Discussion about the Korg PolySix synthesizer

Hi!

I have just released with my MG (LFO) seems to be running very slowly 
with the frequency knob at 10. Before I dip in with my soldering iron 
does anyone know how the MG oscillator works? I have studied the 
schematic and there is nothing in the design which looks remotely like 
one! I believe the MG should run at 50Hz flat out. My 'scope says just 
over 3Hz when I measure at the VCA MOD test point. There is an RC 
network connected to the gate of Q1 on the KLM-367 which possibly 
triggers a voltage at the drain over a time period. The RC nework 
values (R27, C19) 10K, 0.33uF equate to a frequency of just over 48Hz, 
which looks promising. Can anyone shed any light on how this bit of the 
P6 works?

Thanks,
(A)

The oscillator design is of the multivibrator type. The not so common
thing is that it is voltage controlled.
The control voltage is buffered by IC13. it is converted to current
by R16, R69, TH1 combination (temperature compensated by use of TH1).
This current is mirrored by dual-transistor (for good match) Q3. The
mirrored current is fed into control input of transconductance amplifier
1/2 of IC21. Download the data sheet of this IC to find out how it
works. LM13700, NE5517, CA3080 are similar devices. It is an amplifier
with voltage input (high impedance) and current output (high impedance),
hence the name transconductance (current divided by voltage is called
conductance, the reciproke of resistance). So you input a voltage and
get a current on the output. That is the reason why timing capacitor
C19 does not have any resistor associated with. The transconductance
(= similar to gain of opamps) is controlled by the current which is
drawn out of the control pin (pin16).
One half of IC16 is the other vital part of the multivibrator circuit,
it acts as the schmitt trigger (positive feedback).
How does it work?
Lets assume any voltage level on IC16/pin6 after power on. It will be
more or less than the voltage on IC16/pin5. Depending whether it is
more or less the output of IC16 will be positive or negative saturated,
which is about +Vcc -1.5V or -Vcc +1.5V (+/-13.5V). This voltage is
divided by resistor divider R53 - R52 and fed into the transconductance
amp. Depending on the control current the output of IC21 delivers a
proportional current. This current charges the timing capacitor C19
developing a linear ramp voltage. The high impedance voltage is buffered
by JFET Q1 and fed back to IC16 input. When it reaches the same level
as voltage on IC16/pin5 the output switches to other saturation level
and C19 is being discharged until IC16/pin6 trips pin5 and the output
reverses polarity again (multivibrator principle). Since the output
current of IC21 is controlled by the control current, which in turn is
determined by the "MP SPEED" control voltage, the oscillation frequency
is controlled (larger current charges/discharges the timing capacitor
faster). Is it clear?
What can go wrong? - Anything 8-/.
I assume you did the adjustment procedure for MG speed (VR5). The
service manual says, "If these values (50Hz) cannot be obtained after
replacing IC LM13600, adjust R78 and R52 respectively."
First check the control voltage (IC13/pin7). It should reach -5 .. +5V.
Next check the output swing of IC16 (pin7) it should be at least +/-13V.
Next compare voltages before and after the buffer Q1; should be (nearly)
equal.
More measurements are not so easy because it is all current driven; you
cannot simply measure voltages. Best thing to try is replacing components
(C19, IC21, Q3, maybe TH1). If you dont find a dual transistor you can
use two discretes of the same type (would be good of the same lot).

Good luck!

Johannes


Andrew Jury wrote:

> Hi!
> 
> I have just released with my MG (LFO) seems to be running very slowly 
> with the frequency knob at 10. Before I dip in with my soldering iron 
> does anyone know how the MG oscillator works? I have studied the 
> schematic and there is nothing in the design which looks remotely like 
> one! I believe the MG should run at 50Hz flat out. My 'scope says just 
> over 3Hz when I measure at the VCA MOD test point. There is an RC 
> network connected to the gate of Q1 on the KLM-367 which possibly 
> triggers a voltage at the drain over a time period. The RC nework 
> values (R27, C19) 10K, 0.33uF equate to a frequency of just over 48Hz, 
> which looks promising. Can anyone shed any light on how this bit of the 
> P6 works?
> 
> Thanks,
> (A)
>

Johannes,

This is an extremely comprehensive answer. Many thanks for taking the 
time to explain. Let me read carefully, disgest and perform some more 
tests on the LFO. I'll let you know what I find!

