Hi, all! I have bought D6C with CX20084 problems. Looking at this forum, I think about design and production CX20084 replacement module. It will be a small PCB, as small as SO-16 case, pin-to-pin compatible with original IC - one sot23-5 opamp, one sot-23-5 LDO, and 3x3 mm QFN MCU; all regulation curves will be implemented by firmware. What do you think about this idea? Another idea: if CX20084 died from overvoltage, possible die only it's internal LDO, and all other may be in working condition. All three wires of internal LDO have corresponding pins - Input (pin 8), Enable (pin 7), Output (pin 5). So, we can disconnect pin8 from power line and try to place external LDO, such as TPS76318 or TPS76316.
Hello and welcome to the forum @radionick ! I had this idea before and have presented it in other threads regarding CX20084. Mine was to built it in the analog domain with as few transistors as possible (but even with 4-5 transistors the footprint is already large). Doing it with an MCU seems like a better option given how advanced some of them are nowadays, but there are some challanges as well. How would the firmware be implemented ? 1. How are you going to read the FG singal: by ADC or by converting it to squarewave (you will need an additional comparator, internal or external to do this); 2. How are you going to output the motor drive ? PWM or DC ? PWM can be generated directly from the MCU, but there are potential problems in doing that which need adressing (mainly interference, but not limited to it); For DC you either need a DAC, or a low pass filter for the PWM signal (a 2-pole LC filter can do the job, but requires additonal components and the output won't be perfectly clean). As for the second idea, from what I see on the schematic, the output of the internal regulator (pin going down) is internal and that's what you will need to use an external regulator. So even if it would be just the regulator burnt, the problem cannot be fixed by an external one.
It is an interesting idea but I think you would be limiting yourself by using such a small Micro. Once you get up to larger ones (e.g. in a 5 x5 mm package). they come with things like ADCs and DACs built in. This was just the result of a simple Google search https://www.analog.com/media/en/new...ighlight/precisionmicrocontrolleroverview.pdf Reading her latest post, I guess someone like @Daniela would have a better idea of the feasibility of writing software that did the required things. Just thinking about it, if you could get the update rate beyond 20KHz you would never hear it so the only other problem would be making the updates accurate.
Of coarse, it's not simple. As I understand, CX20084 work with FG as with classic tacho signal, by amplitude, not by frequency. Another IC, with crystal, work as additional phase comparator and PLL. So, if I want to do all as it done in 20084, I need to use ADC. Output may be high frequency PWM with simple RC filter. Output of internal regulator come to internal circuits, and to pin 5. So, it's possible to use this output pin as input for stabilized voltage from external LDO.
I always start with the scheme .... too much theory then I get lost ... as we all know the cx 20084 in this walkman is driven by another cip ic 701 controlled by a quartz .. so on its output we will have a very precise frequency to drive the cx on pin 4 .... But the cx 20084 can also work without this quartz stage, as you can see from the diagram, just move the switch and go to a manual adjustment ... now I don't know what frequency it is in this case on pin 4 ... of course by varying the 20k potentiometer Rv 602 .. .you'll give an example a minimum of 1000 hz and a maximum of 2000 hz (I don't know the right frequency) on pin 4 ... all this to say you can design a circuit with three or four transistors in smd to make it even more small and make it stay in place of the original chip ?? .... the speech and make it as simple as possible and accessible to all ... do it already with a couple of chips you talk about with a lot of programming will not be easy to build for an enthusiast ... second point to the ute nte final what would it cost to do it as you say ... maybe at the beginning it will cost 20 euros? and then ?? knowing the Ebay prices will only increase like the original cx20084
@radionick Indeed you are right about the internal regulator, it comes out at pin 5 as well, I did not pay enough attention. In this case, it's something that should be tested first: if it is indeed just the internal regulator that is a problem, putting an external regulator is a simpler solution than redisigning the entire circuit. I don't have any D6C with burnt CX20084 at this moment, so I can't test myself, but you can do it on your walkman and let us know the results. As for the feedback signal FG, it's the frequency that we are interested in (as that is proportional to capstan speed), not the amplitude.That is why I presented the option of converting into a squarewave, as that can be read through a digital I/O pin, eliminating the ADC conversion time. In regard to the output, PWM with a simple RC filter will have a lot of high frequency content on it which can create interference. I would prefer to use a DAC (especially if it's already built-in), as the output will be clean and will only need an output buffer to drive the power transistor. Theoretically, if the frequency is high enough it may be ok, but we need to keep in mind the sensitive Dolby circuits that may work incorrectly when high frequency noise is present on supply lines (even though we use a local regulator, noise will exist on GND). Overall the idea is good, but at the end of the day it depends if its worth the effort. Let's see what @Daniela has to say about this, as she has a lot of patience in regard to re-designing/reverse engineering. @Silver965 If it were to be done using an MCU, the firmware can be open source. And it would be produced to be sold, so people can buy it as a plug-and-play replacement, that's the main idea. The original circuit does indeed work without the PLL, there is even a pad on the PCB that can be desoldered. In original circuit however, the wow&flutter is significantly higher without the PLL.
