WM-DD33 amorphous head mod

This thread covers a mod of the WM-DD33 I did for @Emiel, as he wanted an improved DD33 in terms of sound quality.

We exchanged ideas and settled on:
1. Head swap: from the 42311 original to an amorphous 35711 head from a donor WM-D3
2. Upgraded capacitors - initially, we wanted to include audio grade capacitors.

A. Capacitors upgrade

Let’s first tackle the capacitors.
Audio grade caps are no longer manufactured with the required small footprints.
Nevertheless, the capacitors were all replaced with those from the kit from Marian Mihokm: https://fixyouraudio.com/product/sony-wm-dd33-capacitor-upgrade-kit/.
The kit also includes higher values for the preamp feedback loop (47μF instead of 10μF) and output coupling ones (560μF instead of 220μF).

I did listening tests and measurements both before and after installing the 560μF output coupling caps.
These make little difference even with 16Ω headphones, and I don't have ones with lower impedance to test.
This is to be expected since there are 2 more impedances in series with the output:

1. The TA7688F output impedance. The datasheet does not specify a value but given it's an older design I will assume it in the order or 1-3Ω.
2. The 3.9Ω series resistors R115 and R215.

If we add the 3.9Ω with another let's say 1.5Ω, this gives us an impedance of 5.4Ω.
Without considering the ESR, a 220μF capacitor has an impedance of 14.5Ω at 50Hz, while a 560μF capacitor has an impedance of 5.7Ω at 50Hz.
Despite there is a significant difference, the overall output impedance at 50Hz is still relatively high even with the 560μF: 5.4Ω + 5.7Ω = 11.1Ω.

Conclusion:
1. 16Ω or lower impedance headphones/IEMS: biggest difference;
2. 16-50Ω, the difference is minimal, but still worth the upgrade;
3. 50Ω or higher load impedances, difference is non-existent;

When you buy the kit it's worth replacing those caps anyway since they come in the kit. In this small footprint they're pretty hard to find otherwise and if you consider shipping costs alone
from a large distributor Marian’s kit is simply cheaper even if we are just to replace the 4 mentioned capacitors and nothing else.

NOTE: Measurements are done with Dolby NR off, EQ switch on metal position (BASF Metal Maxima used for test), MegaBass on NORMAL position.

Overall, the upgraded capacitor values make the sub-bass present, while with original values the sub-bass is dull.
Now, depending on the headphones you use you might want to drop the feedback loop value to 33μF if the headphones already have a bass boost or keep it at 47μF if you use more neutral headphones.
Of course, it is to be noted the flat response of the overall system does depend on a properly recorded cassette!

 

B. Head swap

Now, going to the head itself. Before swapping the head, I measured its DC resistance and inductance and the values are summarized below:

Permalloy head 42311 DC resistance:
L - 287Ω
R - 288Ω
Amorphous head 35711 DC resistance:
L - 296Ω
R - 310Ω

Permalloy head 42311 inductance (1kHz):
L - 125mH
R - 129mH
Amorphous head 35711 inductance (1kHz):
L - 116mH
R - 117mH

From the measurements we can observe 2 aspects:
1. Both the inductance and the DC resistances of the 2 heads are close enough to each other for the swap to work without any additional modifications.
2. The amorphous head has tighter tolerances as far as DC resistance goes, hence why the later models do not have Dolby level adjustments anymore.

In this case, the Dolby levels were within specification and did not require any fine tuning.

NOTES:
1. The bandwidth I consider is the frequency where the response drops by 3dB.
2. It is to be noted there is a certain margin of error within the measurement, so 16kHz @ -3dB is only an approximation based on averaged measurement.
3. Both heads were lapped and de-magnetized before doing measurements.
4. Measurements are done with Dolby NR off, EQ switch on metal position (BASF Metal Maxima used for test), MegaBass on NORMAL position.

After measuring the frequency response, an improvement of 1kHz in the overall bandwidth can be seen.

