Topics

Weak 9872CT Plotter Platen

Martin Hepperle
 

My 9872 sucks! ... not enough!

After fixing the mechanical problem with the bent chassis I now have a (hopefully) final problem.
 
The electrostatic platen does not work properly.  It provides only a very weak suction force so that the paper starts to slide. Also the force seems to be a bit higher when switched on and then quickly drops to almost zero within a few seconds.
 
Of course I had to take the platen off when I was taking the plotter apart for fixing the chassis but I was very careful.
The white cover sheet already had and still has a 2-3 deeper scratches going down to the copper/epoxy plate. 
The only modification from my side was to glue the lower right corner (about 20x20mm) of the white sheet down with rubber cement because it was starting to lift off the Epoxy board.
(BTW: this was not the best idea because the solvent in the rubber contact cement caused some small wrinkles in this corner of the sheet.)
 
When I measure the voltage with and without the platen connected I obtain about 480 Volts when ON, dropping to 20 Volts and then dropping further when switched OFF. This seems to be normal.
I also tested most of the components on the platen control board so I assume that the power supply part is fine.
 
Across the platen connector I measure infinite resistance (when plugged out).
I am not able to measure the capacity lacking test equipment for such low values.
There is also infinite resistance between the platen connector wires and the metal frame of the platen.
There the comb-like copper traces are connected to the wires - I measure zero resistance between one wire and one comb
All this seems to be reasonable.
 
The platen is built of interleaved +/- combs on the Epoxy PCB board (covered with the white plastic sheet).
 
There are a 2-3 deeper scratches in the plastic cover sheet so that in two or three spots the copper trace is open to the atmosphere.
Could this be a problem, even if a smaller sheet of paper is not reaching these open spots?
 
Unfortunately the theory of operation of the platen is not explained in the handbooks or in the HP-Journal article for the 9872/7221 plotters.
Is there someone who knows how it actually works and whether it is sensitive to cover sheet damage?

Any ideas or experience with defective platens for 9872/7221 plotters?

Thanks,
Martin

 

Martin,

The platen works electrostatic.
When you want to measure the isolation resistance you'll a Megger (isolation resistance meter) those instruments use 500 or 1000V DC to measure the resistance. This gives you a good idea about the condition of the platen. Next thing to check is the leak current of the capacitor (s) because the platen supply only can source a very little current every leak will decrease the output voltage.

-Rik

-Rik




On Fri, Jan 17, 2020 at 12:27 PM +0100, "Martin Hepperle" <Martin.Hepperle@...> wrote:

My 9872 sucks! ... not enough!

After fixing the mechanical problem with the bent chassis I now have a (hopefully) final problem.
 
The electrostatic platen does not work properly.  It provides only a very weak suction force so that the paper starts to slide. Also the force seems to be a bit higher when switched on and then quickly drops to almost zero within a few seconds.
 
Of course I had to take the platen off when I was taking the plotter apart for fixing the chassis but I was very careful.
The white cover sheet already had and still has a 2-3 deeper scratches going down to the copper/epoxy plate. 
The only modification from my side was to glue the lower right corner (about 20x20mm) of the white sheet down with rubber cement because it was starting to lift off the Epoxy board.
(BTW: this was not the best idea because the solvent in the rubber contact cement caused some small wrinkles in this corner of the sheet.)
 
When I measure the voltage with and without the platen connected I obtain about 480 Volts when ON, dropping to 20 Volts and then dropping further when switched OFF. This seems to be normal.
I also tested most of the components on the platen control board so I assume that the power supply part is fine.
 
Across the platen connector I measure infinite resistance (when plugged out).
I am not able to measure the capacity lacking test equipment for such low values.
There is also infinite resistance between the platen connector wires and the metal frame of the platen.
There the comb-like copper traces are connected to the wires - I measure zero resistance between one wire and one comb
All this seems to be reasonable.
 
The platen is built of interleaved +/- combs on the Epoxy PCB board (covered with the white plastic sheet).
 
There are a 2-3 deeper scratches in the plastic cover sheet so that in two or three spots the copper trace is open to the atmosphere.
Could this be a problem, even if a smaller sheet of paper is not reaching these open spots?
 