Best wishes,
Andy

--- In PolySix@yahoogroups.com, Johannes Hausensteiner <johau@...> 
wrote:
>
> The oscillator design is of the multivibrator type. The not so 
common
> thing is that it is voltage controlled.
> The control voltage is buffered by IC13. it is converted to current
> by R16, R69, TH1 combination (temperature compensated by use of 
TH1).
> This current is mirrored by dual-transistor (for good match) Q3. The
> mirrored current is fed into control input of transconductance 
amplifier
> 1/2 of IC21. Download the data sheet of this IC to find out how it
> works. LM13700, NE5517, CA3080 are similar devices. It is an 
amplifier
> with voltage input (high impedance) and current output (high 
impedance),
> hence the name transconductance (current divided by voltage is 
called
> conductance, the reciproke of resistance). So you input a voltage 
and
> get a current on the output. That is the reason why timing capacitor
> C19 does not have any resistor associated with. The transconductance
> (= similar to gain of opamps) is controlled by the current which is
> drawn out of the control pin (pin16).
> One half of IC16 is the other vital part of the multivibrator 
circuit,
> it acts as the schmitt trigger (positive feedback).
> How does it work?
> Lets assume any voltage level on IC16/pin6 after power on. It will 
be
> more or less than the voltage on IC16/pin5. Depending whether it is
> more or less the output of IC16 will be positive or negative 
saturated,
> which is about +Vcc -1.5V or -Vcc +1.5V (+/-13.5V). This voltage is
> divided by resistor divider R53 - R52 and fed into the 
transconductance
> amp. Depending on the control current the output of IC21 delivers a
> proportional current. This current charges the timing capacitor C19
> developing a linear ramp voltage. The high impedance voltage is 
buffered
> by JFET Q1 and fed back to IC16 input. When it reaches the same 
level
> as voltage on IC16/pin5 the output switches to other saturation 
level
> and C19 is being discharged until IC16/pin6 trips pin5 and the 
output
> reverses polarity again (multivibrator principle). Since the output
> current of IC21 is controlled by the control current, which in turn 
is
> determined by the "MP SPEED" control voltage, the oscillation 
frequency
> is controlled (larger current charges/discharges the timing 
capacitor
> faster). Is it clear?
> What can go wrong? - Anything 8-/.
> I assume you did the adjustment procedure for MG speed (VR5). The
> service manual says, "If these values (50Hz) cannot be obtained 
after
> replacing IC LM13600, adjust R78 and R52 respectively."
> First check the control voltage (IC13/pin7). It should reach -5 .. 
+5V.
> Next check the output swing of IC16 (pin7) it should be at least +/-
13V.
> Next compare voltages before and after the buffer Q1; should be 
(nearly)
> equal.
> More measurements are not so easy because it is all current driven; 
you
> cannot simply measure voltages. Best thing to try is replacing 
components
> (C19, IC21, Q3, maybe TH1). If you dont find a dual transistor you 
can
> use two discretes of the same type (would be good of the same lot).
> 
> Good luck!
> 
> Johannes
> 
> 
> Andrew Jury wrote:
> > Hi!
> > 
> > I have just released with my MG (LFO) seems to be running very 
slowly 
> > with the frequency knob at 10. Before I dip in with my soldering 
iron 
> > does anyone know how the MG oscillator works? I have studied the 
> > schematic and there is nothing in the design which looks remotely 
like 
> > one! I believe the MG should run at 50Hz flat out. My 'scope says 
just 
> > over 3Hz when I measure at the VCA MOD test point. There is an RC 
> > network connected to the gate of Q1 on the KLM-367 which possibly 
> > triggers a voltage at the drain over a time period. The RC nework 
> > values (R27, C19) 10K, 0.33uF equate to a frequency of just over 
48Hz, 
> > which looks promising. Can anyone shed any light on how this bit 
of the 

> > P6 works?
> > 
> > Thanks,
> > (A)
> >
>

Johannes,

After a major fault finding session I would like to report back the 
following:

The first thing I did was be rather lazy and swap the transconductanc 
op-amp IC21 to see if it made any difference; no! I then checked the 
FET Q1 s-g, g-d and g-d. Similarly I checked the matched pair of bi-
polars at Q3 c-b, b-e and c-e. No problems. I then very carefully 
checked all the resistors in circuit, including VR5, which all 
checked OK. I tested C19 which 0.339uF. Not bad for an electrolytic 
which is over 25 years old! The MG was still ocillating flat out at 
3.4hz even with VR5 at it uppermost setting. I then checked the dc 
voltage present at pin 7 of IC13. Changing the MG frequency control 
from 0 (+ 4.98V) to 10 (-0.17v). But wait? Can this be right? I 
checked pin 12 of IC19 which is the mux output X3; same voltage 
reading. Working back to IC2 on the KLM-370 board I checked the mux 
input X2 at pin 15. The voltage swings accurately around the 
frequency pot wiper from +5 to -5v. However, taking a reading at the 
muxed data output on pin 3 of IC2 the voltage will only swing as 
mentioned before from +4.98 to -0.17v. Now forgive me if I am wrong 
but I understood the HA14051 worked like this. You appliy a voltage 
to one of the 8 mux inputs marked X0-7. The voltage change is 
detected and a bitmap is set on data address lines A, B and C 
according to which pin was strobed. At the other end this bitmap is 
decoded the data read on the receiving mux and the appropriate signal 
(in this case a voltage) will appear at the correct pin as determind 
by the bitmap. The fact that the voltage doesn't swing correctly 
between the two poles of the pot would suggest a fault with the 
transmitting HA14051? Am I right in assuming this? If the address 
lines were cut/missing then some of the muxed output would be never 
selected and hence always at 0 volts. This might also account for the 
other problem in that the ADS bit of the EG works (sort of!) but the 
release pot does nothing! (Again the voltage swings about the same as 
the MG frequency control)

Your expert opinion on this matter is greatly anticipated!

