A lot of the custom Ensoniq stuff like OTTO and ESP chips are (generally) in sockets too, and can be found in other Ensoniq gear.

Additional Edit:

As described above, jumping pin 17 to 19 on the mainboard keyboard header will essentially make the VFX operate in rack mode. MIDI events can still control the VFX, so patch selection and playback is entirely possible even without a calibrated/functional/connected keyboard. Likewise, the mainboard does *not* actually require a serial communication response from the display/keypad board to actually boot, although without a proper serial Tx from the keypad PCB the VFX interface will hang at the KEYBOARD CALIBRATION ERROR screen.

Further Edit (3/2017): While auto-bypassing the keyboard with a jumper will boot the VFX enough for testing MIDI and the sound engine, it also essentially disables the front panel keypad, so the VFX is not fully functional while in "rack mode" and it is definitely not suggested as a long term solution.

However, it should also be noted that the VFX supports keypad simulation over MIDI, so even if you have a dead keypad/keyboard then the VFX can still be fully controlled with a SysEx control panel and external keyboard.

Rasputin wrote:

Hyde wrote:I have an EPS16+ with the same self test mode problem. I have a rack mounted unit, as well. I plan to side by side them to see if I can track down the source of the problem, unless I can find a mainboard before then. diagnostics & repair must be more costly than to buy another unit...

Pretty much all the Ensoniq stuff of that era do the same (annoying) thing. The good thing about the VFX SD is that with the 2.0+ ROMs, the keyboard can be pulled out of the loop and it'll still boot past the display self-test, albeit with a keyboard calibration error. The older VFXs will choke in the same situation.

As mentioned earlier in the thread, a good portion of the chips in Ensoniq stuff are common between various models, so you can steal/borrow/swap in various things from stuff you might own or have access to. For example, you can take the CPU out of an EPS series board and drop it in the VFX to test a dead CPU. A lot of the custom Ensoniq stuff like OTTO and ESP chips are (generally) in sockets too, and can be found in other Ensoniq gear.

Good to know, thanks!

There are about a thousand things that can take the VFX down like that, but assuming you've checked your ribbon cables and your power supply voltages, you'll probably want to start in on a few key mainboard chips with a logic probe.

Motorola 68K [MPU 68000p10] (CPU -- commonplace)
Motorola 68681P - DUART (serial communications -- commonplace) [Philips also makes a 68681]
Upper/Lower EPROMs (v2.0+ on the VFX SD) [27512 style -- commonplace]
Ensoniq ESP [ES5510 / ESPR6 -- 5510000102]
(custom programmable DSP chip; has a yellow capacitor soldered directly to the top if it -- fairly rare)
Ensoniq OTIS [ES5505 OTISR2 -- 609-0381303] (custom synth engine and ADC processing -- fairly rare)

Maybe these too:

256AC (x2) - Patch & Sequencer RAM (socketed for upgrade, powered by battery backup when VFX is off)
256AC (x4) - System RAM (not socketed, permanently installed)

The pinouts in the schematics are pretty hard to read, so here's auxiliary information to get you where you're going...

Standard Motorola 68000, but the one in the VFX is a bit faster than the traditional 8.00 MHz version. Here is the pinout diagram for it:



Now since it's almost impossible to find info about it on the 'net, here is the pinout and a blurb about OTIS:

OTIS is a VLSI device designed in a 2 micron double metal CMOS process. The device is the next generation of audio technology from ENSONIQ. This new chip achieves a new level of audio fidelity performance. These improvements are achieved through the use of frequency interpolation and on board real time digital filters. All calculations in the device are made with at least 16 bit accuracy.

The major features of OTIS are:
- 48 Pin dual in line package
- On chip real time digital filters
- Frequency interpolation
- 32 independent voices (up from 25 in DOCII)
- Loop start and stop positions for each voice
- Bidirectional and reverse looping
- 68000 compatibility for asynchronous bus communication
- On board pulse width modulation D to A
- 4 channel stereo serial communication port
- Internal volume multiplication and stereo panning
- A to D input for pots and wheels
- Up to 10MHz operation

And the pinout (nice and legible):



Now for the info on the ESP chip:

The ESP chip is a custom digital signal processor chip with over 75,000 transistors, used in all of the company's musical instruments and some multimedia products. It is a high-speed microprocessor with an instruction set that is optimized for manipulating audio data, which has typical sample rates of between 10 kHz and 50 kHz. The ESP chip is capable of creating a wide range of digital effects including reverb, delay, echo, flanging, chorusing, harmonizing, equalization, and distortion, and is capable of generating multiple effects simultaneously.

