JVC HR-D750 FF Trips: Requires Reel Motor Repair?

[HamishBoxer]

Check the ESR of all secondary power supply caps when you get odd ball faults.

[Takapuna]

Dave/Everyone thanks for staying with me on this and for the help and suggestions so far. Much appreciated.

This is a partial reply to some of your points. I'll try to get round to the others later.

I have a working machine that I use frequently so I'm reluctant to dive inside or swap any parts as I could end up with two non-workers! However, I dared to remove the cover and 'scope the end sensor signal some time ago whilst winding and as I remember it was noise-free.

The PSU module received a good dose of new electrolytic capacitors when doing the initial fix. Apart from the 1uF start-up capacitor C14 and one other they all checked out OK for both value and ESR. However, once you've removed them it seems sensible to replace them so I have done so. Apart from the PSU and the 3.3uF cap on the drum motor FG signal all of the electrolytic capacitors tested so far have measured pretty close to spec. The capacitor quality in this model appears good. New components don’t measure significantly better and my replacement Panasonic or Rubycon types are unlikely to have longer lives.

Main board capacitors changed to date are: C602 on the 5V output from the series-pass regulator, C604 smoothing the M54647 pin 10 “VR” voltage, C605 smoother on the “Motor 12V” supply (not used by M54647), C606 on the “Unreg 17V” supply to M54647 pins 2&3, and for good measure C607 which is the POR delay to the main controller IC601.

Prompted by your suggestions I’ve taken a closer look at the interface between main controller IC601 and the M54647L reel motor driver. A data sheet for M54647L would be a help but I've been unsuccessful in finding one. The truth table in the manual shows how the High/Low status of the discrete input signals affects the motor power output polarity and the brake function. The other connections between the two ICs have less clear functionality but I assume (fatal!) that they are analogue setpoints to change the motor speed between the various play and wind modes. Available motor torque might be changed by these too as there is no slip in the drivetrain from motor to spool table. Looking around these analogue signals with the ‘scope I found M54647 pin 10 “VR” to be about the only place I could find a trace that reflected the motor commutation noise albeit at a low amplitude. In an effort to reduce this I tagged a few more uF across C604 which made the system unstable, the motor switching rotation direction every two seconds or so.

I don’t think there’s any feedback from the M54647L driver chip to the main controller IC601: I think IC601 speaks and the driver chip listens and carries out the commands.
There is feedback to the controller IC from the take-up spool opto “TU FG” so it does know speed of the spool but not direction. This is one other signal that will appear to be locked to reel motor rotation. I need to think about this a little more…

Thanks,

Phil
----

[G6ONEDave]

Having just searched for a data sheet for the M54647L, the only webpage that shows a pdf symbol by the side of the chip number is 'Jotrin' but you have to sign in with them to get it. Might be worth asking if they actually have it first though.

Dave

[G6ONEDave]

Something else that came up in a conversation this morning is about the lower drum. Does the tape fully unload when in FF or does it remain laced because if so there might be an issue with the lower drum if it's not shiny enough for the tape to ride over. The tape may be snagging on it.

Might well be nothing to do with this, just thought I'd mention whilst fresh in my head.

Dave

[Takapuna]

I've spent some more time with the VCR and these are my latest findings.

I was wondering if the take-up spool PG circuit and system had anything to do with the fault as it produces a signal which is synchronised to the reel motor rotation either directly during FF or via the tape driving the spool in REW. The answer appears to be No. The "TU FG" signal is a healthy (nearly) square wave of 5V amplitude with good sharp edges showing some overshoot similar to that seen when a x10 'scope probe hasn't been matched to the 'scope input. In FF mode the TU-FG pulse frequency remains much lower than the noise spikes that I'm trying to trace. For this reason I'm saying it's OK, even though the envelope extends only 1V positive with the remainder below the 0V line as if it's being clamped somewhere. This is regardless of the source being a device connected between 5V and ground.

