Step-by-step tutorial for restoring a SONY WM-D3 Professional walkman Parts you will need: MECHANICAL: new center gear, new capstan ring, new small reel gear, new counter belt; https://fixyouraudio.com/product/sony-tcm-d3-walkman-new-center-gear-repair-kit/ https://fixyouraudio.com/product/sony-wm-dd3-small-reel-gear/ (it includes the counter belt) https://fixyouraudio.com/product/casptan-ring-wm-d3/ ELECTRONIC: Dolby level tape, speed calibration tape, blank tape for bias and rec level adjustment. https://www.ebay.com/itm/Playback-Dolby-Level-Crosstalk-Test-Calibration-Cassette-315Hz-333Hz-400Hz-1kHz/312997773618 https://fixyouraudio.com/product/speed-calibration-tape/ *For measurements you will need: TrueRMS multimeter/digital oscilloscope, function generator/digital audio player, 47K resistors (standard 5% tolerance are enough) for loading the line out. For ease of measurement you should purchase RCA to BNC adaptors for oscilloscope or BNC to banana adaptors for multimeters (you can make those yourself). NOTE ON PINCH ROLLER: A new pinch roller may also be needed depending on the condition of it: if the roller has indentation marks, pitting, cracking or glaze it should be replaced. https://fixyouraudio.com/product/sony-tps-l2-pinch-roller/ The TPS-L2 type pinch roller fits most older SONY walkmans including: WM-3, WM-D6C, WM-D3, WM-7, DD range and others. It does fit other brands as well. Read this tutorial for pinch roller replacement: https://stereo2go.com/forums/threads/pinch-roller-replacement-tutorial-for-dd-series-walkman.7080/ NOTE: This walkman does not suffer from capacitor problems, so I would not replace any capacitor unless there is a specific problem. The output capacitors on the headphone amplifier can be replaced with 220uF ones if you want a better low-frequency response. The line amplifier and recording amplifier could be replaced with NE5532 for lower noise and lower distorsions. DISASSEMBLY & MECHANICAL REPAIR 1. The first step would be to take the cassette door out. You need to remove the 2 screws on the bottom and open the latch. 2. Then the back cover can come out, it's held in 4 screws. Be mindful of the latch, as it has a spring that will fly out if not careful. One of the screws is longer than the rest, that is why I like to lay them down on the table in the position they fit on the unit, so it's easy to reassemble back. 3. You need to slide the PCB to one side. It's held in 6 screws: 2 of them on the side with the jacks are holding a part of the case (that covers the jacks and pots) that needs to be taken out. These to are self-tapping screws. The rest of 4 go into metal inserts. After taking out the screws, you must also detach the tape that is holding the wires to the PCB. 4. Check that the motor spindle is in full contact with the capstan ring. Mine was half way out and it slipped, making grooves in the capstan ring. I discovered the problem was caused by a bent motor cage. I market on the picture the part that was bent (in the picture it is already straightened). Don't know the cause of the bend, but another forum member had this problem in a DD3. That's why it's something worth noting, as most likely it's not just my unit which had this issue. 5. At this point you can follow the tutorial on replacing the center gear. https://fixyouraudio.com/tutorials/sony-dd-series-step-by-step-repair-guide-for-dummies/
ELECTRONIC ADJUSTMENTS It is critical to do the adjustments in the order presented here, as one is influenced by the other. You cannot set the recording level or bias, if the play level or azimuth is off. TAPE SPEED ADJUSTMENT *If the unit had leaked batteries, it might be necessary to clean the trimpot with contact cleaner and turn it a few times from end to end. Using a 3Khz test tape adjust the trimpot to obtain a 3Khz signal at heaphone output or line out. Can be measured with multimeter, oscilloscope, frequency counter or even a computer software through the line in of the sound card (http://www.techmind.org/audio/#mustun). AZIMUTH ADJUSTMENT On this unit there is more than one adjustment that affects the tape path alignment, so be careful. The tape path adjustment screw is recommended to be left alone, unless the tape is curling when played. I would also suggest if playing a known good tape sounds ok and the lock paint on the azimuth screw has not been tampered with, do not ajust it. METHOD 1: Play a known good tape and confirm that the high frequency response is correct (the tape should NOT sound dull/muffled). If the sound is muffled, adjust until the treble is restored. METHOD 2: Requires a 6.3Khz -10dB test tape. You need to observe the output signal on the scope and adjust until both left and right channel have the maximum amplitude. PLAYBACK LEVEL ADJUSTMENT (RV101, RV201) You will need a TrueRMS multimeter or digital oscilloscope to do this adjustment correctly. *Using a computer sound card and software to make this adjustment is not recommended for 2 reasons: 1. Usually, it has low input impedance, which will be placed in parallel with the loading resistors, lowering the load value considerably. 