The ESP is a VLSI device designed in a 1.0 micron double-metal CMOS process. It is optimized for signals in the audio frequency range. The multiplicity and flexibility of the data paths in the ESP allows many DSP operations to be accomplished in a minimum number of microinstruction steps. Its nominal instruction cycle is 250 ns, yielding program lengths from about 64 to 160 microinstructions at typical sample rates. Because the ESP chip is fully programmable, the range of effects is unlimited. (Competitive DSP products are generally limited to their specific design functions.)

The major features of the ESP chip are:

48 Pin DIP or 52 Pin PLCC
- Separate Address Generator ALU
- 4 Programmable Serial I/O Channels (I2S or Sony Format)
- On-Chip Data and Microprogram Memory
- 8-Bit Address/Data Multiplexed Host CPU Interface
- External Sample Rate Synchronization
- Multiplexed Addressing for Simple DRAM Interface
- Host Access to ESP DRAM

The architecture of the ESP chip is implemented by the following major components:

ALU - 24-bit wide, capable of 16 different instructions
- Multiplier - 24x24 bit with dedicated 48 bit accumulator
- Separate Address Generator ALU
- Microinstruction Memory Array (160 x 45 bits)
- General Purpose Register Array (192 x 24 bits)
- 23 Special Purpose Registers
- Three 24-bit wide data paths
- Serial Digital I/O (4 stereo channels, I2S or Sony)
- Host interface

Here is the pinout in the case of the VFX:



Note that this is different from 5510 ESP revision 7 (DP2 & DP4, etc.) which can be viewed for comparision here:



As far as the UART goes, here is a nice clean diagram of the pinouts:



If you're looking for gear to swap around some of these chips for testing or to cannibalize, then the Ensoniq EPS, EPS-16 Plus, SD-1, SD-1 32, SQ-1, SQ-R, SQ-1 Plus, SQ-2, and KS-32 should have some major chips in common with the VFX/SD.

Here's how the serial communication is laid out:

68681 (DUART) pin 11 is SERIAL OUT which transmits to Display via 4-pin jack pin 2

SERIAL OUT is processed by the Display and comes back as DISPLAY OUT through 4-pin jack pin 1 and goes to Keyboard MCU input [ultimately pin 42 (RxD) of 68HC11A1] via 20-pin jack pin 19

DISPLAY OUT is processed by the Keyboard and is returned to 68681 (DUART) pin 10 as SERIAL IN by Keyboard MCU output [ultimately pin 43 (TxD) of 68HC11A1] via 20-pin jack pin 17

68681 (DUART) pin 15 then sends a SERIAL ACK to Keyboard MCU [ultimately pin 6 (PA2) of 68HC11A1] via 20-pin jack pin 4

Keep in mind that there are all sorts of things that can keep the VFX from booting, thereby killing all serial communication. Just because the display is in self-test mode doesn't mean it's a serial communication problem, per se.

Here is the pinout for the Motorola 68HC11A1 MCU which handles the keyboard's serial communication:

The display panel is almost definitely working since it's successfully in self-test, so other than continuity on the wire harness, I doubt there's any reason to poke at it. I think it's almost always the mainboard / PSU that flops in cases like this.

If you want to check the display PCB though, the main chip to check is NCR 6500/11. You'll also see a whole bunch of Texas Instruments - TL5812 chips, but those are just drivers for the fluorescent display and aren't tied into the mainboard.

You can find the diagram for the 6500/11 pinout here:

Here's an extra document regarding the 6500 MPU which is more in-depth:


Make sure there's serial traffic on 6500/11 Port A by checking:
Pin 23 [PA7] - Display IN (*from* pin 2 of header)
Pin 24 [PA6] - Display OUT (*to* pin 1 of header)

UART TROUBLESHOOTING

Seeing as the UART is the hub of communications activity, and the display self-test is an indication of lack of communications, it's not a bad place to probe (once you can establish voltage/clock/reset on the CPU, at least).