Using Dave's suggestion of applying 10nF between tape end phototransistor and ground (thanks Dave) has indeed reduced the amplitude of the sharp leading-edge spike and calmed things down significantly. The tape that would wind "reliably" to around the 40 min time will now wind all the way. That "knowedge" (for want of a better description) of how far it has wound is still there in that the signal across the phototransistor shows no disturbance at the start of the tape looking like a dc level. As more tape winds the signal takes on the shape of a triangular waveform with a sharp negative-going leading edge and a somewhat curved R-C style return to the high level. The amplitude of these increase as more tape is wound but are never sufficient to cause a tape-end trip. Every so often a spike of significantly greater amplitude appears and these are the ones that will cause the trips. There are a couple of these every second or so and more interestingly one is generated each time the reel motor does the stepped-acceleration thing during a wind. This suggests, again, that the reel motor and its drive system are the cause of the problem...but I've spent a lot of time there already.
I've allowed the tape to wind to the end a couple of times and it's unclear as whether the sensor catches the tape end in time or not having delayed the event a little by the addition of the capacitor. It sounds OK and doesn't appear to snatch. The Philips tape that would not FF at all now winds nearly all the way to the end. The difference with this tape is that a triangular signal of significant amplitude is present across the sensor right from the start, getting even higher as the wind progresses.

A small capacitor across both tape end sensors is almost a fix for this machine, but it really is a work-round. Not finding the true cause would be a bit of a cop-out.

Searching around with the scope I found that during FF and REW the CAPSTAN motor windings carry what can only be described as a descending staircase signal of significant amplitude, the highest step being at around 6V and the lowest just below zero. I'll re-check this last point... The interesting thing is that each of the step edges is synchronous with the noise on the end sensor etc. There is obviously some energy/noise at this frequency from somewhere. The capstan motor command lines obey the truth table in the manual, which is not surprising as all REC/Play modes work as they should. In the stop mode both control lines are low, leaving the motor windings open.
Being careful not to select PLAY I tried a 4.7uF capacitor across the capstan motor windings and this made little if any difference. Next I tied one side of the windings, still shorted by the capacitor, to ground via a 1K resistor. This removes the staircase but leaves sharp rising edges in time with the interference. Not quite sure what, if anything to make of this. I'm sure all of this will disappear once the source of the fault has been identified. These lines are perfectly quiet in STOP mode.

Regarding lower drum tape wrapping, this is a model that has a half-load to allow it to display real time and Index mark search. It does not leave the tape around the drum during winding, rather a small pole pulls tape out far enough to contact the audio/control head. I can't think of a mechanism, but wonder if this wrapping has anything to do with noise signal generation. It is obviously the means by which I know that the "good" tapes wind consistently for 40 min or so but don't think the intelligence of an index mark search or similar is in play and commanding the trips as the clear service cassette produces some noise spikes without any tape.

I have a new STK5490 regulator chip available and could try that as things are getting desperate. As mentioned previously, I'm trying not to fit it as it doesn't appear to play a part in any of the reel motor or sensor supplies.

Please feel free to discuss and dismiss as you wish.

Thanks.

Regards,

Phil
----

[G6ONEDave]

The 4.7uf cap is probably a bit too large in value to respond fast enough to the spike pulses, that's why I suggested a 10nf cap. My next suggestion was to try the 10nf across the reel motor drive, or to fit a 10nf cap from each side of the reel motor drive and chassis (0v).

However I now think that the problem is actually power supply related. Try scoping the motor 12v supply output to see if you have all those spikes at that location. I'm tipping that they will be there and of a higher amplitude. You could try the 10nf across the power supplies motor supply output to see if it cures the problem. If it does dramatically improve things then it suggests that the actual problem is being caused by the STK5490. This regulator chip is probably a switch mode design internally as opposed to a linear design. I've not checked the data sheet to confirm this yet but feel that this may well be the case.

Dave

[Takapuna]

I've revisited the power rails with my scope. Tried looking differentially with a pair of probes too, but this makes little difference perhaps due to one side of the signal being ground anyway. Initially everything looked quiet with just the usual hash showing up as channel gain is increased. There are a couple of tone-bursts flying around too bit I think these are from the mains or airborne.

The 10nF capacitor has remained connected across the end-sensor transistor during all of this testing as it allows FF to work almost correctly as described in my previous submission. Also note that FF mode needs to be engaged for "problem" noise to be generated.