Best regards,
Andy

--- In PolySix@yahoogroups.com, Johannes Hausensteiner <johau@...> 
wrote:
>
> The oscillator design is of the multivibrator type. The not so 
common
> thing is that it is voltage controlled.
> The control voltage is buffered by IC13. it is converted to current
> by R16, R69, TH1 combination (temperature compensated by use of 
TH1).
> This current is mirrored by dual-transistor (for good match) Q3. The
> mirrored current is fed into control input of transconductance 
amplifier
> 1/2 of IC21. Download the data sheet of this IC to find out how it
> works. LM13700, NE5517, CA3080 are similar devices. It is an 
amplifier
> with voltage input (high impedance) and current output (high 
impedance),
> hence the name transconductance (current divided by voltage is 
called
> conductance, the reciproke of resistance). So you input a voltage 
and
> get a current on the output. That is the reason why timing capacitor
> C19 does not have any resistor associated with. The transconductance
> (= similar to gain of opamps) is controlled by the current which is
> drawn out of the control pin (pin16).
> One half of IC16 is the other vital part of the multivibrator 
circuit,
> it acts as the schmitt trigger (positive feedback).
> How does it work?
> Lets assume any voltage level on IC16/pin6 after power on. It will 
be
> more or less than the voltage on IC16/pin5. Depending whether it is
> more or less the output of IC16 will be positive or negative 
saturated,
> which is about +Vcc -1.5V or -Vcc +1.5V (+/-13.5V). This voltage is
> divided by resistor divider R53 - R52 and fed into the 
transconductance
> amp. Depending on the control current the output of IC21 delivers a
> proportional current. This current charges the timing capacitor C19
> developing a linear ramp voltage. The high impedance voltage is 
buffered
> by JFET Q1 and fed back to IC16 input. When it reaches the same 
level
> as voltage on IC16/pin5 the output switches to other saturation 
level
> and C19 is being discharged until IC16/pin6 trips pin5 and the 
output
> reverses polarity again (multivibrator principle). Since the output
> current of IC21 is controlled by the control current, which in turn 
is
> determined by the "MP SPEED" control voltage, the oscillation 
frequency
> is controlled (larger current charges/discharges the timing 
capacitor
> faster). Is it clear?
> What can go wrong? - Anything 8-/.
> I assume you did the adjustment procedure for MG speed (VR5). The
> service manual says, "If these values (50Hz) cannot be obtained 
after
> replacing IC LM13600, adjust R78 and R52 respectively."
> First check the control voltage (IC13/pin7). It should reach -5 .. 
+5V.
> Next check the output swing of IC16 (pin7) it should be at least +/-
13V.
> Next compare voltages before and after the buffer Q1; should be 
(nearly)
> equal.
> More measurements are not so easy because it is all current driven; 
you
> cannot simply measure voltages. Best thing to try is replacing 
components
> (C19, IC21, Q3, maybe TH1). If you dont find a dual transistor you 
can
> use two discretes of the same type (would be good of the same lot).
> 
> Good luck!
> 
> Johannes
> 
> 
> Andrew Jury wrote:
> > Hi!
> > 
> > I have just released with my MG (LFO) seems to be running very 
slowly 
> > with the frequency knob at 10. Before I dip in with my soldering 
iron 
> > does anyone know how the MG oscillator works? I have studied the 
> > schematic and there is nothing in the design which looks remotely 
like 
> > one! I believe the MG should run at 50Hz flat out. My 'scope says 
just 
> > over 3Hz when I measure at the VCA MOD test point. There is an RC 
> > network connected to the gate of Q1 on the KLM-367 which possibly 
> > triggers a voltage at the drain over a time period. The RC nework 
> > values (R27, C19) 10K, 0.33uF equate to a frequency of just over 
48Hz, 
> > which looks promising. Can anyone shed any light on how this bit 
of the 

> > P6 works?
> > 
> > Thanks,
> > (A)
> >
>

--- In PolySix@yahoogroups.com, "Andrew Jury" <andy@...> wrote:
>
> Now forgive me if I am wrong 
> but I understood the HA14051 worked like this. You appliy a voltage 
> to one of the 8 mux inputs marked X0-7. The voltage change is 
> detected and a bitmap is set on data address lines A, B and C 
> according to which pin was strobed. At the other end this bitmap is 
> decoded the data read on the receiving mux and the appropriate signal 
> (in this case a voltage) will appear at the correct pin as determind 
> by the bitmap. The fact that the voltage doesn't swing correctly 
> between the two poles of the pot would suggest a fault with the 
> transmitting HA14051? 

That's how it works, but before you go suspecting the mux, are you sure you're getting 
the full range of digital data on A, B and C at both ends?  The at both ends bit being really 
important, on one of my KLM-367s I had the right signals present right up to within 1mm 
of the connector, at which point one of my mux selection signals became something 
entirely different thanks to the wonderful way the nicad redesigned the board.

Sorry about the over-zealous quote snipping, btw.

