Hello,
did anybody open the Teac Wap2200 to see how can it be tweaked for better audio quality?
This streamer looks very interesting for various reasons:
-Remote with built in display
-Plays directly from USB storage, no need for a music server
-pretty low cost
The cons are (from what I can see):
-only optical SPDIF
-not the best audio quality
If the first issue looks easily solvable, a serious audio tweaking (clock slaving, IIS tapping) needs the knowledge of the inside.
So, before buying one, I'm asking if someone has already opened one and maybe knows which dac is used, or maybe can post some photos...
Ciao
Andrea
did anybody open the Teac Wap2200 to see how can it be tweaked for better audio quality?
This streamer looks very interesting for various reasons:
-Remote with built in display
-Plays directly from USB storage, no need for a music server
-pretty low cost
The cons are (from what I can see):
-only optical SPDIF
-not the best audio quality
If the first issue looks easily solvable, a serious audio tweaking (clock slaving, IIS tapping) needs the knowledge of the inside.
So, before buying one, I'm asking if someone has already opened one and maybe knows which dac is used, or maybe can post some photos...
Ciao
Andrea
Somebody already opened a wap4500 which i also have. Same thing with ethernet and wlan support. He changed the output caps (elco) with something better. Google for WAP4500 - Tuning.pdf. Some pictures in there too.
As for the DAC, it's a Wolfson WM8750BL. Not the greatest indeed. It's a codec (ADC/DAC with headphone amp).
I guess the 2200 is equal in that respect.
I'm thinking of integrating one into a diy dac, but i'm not shure it is possible (clock setup).
As for the DAC, it's a Wolfson WM8750BL. Not the greatest indeed. It's a codec (ADC/DAC with headphone amp).
I guess the 2200 is equal in that respect.
I'm thinking of integrating one into a diy dac, but i'm not shure it is possible (clock setup).
From what I can see the Wolfson codec uses a internal PLL to generate the needed frequencies starting from (generally) a multiple of 48kHz but it mostly depends on the actual implementation. I see 2 crystals on the left of the bigger photo but I can't see their frequencies. If one of them is multiple of 44.1kHz it means that the DSP "switches" the correct xtal and it might be possible to inject a external, low jitter clock.
As for I2S tapping, the DAC supports it along with other data formats, so it depends again from the actual implementation if it is possible to avoid SPDIF...
Ciao
Andrea
As for I2S tapping, the DAC supports it along with other data formats, so it depends again from the actual implementation if it is possible to avoid SPDIF...
Ciao
Andrea
I'm reading the datasheet too... 🙂
Question one, is the dac master or slave on the i2s signal. As the xtals are not near the dac, i'm now guessing it's slave mode. So than it's not interesting what clock is used for the dac's master clock. It's the clock of the dsp which generates the i2s signal which is of interest.
I'm not afraid if the signal itself is not i2s. Glue logic should be no problem. But i would also need to look at the sample rate as a pmd100 (the filter i have here for a dac) does not take 96kHz (55max). I would have to bypass the filter if the rate is higher than 48kHz i guess. More glue logic... Or downsample if i have 96kHz material.
Or still use spdif (still with the master clock in the dac) as it's output is 44.1 or 48 according to the manual. Makes me wonder if the dac would receive 96kHz with 96kHz material or also a downsampled version....
I'm currently putting all my CD's on my nas. At least that's all 44.1.
I guess i need a nice logic analyser for serial signals to check out the i2s signal (or whatever it is) and also the control signals to the dac. And then try material with different samplerates and number of bits. Or buy a Linn 😀
I remember you 're using the PMD100 together with the TDA1541A (which is also my favourite DAC).
Of course "our" DAC is limited to 44.1/16bit but most of the "software" is 44.1/16. And for the DAC I guess that a 96k/24bit IIS stream is equivalent to a 2x oversampled 16 bit stream (further bits are ignored IIRC).
And I have a ES9008 on its way...
Ciao
Andrea
Of course "our" DAC is limited to 44.1/16bit but most of the "software" is 44.1/16. And for the DAC I guess that a 96k/24bit IIS stream is equivalent to a 2x oversampled 16 bit stream (further bits are ignored IIRC).
And I have a ES9008 on its way...
Ciao
Andrea
I'm not using it at the moment as the (test) setup got unreliable. If you've seen the pictures of the insides of that 650 you know why
There is no DAC yet, but i'm working on one (or planned to). I have some output stages for a pcm63 setup (D1 and the old "tent" dac tube stage nearly finished) and enough pcm63's to build something. But as i'm now transfering to all music to harddisks, i stopped playing with cd players for now.
A Linn 24bit 88.2kHz flac testtrack plays fine on the WAP4500. It's 88.2kHz according to the remote, but what actually goes to the dac....
I have a WAP2200. For playing music in the garden or in the kitchen without having to run mains cables, it is great. For audiophile use it can improve a bit.
The built-in DAC is not brilliant, and output level is low (limited by 5V supply voltage). Probably also with SMD decoupling caps (most likely ceramic MLCC). Digital (SPDIF) out is Toslink and not BNC. No I2S out direct. Power supply is switch mode. It consumes almost 1A at 5V. The thing gets really warm.
The uP has IMHO too many tasks to do -- get data from USB, decode various music formats (almost anything you can think of), talk to the remote with lots of complex information exchange both ways, stream data, ...... I doubt the data is still jitter free. MP3 fans would love it. To be fair I have only tried the analogue output and not (yet) Toslink.