2. The voltage reference on the ADC is not accurate enough for making precise measurements. SPECIFICATION & UNIT CONVERSION: SONY specifies the output level using a P-4-L300 tape (315Hz, 160nWb/m DIN). A Dolby tape is 200nWb/m ANSI (218nWb/m DIN), which is 2.8dB higher than the P-4-L300. * I have read here and on tapeheads.net some members saying the DIN standard was not accurate. I can say from adjusting SONY units that all of them had big L/R imbalance (which was consistent between the units). This made me believe that most likely the unit was adjusted from the factory with an incorrectly recorded test tape. * I have tested the unit with pre-recorded Dolby encoded tapes and it plays them perfectly, so I firmly believe that using a Dolby tape is the correct way to do the ajustment. These are the levels you should be getting at the LINE OUT, terminated with 47K resistors (note that the reading will be higher if the output is measured on high impedance, like 1M on oscilloscope or 10M on multimeters): - with P-4-L300 tape or equivalent: 250mV RMS - with Dolby tape: 340mV RMS * the VU-meter on this unit will read -5dB when PB level set correctly (the Dolby mark is below 0dB). That is when using a Dolby tape. From my experience, the PB levels are usually lower than spec, hence the loss of high frequencies most people encounter and blame it on Dolby itlsef being junk technology. After doing this adjustment, the unit should play a Dolby encoded tape completely neutral, without any high frequency loss. RECORD LEVEL ADJUSTMENT (RV102, RV202) NOTE: It is important to use a new tape when making this adjustment. Also important is to choose a tape that you will want to record on using the unit, as the magnetic properties of different tapes vary. Usually if the adjustment is done with a good tape, the results will work with most of the tapes out there, with minimal differences (the true chrome tapes like BASF Chromdioxid are an exception though). Before you begin: You will need to generate a 315Hz sinewave to do this adjustment. You can do that with a function generator, a digital audio player or computer sound card. Given that this unit does not output signal on LINE OUT during recording, you will need to monitor the output signal through the headphones output. INPUT - the signal monitored through headphones output during recording OUTPUT- the signal monitored through headphones output during playing back the recorded tape The goal of the adjustment is to make the INPUT the same as OUTPUT (ignore any L/R imbalance). The level of the signal recorded is not critical, it can be anything between -10dB and 0dB, but not higher than 0dB. You can adjust that using the internal VU-meter. ADJUSTMENT PROCEDURE: - take a piece of paper and note the INPUT levels: usually there will be a L/R imbalance even with perfectly matched input signals, but that's ok, because it's only the ratio that matters, not the absolute value. - record the tape for about 10-20 seconds. Rewind, play back and note the OUTPUT levels and write the down. - compare the INPUT vs OUTPUT and adjust the trimpots accordingly. - record again and compare. - repeat the procedure until the difference becomes minimal (a few mV RMS is ok). You will need to repeat the procedure a few times to get is right. BIAS ADJUSTMENT NOTE: As in the previous adjustment, you also need to use a new tape. The tape type is more important at this step, as the bias adjustment depends more on the magnetic properties of the tape. Use the same tape for both rec level and bias. In my experience, using a good tape will yield good results on most tapes out there with minimal differences. Since the signal recorded at this step will be at 10Khz, it's best to use a chrome tape because of the better high frequency response. Before you begin: You will need to generate a 10Khz -10dB signal. The exact level is not critical, but it needs to be a low-level, since the head cannot record high levels in the treble range. You can use the internal VU-meter and adjust your signal source so that the -10dB led if fully lit. The adjustment on this unit is done by soldering pads which connects different capacitor values in the bias oscillator circuit. The values are between 15pF and 27pF; there's also a 2.2pF for fine adjustment. Since there are only incremental capacitance values, do not expect the final result to be perfect, beacause it won't be (but you can get it very close). Multiple capacitors can be soldered if necessary. If the output is higher than the input, you need to increase the capacitance. If the output is lower you need to decrease it. You can solder more than one capacitor if needed. The goal of the adjustment is to make the INPUT the same as OUTPUT (ignore any