You have lost another important connection, from pin 9 of CX20084. Pulses from this pin go to IC701, where its phase compared with phase of reference oscillator; phase error pulses from pin 5 IC701 go to DC speed regulator (R715C605 working as integrator).
No. Look at internal structure of IC: for frequency/phase regulator sawtooth wave generator is not needed. If there is classical inductive tachogenerator, its amplitude and frequency both will depend of speed. Phase compare done outside of CX20084; it only formed pulses for external PLL (IC701, look at previous reply). Tomorrow I will start to repair my D6c; so, if there is internal LDO problems, I'll try. About 3 or 4 transistors schematic - I have seen it; it's another way to do replacement module; but it's too hard to do it as small as SO-16.
I am delighted that it can be done as you say ... but I would also settle for three transistors hidden in the case that at least allow me to use it ... rather than keeping it in a box to rot ... waiting for a spare cx pin 9 ... I have not considered it because by switching to manual mode that part of the circuit is excluded by disconnecting the connection that comes from pin 5 of the cip 701 ... But I repeat, happy that it can be done
This simple transistor circuit is not so simple as it seems at a first look. Devil in the details. In this circuit thermal instability of each element is compensated by all other. So, if you use not the same type transistors, in the same cases, and with same PCB design, you must be ready to thermal troubles. In the middle of 90's I have some experience in design discreet TCXO's; it's interesting, but if it's possible, I'll newer do it more, and use integrated TCXO. Same situation with precision amplifiers, comparators, etc. For example, if you use matching pair of transistors at input stage, temperature difference between transistors about 0,1 degree produce the same error as temperature change about 100 degrees, but together. So, if try to make analogue equivalent of CX20084, I'll prefer to use opamps, such as AD8605-AD8608.
You are right, and I was wrong, sorry. It's really work only in time domain, not in amplitude. I have done my D6C - problem was only in the hands of it's previous owner (as Russians say, he is "rukojop" - "hands_from_ass_man"). Done, and start to analyze signals on the pins of CX20084. It's very, very interesting! I'll try to do fully analogue replacement module.It's not needed for me now, but interesting. May be, will use it later in other brushed DC motor drivers. Little trick about speed adjustment: crystal oscillator run at 34700 Hz, so, at normal speed there must be 1084,75 Hz (34700:32) from FG sensor. It can be measured with high impedance frequency counter at pin 3 of CX20084 (use oscilloscope probe cable with 1:10 divider). So, it's possible to do correct speed adjustment with no special test cassette. Do all in the same manner as described in manual: insert cassette with your favorite music, speed tune to off, remove solder bridge and turn RV601 to set about 1115 Hz at pin 3 (at 3 kHz test signal exceeding the frequency must be about 90 Hz, 3090 instead of 3000; so, at near 1 kHz it must be triple smaller, about 30 Hz); then restore soldering bridge, set speed tune to the middle position, and turn RV603 to set 1084,75 Hz at pin 3 (tolerance is triple smaller: +- 3 Hz instead of +-10 Hz for 3 kHz test signal).
wow, that's a project. To redesign a custom chip is much harder than redesigning a microcontroller (as I did) since it's mainly software based. I reckon the major problem in this case is the size of any kind of replacement, to fit everything into the narrow case of a walkman. Have not found a datasheet which explains the features and I doubt that such a datasheet got published. Would do it the same way as radionick mentioned. Sounds very promising, although still a huge challenge. Good luck!
A datasheet has never been published for sure, as it's a proprietary design (as all CX type ICs). But the internal block diagram is available in the service manual and I reckon that is enough to fully understand the design when waveforms at different pins are also observed on oscilloscope in different working conditions. I have looked at the waveforms in the meantime and I maintain my opinion about doing it fully analog and @radionick seems to agree. I have thought about the footprint more since the last post and I have a different idea (as a SO-16 footprint seems less feasable to me): on the back of the PCB there is a lot of empty space that can be used to mount a custom solution. Then the connection can be made through-hole, by drilling 0.5mm or smaller holes in the middle of the SO-16 pads. This way, one, two or even 3 vertical boards can be mounted in that space, eliminating the size constrainst completely. Sure, it won't be an exact plug and play replacement, but I think it's a good compromise, as drilling the additional holes is something that can be done easily. @radionick Good trick about adjusting the speed, have not though about that before. As of accuracy, from what I have seen the crystal oscillator always runs a little below 34.7kHz, meaning the tape speed will always be a little bit lower than 3000Hz (but witin 6Hz or so). This is the case on all SONYs that have a PLL (Quartz lock), not just on the D6C.
Fully understand... ))) Now I think about signal on pin 9. Short pulses (about 10...20 us) with the same frequency as FG signal, duration varying about 50% in normal run (playback) mode. Frequency-to-voltage converter with varying period and varying duty circle? Compare non-linear sawtooth with internal reference?.. As a result, very specific V/F dependency? Interesting...
Yes, it's one of the major problem. But modern QFN, DFN, TSSOP and DSBGA cases is a way to do PCB practically as small as old SO-16, and pin-to-pin compatible. I think, near a half of functional analog module may be designed with LMC555, which available in DSBGA 1,5x1,5 mm footprint.