Unfortunately the theory of operation of the platen is not explained in the handbooks or in the HP-Journal article for the 9872/7221 plotters.
Is there someone who knows how it actually works and whether it is sensitive to cover sheet damage?

Any ideas or experience with defective platens for 9872/7221 plotters?

Thanks,
Martin

 

The platen supply has a voltage doublet and a very high impedance. Below the diagram.
Image

-Rik



On Fri, Jan 17, 2020 at 2:01 PM +0100, "Rik Bos" <hp-fix@...> wrote:

Martin,

The platen works electrostatic.
When you want to measure the isolation resistance you'll a Megger (isolation resistance meter) those instruments use 500 or 1000V DC to measure the resistance. This gives you a good idea about the condition of the platen. Next thing to check is the leak current of the capacitor (s) because the platen supply only can source a very little current every leak will decrease the output voltage.

-Rik

-Rik




On Fri, Jan 17, 2020 at 12:27 PM +0100, "Martin Hepperle" <Martin.Hepperle@...> wrote:

My 9872 sucks! ... not enough!

After fixing the mechanical problem with the bent chassis I now have a (hopefully) final problem.
 
The electrostatic platen does not work properly.  It provides only a very weak suction force so that the paper starts to slide. Also the force seems to be a bit higher when switched on and then quickly drops to almost zero within a few seconds.
 
Of course I had to take the platen off when I was taking the plotter apart for fixing the chassis but I was very careful.
The white cover sheet already had and still has a 2-3 deeper scratches going down to the copper/epoxy plate. 
The only modification from my side was to glue the lower right corner (about 20x20mm) of the white sheet down with rubber cement because it was starting to lift off the Epoxy board.
(BTW: this was not the best idea because the solvent in the rubber contact cement caused some small wrinkles in this corner of the sheet.)
 
When I measure the voltage with and without the platen connected I obtain about 480 Volts when ON, dropping to 20 Volts and then dropping further when switched OFF. This seems to be normal.
I also tested most of the components on the platen control board so I assume that the power supply part is fine.
 
Across the platen connector I measure infinite resistance (when plugged out).
I am not able to measure the capacity lacking test equipment for such low values.
There is also infinite resistance between the platen connector wires and the metal frame of the platen.
There the comb-like copper traces are connected to the wires - I measure zero resistance between one wire and one comb
All this seems to be reasonable.
 
The platen is built of interleaved +/- combs on the Epoxy PCB board (covered with the white plastic sheet).
 
There are a 2-3 deeper scratches in the plastic cover sheet so that in two or three spots the copper trace is open to the atmosphere.
Could this be a problem, even if a smaller sheet of paper is not reaching these open spots?
 
Unfortunately the theory of operation of the platen is not explained in the handbooks or in the HP-Journal article for the 9872/7221 plotters.
Is there someone who knows how it actually works and whether it is sensitive to cover sheet damage?

Any ideas or experience with defective platens for 9872/7221 plotters?

Thanks,
Martin

Martin Hepperle
 

Rik,

 

thank you for your helpful comments.

 

Unfortunately I am not equipped for such insulation resistance tests – maybe I have to ask the electricians who from time to time test the equipment in our offices for faults with grounding.

 

I see some small bits of copper corrosion in the open spots of the platen, so I will first try to enlarge the scratches a tiny little bit for better accessibility and clean the part in these spots. This is not heavy corrosion, only a tiny little bit (in the visible area).

 

In the extreme I could rip of the plastic film and clean the copper surfaces. But then I would have to apply a new plastic sheet and the current sheet seems to be considerably thinner than the self-adhesive plastic films I know (e.g. DC-Fix, a German brand).

 

I tested the capacitors only for capacity which is as specified, but not the leak current. I think my meter also shows the ESR value.

Replacing the capacitors would be no problem – maybe something to try before tinkering with the platen.

 

Martin

Kuba Ober
 

400V seems low in general, but perhaps that’s what this model was designed for? I had a Roland DXY plotter with electrostatic platen and the voltage was a couple times higher IIRC.

The insulator is there for user protection only, and it is otherwise necessary. It is “in the way”, in fact. 