L/R imbalance). ADJUSTMENT PROCEDURE: - take a piece of paper and note the INPUT level. - record on the tape - compare the INPUT vs OUTPUT. Write down the values. - unsolder the current capacitor and solder another one accordingly - record again and check - repeat the entire procedure until you get the input close to the output (a few mV is enough) NOTE ON HEAD WEAR: If the output is still smaller than the input, even when only the 15pF is connected, the head needs lapping or replacement. These amorphous heads do wear out, especially from cheap poor-quality TYPE I tapes. Usually they can be brought back to life with a lapping, but that's not always the case. The most common indicator of head wear are pits in the area where the tape contacts the head. If the pits are not near the head gap, the head can perform well even without lapping. RECORD LEVEL RE-CHECK After doing the bias adjustment, you will need to check the record level again. The bias adjustment does influence the recording level. If the head does not have considerable wear and the bias adjustment was not huge, the record level will be close and will only need a tiny adjustment.
I appreciate your effort for this topic. This can serve as a reference if you edit it and make some changes. the way it is all typed in one post makes people read a little and skip the rest. I have few suggestions: 1- separate each step with a unique relevant picture is a separate post. 2- highlight the title for each step with BOLD font to make it easier to navigate through posts. 3- makes the posts very understandable for newbies.
@Boodokhan Thank you for the suggestions ! I want to make the thread more readable, as I also think it's not easy to read. Will take more pictures to make a relevant one for each step and edit in the following days. @CDV Given that I do not have a wow&flutter meter I cannot say what the value is, but there is no audible wow&flutter. Listened to different types of music, including ones with piano passages, where wow&flutter would be more noticable.
All you need is a computer and a testing cassette. I use a non-professional cassette, so the numbers I get are likely higher than they would be if I used a professionally-produced cassette, but it is good enough. Anyway, I was just wondering what the ballpark number is.
@CDV You made me curious to measure it. I also have a test tape recorded on a direct drive quartz-locked deck, which I think is far from perfect for this kind of measurement, but good enough for speed adjustments (which is what I mainly use it for). I can measure both the deck it was recorded on and the WM-D3 and maybe substract the 2 values to give a more realistic reading. What software do you use for measuring wow&flutter ?
I use WFGUI by Alex Freed. It is a Windows program, but people were able to run it on other systems using a VM. If I understand correctly the numbers it shows, the "DIN RMS" number essentially means WRMS, while the bouncing needle at the bottom right ("DIN PEAK") means DIN value. Usually DIN PEAK is about twice higher than DIN RMS.
I am suprised by what I have found. With my non-professional test tape, I obtained 0.14% DIN, compared to 0.13% specified in the manual. So in a worst case scenario (in which the test tape I made myself is actually better than I think), the unit is extremely close the original specification, possibly lower. The pinch roller was not replaced in this unit, because it is in very good condition. But I bet a new pinch roller would lower the w&f even more. I was curious: measured my WM-D6C and it reads 0.11% DIN (compared to 0.14% specification). The unit has new rubber (including a new pinch roller, which I do think has a big contribution to this low reading) and is lubricated properly (but the quartz PLL is disconnected, since it drifted on mine). Gave it a try to the TC-D5M, equipped with original rubber (which I hope will get a replacement when one will be available from Mihokm) and it reads 0.05%-0.06% (specification is 0.06%). This unit does have a new pinch roller. I am really not sure though what value is correct: it seems that my reading is too close to the spec, especially with a non professional tape. If the software reads in WRMS as you state, that would be more realistic. The problem is in the software interface it's written DIN and in the service manual DIN is a spec separate from WRMS. WRMS is obviously not a standard (I guess it means weighted RMS), so it could be measured using different standards. I would really like to know what is the value I should be comparing my readings with. So if someone with more knowledge about this knows the answer, please post it. EDIT: 1. I think the DIN value in the service manual should be compared to the peak reading in the software. 2. The WRMS value can be roughly compared to the WRMS (NAB) value from the service manual. Don't know what the difference is between the NAB and DIN standards.
