Chasing the Sony CX20084 speed chip for D6C walkman

One thing that catches my attention is how this circuit is powered: it's using the +11.2V rail from the DC-DC convertor and regulate it down to 5V.
The motor draws ~100mA with load, not sure if that DC-DC can actually supply that much extra current (without significant ripple) since it wasn't designed to power the servo or the motor, but just audio circuitry.

I think it's a better idea to power it directly from the 6V rail with some protection which has to be low-dropout given minimum supply voltage of BA6301 is 4.5V.

As for the FF/REW auto-off, that has to remain.

Wonder if the other Chinese replacement (https://www.goofish.com/item?&id=766454209480) is also based on the BA6301.
In this case only the IC is powered through a regulator from the 11V rail, not the motor.

 

I feel the same, I have already written off this board as a lab experiment in what NOT to do. Last evening I modified it heavily by connecting both motor and FG directly (contrary to Chinese instructions) and FG noise went way down. I also twisted the FG wires like a twisted pair to maybe reduce some motor noise. Finally I installed a (larger, throughole but more suitable) transistor and to suspend the motor from the 6V rail. I think LeKos forgot how weak the 10.8V rail is AND the motor noise you get in audio as well. He has a second problem, tying pin 5 of the CX20084 (or pad 5 with that chip gone) directly to 5V. Not sure how that works but it certainly exceeds the 3.15V(brown PCB) 4V(Green PCB) Electrolytic rating on that pin. My Chinese board has a trimmer that biases pad 5 between 0-5V (factory set to 3.65V). It’s totally unstable however. Sony indicates that pin as 1.65V. Any opinions on how to determine that?

I think the other Chinese board looks suspiciously like the Korhonen board for early brown PCB units only, where you must remove the components around the CX20084 as well to attach the board to the battery solder tabs. I think the solution (for later units anyway) will be a more rectangular board that fits in the bottom of the cabinet (top of the unit in use) beneath the DC-DC converter. At least on the D6C I have there is plenty (well some room down there.

 

It seems you are looking down on the Ba6301 proposal so much that you didn’t even bother to properly look into it, while also doubting Lekos's results. You also seem confused about the schematic provided by Lekos.

I would like to assert that the schematic is completely rational, fully compliant with standard specifications, and extremely close to the original diagram. It operates with 100% stability (I have been running the Ba6301 for two years now and it runs stably, even better than the original due to the high-quality capacitors and resistors).

And please note, so there is no confusion: the motor power is drawn directly from the battery and is completely unrelated to the 11V power supply. The 11V supply is converted to 5V solely to power the IC and serve as a reference voltage (which is why it doesn't drift and is extremely accurate, exactly the way the CX20084 operates).

I don’t know how the Chinese circuits are designed, but Lekos’s schematic is definitely the most perfect solution for even the ultimate perfectionist. It could have a simpler version where you just need to supply 6V battery power, 11V, and two pairs of wires for the motor and FG to get it running (omitting the speed adjustment and quartz lock functions, which do not affect Wow and Flutter).

Please look into it carefully and try it out—then you will see just how perfect the results are.

 

I assure you I do not look down my nose at the LeKos design or the chip he uses. I am in fact starting a project using it. I referred to the Chinese module I ordered as deeply flawed and I have reached out to LeKos with questions that came to me as I studied both his original and the knockoff. I think my concerns are genuine but will be the first to celebrate if I am wrong.

I do now see he tied the motor to 6V rail, I misunderstood and it was unclear from his schematic (the Chinese engineer tied it directly to 10.8v which was a terrible mistake. I hope I’m also wrong that LeKos is tying Pad 5 of the CX20084 to 5v, the cap on that pad is rated much lower.

 

Sorry about that, I lost my cool for a second.

 

No problem.. Did you build with his PCB? I am going to order a few and see if I can build it though I have little surface mount experience. Any tips gratefully accepted.

 

@Hn14197 Ok, I made a mistake regarding LeKos schematic: it's only the Chinese that has the mentioned problem. Edited my post to correct it.
But I think it's more helpful to just point out that I made a mistake, rather than assuming things about me: that I look down on people, that I look down on the BA6301 proposal itself.
Also never said LeKos solution isn't working or isn't up to spec (it's clear that it is, post #119 shows it), rather I was wondering why Chinese one doesn't since they are using the same IC.