The CPU sends a communication byte into the UART on D0~D7 (one pin per bit; eight pins per byte = parallel). Once the UART receives this, it constructs a serial frame and sends it out one bit at a time over TxDB (pin 11; bits transferred sequentially over this single pin = serial) as the SEROUT signal which is supposed to go to the display/keypress unit.

TxDA (pin 30) and TxDB (pin 11) are high at idle. As such, they default to being high right after the UART gets a RESET signal (proper reset is a low blip on pin 34 which is then stuck high the whole time the UART is powered up).

Incidentally, TxDA is the MIDI output. Although it isn't relevant to the VFX being stuck in self-test mode, the behavior of pin 30 can help you diagnose if the UART is working to some degree.

As an example: If the UART has +5V on pin 40, ground on pin 20, has a pulsing clock on pin 32, a proper reset on pin 34, and pin 30 is stuck high then that's a good sign.

If pin 11 is floating or stuck low then it's an almost definite indication that the UART is messed up.

If the situation is reversed in that pin 11 is typically high or pulsing and pin 30 is stuck low or floating, then you might have a booting VFX but MIDI out isn't working anymore.

If the UART looks fine so far, but pin 11 and pin 30 are permanently high (not even occasionally pulsing) then we want to look at D0~D7. If you see pulsing activity on D0~D7 then that should indicate the CPU is trying to communicate with the UART. In that case, we would expect pin 11 to be pulsing with activity because if the CPU is working enough to try to send something out through the UART then there should also be attempted serial communication traffic.

If D0~D7 show traffic but pin 11 and pin 30 never show any activity (stuck high and nothing else) then the UART is worth further investigation.

In general, you'll want to check the UART in this order, for these results:

Pin 20 [GND] - Ground (stuck low, permanently tied to digital ground)
Pin 40 [VCC] - +5VDC (stuck high, permanently tied to voltage)
Pin 32 [X1-CLK] - Clock input (pulsing [@ 5 MhZ], permanently tied to one of the crystals)
Pin 34 [RES] - should briefly hit low on power up, then stay high
Pin 35 [CS] - should be low if chip is enabled, so probably pulsing but maybe low (stuck high or floating = bad!!!)
Pin 10/11 [Serial RX/TX] - should show pulsing traffic, or at least be high at idle (stuck low or floating = bad!!!)
Pin 30/31 [MIDI TX/RX] - same as serial RX/TX above
Pins 16~19 / 22~25 - [D0~D7] - pulsing
Pin 8 [R/W] - pulsing (possibly mostly high with negative blips)

Permanently floating pins are a giant red flag. Pins that are stuck high or low might not be the chip itself, but something connected to the chip instead. For example, something that's generally high like a Tx/Rx pin could be grounding out somewhere else, or D0~D7, R/W, CS, etc. could be improperly fed by the CPU, GLU, or SUPER GLU.

DISPLAY / KEYPAD / DATA ENTRY SLIDER / VOLUME SLIDER TROUBLESHOOTING

If your display looks dead in the middle with flickering behavior on the far left & right sides then you might have blown two fuses, usually by disconnecting the music keyboard and then hooking up the music keyboard ribbon backward. A poorly connected ground pin on the keyboard connector might cause this as well. The two uppermost fuses on the PSU are easy to blow and they almost always go at exactly the same time.

The display/buttons themselves can be tested by putting the display/keypad into self-test mode by unhooking the 4-pin display connector from the mainboard. Hitting the buttons will respond with numbers/letters and cursor movement. Obviously, the wider ribbon cable going to the display/keypad needs to still be connected, as it supplies the power.

The behavior of the 4-pin display connector is pretty simple to examine. Only one of the pins is an input to the display, the other three are outputs from the keypad board.