On this screenshot RMS (0.09%) means WRMS when DIN is selected from the dropdown. PEAK (0.12%) is basically DIN. See this thread for more info. Quoting A.N.T.: "When you use the DIN mode with a 3kHz tape the RMS figure is JIS" and "JIS = WRMS". I wish my deck could have numbers like this: I figured for myself that for listening to pre-recorded cassettes, which have very low baked W&F, anything less than 0.1% (all numbers are WRMS) is good for pretty much any content. 0.1%-0.15% is a gray zone. 0.15%-0.2% is ok for pop stuff, but not good for classical or any other music with slow piano and strings. 0.2%-0.25% is barely acceptable, and above 0.25% is unacceptable. Funny that a $500 TEAC "professional" cassette deck that TEAC sells these days is spec'd at 0.25%. What irony.
Ok, I understand now. In this case, my WM-D3 measures a little above 0.20%, which I still consider very good comparing to the 0.13% spec. The deck that I recorded the test tape on has a 0.05% w&f (it's the spec, not an actual measure, but given it was serviced and has a new pinch roller, I expect it to be close to the spec), which confirms because the peak value varies between 0.20%-0.25%. So I am pretty confident that the actual value is in the neighbourhood of 0.20%.
I can say that when I did the test, I could clearly hear the wow&flutter with the 3Khz test tone. But when listening to music it's not noticeable, at least not obvious. It's certainly related to the fact that I don't listen that much to music with slow piano and strings. I must admit I did not pay a lot of attention to this parameter because I appreciate these old devices for what they are and don't have extremely high expectations from them. But certainly became more interested now. I think that 0.25% was pretty much the standard for consumer players and recorders using belts. I think most walkmans are above that value. The TPS-L2/WM-3 doesn't even have a spec for w&f. The TEAC is like that because the only mechanism that a manufacturer can get nowadays is the Tanashin (Tanashin - the last cassette mechs & Aldi's 2019 Boombox - YouTube), which is far from a high-end one (it's belt driven and the motor doesn't even have a servo control). The decks and walkmans made nowadays are mostly junk in my opinion. That's why I am only interested in vintage equipment, especially from 80s. What deck obtained that 0.021% w&f in your picture ?
Here is my WM-FX407, which shows about 0.17%-0.20% WRMS, but occasionally jumps to 0.22% area. Feel free to skip to 0:37 for the W&F test and a test of a classical piece. What do you think? My best cheap walkman, WM-FX435, shows about 0.15%. Right, which is why I find $500 price outrageous, and the whole idea of this deck being "professional" laughable. I got a 40-year old Panasonic low-end deck cheap, it was spec'd to 0.05% WRMS, I tested it to have 0.07% using a non-professionally made tape. It is Revox B215. I grabbed this picture on TapeHeads.net.
Regarding to your WM-FX407, I think the w&f is audible with the music you are playing, but I would not consider it bad. It's just that it is there and it's clearly more obvious with the kind of music you're playing. But the value you are obtaining is typical for a walkman: I measured a couple of mine and the better ones are in the 0.20% WRMS area, the less good ones in the 0.30% area. I put a picture with the WM-D3 measurement. I remember someone on the forum suggested that sanding the capstan rubber ring will lower the w&f. I did not do that on mine, but I guess that doing that and also replacing the pinch roller will probably bring the value down close to the specification. Edit: I decided to lubricate the bearing on the back of the motor (the front one was already done). The result is a significant drop in wow&flutter. There was also a slight motor hum that could be heard in the headphones at minimum volume, that is also gone.
Nice! What did you lubricate it with? Liquid oil or grease? Which one? I bought synthetic grease and watch oil, but I haven't used the oil yet. I've read that unlike traditional oil that is designed to penetrate through seals, watch oil is designed to stay where it is applied, so to use it correctly I need to disassemble the motor, which I don't want to do. Usually I break things when I take them apart P.S. Weird frequency value: 105 Hz.