 

I think perhaps I started this by reopening an "older" post and dissing the Chinese module I bought. In fairness to the Chinese designer, only one side of the motor is shown on the LeKos schematic and he made the wrong choice for the other end (10.8V). I think we can all agree that LeKos did us all a great service with his design. I for one plan to try to fix my D6C by building his board. I reached out to him by DM to ask if he has any updates to his files or I should just start there, no word yet.

Valentin, what little I know of the D6C I learned from your wonderful tutorials here (shoutout to Deb as well) and I appreciate the time and knowledge you share with us.
Hn14197, I think you first mentioned the BA6301 on this thread, thanks for that, it is incredibly close to the CX20084... let's hope they continue to be available, SMD packaging would be even better.

This is a great thread on a great board. I'm learning a lot from everyone.

 

I sincerely apologize for my hasty words earlier. First, regarding the purely analog solution, we already have Lekos's excellent approach. As for the MCU-based solution, even though I might have a slight prejudice against it, I still fully support it and look forward to seeing the results.
Although my knowledge is limited (as my major is mechanical manufacturing), based on the time I've spent researching this IC and the CX20084 circuit of the D6C, I would like to point out a few things according to Lekos's guide as I understand them. I hope this helps you guys visualize and review the setup, and proves useful to some of you. Please feel free to correct me if there are any mistakes:

1. Wire from "To pin 4 of CX20084" connects to the pad for pin 4 of IC601. Since both ICs have an output pin from the sawtooth wave generator block, this pin receives a reference voltage to determine the capstan motor speed—pin 9 for the BA6301 and pin 4 for the CX20084. On the D6C circuit, a reference voltage is already present at the solder pad for pin 4 of the CX20084. This reference voltage incorporates the results of both the speed tune function and the quartz lock. The voltage from this pin 4 solder pad is supplied to pin 9 of the BA6301 to provide the reference voltage for the IC.

2. Wire from "+5V" pin on the replacement board connects to the pad for pin 5 of IC601. Inside the CX20084, there is an integrated voltage regulator block, and pin 5 is its output pin. It supplies a stable voltage to the aforementioned reference voltage generator circuit. When switching to Lekos's circuit, we supply a stable 5V to the pin 5 solder pad. However, due to this voltage difference from the original design and compatible with the new IC, R603 and R604 must be changed. Additionally, a few minor adjustments (RV603) need to be made so that it runs at the correct speed when the speed tune knob is centered and there is no intervention from the quartz lock circuit (Speed Tune ON).

3. Brown wire #15. (OR the brown wire from the replacement board can be soldered directly to PAD 7 of the CX20084, but resistor R605 must be removed and do not remove wire 15 (brown) of the D6C).
This is the signal for the auto-stop circuit.

4. Blue wire #17 . (OR the blue wire from the replacement board can be soldered directly to PAD 9 of the CX20084 and do not remove wire 17 (blue) of the D6C ).
The quartz lock circuit needs to sample the Sawtooth wave from the FG coil to compare it with the quartz crystal frequency, then make adjustments by acting on the reference voltage. When the CX20084 is removed, this signal is no longer supplied, so it will be taken from pin 6 of the BA6301 IC (after the IC's Hysteresis Amp block).
5. Red wire from motor solder to the same place as brown wire #18.
To make it easy to understand, the Red motor wire is to be connected to the positive (+) terminal of the 6V battery pack.

And from what I can see, with certain circuit design modifications, a simplified replacement board could be made that just needs wiring to the CX20084 solder pads, with minimal modifications to the stock D6C circuit

 

Do have a certain bias towards analog solutions myself, although there's nothing inherently wrong with a digital approach either.

In my view there are 3 main approaches to doing this, each with some benefits and downsides:

1. A fully discrete solution, built just with transistors, resistors, capacitors, diodes
Advantages: it's cheap and future proof (can be built only using jellybean parts, doesn't need specialized ICs).
Disadvantage: With a minimalist design (3-5 transistors) likely won't obtain the original W&F. Also due to limited PCB space, the number of components can't be too high.

2. A solution based on an analog servo IC (such as BA6301)
Advantages: can match or exceed the original spec with a relatively low cost and small board footprint. Likely the best price/performance ratio.
Disadvantage: depends on a specialized IC that I think is no longer made (please correct me if I'm wrong). There is still plenty of stock, so there's no immediate concern regarding this, but it's not as flexible as 1.