--
phil

Remember that MG is connected with frequency knob. If it runs low on
max setting then first try to look at dead lfo capacitors

Andrew,

Yes, the voltage must swing between -5V and +5V. It is not so easy
to tell whether it happens on the input side (KLM-369/370) or the
output side (KLM-367). Do the following:
* connect CH2 of your scope to testpoint "TP4 CLOCK" on KLM-367.
   It is right beneath the 8048 CPU, IC22. Set the trigger to CH2
   and adjust the trigger level to see a narrow pulse every 20msec.
* connect CH1 of scope to test point "TP2", which is right beneath
   the "RESET ADJ" poti, VR1.
* select MANUAL on the Polysix front panel and turn all knobs
   fully left. The scope reading should be mainly -5V. There may be
   some garbage at the end of the cycle.
* now turn each knob fully up and down again. You should see a
   small "bar" going from -5V to +5V. Depending on the knob the
   position within the cycle is different. The order is according
   to the outputs of 4051 multiplexers IC18, IC19.
   What you see is the analog CV /output/
* Move the CH1 probe to CN11/pin2, labeled "DATA". Here you will
   see a similar thing; the bars will be different in length, but
   the principle is the same. Again turn each knob up and down.
   Observe that each knob reaches the full span of -5 .. +5V.
   This is the /input/ side.
Now you can judge if the fault is a 4051 on KLM-369/370 or KLM-367.
If you have problems to interpret the scope screen take pictures and
email them. I will try to figure out what's wrong.

Note: I had it once or twice that the leaked acid from the NiCd
battery crawled via the GND connection (thick black wire) from
KLM-367 to KLM-370 and did damage there. Check for this!

Good luck !

Johannes


Andrew Jury wrote:

> Johannes,
> 
> After a major fault finding session I would like to report back the 
> following:
> 
> The first thing I did was be rather lazy and swap the transconductanc 
> op-amp IC21 to see if it made any difference; no! I then checked the 
> FET Q1 s-g, g-d and g-d. Similarly I checked the matched pair of bi-
> polars at Q3 c-b, b-e and c-e. No problems. I then very carefully 
> checked all the resistors in circuit, including VR5, which all 
> checked OK. I tested C19 which 0.339uF. Not bad for an electrolytic 
> which is over 25 years old! The MG was still ocillating flat out at 
> 3.4hz even with VR5 at it uppermost setting. I then checked the dc 
> voltage present at pin 7 of IC13. Changing the MG frequency control 
> from 0 (+ 4.98V) to 10 (-0.17v). But wait? Can this be right? I 
> checked pin 12 of IC19 which is the mux output X3; same voltage 
> reading. Working back to IC2 on the KLM-370 board I checked the mux 
> input X2 at pin 15. The voltage swings accurately around the 
> frequency pot wiper from +5 to -5v. However, taking a reading at the 
> muxed data output on pin 3 of IC2 the voltage will only swing as 
> mentioned before from +4.98 to -0.17v. Now forgive me if I am wrong 
> but I understood the HA14051 worked like this. You appliy a voltage 
> to one of the 8 mux inputs marked X0-7. The voltage change is 
> detected and a bitmap is set on data address lines A, B and C 
> according to which pin was strobed. At the other end this bitmap is 
> decoded the data read on the receiving mux and the appropriate signal 
> (in this case a voltage) will appear at the correct pin as determind 
> by the bitmap. The fact that the voltage doesn't swing correctly 
> between the two poles of the pot would suggest a fault with the 
> transmitting HA14051? Am I right in assuming this? If the address 
> lines were cut/missing then some of the muxed output would be never 
> selected and hence always at 0 volts. This might also account for the 
> other problem in that the ADS bit of the EG works (sort of!) but the 
> release pot does nothing! (Again the voltage swings about the same as 
> the MG frequency control)
> 
> Your expert opinion on this matter is greatly anticipated!
> 
> Best regards,
> Andy
> 
> --- In PolySix@yahoogroups.com, Johannes Hausensteiner <johau@...> 
> wrote:
>> The oscillator design is of the multivibrator type. The not so 
> common
>> thing is that it is voltage controlled.
>> The control voltage is buffered by IC13. it is converted to current
>> by R16, R69, TH1 combination (temperature compensated by use of 
> TH1).
>> This current is mirrored by dual-transistor (for good match) Q3. The
>> mirrored current is fed into control input of transconductance 
> amplifier
>> 1/2 of IC21. Download the data sheet of this IC to find out how it
>> works. LM13700, NE5517, CA3080 are similar devices. It is an 
> amplifier
>> with voltage input (high impedance) and current output (high 
> impedance),
>> hence the name transconductance (current divided by voltage is 
> called
>> conductance, the reciproke of resistance). So you input a voltage 
> and
>> get a current on the output. That is the reason why timing capacitor
>> C19 does not have any resistor associated with. The transconductance
>> (= similar to gain of opamps) is controlled by the current which is
>> drawn out of the control pin (pin16).
>> One half of IC16 is the other vital part of the multivibrator 
> circuit,
>> it acts as the schmitt trigger (positive feedback).
>> How does it work?
>> Lets assume any voltage level on IC16/pin6 after power on. It will 
> be
>> more or less than the voltage on IC16/pin5. Depending whether it is
>> more or less the output of IC16 will be positive or negative 
> saturated,
>> which is about +Vcc -1.5V or -Vcc +1.5V (+/-13.5V). This voltage is
>> divided by resistor divider R53 - R52 and fed into the 
> transconductance
>> amp. Depending on the control current the output of IC21 delivers a
>> proportional current. This current charges the timing capacitor C19
>> developing a linear ramp voltage. The high impedance voltage is 
> buffered
>> by JFET Q1 and fed back to IC16 input. When it reaches the same 
> level
>> as voltage on IC16/pin5 the output switches to other saturation 
> level
>> and C19 is being discharged until IC16/pin6 trips pin5 and the 
> output
>> reverses polarity again (multivibrator principle). Since the output
>> current of IC21 is controlled by the control current, which in turn 
> is
>> determined by the "MP SPEED" control voltage, the oscillation 
> frequency
>> is controlled (larger current charges/discharges the timing 
> capacitor
>> faster). Is it clear?
>> What can go wrong? - Anything 8-/.
>> I assume you did the adjustment procedure for MG speed (VR5). The
>> service manual says, "If these values (50Hz) cannot be obtained 
> after
>> replacing IC LM13600, adjust R78 and R52 respectively."
>> First check the control voltage (IC13/pin7). It should reach -5 .. 
> +5V.
>> Next check the output swing of IC16 (pin7) it should be at least +/-
> 13V.
>> Next compare voltages before and after the buffer Q1; should be 
> (nearly)
>> equal.
>> More measurements are not so easy because it is all current driven; 
> you
>> cannot simply measure voltages. Best thing to try is replacing 
> components
>> (C19, IC21, Q3, maybe TH1). If you dont find a dual transistor you 
> can
>> use two discretes of the same type (would be good of the same lot).
>>
>> Good luck!
>>
>> Johannes
>>
>>
>> Andrew Jury wrote:
>>> Hi!
>>>
>>> I have just released with my MG (LFO) seems to be running very 
> slowly 
>>> with the frequency knob at 10. Before I dip in with my soldering 
> iron 
>>> does anyone know how the MG oscillator works? I have studied the 
>>> schematic and there is nothing in the design which looks remotely 
> like 
>>> one! I believe the MG should run at 50Hz flat out. My 'scope says 
> just 
>>> over 3Hz when I measure at the VCA MOD test point. There is an RC 
>>> network connected to the gate of Q1 on the KLM-367 which possibly 
>>> triggers a voltage at the drain over a time period. The RC nework 
>>> values (R27, C19) 10K, 0.33uF equate to a frequency of just over 
> 48Hz, 
>>> which looks promising. Can anyone shed any light on how this bit 
> of the 
>>> P6 works?
>>>
>>> Thanks,
>>> (A)
>>>
> 
> 
> 
> ------------------------------------
> 
> PolySix "Digiest" Page: http://www.acc.umu.se/~amber/Poly6Yahoo! Groups Links
> 
> 
>