I lubricated it with watch oil (Moebius 9010). Grease is not suitable for motors, it's too thick and tends to dry over time. The watch oil is very fluid (low viscosity) and has a high viscosity index (viscosity does not increase much with drop in temperature). It does have a good grip to the material, but that does not mean it does not flow. I attached the datasheet for anyone interested. I did not disassemble the motor, it has a plastic cover that I removed. I would not disassemble such a tiny motor, as it's likely to ruin the brushes in the process. The motor does not have any seals: it either has an opening on the back where you can put oil in (does not depend on the type of oil used) or it does not, in which case it needs to be disassembled. Big motors can be disassembled without any damage. In smaller ones usually I only oil the front bearing which is accessible and in case of these type of SONY motors, they have this opening in the back. Otherwise I live with it as is, as it would be very hard to find such motors nowadays. The 105Hz reading is a glitch in the software. It does that when you stop the unit and the software keeps running with no signal.
Hi there, Thanks so much for the step-by-step tutorial! One question - I’ve been trying desperately to understand importance of matching my test gear’s input and output impedance against those specified in the WM-D3 manual (and those for other units I’m looking to calibrate as well). I noticed you mention terminating each of the WM-D3’s output channels with a 47k resistor, but nothing about resistors for adjusting input impedance of the WM-D3 based on what’s been specified in the manual. I’m using a Rolls MO2020 Testoscillator as my signal generator, which has an output impedance of 100 ohms. My understanding is that a 600 ohm output impedance from a signal generator is usually what’s expected by calibration instructions, and so I was wondering how important you thought it was to somehow adjust my input impedance accordingly. I also see that the manual shows a 10k ohm resistor in series with either the channel’s positive or negative lead, in addition to the parallel 600 ohm resistor. I’m curious why that series resistor has been included, and if this should also be accounted for in my current test set-up? If so, should it be added to the positive or negative lead (or doesn’t it matter)? Ultimately, how important are the input / output impedances of the test gear itself specified in service manuals when it comes to proper calibration? Again, really appreciate the information! Thanks, John
Hi John, I'm glad the tutorial is helpful ! Getting to your question, the answer is pretty straightforward: when the signal you are generating is a voltage, matching of the input and output impedances is not what you want to do. Instead, you need a low source impedance and a high "destination" impedance so to say. In this particular case, the LINE IN imepdance is 47Kohms, which is about 2 orders of magnitude (100 times) higher than 600 ohms. Older analog function generator or audio generators have a 600hm output impedance. Newer digital ones have 50hms. A digital audio player has a very low output impendance (a few ohms). The idea is that you want the source to be able to supply current without a significant drop in voltage (low impedance) and the input to draw at little current as possible, so it does not load the source so much that it causes a drop in voltage (high impedance). Impedance matching is only needed when you want to maximize the power transfer, as in RF applications. In audio, the signals are voltages, so the technique called IMPEDANCE BRIDGING is used. You can read more about it here: https://en.wikipedia.org/wiki/Impedance_bridging Given what I have stated, any source impedance of 600 ohms or lower will work just fine. That's why I said you can use either a computer sound card, a digital audio player or function generator (because not all people have function generators). The SONY service manual is not very straight-forward when it comes to rec level and bias adjustment. They state the 10K in series with the source to create a 10K source impedance; that is because they want you to use the MIC IN (which has only 300 ohm input impedance). There are 2 problems with using the MIC IN: you need a very low input signal (which you won't get without an attenuator). That causes a lot of noise to be present on the signal. The second problem is the MIC IN has plug in power, which makes it unsuitable for coupling it directly to a signal generator output without a DC blocking capacitor in series. This is the reason I created a different, more straight-forward approach to making these adjustments. So the answer to your question is no, the source impedance does not matter. This is because in my procedure, the input voltage is not relevant (you put in whatever voltage gives a certain output) and I also state using the LINE IN, not the MIC IN. SONY manual on the other hand, states a very specific input voltage. In my opinion, this is less accurate, because it does not take into consideration tolerances of components; The variances of component values will generate slightly different outputs on different units, even with the same inputs. You can use your Rolls MO2020 signal generator without any addional resistances. Just connect it to the LINE IN. The output loading is important because if not loaded, the voltage will be lower than it needs to be. This is critical only when adjusting the PB level. In REC and BIAS adjustment, you are only making relative measurements and they're done on the headphone out anyway (which gain is not as precise as in the line out amplifier).