3. A microcontroller based solution
Advantages: can match or exceed original spec, loop parameters can be tuned in software, no aging/drift.
Disadvantages: cost and PCB footprint higher than previous solutions, also depends on a somewhat specialized IC.
While less likely a specific microcontroller won't be available in the future, manufacturers chaning process nodes can lead to this, requiring adjustments to the code and PCB.

Another thing to keep in mind is there are 2 versions of motors used in the D6C, 2 PCB versions and 2 servo ICs: CX20084 and CX069 in late models (latter less likely to fail).
Solution 3 gives some advantage regarding the 2 motor versions compared to an analog solution which would require different components.

The simplified BA6301 sounds good, this would likely solve the THT/SMD PCB problem also.

 

Anyone who has been following this thread will know I am in the process of building the LeKos board. I am still waiting for 2 components but the PCBs just arrived and due to a mixup, I got 16 of them.
So, if anyone is interested, I will be happy to send you one.

In the US, just DM me for an address. Then send me a stamped envelope addressed to you and I'll be happy to pop a PCB in there and mail it back. They are truly tiny

Also, I would not try to solder these by hand. I am planning to write up some docs for the whole process including a Mouser BOM to help you order parts. From LeKos's post I have written instructions
for both brown and green PCB models. After I verify by building mine I will share those here if anyone is interested. I'll also show the inexpensive solder paste and hot plate I use to make surface mount easy. -Tim

 

Hello everyone!

I just wanted to clarify some misunderstandings regarding my notes about the installation of the replacement board. Hn14197's last message was absolutely correct. The board uses +10V8 only to power the BA6308 via the 78L05. The motor receives power directly from the battery pack or input jack! When designing the replacement board, I tried to use as many original components and wires as possible, so I connected the sawtooth generator circuit to the main board. I wanted to maintain the original function of optional speed tuning, and I wanted to use RV601 and RV603 components like they were in the original design. The board could be simplified if all passive components needed for the sawtooth generator, including potentiometer for speed adjustment, would be puted on the new board. This would make installation easier, but it would also remove the speed tuning features. Additionally, the part of the circuit that inverts the auto stop signal could be removed, although I wouldn't recommend doing so. This is a simple inverter involving just one generic transistor and a few resistors, and, in my opinion, the auto-stop on FF/REW is an important feature that should be preserved.

Also, I have read the latest messages regarding the design of a replacement board using an MCU. If it works, I believe it would be a better option than the BA6301, as it will use components that are manufactured today and will continue to be produced in the future.

I cannot guarantee that the BA631 will be easily obtainable. And at the current time, when Boyeen is awaiting the arrival of the final components and has not tested the board yet, I can't guarantee that this board will function as it is in another unit. I've only tested it on my D6C, so maybe some changes in component values will be needed for another Walkman.

There is one important point I want to emphasize. It's not about correcting schematic or PCB design files. Rather, it's about the different component values. The schematic is correct and functional!

I am looking forward to receiving Boyeen's messages about installing my board. I'm a little nervous about it. Will it work well in another unit after assembly without any problems? Will there be any changes to the components? We'll see. I will try to help. Feel free to ask me here or in inbox messages. My responses may be delayed. Apologies for that. English is not my native language, so it takes me time to respond.

Best regards, LeKos

 

Another interesting thing about CX20084 is that before I decided to create a replacement board, I tried to figure out which part of the IC was dead and which was working. In my case, it seems like the only part that was not functioning properly was the Phase Corrrector, an OP-AMP connected to pins 13-15 of CX20084.

I tried to bypass it by connecting another OP-AMP to the same pins and it worked for about a week. It was about a year ago, so I don't remember exactly why I eventually decided to move on to a new board. I do recall that I might have burned down my CX20084 in some other way, which caused it to completely stop working.

Here are some photos of the external OP-AMP in the attachments. The Q601 was not the original one - the previous technician that tries to repair this unit decided to install a convenient IC for stabilizing the motor speed, but it worked terribly - W&F around 0.2 RMS and speed drift when carrying the device in hands. So, that technician removed Q601, that`s the reason why I installed a different transistor.

What will you do with that information? I'm not sure, but this is a possible way to "fix" the original IC if it dies in the same way as mine did.