The metal electrode polarizes the plastic sheet, displacing the charges on the facing surfaces. Say, a positive electrode will attract the electron clouds in the insulator atoms, so and this will propagate through the thickness, and the nuclei close to the surface will be less shielded by their electrons, “exposing” a positive charge. If the insulator molecules are also polar, then this will bolster the effect: they’ll align along the gradient of the electric field.

The electric field of the electrode is somewhat suppressed by the opposing charge at the facing surface of the insulator, while the charge at the exposed surface of the insulator regenerates the electric field. 

But there’ll be trouble if there’s leakage onto the surface of the insulator, eg if it was “cleaned” with something that leaves conductive residue. The surface charge can then be neutralized by a current flowing through the residue. The residue acts like a low conductance capacitor plate then. If there are cuts or pinholes through the insulator, they’ll act as local shunts, actually improving the performance; the insulator is there for our safety after all. 

Here are the problems I can imagine (that’s by no means exhaustive, just a brain dump):

1. Incorrect voltage measurement technique. The measurement of may require a high voltage probe (or a homebrew approximation with high voltage high resistance resistors, with no concern for AC compensation – see eg joe smith’s YouTube videos). Typical good hand-held multimeters become low impedance devices above 1.5kV or so, and cheaper ones will do it well below that voltage. The voltage source in the plotter does not have enough oomph to overcome this shunting behavior (it’s not a maybe, it’s most certainly true unless the design is bonkers unsafe). So even with the source fully operational, a direct multimeter measurement is likely to give misleading results.

Diagnosis of the issues below requires a proper voltage measurement technique, then!

2. Supply fault: something within the supply causes it to produce too low of a voltage. Say, a shorted diode in the voltage doubler would cut the voltage in half IIRC. Leaky diodes will cause low voltage output. Leaky capacitors will do that too. 

This would need to be tested with the supply disconnected from the electrode, to prevent any problems upstream of the supply to confuse the results. I doubt that a feedback-stabilizing high-voltage capacitor would be necessary to stand-in for the electrode. The supply shouldn’t be so fancy as to need that. It probably works open-loop with regard to high voltage output.

3. Excessive electrode-side leakage pulling the supply down. Remove the electrode assembly from the unit, place on insulators, and re-measure the voltage. If the voltage is now correct and stable, then look for leakage around the mounting points for the platen, and wherever else it may be making contact with the rest of the plotter.

4. Excessive insulator surface leakage discharging the outer surface of the insulator. This would neutralize the surface charge and thus disable the electrostatic holding effect. Place a tinfoil plate on the top of the insulator (with a 2” margin from the ends of the insulator surface), ground it for a minute, then unground and power the supply up. Measure voltage 10s after power up, with meter in peak hold/max mode, with resolution set to minimum if possible (fastest refresh). Power off, then repeat the procedure but measure 120s after. The voltages should be very close. If they are not, then there’s surface leakage. Cleaning the insulator and the insulator-to-electrode transition may be necessary. IPA and deionized water at minimum.

This would need to be rechecked with the platen back in the unit, as an additional leakage path may then be present. 

5. Excessive insulator conductance shouldn’t be a problem though: conductors polarize much better than insulators do. My DXY plotter has a virgin Teflon insulator that shouldn’t degrade with time unless some Teflon-incompatible ionic or polar cleaners were used.

6. Paper that is not conductive enough. The electrostatic hold works by displacing the charges in the paper and creating an opposing surface charge on the side of the paper that faces the insulating sheet. Conductors polarize much better than insulators do, because their interior charge is quite mobile and will gladly move across the entire thickness of material, whereas an insulator’s interior charge can only be displayed at atomic- to molecular-scale.

Use a piece of tinfoil in place of paper to get an in-the-limit idea of how good the hold may be. Anything less conductive will be held with less force, at least initially. The surface polarization may build up a bit over time, though – it really depends on the makeup of the material.

I hope this helps in the diagnosis, at least a bit. 

Cheers, Kuba Ober


17 jan. 2020 kl. 6:27 fm skrev Martin Hepperle <Martin.Hepperle@...>:


My 9872 sucks! ... not enough!