The UNREG 17V rail comes direct from the PSU and shows around 30mV of negative-going noise when FF is selected. Adding 1nF disc + 150nF polypropylene + 47uF electrolytic to this rail as close to the reel motor driver IC as possible reduces this only slightly.
Adding 10nF to ground on either side of the reel motor supply at the main PCB pins (as this is the easiest place to apply them) had little effect on the sensor noise. If it did anything it made the noise slightly worse.

It was at this point that I turned the 'scope brightness up to a level that looked like it will burn the phosphor and noted exceptionally brief bi-directional spikes of some 2.5V amplitude on the motor drive signal. Looking again on the supplies I could see that they were present to some degree or another on all of them. They are locked to the reel motor speed.

At this point I decided it was time to replace the STK5490 regulator. I had a new one to hand so I fitted that. The result is no change. The fault is still present as before, as are the various fast spikes.

Having a working version of this deck to hand I decided that exchanging the PSU assemblies was a low-risk task, requiring the removal only of two screws, two connectors and one soldered joint. It would allow me to eliminate the input PSU assembly from the equation. After carrying out this transplant there is no change in symptoms or behaviour. I've adjusted the output to 15.3V as instructed and left the removed PSU fitted to my in-use machine as I see little point swapping them back. As mentioned in an earlier post, I don't want to swap any more parts and end up with two faulty VCRs!

The remaining unsolved clue is the mechanism by which the noise signal increases as a tape FFs from start to end. The clear service cassette shows little disturbance AND applying drag to the rotating take-up spool, almost to stalling point, only increases noise amplitude slightly. The "good" tapes I insert show a noise signal that increases from zero at the start to a considerable amplitude at the end. The "bad" (Philips) tapes show a high signal amplitude from the start, increasing sufficiently to trip and change to REW just before the tape end is reached. This behaviour is with the 10nF filter capacitor across the sensor.

FF mode in my working machine does show a few small spikes across the sensor now I know how to look for them. However, they are of around 150mV amplitude and hence nowhere near to generating a trip.

Thanks,

Phil
----

[G6ONEDave]

I was almost convinced that it was the psu causing a problem but since you not only replaced the STK chip and then swapped the 2 psu's around to no avail, that confirms that the psu was not responsible for the trouble.

Although frustrating that the trouble still exists, there is some progress being made. We now know that the psu is OK, we also know that the reel motor is OK and also the reel motor driver chip is OK.

The intereference spikes are being caused by the reel motor drawing extra current to drive the tape from the supply spool to the take-up spool. This is confirmed by the fact that they don't exist to the same extent with the blank tape. I am a little surprised that the filter caps fitted directly to the reel motor had no effect and this brings me to my next suggestion.

I wonder what would happen if the chassis/earth/0v rail was slightly resistive. It could be worth checking all the 0v lines from the psu to the main pcb, the end sensor, deck terminal pcb and then to the mechanism metalwork. The ohms readings should be 0.1 perhaps up to 0.3 ohms depending on the resistance of the meter leads. Certainly not anything above 1 ohm. You could also try running a wire in parallel with the existing 0v lines. Also remember to check the pcb tracks involved with the 0v lines.

The only other item not swapped in the involved circuits is the syscon chip but let's not go there just yet.

Dave

[HamishBoxer]

What sort of mode switch does this have? Are the brakes sticking?

[Takapuna]

Hamish,
The VCR is fitted with the "usual" 3-pole sliding mode switch. However, it moves through all modes properly. The fault I'm chasing is generating a false tape-end signal during FF, causing it to move to STOP and then REW modes. I think the brakes are OK and don't believe this to be a mechanical fault.

Dave,
I agree this fault feels like a grounding, signal pick-up or power rail fault. As far as I can tell, the grounds appear sound. A quick run around with the AVO on Ohmsx1 has not identified anything yet.
Alongside the plug/socket PSU output connections JVC added a supplementary ground wire. It is soldered to the power supply reservoir cap negative leg and has a push fit tag onto one of three blades screwed to the deck mechanism. The other blade connects back to the mainboard as a parallel or back-up to the ground reaching the mainboard from the PSU through the harness. You can remove and replace either or both of these with the deck operating and it makes no perceivable difference to the fault or to machine operation.
As mentioned previously the grounding screw for the deck terminal, complete with star-washer, is essential for end sensor operation. It is the 0V return for the phototransistor. The noise generated by the sensor goes away and FF is maintained without this ground connection. Removing the screw gives the same result as removing the wire from sensor.
I would dig deeper on these end-sensor connections but the tape start-sensor exhibits similar noise problems, albeit to a lesser degree or rather a lower amplitude such that false trips are rare. The start sensor phototransistor is part of the PCB on the cassette cage and receives its connections and ground through a different loom.
I'll remove the mechanical deck once more for a better look at the PCB tracks and connections therein but there's still a number of tracks obscured by motors and mechanism.