 

C. Is the difference between the 2 heads audible ?

To my ears while listening on Beyerdynamic DT900 Pro X headphones I would say no.
Why ? because of 3 main reasons:
1. Human hearing is not that granular at high frequencies, so for such an improvement to be audible we likely need a larger increase in bandwidth. See linked video for more details.
2. There is very little music content above 10kHz, even more so above 15kHz. So while differences do exist, they're minimal and pretty subtle.
3. The threshold of human hearing at high frequencies is pretty high, meaning you will only hear loud sounds.
And in music the details at those frequencies are of a low volume, meaning they're likely not that audible to begin with.

NOTES:
1. Regarding point 1, it is to be noted on walkmans having an 18kHz frequency response, I do hear a difference compared to a WM-DD33 (15kHz).
The differences are even higher if we are to use Dolby encoded tapes.
2. This was an experiment more than anything else. Of course, for an in-depth comparison other parameters like crosstalk, THD and IMD should also be measured.
3. Video link: Understanding Audio Frequency Response & Psychoacoustics - YouTube

 

@Valentin : great description of our experiment!
Did I already mention it looks great with the blue LED? Well, it looks great with the blue LED!

 

Thank you for a thorough and detailed write up on your experiment. One of the best technical things on analog audio repair/mod I’ve read in a while.top notch!

 

they were doing this 20 years ago and it does improve things....great work guys

they were using higher end aiwa walkmans with dolby c but im, not sure what difference that would make....i had an aiwa from uk and one from japan both recapped but the japan 1 had a higher head as standard from japan....the japan one sounded better lads it just did

now i doubt i could tell as my hearing is nothing like it used to be, great work lads

 

Because a WM-D3 is able to make recordings, I presume Sony used a head that has a large gap, which is nessesary for recording.
Such a head theoretically can't reach frequencies as high as a dedicated playbackhead with a narrow gap.
That is why a 3-head cassettedeck usually has a better frequency response compared to a 2-head deck, recording and playing use their own heads.

I don't know if Sony equipped their Walkmans with dedicated playbackheads. Because Walkmans are not really 'Hifi', I doubt it.

It would be an interesting experiment to try a head from a 3-head (Sony) cassettedeck (if it would fit a Walkman at all) and see how the Walkman performs with it.

 

The frequency response of most of the DD and predecessor models seems to be the same: 40-15000Hz. See the table in the DD genealogy thread in my signature.
The DX100 and DD9 clearly stand out.
I can’t comment on the fit of the heads that were more commonly used in decks.

 

Unlike a 3-head recorder, on a 2-head recorder the head is a compromise between recording and playback.
So the gap is not as large as on a record-only head used in 3-head decks.
A smaller gap in the record head can be compensated for with a stronger bias singnal, so there is some room for compromise.

Don't know if SONY used dedicated playback heads. But what we do know is the 35711 is not used in any playback-only device.

Models having a response going higher than 15kHz all have a slim-type head which very likely has a narrower gap.
It's to be noted none of the these models use the BA3304F head preamp. This IC might be a limitation itself, apart from the head gap.

As for doing an experiment with a head from a 3-head deck, that's an interesting idea.
Don't know if one can be fitted to a walkman and if it would sit at the correct height.
Unfortunately I do not have any such spare head at the moment.

 

Great experiment and overall discussion!

 

It's not really necessary to use a head from a 3-head deck. There are many twoheaders reaching 20 kHz in playback.
Any entry-level Nakamichi model will reach that frequency.
From the BA3304F datasheet (attached) I do not conclude that this IC will be the limiting factor.

 

Indeed the BA3304F is not the limiting factor as the closed loop voltage gain remains contant until above 30kHz.

Looking at the datasheet of TA7688F I can say the headphone amp is not a limitation either.
The NJM2063 Dolby B NR IC: the datasheet doesn't offer enough information, but it's unlikely this is a limitation.
The NJM2106 MegaBass amp is not a limitation either, the datasheet shows a flat frequency response beyond 20kHz.

As far as a test with a better head, will consider doing this in the future. At the moment I don't have any suitable spare head for this being it a single or double head.

 

Great (color) combination with the dark blue planar in-ears from Hidizs!