Phil,

No, please do quote away! This is interesting because the KLM-367 
board on this P6 was pretty badly burnt with blue nickel cadmium 
pixie dust. It would make sense it suspect further broken tracks 
which I might of missed. I will check the data line and the inhibit 
continuity for the whole circuit and see if there are any breaks. Can 
you just confirm that the receiving mux's output should swing between 
+5 and -5 and not +5 and -0.17v! I guess I know what you are going 
say...

In my case I am assuming this is why I can't get the MG to oscillate 
above 3.4Mhz. If the swing reached -5v then I might get the full 50hz.

Thanks and I will let you know what gives...

Cheers,
Andy



--- In PolySix@yahoogroups.com, "synx508" <philyahoo@...> wrote:
>
> --- In PolySix@yahoogroups.com, "Andrew Jury" <andy@> wrote:
> >
> > Now forgive me if I am wrong 
> > but I understood the HA14051 worked like this. You appliy a 
voltage 
> > to one of the 8 mux inputs marked X0-7. The voltage change is 
> > detected and a bitmap is set on data address lines A, B and C 
> > according to which pin was strobed. At the other end this bitmap 
is 
> > decoded the data read on the receiving mux and the appropriate 
signal 
> > (in this case a voltage) will appear at the correct pin as 
determind 
> > by the bitmap. The fact that the voltage doesn't swing correctly 
> > between the two poles of the pot would suggest a fault with the 
> > transmitting HA14051? 
> 
> That's how it works, but before you go suspecting the mux, are you 
sure you're getting 
> the full range of digital data on A, B and C at both ends?  The at 
both ends bit being really 
> important, on one of my KLM-367s I had the right signals present 
right up to within 1mm 
> of the connector, at which point one of my mux selection signals 
became something 
> entirely different thanks to the wonderful way the nicad redesigned 
the board.

> 
> Sorry about the over-zealous quote snipping, btw.
> 
> --
> phil
>

Hi,

Seem to be getting nowhere fast with this one. I made the 
observations with the oscilloscope you mentioned and seemed to be 
able to reproduce the waveforms you described, if a little on the low 
voltage side.