Pin 1 is the keypad serial communications output, so if you're monitoring pin 1 with a logic probe and you push a keypad button, you should get a blip on your logic probe. If a button works in self-test mode but doesn't cause a blip on pin 1 then probe pin 24 on the 6500/11 chip. If there isn't activity on that pin then the 6500 is probably bad. If there is activity on pin 24 but none on pin 1 on the 4-pin header then it's probably a continuity issue.

Pin 3 is connected directly to the volume slider output. It's analog, so you should measure a logic low when it's down all the way, and a high when it's all the way up. Anywhere in the middle and there will be varying results between low, high, and floating.

Pin 4 is the data entry slider output. It works exactly like the volume slider. Easy to test.

Pin 2 is the solitary input for the display/keypad board. It's the serial communication that's sent by the DUART, so if you probe it during the VFX boot sequence you should see a few blips. If you get a couple blips at pin 2 on the display side but nothing on the display, check pin 23 on 6500/11A. If there's nothing at pin 2 on the display side, check pin 2 on the display header on the mainboard. If nothing is present there, check pin 11 on the DUART.

If there's nothing on pin 11 on the DUART then see the previous UART troubleshooting section.

This isn't directly related to the OP's problem, but it falls under mainboard troubleshooting, so...

MIDI TROUBLESHOOTING

Fortunately, if the VFX is otherwise booting up and acting normally, there is little that can go wrong with the MIDI ports. It's very likely to be either one bad pin on the UART/7407/optocoupler chips, or something physically simple (and visually obvious) like a bent pin in the MIDI jack.

If both MIDI IN *and* MIDI OUT are not working:

It is wise to triple check any MIDI cables and channel settings, etc. The most significant common part between MIDI in/out is the UART, but a UART that is entirely bad will prevent the VFX from booting at all, so it's fairly unlikely that the UART would partially fail in a way that kills both MIDI IN *and* MIDI OUT. However, check UART pins 30 and 31. If either or both are stuck low or floating then the UART is highly suspect.

If there is no MIDI OUT (but MIDI IN still works):

Test the MIDI OUT port by checking to make sure pin 2 is grounded, and that pin 4 is showing between 3-5 volts. If pin 4 isn't showing voltage, check the voltage at FB4 (it's a ferrite bead that looks like a graphite cylinder).

If everything looks correct so far, check MIDI OUT pin 5. It should ordinarily be the same voltage as pin 4, but should dip low or pulse when MIDI events are generated (try constantly moving the MOD WHEEL). If the voltage is either floating, stuck low, or stuck high when the keyboard / mod wheel is in use then start looking at the 7407.

7407 pin 11 should show activity during a MIDI event. If it isn't, check UART pin 30. If there still isn't activity then the UART is very likely bad. Either troubleshoot the UART more or replace it.

If there's activity on 7407 pin 11 but not 7407 pin 10 then the 7407 is partially bad. Route pin 11 to pin 13 and pin 10 to pin 12 on the 7407 and then test to see if there's activity on pin 12 during MIDI activity. If so, MIDI out is fixed.

If there is no MIDI IN (but MIDI OUT still works):

First see if MIDI THRU is still working. If MIDI THRU works, but not MIDI IN then check UART pin 31. If it's floating, stuck low, or is *always* high despite MIDI IN traffic then the UART is likely bad; either troubleshoot the UART more or replace it. Otherwise, continue to the next problem.

If MIDI IN and MIDI THRU both do not work:

Test the MIDI IN port by checking the optocoupler (6N138) pin 8 for +5V and pin 5 for ground. If the voltage is not present or significantly wrong then investigate the decoupling capacitor next to 6N138. If +5V and GND look good then check 6N138 pin 6 to see if it is floating or stuck low. If so, then 6N138 is likely bad.

If 6N138 pin 6 is high, check pins 4 and 5 of the MIDI IN port for 3-5 volts. Both pins should be the same voltage. If not, check 6N138 pins 2 and 3. If one or both are floating or stuck low then 6N138 is likely bad. If MIDI IN pins 4 and 5 are the same voltage (3-5v) then check 6N138 pin 2 when MIDI IN is being sent traffic. If it's *always* high then check the diode (CR1) 6N138 pins 2 and 3. If the diode is good then 6N138 is likely bad.