After fixing the mechanical problem with the bent chassis I now have a (hopefully) final problem.
 
The electrostatic platen does not work properly.  It provides only a very weak suction force so that the paper starts to slide. Also the force seems to be a bit higher when switched on and then quickly drops to almost zero within a few seconds.
 
Of course I had to take the platen off when I was taking the plotter apart for fixing the chassis but I was very careful.
The white cover sheet already had and still has a 2-3 deeper scratches going down to the copper/epoxy plate. 
The only modification from my side was to glue the lower right corner (about 20x20mm) of the white sheet down with rubber cement because it was starting to lift off the Epoxy board.
(BTW: this was not the best idea because the solvent in the rubber contact cement caused some small wrinkles in this corner of the sheet.)
 
When I measure the voltage with and without the platen connected I obtain about 480 Volts when ON, dropping to 20 Volts and then dropping further when switched OFF. This seems to be normal.
I also tested most of the components on the platen control board so I assume that the power supply part is fine.
 
Across the platen connector I measure infinite resistance (when plugged out).
I am not able to measure the capacity lacking test equipment for such low values.
There is also infinite resistance between the platen connector wires and the metal frame of the platen.
There the comb-like copper traces are connected to the wires - I measure zero resistance between one wire and one comb
All this seems to be reasonable.
 
The platen is built of interleaved +/- combs on the Epoxy PCB board (covered with the white plastic sheet).
 
There are a 2-3 deeper scratches in the plastic cover sheet so that in two or three spots the copper trace is open to the atmosphere.
Could this be a problem, even if a smaller sheet of paper is not reaching these open spots?
 
Unfortunately the theory of operation of the platen is not explained in the handbooks or in the HP-Journal article for the 9872/7221 plotters.
Is there someone who knows how it actually works and whether it is sensitive to cover sheet damage?

Any ideas or experience with defective platens for 9872/7221 plotters?

Thanks,
Martin

Martin Hepperle
 

Kuba,

thank you for the many hints and the good explanation of the theory of operation.
I'll check what happens when I place some aluminum foil on the platen.
Next week i will talk with my colleagues in the electronics lab whether they have suitable test equipment - I only have a cheap DVM and some HP 3456 and 3478 voltmeters.

BTW: I still have my Roland DXF-990 flatbed plotter which I bought in the early 1990s - HP plotters were too expensive for me and the Roland has a nice X-Y-coordinate display.

Martin

 

You could see if you can get some Monokote film, which is used for model aircraft. It is reported as being suitable as a replacement for the bed covering. I intend to do this for one of my 9872C plotters. It was recommend to partially preshrink the film before applying to the bed.

Vassilis PREVELAKIS (series80.org)
 

Hi,

My 7220C (the serial version of 9872C) with a working electrostatic panel, shows slightly over 500V when the CHART HOLD button is pressed.

Regards

**vp

www.series80.org

Kuba Ober
 

So, 400V on the problematic plotter may be at least in the right ballpark, then. My old Roland would make a nice corona noise when you placed paper on it, not unlike what you’d sometimes get when you powered off (or on) a CRT monitor. I don’t know whether that’d be audible at 400-500V.

Cheers, Kuba

18 jan. 2020 kl. 5:02 fm skrev Vassilis PREVELAKIS (series80.org) <series80@...>:

Hi,

My 7220C (the serial version of 9872C) with a working electrostatic panel, shows slightly over 500V when the CHART HOLD button is pressed.

Regards

**vp

www.series80.org


Martin Hepperle
 

Some update: I made some test with aluminum foil but the result was similar to paper - not enough suction.
I also washed the platen as well as the high voltage PCB with a PCB cleaner and Isopropanol to make sure no dirt conducts my valuable electricity away.

Mine is the "C" versions of this plotter (9872/7720) which has a different schematics that the "A" model shown by Rik.
I am not sure what I should expect at points (A), (B), (C).
I could not find a replacement type for U102, the HP number does not appear in my X-Ref lists.

Here is the relevant part (with my blue scribbling on top, check marks indicate that I desoldered one side of the component to test it):




Martin Hepperle
 

... schematics in true size + platen schema

 

I have the service manual and two Cs, one of which works and the other doesn't for the chart hold down voltage. If someone doesn't beat me to it, I can make some measurements tonight.