However, careful study of the schematic reveals that the tansmitting 
and receiving mux are not directly connected. Which adds to the 
confusion! The data from the pin 3 of the 14051 on the control panel 
terminates on the non-inverting input of comparator IC6. The voltage 
being compared is that fed back from from the output of the D/A 
converter IC33. My guess is that when you turn one of the pots on the 
multiplexed bus the actually control is provided by the logic of the 
CPU programmed via pin T0. This would mean that the actually 
multiplexed voltages are actually controlled by the D/A converter 
rather than receiving the raw voltages directly from the transmit end 
(if that makes sense). Now here is the bombshell... The DC voltage at 
pin 4 of the D/A converter is always 0v (strictly 0.00v, like it is 
grounded, which it isn't!) The output of op-amp IC27 varies only a 
matter of millivolts when the controls are turned, which would 
explain the weird MG control and lack of release on the VCA. So I am 
guessing that the data bus d0-d7 is low/missing? I noticed on this 
board the 1408 has been replaced in the past with a DAC08. Looks pin 
compatible, but maybe not.

Oh, and you'll like this! I managed to purloin a working KLM-367 from 
a friend and transplanted it into my P6. It works fine showing all 
the peripheral circuitry is working ok. Back to the continuity 
testing...

Cheers,
Andy

Of course, this is the reason fo rall that multiplex stuff. The
analog controlo voltages are digitised (via the D/A converter)
under software control and saved in the battery backed memory.
Otherwise the patches could not be saved and recalled. And yes,
(of course) this is all under software control by the 8048 CPU.

The 1408 DAC chip (it IS compatible with the DAC08) has, like
many DACs of this design a current output. You will notice that
there is no resistor in series to the inverting input of the
following opamp as it is normally the case for an inverting
amplifier. IC27/pin2 (which is the same point as DAC/pin4) is
the /virtual ground/ of the opamp; this means the opamp does all
it can (like all opamps with negative feedback) to make the
voltage difference between its two inputs zero. So we know from
your measurement 0V (few mV) at DAC/pin4 that opamp IC27 is
working.
Comparator IC6 works in conjunction with the software and the DAC
similar like a neg. feedbacked opamp: the voltage between its two
inputs is minimised. It is not perfect (by design) because it is
a digital feedback loop, which means it has discrete steps of
+5-(-5)V / 256 steps, which happens to be ca. 39mV.
So the reason of your problem may lie in IC6, the DAC, or associated
components or the software. Since we know that the software works
(what a rare case!) it only can be that the data bus is corrupted,
which is the most likely reason in the case of a battery leaked
Polysix.

Do you get the full span (-5 .. +5V) for other control voltages?
What is the voltage difference between pin2 and pin3 of IC6?
Are there good logic levels on CPU/pin1 and on all of the 8 data
lines of the DAC?
Is the reference voltage of the DAC (pin14; generated by 2nd half
of IC27) OK?
In any case, if there is a problem with the DAC or comparator or
data lines /all/ CVs are affected. If it is only certain CVs then
the problem lies in the multiplexer and demultiplexer circuits
(4051 chips on KLM369/370 and KLM367).

Johannes

Andrew Jury wrote:

> Hi,
> 
> Seem to be getting nowhere fast with this one. I made the 
> observations with the oscilloscope you mentioned and seemed to be 
> able to reproduce the waveforms you described, if a little on the low 
> voltage side.
> 
> However, careful study of the schematic reveals that the tansmitting 
> and receiving mux are not directly connected. Which adds to the 
> confusion! The data from the pin 3 of the 14051 on the control panel 
> terminates on the non-inverting input of comparator IC6. The voltage 
> being compared is that fed back from from the output of the D/A 
> converter IC33. My guess is that when you turn one of the pots on the 
> multiplexed bus the actually control is provided by the logic of the 
> CPU programmed via pin T0. This would mean that the actually 
> multiplexed voltages are actually controlled by the D/A converter 
> rather than receiving the raw voltages directly from the transmit end 
> (if that makes sense). Now here is the bombshell... The DC voltage at 
> pin 4 of the D/A converter is always 0v (strictly 0.00v, like it is 
> grounded, which it isn't!) The output of op-amp IC27 varies only a 
> matter of millivolts when the controls are turned, which would 
> explain the weird MG control and lack of release on the VCA. So I am 
> guessing that the data bus d0-d7 is low/missing? I noticed on this 
> board the 1408 has been replaced in the past with a DAC08. Looks pin 
> compatible, but maybe not.
> 
> Oh, and you'll like this! I managed to purloin a working KLM-367 from 
> a friend and transplanted it into my P6. It works fine showing all 
> the peripheral circuitry is working ok. Back to the continuity 
> testing...
> 
> Cheers,
> Andy
> 
> 
> ------------------------------------
> 
> PolySix "Digiest" Page: http://www.acc.umu.se/~amber/Poly6Yahoo! Groups Links
> 
> 
>

Hi,

This one is really turning into a pig! 

Here's what I have done and still I have a sweep voltage on each 
demux output from -4.58 to 0.16v. Just imagine what this P6 sounds 
like with only partial control over it's functions!

1. An agressive continuity check on data bus lines DO0-7. All OK
2. The following ICs replaced IC6, IC33, IC27, IC18, IC19
3. The P6 works fine when a replacement KLM-367 is fitted from a 
working unit.

I must admit it sounds like one of the positive supply rails is 
missing. However, all the silicon seems to be supplied with the right 
voltages. All I can assume is that the inbound muxed voltage lines at 
X (pin 3) are wrong and this implies that the d/a converter is not 
playing back the correct voltage steps when the various different 
bitmaps are applied to the data bus. 