Thanks,

Phil
----

[G6ONEDave]

Whist checking the 0v wiring, remember to check the plug/socket joints as they can become tarnished and go high resistance.

Dave

[Takapuna]

I’ve continued trying to fix this VCR and have spent many hours on it since my last post way back in December.
Finally, I have good news as you will read later. Before the happy event though I was convinced I’d reached the end of my efforts to fix it and had prepared the draft presented below to document and ask for any last ideas. I’ve decided to post my recent efforts for the record. They might help anyone suffering with insomnia. Otherwise, the last few paragraphs reveal the outcome...

After searching and experimenting I think I better understand the reason for most of the VCR’s behaviours but cannot isolate the cause.
To recap…Noise spikes present across both tape end and tape start sensor phototransistors are of sufficient amplitude to trip the system, especially during Fast-Forward. The tape-end sensor is especially sensitive, tripping most often and resulting in the STOP-REWIND behaviour as programmed in the firmware. These noise spikes are time-coincident with edges of the commutation noise from the reel motor.

My final investigations and some findings were/are as follows:
1). I removed the tape dack from the case once more and inspected connections and re-soldered many joints, all of which looked excellent in the first place! The logic behind doing this was poor when I consider that the tape-start sensor suffers the same noise signals, albeit to a lesser degree, and is on a separate PCB on the cassette housing. This is connected by its own loom and multi-pin connector and has different signal paths to the troublesome end sensor.

2). Taking a heavy-gauge ground connection from the power supply and briefly connecting it to each of the PCB ground lands that connect reel motor power returns or sensor returns has little effect on the spurious signals. The same is true when taking the 5V supply directly from the emitter of the series pass regulator Q601 and “shorting” it directly to lands on the PCB supplying this supply to the various sensor pull-up resistors etc.

3). Another idea was that noise might be coming out of the pin(s) of the mech-con chip. Hence, I removed the series 4k7 resistor from between the end-sensor pull-up resistor and IC input pin, pulling the IC pin high directly via a separate 120k resistor to keep it happy. No change. The end sensor continues to generate noise signals in FF mode…even when it is not connected to anywhere…the circuit comprises now only a 5V supply connected to the collector of the phototransistor through a 120k pull-up resistor. The emitter of the transistor is grounded to deck metalwork. As a sanity(?) check, I wired my spare end-sensor phototransistor assembly via a 120k pull-up directly to the 5V rail and positioned it on the top of the main board. This was done birds-nest style with short leads to the sensor in mid-air. This arrangement also generated noise at the collector when the FF mode was selected, albeit of a much-reduced amplitude compared to the installed one.

4). Next, I disconnected the cassette compartment LED (the cassette “lamp”). This brought-about a significant reduction in the amplitude of the noise signal, which was unexpected and initially difficult to explain. The machine exhibits some mechanism that allows it to “know” how far certain E180 cassettes have been wound from the start, the noise increasing as the wind progresses and tripping 40min-in. Two E120 cassettes refuse to FF from the start at all.

I’ve concluded the explanation for this is that the phototransistor(s) are never completely blinded by the tape in the cassette and some miniscule amount of light from the LED reaches them. In reality, the transistors never turn off fully and although the collector rises to somewhere near the 5V supply rail the phototransistors are still in their active region and consequently sensitive to the noise from the reel motor. Blinding the phototransistors with insulating tape reduces the noise amplitude but does not remove it completely. The fully rewound E180 has a full supply spool initially, hence the tape is closer to the window in the cassette and there are fewer paths for any light leakage and little interference is generated across the sensor. As the tape winds the radius at which it leaves the spool decreases and light leakage transmission paths from LED to external phototransistor through the cassette increase. As the amount of light reaching the transistor increases then the transistor becomes more susceptible to the “noise” from the reel motor, hence the noise signal amplitude increasing as the tape winds. At 40mins into the tape the phototransistor noise is sufficient to cause trips. The two Philips E120s are filled with less tape than the E180s hence even fully rewound they have sufficient light leakage to cause trips and will not FF at all. I’m not talking about much light leakage here as the scheme works well in the 1000’s of models employing this deck. For some reason it is a problem in this machine.