On Jan 20, 2020, at 6:22 AM, Martin Hepperle <Martin.Hepperle@...> wrote:

<9872CT-Chart-PSU-Schematics.png>

Paul Berger
 

Martin,

At "A" I would expect 5V pulses as long as the voltage on pin 2 of U102 is less than 5V.

At "B" you should see +34V

What you will see at "C" would depend on the state of chart hold, if chart hold is off then "C" will be around 34V if chart hold is on then it should be pulsed.

It would appear that when chart hold is off pin 4 of U103 should be low and pin 6 would go to open (7406 has open collector outputs) this would inhibit Q101 which is the switch transistor.  When chart hold is on, pin 4 should go to open state and allow the 20 KHz   to reach pin 5 which would cause the output to switch at a 20KHz rate and in turn "C" would be pulsed at that rate as long as the base of Q101 is positive (around +5V) this is controlled by Q102 that has its collector tied to +5V the base Q102 should be positive and turn on Q102 as long as the voltage on pin 2 of U102 is less than pin 3.  Pin 3 of U102 is tied to a +5 reference voltage and should be very close to 5V.

In your picture U102 does not have a manufacturers logo on it and I cannot find that part number in any cross reference I have so perhaps it is a custom HP made part but in this application it is likely to be an op-amp or a comparator.  Is the 50V above C108 the voltage rating of C108? and is the supply voltage to pin 7 of U102 correct?

Paul.


On 2020-01-20 9:22 a.m., Martin Hepperle wrote:
Some update: I made some test with aluminum foil but the result was similar to paper - not enough suction.
I also washed the platen as well as the high voltage PCB with a PCB cleaner and Isopropanol to make sure no dirt conducts my valuable electricity away.

Mine is the "C" versions of this plotter (9872/7720) which has a different schematics that the "A" model shown by Rik.
I am not sure what I should expect at points (A), (B), (C).
I could not find a replacement type for U102, the HP number does not appear in my X-Ref lists.

Here is the relevant part (with my blue scribbling on top, check marks indicate that I desoldered one side of the component to test it):




Paul Berger
 

Opoops I meant to change my statement about "A" it should be steady near 5V as long as pin 2 of U102 is less than pin 3. 


On 2020-01-20 11:18 a.m., Paul Berger via Groups.Io wrote:

Martin,

At "A" I would expect 5V pulses as long as the voltage on pin 2 of U102 is less than 5V.

At "B" you should see +34V

What you will see at "C" would depend on the state of chart hold, if chart hold is off then "C" will be around 34V if chart hold is on then it should be pulsed.

It would appear that when chart hold is off pin 4 of U103 should be low and pin 6 would go to open (7406 has open collector outputs) this would inhibit Q101 which is the switch transistor.  When chart hold is on, pin 4 should go to open state and allow the 20 KHz   to reach pin 5 which would cause the output to switch at a 20KHz rate and in turn "C" would be pulsed at that rate as long as the base of Q101 is positive (around +5V) this is controlled by Q102 that has its collector tied to +5V the base Q102 should be positive and turn on Q102 as long as the voltage on pin 2 of U102 is less than pin 3.  Pin 3 of U102 is tied to a +5 reference voltage and should be very close to 5V.

In your picture U102 does not have a manufacturers logo on it and I cannot find that part number in any cross reference I have so perhaps it is a custom HP made part but in this application it is likely to be an op-amp or a comparator.  Is the 50V above C108 the voltage rating of C108? and is the supply voltage to pin 7 of U102 correct?

Paul.


On 2020-01-20 9:22 a.m., Martin Hepperle wrote:
Some update: I made some test with aluminum foil but the result was similar to paper - not enough suction.
I also washed the platen as well as the high voltage PCB with a PCB cleaner and Isopropanol to make sure no dirt conducts my valuable electricity away.

Mine is the "C" versions of this plotter (9872/7720) which has a different schematics that the "A" model shown by Rik.
I am not sure what I should expect at points (A), (B), (C).
I could not find a replacement type for U102, the HP number does not appear in my X-Ref lists.