Why this one sided representation of the input voltage? Bah! Will my 
hair ever grow back!?

Cheers,
Andy

--- In PolySix@yahoogroups.com, Johannes Hausensteiner <johau@...> 
wrote:
>
> Of course, this is the reason fo rall that multiplex stuff. The
> analog controlo voltages are digitised (via the D/A converter)
> under software control and saved in the battery backed memory.
> Otherwise the patches could not be saved and recalled. And yes,
> (of course) this is all under software control by the 8048 CPU.
> 
> The 1408 DAC chip (it IS compatible with the DAC08) has, like
> many DACs of this design a current output. You will notice that
> there is no resistor in series to the inverting input of the
> following opamp as it is normally the case for an inverting
> amplifier. IC27/pin2 (which is the same point as DAC/pin4) is
> the /virtual ground/ of the opamp; this means the opamp does all
> it can (like all opamps with negative feedback) to make the
> voltage difference between its two inputs zero. So we know from
> your measurement 0V (few mV) at DAC/pin4 that opamp IC27 is
> working.
> Comparator IC6 works in conjunction with the software and the DAC
> similar like a neg. feedbacked opamp: the voltage between its two
> inputs is minimised. It is not perfect (by design) because it is
> a digital feedback loop, which means it has discrete steps of
> +5-(-5)V / 256 steps, which happens to be ca. 39mV.
> So the reason of your problem may lie in IC6, the DAC, or associated
> components or the software. Since we know that the software works
> (what a rare case!) it only can be that the data bus is corrupted,
> which is the most likely reason in the case of a battery leaked
> Polysix.
> 
> Do you get the full span (-5 .. +5V) for other control voltages?
> What is the voltage difference between pin2 and pin3 of IC6?
> Are there good logic levels on CPU/pin1 and on all of the 8 data
> lines of the DAC?
> Is the reference voltage of the DAC (pin14; generated by 2nd half
> of IC27) OK?
> In any case, if there is a problem with the DAC or comparator or
> data lines /all/ CVs are affected. If it is only certain CVs then
> the problem lies in the multiplexer and demultiplexer circuits
> (4051 chips on KLM369/370 and KLM367).
> 
> Johannes
> 
> Andrew Jury wrote:
> > Hi,
> > 
> > Seem to be getting nowhere fast with this one. I made the 
> > observations with the oscilloscope you mentioned and seemed to be 
> > able to reproduce the waveforms you described, if a little on the 
low 
> > voltage side.
> > 
> > However, careful study of the schematic reveals that the 
tansmitting 
> > and receiving mux are not directly connected. Which adds to the 
> > confusion! The data from the pin 3 of the 14051 on the control 
panel 
> > terminates on the non-inverting input of comparator IC6. The 
voltage 
> > being compared is that fed back from from the output of the D/A 
> > converter IC33. My guess is that when you turn one of the pots on 
the 
> > multiplexed bus the actually control is provided by the logic of 
the 
> > CPU programmed via pin T0. This would mean that the actually 
> > multiplexed voltages are actually controlled by the D/A converter 
> > rather than receiving the raw voltages directly from the transmit 
end 
> > (if that makes sense). Now here is the bombshell... The DC 
voltage at 
> > pin 4 of the D/A converter is always 0v (strictly 0.00v, like it 
is 
> > grounded, which it isn't!) The output of op-amp IC27 varies only 
a 
> > matter of millivolts when the controls are turned, which would 
> > explain the weird MG control and lack of release on the VCA. So I 
am 
> > guessing that the data bus d0-d7 is low/missing? I noticed on 
this 
> > board the 1408 has been replaced in the past with a DAC08. Looks 
pin 
> > compatible, but maybe not.
> > 
> > Oh, and you'll like this! I managed to purloin a working KLM-367 
from 
> > a friend and transplanted it into my P6. It works fine showing 
all 
> > the peripheral circuitry is working ok. Back to the continuity 
> > testing...
> > 
> > Cheers,
> > Andy
> > 
> > 
> > ------------------------------------
> > 
> > PolySix "Digiest" Page: http://www.acc.umu.se/~amber/Poly6Yahoo! 
Groups Links

> > 
> > 
> >
>

Hi,

If you have been following the sorry tale of my P6 repairs then a 
small word of warning for you. If you ever need to change the d/a 
converter (IC33) or come across a non-working one where it has been 
changed beware! My KLM-367 had a DAC08 fitted. This is supposed to be 
pin compatible with the HA17408P which it replaces. IT ISN'T!

Pin 1 on the original chip is not connected, however, on the 
replacement it needs to be grounded! A small blob of solder 
connecting pins 1 and 2 and the KLM-367 is now back in full working 
order.

One for the field notebook I fear...

Cheers,
Andy

P.S. The 'Alldatasheet.com' website is a lifesaver!