If there is no MIDI THRU yet MIDI IN still works:

Check for +5V on pin 4 of the MIDI THRU port and ground on pin 2. If so, check FB3 (again, a ferrite bead that looks like a graphite cylinder) for activity when MIDI IN traffic is being received. If there is traffic then check for a faulty MIDI THRU pin 5. If there is no traffic, check 7407 pin 8. If it's floating or stuck low, or *always* high even during MIDI IN traffic then route 7407 pin 9 to pin 13 and 7407 pin 8 to pin 12. Check again for traffic on MIDI THRU pin 5. If so, MIDI THRU is fixed.

I think I got that all correct, but you never know. Don't take it as gospel.

Rasputin: that's impressive info you give there.

POWER SUPPLY (PSU) TROUBLESHOOTING

There is nothing particularly complicated about the VFX / VFX SD / SD-1 power supply, so some of this is just generic knowledge geared toward any linear power supply. Also, a lot of the Ensoniq power supplies are virtually identical or at least very similar to the VFX series, so a lot of this applies to additional machines such as the SQ80, etc.

Hopefully there's no need to cover checking the main fuse or the fuses on the PSU PCB, or such things as continuity on the power switch with any specificity. If you can't figure that out, well... you shouldn't be servicing gear, let alone power supplies as you're likely to get yourself killed or maimed.

The service manual contains the basic info needed to check the power supply, so reference the service manual to check whether the AC and DC voltages are in spec or not. The transformer outputs three separate sets of AC voltages, each one is a "triad" of wires with the middle wire being the common wire. The triads are red/black/red, yellow/white/yellow, and brown/orange/brown. Use a multimeter to measure the AC voltage between the two reds, the two yellow, and the two browns. Make sure the results reflect what is in the service manual.

The red triad ultimately supplies the digital components (CPU, UART, ESP, RAM, EPROMs, etc.) with the +5V DC the need to run (meaning they're all TTL level chips). The yellow triad ultimately supplies the analog circuit (Data Entry, Volume, Mod Wheel, Pitch Bend, Main/Aux/Headphone Output, etc.) with the +12/-12V DC that it needs. The brown triad supplies about 9VAC to the display.

Most of the AC voltage supplied by the transformer goes through two points: rectifier diodes and fuses. The AC bound for the digital board goes through fuses F3/F4 and then through the first two 1N5400 diodes (CR1/CR2; there are six total, named CR1-CR6). That's why the F3/F4 fuses are very prone to popping and they'll always go if you put in the keyboard ribbon cable backward and other such things.

The other four rectifier diodes (CR3-CR6) are 1N5400 as well and (as a group) form a full-wave bridge rectifier. The yellow triad from the transformer is sent through fuses F1/F2 which then goes through the bridge rectifier and eventually onward to the analog circuit.

The brown triad stays AC so it does not go through any rectifier diodes and simply goes through fuse F5 before being sent out to the display filament. There really isn't much to troubleshoot as far as the 9VAC on the PSU goes, as it's just delivered hot off the transformer with a fuse thrown in for protection.

The DC voltages should be checked with the method given in the service manual. The PCB has each pin silk-screened, so it should be very obvious what to check for. The biggest thing to watch out for is the display filament AC output on the display header (pins 5 and 6), as everything else is DC voltage on the PSU output headers.

Assuming the values at all these points fall within the accepted ranges specified by the service manual then all is probably well. However, there are a few other points to consider. The first one is that measuring the DC voltages with a multimeter isn't necessarily accurate because there can be ripple (AC superimposed on the DC lines, basically) which will skew the results. The second one is that the PSU may test fine when all or some of the output headers are disconnected from the PSU PCB and then fall out of spec when the output headers are reconnected to the mainboard, display board, and/or floppy.

If things aren't measuring correctly, disconnect the output headers and measure again. If the problem goes away then there are a couple things that could be going on. It could be that there's a problem with whatever board(s) are connected to the power supply and that's causing the voltage(s) to drop, or there could be a problem with the power supply. One way to narrow that down is to test the PSU for ripple. The "correct" way to do that is to use an oscilloscope, but the lazy/cheap way is to set your multimeter to measure AC voltage at the same points the DC voltages are checked. In other words, check for AC voltage where there is only supposed to be DC voltage present. In an ideal world, the AC voltage would show exactly 0.0 volts, so if the AC voltage shows anything greater than about half a volt then investigate the power supply further.