Here is the relevant part (with my blue scribbling on top, check marks indicate that I desoldered one side of the component to test it):




Kuba Ober
 

Aaah. Now that’s something I did not realize: this is an interdigitated electrostatic platen. The one I saw a while ago was a continuous electrode, not two interleaved ones. I have to think of what does this bring to in possible troubleshooting. – Kuba

20 jan. 2020 kl. 8:32 fm skrev Martin Hepperle <Martin.Hepperle@...>:

... schematics in true size + platen schema
<9872CT-Chart-PSU-Schematics.png>
<9872CT-Platen.pdf>

 

I can confirm what Paul said about the circuit. I get 525 V when the chart hold is activated with good adhesion of the paper.

Paul Berger
 

One thing I might add is you might want to measure the voltage at the cathode of CR103 to ground or disconnect the platen for the first try.  There are very large resistors in series with the connections to the platen, and if some contamination happened to get in through the damage, it would not take much to pull down the voltage at the platen.

Paul.

On 2020-01-21 10:13 p.m., Craig Ruff wrote:
I can confirm what Paul said about the circuit. I get 525 V when the chart hold is activated with good adhesion of the paper.

yann Minder
 

Hello, do not forget your voltmeter input impedance forms a voltage divider with the high value resistors at the votage multiplier output. Different models will show different results...

yann




On Wed, Jan 22, 2020 at 4:16 AM +0100, "Paul Berger" <phb.hfx@...> wrote:

One thing I might add is you might want to measure the voltage at the 
cathode of CR103 to ground or disconnect the platen for the first try.  
There are very large resistors in series with the connections to the 
platen, and if some contamination happened to get in through the damage, 
it would not take much to pull down the voltage at the platen.

Paul.

On 2020-01-21 10:13 p.m., Craig Ruff wrote:
> I can confirm what Paul said about the circuit. I get 525 V when the chart hold is activated with good adhesion of the paper.
> 
>



Martin Hepperle
 

Thank you all for your comments and hints. I guess my test equipment is not perfect for these tests and now I am not sure that the platen itself is the problem.

I measured the voltages going into the comparator/op-amp U102.
a: UREF 5V is actually 5.21V in my case (measured at right side of R119)
b: behind R119, at (+) of U102 I see 5.04V (measured at left side of R119)
c: between R109 and R110, at (-) of U102 I see 5.20 V (measured at left side of R109)
(right and left as seen from the rear of the open plotter)

My plan is now to first correct UREF to be closer to 5V. In fact this may be my fault as I replaced the broken 4.99V Zener VR12 on the main PSU board by a 5.1V Zener because I did not have a 4.99V at hand. I believed that with the tolerances this might be o.k. but just believing is only an alternative fact ... So I'll order some 4.9 and 5.0 V Zeners to see what happens .next.

Martin

Paul Berger
 

Martin,

The reference being slightly higher would only mean the high voltage would be a bit higher.  The fact that you see 5.2 volts at the - input of U102 would suggest that there is high voltage since it is fed by a voltage divider with approximately 90 to 1 ratio . What do you get at the cathode of CR103?  What sort of a meter are you using?  If it is a digital meter they usually have an input impedance of 10M especially on the higher voltage ranges.

Paul.

On 2020-01-22 6:53 a.m., Martin Hepperle wrote:
Thank you all for your comments and hints. I guess my test equipment is not perfect for these tests and now I am not sure that the platen itself is the problem.

I measured the voltages going into the comparator/op-amp U102.
a: UREF 5V is actually 5.21V in my case (measured at right side of R119)
b: behind R119, at (+) of U102 I see 5.04V (measured at left side of R119)
c: between R109 and R110, at (-) of U102 I see 5.20 V (measured at left side of R109)
(right and left as seen from the rear of the open plotter)

My plan is now to first correct UREF to be closer to 5V. In fact this may be my fault as I replaced the broken 4.99V Zener VR12 on the main PSU board by a 5.1V Zener because I did not have a 4.99V at hand. I believed that with the tolerances this might be o.k. but just believing is only an alternative fact ... So I'll order some 4.9 and 5.0 V Zeners to see what happens .next.

Martin