--- In PolySix@yahoogroups.com, "Andrew Jury" <andy@...> wrote:
>
> Hi,
> 
> This one is really turning into a pig! 
> 
> Here's what I have done and still I have a sweep voltage on each 
> demux output from -4.58 to 0.16v. Just imagine what this P6 sounds 
> like with only partial control over it's functions!
> 
> 1. An agressive continuity check on data bus lines DO0-7. All OK
> 2. The following ICs replaced IC6, IC33, IC27, IC18, IC19
> 3. The P6 works fine when a replacement KLM-367 is fitted from a 
> working unit.
> 
> I must admit it sounds like one of the positive supply rails is 
> missing. However, all the silicon seems to be supplied with the 
right 
> voltages. All I can assume is that the inbound muxed voltage lines 
at 
> X (pin 3) are wrong and this implies that the d/a converter is not 
> playing back the correct voltage steps when the various different 
> bitmaps are applied to the data bus. 
> 
> Why this one sided representation of the input voltage? Bah! Will 
my 
> hair ever grow back!?
> 
> Cheers,
> Andy
> 
> --- In PolySix@yahoogroups.com, Johannes Hausensteiner <johau@> 
> wrote:
> >
> > Of course, this is the reason fo rall that multiplex stuff. The
> > analog controlo voltages are digitised (via the D/A converter)
> > under software control and saved in the battery backed memory.
> > Otherwise the patches could not be saved and recalled. And yes,
> > (of course) this is all under software control by the 8048 CPU.
> > 
> > The 1408 DAC chip (it IS compatible with the DAC08) has, like
> > many DACs of this design a current output. You will notice that
> > there is no resistor in series to the inverting input of the
> > following opamp as it is normally the case for an inverting
> > amplifier. IC27/pin2 (which is the same point as DAC/pin4) is
> > the /virtual ground/ of the opamp; this means the opamp does all
> > it can (like all opamps with negative feedback) to make the
> > voltage difference between its two inputs zero. So we know from
> > your measurement 0V (few mV) at DAC/pin4 that opamp IC27 is
> > working.
> > Comparator IC6 works in conjunction with the software and the DAC
> > similar like a neg. feedbacked opamp: the voltage between its two
> > inputs is minimised. It is not perfect (by design) because it is
> > a digital feedback loop, which means it has discrete steps of
> > +5-(-5)V / 256 steps, which happens to be ca. 39mV.
> > So the reason of your problem may lie in IC6, the DAC, or 
associated
> > components or the software. Since we know that the software works
> > (what a rare case!) it only can be that the data bus is corrupted,
> > which is the most likely reason in the case of a battery leaked
> > Polysix.
> > 
> > Do you get the full span (-5 .. +5V) for other control voltages?
> > What is the voltage difference between pin2 and pin3 of IC6?
> > Are there good logic levels on CPU/pin1 and on all of the 8 data
> > lines of the DAC?
> > Is the reference voltage of the DAC (pin14; generated by 2nd half
> > of IC27) OK?
> > In any case, if there is a problem with the DAC or comparator or
> > data lines /all/ CVs are affected. If it is only certain CVs then
> > the problem lies in the multiplexer and demultiplexer circuits
> > (4051 chips on KLM369/370 and KLM367).
> > 
> > Johannes
> > 
> > Andrew Jury wrote:
> > > Hi,
> > > 
> > > Seem to be getting nowhere fast with this one. I made the 
> > > observations with the oscilloscope you mentioned and seemed to 
be 
> > > able to reproduce the waveforms you described, if a little on 
the 
> low 
> > > voltage side.
> > > 
> > > However, careful study of the schematic reveals that the 
> tansmitting 
> > > and receiving mux are not directly connected. Which adds to the 
> > > confusion! The data from the pin 3 of the 14051 on the control 
> panel 
> > > terminates on the non-inverting input of comparator IC6. The 
> voltage 
> > > being compared is that fed back from from the output of the D/A 
> > > converter IC33. My guess is that when you turn one of the pots 
on 
> the 
> > > multiplexed bus the actually control is provided by the logic 
of 
> the 
> > > CPU programmed via pin T0. This would mean that the actually 
> > > multiplexed voltages are actually controlled by the D/A 
converter 
> > > rather than receiving the raw voltages directly from the 
transmit 
> end 
> > > (if that makes sense). Now here is the bombshell... The DC 
> voltage at 
> > > pin 4 of the D/A converter is always 0v (strictly 0.00v, like 
it 
> is 
> > > grounded, which it isn't!) The output of op-amp IC27 varies 
only 
> a 
> > > matter of millivolts when the controls are turned, which would 
> > > explain the weird MG control and lack of release on the VCA. So 
I 
> am 
> > > guessing that the data bus d0-d7 is low/missing? I noticed on 
> this 
> > > board the 1408 has been replaced in the past with a DAC08. 
Looks 
> pin 
> > > compatible, but maybe not.
> > > 
> > > Oh, and you'll like this! I managed to purloin a working KLM-
367 
> from 
> > > a friend and transplanted it into my P6. It works fine showing 
> all 
> > > the peripheral circuitry is working ok. Back to the continuity 
> > > testing...
> > > 
> > > Cheers,
> > > Andy
> > > 
> > > 
> > > ------------------------------------
> > > 
> > > PolySix "Digiest" Page: 
http://www.acc.umu.se/~amber/Poly6Yahoo! 

> Groups Links
> > > 
> > > 
> > >
> >
>