See those four black rectangles with the metal "fins" that are lined up and screwed into the back of the PSU? Those are voltage regulators. Ultimately, they are supposed to take voltage that is higher but imperfect and drop it down into a lower but perfect voltage. These are robust and fairly hard to kill (they will generally survive being shorted out, overheated, and otherwise tortured), but that doesn't mean they never go bad. If one of the DC voltages is bad or missing then they're worth checking into.

Here's what the four regulators are named and what they do:

7805 - This is a super common regulator that can be found in a bazillion power circuits. It's job in the VFX SD is primarily to supply +5VDC to the display and floppy drive.

7812 - This is like the 7805 but supplies +12VDC to the floppy and the positive half of the analog circuit.

7912 - This is like the 7812 except it is NEGATIVE (-12VDC) and only supplies to the negative half of the analog circuit.

LM2925 - This is the only one that has five pins, all the rest have only three. While these were easy to obtain a couple decades ago, they're fairly hard to find these days. It's job in the VFX is two-fold: supply +5VDC to the digital circuit and also supply +5VDC to the RESET line *and* provide a reset pulse to boot the CPU (and some other digital chips).

If +5D or VRES is bad or not present then LM2925 is a good place to poke. If +12A is bad or missing then 7812 is a good place to poke, and so on. The output voltage on 7805 and 7812 can be measured by putting the black probe on the center pin (pin 2) and the red probe on the right pin (pin 3). The output voltage on 7912 is measured differently! The black probe goes on the left pin (pin 1) and the red probe is on the right (pin 3) [remember, the voltage will be negative on this regulator]. LM2925 is measured with black probe on pin 3 (ground), and either red probe on pin 5 (+5VDC for the VRES line) or pin 2 (+5VDC for the +5D line).

Should the output voltages measure as bad then check the input voltages being supplied to the regulators by moving the red probe to pin 1 on 7805/7812/LM2925 and pin 2 on 7912. The voltages need to be significantly higher on the input for a regulator than what they are intended to output, so the inputs should be in the ballpark of 5 volts higher. If the input is less than 3-4 volts higher than what the regulator is intended to output then the regulator won't have enough voltage to work with and all h**l will break loose.

Assuming the AC voltages off the transformer measure correctly, if either the input or the output voltages are incorrect then there are a few things it could be. Before the AC voltage hits the regulators, the AC voltages goes through rectifier diodes and smoothing capacitors which clip and smooth the AC voltage into DC voltage. If either one of those systems aren't working then the voltage regulator won't be able to make the nice clean voltage it is designed to put out to make the other PCBs happy.

Symptoms and probable causes:

If both the 7812/7912 are giving weird results then test all four diodes in the bridge rectifier (CR3,4,5,6).
If only 7812 is giving weird results, look at C18 (one of two 25V, 3300uF electrolytic capacitors on the PSU).
If only 7912 is giving weird results, look at C14 (same kind of cap as C18).
If either 7805 or LM2925 is giving weird results, look at the larger 16V rated electrolytic capacitors on the PSU.

Note: +VU isn't regulated so it may have about a three volt swing.

In the event that the large electrolytic capacitors and regulators are replaced but the regulators still aren't giving the proper measurements, it's possible that the small capacitors that "sandwich" the misbehaving regulator are to blame. There is one small cap before and after each regulator (usually a 0.1uF before and a 10uF after) which are there to keep that regulator from going haywire. It's unlikely but entirely possible that one of those caps are bad.

Some other symptoms:

If your Data Entry slider is working but the values are dancing around by themselves, check for ripple on the +12 and -12 analog lines. This is usually coupled with excessive buzzing or other anomalies on the audio outputs. This would also affect the Volume slider, Mod Wheel, and Pitch Bend, although it's harder to detect. If you have all those symptoms but the floppy drive seems to work perfectly then you might suspect -12 is wonky, whereas if you have all those symptoms and the floppy drive doesn't seem quite right then you might suspect +12 is out of sorts.