Hi Valentin, This is seriously great information - so appreciate your taking the time to thoroughly explain. It sounds like, while not critical, it couldn't hurt to continue using the 47k resistors as dummy loads in-between the Line Out and Oscilloscope when adjusting azimuth as well, correct? I actually have a scope on the way, and had up until now been using a software scope via Focusrite Scarlett USB interface inputs. While I'm looking forward to learning the 'scope, it sounds like it's also the case that the input impedance of something like a USB interface (~60kohms) should also pair OK with the D3's 4.7kohms (is that right?) Line Out impedance when performing azimuth and other adjustments where the D3's Line Out is connected to the Scarlett's Line In. Last question for now - and this will make it clear that I'm very much still learning - Even though in this case it's not going to be necessary to add these resistors to my line input test signal, why does the manual show a parallel resistor in some cases (600ohm parallel resistor on output from signal generator, 47kohm parallel resistor on output from D3), while in other cases show a series resistor like the 10kohm resistor in series on line input? And do these various resistors affect both the output and input impedances of the connected equipment? i.e. do those 47k output resistors adjust the D3's line output as well as the oscilloscope's line input? Felt like this was my opportunity to ask some questions that have been burning a hole in my brain for the past week now - I'm grateful for the opportunity to clear these up. Thanks again, John
Yes, it is true that using the 47K load when adjusting the azimuth can help stabilize the output by having a constant load. However, the left and right peaks are usually not the same anyway, because of tolerances in component values and difference of the head L/R coils resistance and inductance. But given your measurement unit already has a 60K ohms input impedance, loading the output with 47K would not be necessary (60K is very close to 47K anyway). So yes, 60K is ok for connecting to the LINE OUT or HEADPHONE OUT. Keep in mind that if you load the LINE OUT with 47K and also use the USB interface with 60K input impedance, the total impedance will be 47K in parallel with 60K, which give about 26K (lower than the 47K). If you make the adjustment like this, you will load the output more than needed and make the voltage higher than the spec. Making the Dolby level higher, the treble will be emphasized (it will sound like Dolby does not have much effect at all). So if you use the USB interface to make the adjustment, use it without any dummy load. But if you have a scope on the way, wait for that. It's much more precise than the USB interface. The 47K dummy load only works when the input impedance of the measuring instrument is much higher (some orders of magnitude at least) than 47K. For example an oscilloscope has 1Megaohm input impedance and a multimeter has 10Megaohms input impedance: this will not not change the load resistance value in any significant way. Regarding your other question with the source stated in the service manual... It goes like this: - that 600 ohm resistance is parallel with the output is there because the AF attenuator does not have a constant output impedance (it varies with attenuation level). So that is used to create a constant load at the attenuator output. My procedure does not require an attenuator, that is why I have not mentioned anything about that. - the 10K in series, although the service manual shows it when connecting to LINE IN, I think they want that resistor there when using the MIC IN (a few kilo ohms input impedance). Note that modern function generators have even lower impedance than 600 ohms (50 ohms) and a Digital Audio Player has an output impedance lower than 10 ohms. So, in reality, that 10K in series may be useful when using older equipement with 600 ohms input imepdance, but with modern test gear it is not needed at all. My procedure does not follow the principle: put X level in, get Y level out, so even if there are minor loading effects, they will be cancelled out by the fact that measurements are relative in my procedure. That's why I ignored that part completely, because it's another step that just complicates the process without any real benefit. The answer to your last question is no, the various resistors do not change either the LINE OUT impedance or the oscilloscope input impedance. What it does is it loads the LINE amplifier to a certain level at which it needs to be measured to give to correct Dolby level. It's like an audio power amp, where it gives a certain power when loaded with 8ohm speakers and another power if you use 4ohm speakers.