For the record, the components that have been either substituted or replaced include: both end sensor transistors, the cassette compartment LED, reel motor, reel motor drive IC, voltage regulator IC, a number of electrolytics (most of which were OK) and the 5V series-pass regulator transistor Q601. The main PSU unit has been proven by substitution. I’ve also had to replace a couple of the so-called circuit protectors which are very quick to open when an earth lead or similar accidentally brushes against somewhere that it shouldn’t!

5). Using a pick-up coil in the form of a 1mH axial choke connected to a ‘scope channel shows there to be a high level of electrical/electromagnetic noise around the reel motor when it runs. What I don’t know is if this is normal for the motor and the problem is that I have a deck with heightened sensitivity, or if there remains an issue with the motor or its connections or drive. The reel motor is operational in the play and trick modes but is supplied with a much lower voltage/current and hence the noise generated is much less.

6). Experiments show that it may be possible to quieten things down by adding motor suppression if I can squeeze it in close to the motor terminals. An 470uH inductor in each leg and a 10n/47n disc or similar either side of the inductor and a 10nF from each motor terminal to motor case calms it down significantly. Earlier trials with 10nF on the motor drive at the main PCB made little difference but there is a long connection path between that point and the motor. I’ve re-checked this and it remains true. The motor suppression along with 10nF or similar across each phototransistor may prove a working cure. Reducing the phototransistor collector resistor by a factor of 10 or so helps as it reduces the transistor gain so that small amounts of light leakage are insufficient to allow the output to fall 0.5V or so below the supply rail at the tape ends.

Sadly, all of this constitutes a “fix” rather than a repair.

-------------------------------------------------------------

So, to the good news. There is a saying that only a mad person continues to repeat the same experiment expecting a different result… In the end my thoughts remained as they were at the start: a noisy reel motor is to blame. This is in spite of replacing it twice already. I’ve obtained and fitted another used motor and it has fixed the problem! The machine functions reliably and no noise spikes are generated across either end sensor. The sensor lines remain to all intents and purposes perfectly quiet.

In the end it was not an electronic problem, although after servicing the original motor and then fitting a replacement I’d convinced myself it was. Finding a replacement main board would not have produced a result. I’m curious to know what the physical fault is with the first two motors, especially as the one I disassembled looked so clean inside. In post #6 “jayceebee” John stated that “the PU59926V reel motor was used in many models and quite a common failure”. John, you were correct. Both motors pass the spin test however.

Happy to have fixed it but disappointed at having replaced perfectly good ICs and capacitors and re-soldered many good joints and left some tidying-up to do.

Not sure if I should look out for another one of these to put on the spares shelf or not😊.

Thanks to everyone who offered help and ideas, it was you that kept me going.

Kind Regards,

Phil
----

[ben]

Blimey, this should be made a sticky thread to give people proof that persistence pays off!
The sort of fault that used to be in Television magazine 'Test Case' column.

Just shows that often, our instincts have the answer. What are the odds of having more than one duff motor.

A hearty 'Well done' is in order! I shall raise a glass of Spanish red to you!

[Takapuna]

Thanks Ben; I had a glass of white to celebrate.

Looking a little closer with the 'scope shows some spikes on the end sensor lines but of much reduced amplitude to previously and similar to my in-service deck. End-to-end winding is now available. Perhaps the spikes are a measure of motor wear?

Back in the days when spares were plentiful I'm sure I would have fitted a new motor (price permitting) and all would have been well.

I'll put the lid on and use it for a while. Will stay on the lookout for a replacement PU59926V or PU58636W reel motor but understand that ones taken from old VCRs are not guaranteed.

Thanks,

Phil
----

[G6ONEDave]

Glad that you got to a resolution for the JVC. The thing to remember is that just because a part is brand new, it does not mean that it is free from defects. I suspect that if this issue arose when these videos were in normal service life, JVC technical would have either specified a particular motor or a modification for this problem.

Well done for sticking with it.

Dave