If everything seems to work in general, but you experience lots of random crashes, lockups, etc. then you might want to check LM2925. If (while logic probing the CPU, say) the RESET line is erratic or not blipping low and sticking high on boot then also check into LM2925.

And lastly: Don't forget, solid grounds are equally as important as proper voltage. This is an often overlooked fact.

I think that about covers most of the common power supply woes in the Ensoniq VFX.

Addendum: Incidentally, there are a couple other voltage regulators in the VFX that are *not* on the PSU. While +VU is unregulated as it's passed off the PSU, it eventually is regulated on the keyboard processor board by an LM2926 (which is similar to the LM2925 on the PSU) which supplies both +5VDC and RESET to all the chips on the keyboard processor board.

Likewise, there are a few chips on the mainboard which derive their +5VDC from the +12VDC analog rail which is regulated by a 7805 over by the analog circuitry on the right side. One other regulator can be found close to where the power header is on the mainboard.

KEYBOARD TROUBLESHOOTING

The obvious problem is the very well-known calibration error. Fixing that is simply a matter of reading the Service Bulletin that's posted way earlier in this thread. Take off a few center keys, solder flaky pins, replace keys, done. Also, putting the ribbon cable in backward after servicing the keyboard and popping a couple fuses is semi-typical. Those two things cover the good majority of keyboard issues, but I'll expound upon some other issues in detail.

First thing is to make sure there is continuity over the 20-pin keyboard ribbon cable. Test each and every pin on both ends, and make sure the PCB headers on the keyboard processor and mainboard are free of oxidation/corrosion. If that checks out, make sure the keyboard cable is routed away from and around the GLU chip. It's the square chip (gate array) in the brown PLCC socket on the mainboard.

No joy? As mentioned in the last post, +VU is important for the keyboard processing unit. Put your red probe on pin 1 on LM2926 and your black probe on pin 3 to make sure it's 8V DC or higher. Move the red probe onto pin 2 to make sure +5VDC is being output.

Motorola 68HC11A is something to focus on a lot. It's basically a CPU/UART combo unit that controls the processing and communication with the mainboard. If the ribbon cable is good, any bad serial communication with the keyboard is probably going to be traced back to this chip. Don't take that to mean that all serial communication problems are the keyboard's fault, though. Trace the serial path all the way from the DUART on the mainboard to the display header into the keyboard ribbon cable and see if it disappears into the 68HC11. Also, if one or both footswitches are dead then this chip a good place to look.

Here are some key things to focus on with the MC68HC11A1P:

Pin 39 should see the RESET sent by the LM2926. It works the same way as the CPU on the mainboard -- blip low then stuck high when booted.

Pins 48 should be +5VDC, as generated by LM2926. Pin 23 should be grounded.

Pins 42/43 should occasionally show bursts of serial traffic. Pressing buttons on the display keypad should generate some blips on the serial communication lines.

Pin 6 should show the same activity that pin 4 on the keyboard ribbon cable should show which, in turn, should be the same as pin 15 of the DUART on the mainboard.

Pins 29/30 are of critical importance and are tied into the 8 MHz clock from the crystal. Unlike some other MCUs in the VFX, this chip has two clock pins, and so both pins need to be connected to the crystal and pulsing. If not, measure 74HC00 pin 2 for 8 MHz. If that is not present either, there's a problem with the crystal or its related circuit, so look at the crystal itself as well as C1, C2 and R1 which are mostly obscured by the black heatsink that LM2926 is cooled by.

Pin 30 on 68HC11 is XTAL which basically "rings the bell" of the crystal and puts it into oscillation. The crystal then acts as an external clock whose timing is fed back into 68HC11 on pin 29 (EXTAL).

R1 should have one end connected to EXTAL (pin 29), the other end to XTAL (pin 30). The crystal should have one leg connected to one side of R1 (thereby connecting to EXTAL) and the other leg to the other side of R1 (thereby connecting to XTAL). The crystal should also have one leg connected to one side of C1 (and thereby XTAL), and the other leg to one side of C2 (and thereby EXTAL). The other sides of C1 and C2 should be ground.

Assuming that all looks correct, the 8 MHz crystal is cheap and easy to find, so replacement is a trivial issue. One off-the-wall idea to try if the crystal seems dead is to take it out and put it back in reversed. The crystal has no polarity, but might be "shaken up" enough to live again.

As far as the footswitches go, check pins 7 and 9 on the keyboard ribbon cable for activity when a footswitch is in use. If so, check pins 7 and 8 on 68HC11. Both signals should ordinarily be high, but one will go low when the corresponding footswitch is hit. Also double-check that pin 22 is stuck high at +5V and pin 21 is stuck low (grounded) as these two pins are compared to pins 7 and 8 as reference points.

Another semi-unlikely but entirely possible problem is that the keyboard EPROM is shot. Put the EPROM in an EPROM reader and generate its checksum (4B72 on v2.33). If in doubt, the EPROMs for the VFX SD can be found toward the top of this thread.

Probably the most mysterious chip in the whole synth is the Motorola S38BC010PS01 (aka KPC chip). It appears to do most of the low-level interaction with the actual inductance keys themselves (key note value, aftertouch, velocity, etc.), but replacing a defective one will prove to be a chore because they don't seem to exist outside of Ensoniq keyboards.

In order to even operate at all, the KPC needs a few easily checked things, though:

Pin 24 needs to get +5VDC, and pin 12 needs to be grounded.
Pin 8 is the clock input, but verifying clock on 68HC11 should be done before addressing any clock issues on KPC.

If 6HC11 is clocked correctly but there's still no clock on KPC pin 8 then take a close look at 74HC00 because chances are that pin 3 will be floating or stuck.

Edit (3/2017):

There is something very important to note about the 68HC11. While 68HC11 chips are pretty easy to find, the particular revision used on the KPC board contains a built-in EEPROM. What this means is that it has been programmed with special code and some random, generic 68HC11 may not work as a replacement. Unless the 68HC11 already contains the Ensoniq code or the EEPROM inside the OEM EEPROM is read out and programmed into the donor 68HC11 then things may not work as designed.

The good news is that multiple Ensoniq products (SQ80, et al) use a similar or identical 68HC11, so they can be found without too much difficulty. The other good news is that the EEPROM in a generic 68HC11 of the same revision can be rewritten with the Ensoniq code (provided you have access to the original Ensoniq code to write in the first place). Long story made short, the 6HC11 in the VFX/VFX SD/SD-1 series is specialized but not nearly as impossible to replace as the S38BC010PS01.

Also, the ICs on the KPC board do not like to operate without being physically connected to the keybed. Unless you have access to the special keyboard simulation test board then it will very likely be necessary to have the full keyboard assembly connected to the mainboard as you troubleshoot a large portion of the VFX system.


KPC BOARD OPERATION

Motorola 68HC11A1P is the main brain in the keyboard processing board:

Port A selects which group of music keys are currently to be scanned and also reads in the state of the footswitches
Port B is addressing (exclusively)
Port C is address/data lines which gets data to/from the EPROM and KPC chip
Port D is serial transfer to/from the mainboard and the display/keypad PCBs

74LS373 is a latching IC that demultiplexes Port C of 68HC11 into separate data and address streams.

The EPROM (27C256) contains the instructions for the operation of 68HC11.

The KPC CHIP [U3] handles the reading of the actual music keyboard:

F0~F3 are output lines coming from the keyboard, to be decoded by the KPC chip and the result is forwarded to the 68HC11; the result of F0~F3 is dependant upon which keys are currently being driven by 68HC11 Port A (CS0~CS3).

68HC11 processes both the display/keypad button and music key data and then sends it to the mainboard as serial data.

As far as the actual inductance functionality goes, this thread explains the theory of operation:
http://electronics.stackexchange.com/qu ... d-function

Hyde wrote:I have an EPS16+ with the same self test mode problem. I have a rack mounted unit, as well. I plan to side by side them to see if I can track cdg Airport transfer down the source of the problem, unless I can find a mainboard before then. diagnostics & repair must be more costly than to buy another unit...

Hello,
Yes it will cost you much more expensive than bought a new one I think!
Not to mention the loss of time!
Good luck!