My preferred solution is the QA550 SD player. Has everything I want, totally solid state, has I2S out on top of SPDIF, which as stock already sounds great with a simple TDA1543 (just quick and dirty test). Plan to test external clock and I2S reclocking. Need a few more months.
For price and features, the TEAC is hard to beat. If you can read German, there are tons of discussions here :
http://www.hifi-forum.de/viewthread-181-2427-1.html
And pictures of the PCB :
Directupload.net - D5m6jake4.jpg
Directupload.net - Dcfis6np2.jpg
Directupload.net - Dr2fsvvf6.jpg
Directupload.net - Dx5aq8fpg.jpg
It's a very good consumer product, if you do not expect it to do anything high end.
Patrick
The built-in DAC is not brilliant, and output level is low (limited by 5V supply voltage). Probably also with SMD decoupling caps (most likely ceramic MLCC). Digital (SPDIF) out is Toslink and not BNC. No I2S out direct. Power supply is switch mode. It consumes almost 1A at 5V. The thing gets really warm.
The uP has IMHO too many tasks to do -- get data from USB, decode various music formats (almost anything you can think of), talk to the remote with lots of complex information exchange both ways, stream data, ...... I doubt the data is still jitter free. MP3 fans would love it. To be fair I have only tried the analogue output and not (yet) Toslink.
My preferred solution is the QA550 SD player. Has everything I want, totally solid state, has I2S out on top of SPDIF, which as stock already sounds great with a simple TDA1543 (just quick and dirty test). Plan to test external clock and I2S reclocking. Need a few more months.
For price and features, the TEAC is hard to beat. If you can read German, there are tons of discussions here :
http://www.hifi-forum.de/viewthread-181-2427-1.html
And pictures of the PCB :
Directupload.net - D5m6jake4.jpg
Directupload.net - Dcfis6np2.jpg
Directupload.net - Dr2fsvvf6.jpg
Directupload.net - Dx5aq8fpg.jpg
It's a very good consumer product, if you do not expect it to do anything high end.
Patrick
I've also seen the QA550, but it lacks 2 of the wap's best features, i.e. the remote with LCD for comfortable browsing and support for something bigger than a SD card (even tough reading from ss source has its advantages...)
Thanks for the pics😱. For what I can see it has a 11,289MHz crystal, so there is the possibility that such clock is used to stream 44.1kHz data. This shoud be verifiable looking at the relative phase between it 🙂4) and MCLK. If the relative phase remains constant....
I am aware that the Wap is not high end, but I hope it is tweakable to become high end 😀
Ciao
Andrea
A SD card has up to 32GB. I doubt you have the stamina to listen for that long. If you want 1TB of music on one hard disk, feel free to try the WAP2200. With 5 X'tal on one PCB, I am not going to try to figure out what is syn to what.
The tuning in post #2 is just changing the decoupling cap. Doesn't get to the root. I guess better than no change.
Patrick
The tuning in post #2 is just changing the decoupling cap. Doesn't get to the root. I guess better than no change.
Patrick
The xtal on the bottom is called Y7, so there are enough x'tals... But looking at the xtals near the '04 there must be possiblities to feed a clock in. The biggest problem to me seems to do measurements on those tiny pins and tracks 🙄
I'm thinking about still using the spdif (48kHz max) connection and feeding the masterclock back in. Then i don't have to care about the interface between the dsp and dac and converting it. And on how to handle 96kHz material if that would go to the dac as 96kHz over the i2s. And there is optical isolation if i would include that in the clock feed.
The other xtal is 12.295 which is close to the theoretical 12.288, so i think it is the xtal for 48kHz material. As i don't have any 48kHz material i know of, i won't bother with that one.
No SD for me. I have all my media at one nas with a standard upnp/dlna server to serve my tv, computer, audio streamers or whatever in the future. And there are already high end options for music like Linn. I think more will follow.
Guess it's time for a screwdriver 😀
I'm thinking about still using the spdif (48kHz max) connection and feeding the masterclock back in. Then i don't have to care about the interface between the dsp and dac and converting it. And on how to handle 96kHz material if that would go to the dac as 96kHz over the i2s. And there is optical isolation if i would include that in the clock feed.
The other xtal is 12.295 which is close to the theoretical 12.288, so i think it is the xtal for 48kHz material. As i don't have any 48kHz material i know of, i won't bother with that one.
No SD for me. I have all my media at one nas with a standard upnp/dlna server to serve my tv, computer, audio streamers or whatever in the future. And there are already high end options for music like Linn. I think more will follow.
Guess it's time for a screwdriver 😀
While waiting for my Wap I got caught by this sentence: the more I think about it the more I like it, but I still miss some details on how to do it... can you elaborate a bit?
Ciao
Andrea
Well the idea was to have the master clock in the dac and remove the xtal in the teac. Then feed the clock back (another toslink or so) to somewhere near the xtal. But it means hacking the teac. I would like to have the service manual before trying this and guarantee is gone.
So i think i'm going for option two: dac with pll (next to the one in the receiver) and voltage controlled clock. Maybe a bit more expensive, but more universal as it can also be used by other sources. But this will take time...
So i think i'm going for option two: dac with pll (next to the one in the receiver) and voltage controlled clock. Maybe a bit more expensive, but more universal as it can also be used by other sources. But this will take time...
So something like the XO-DAC from TentLabs... interesting and versatile.
Unfortunately such a unit would cost more than the WAP itself.... I think I will hack it (if possible)
Ciao
Andrea
Unfortunately such a unit would cost more than the WAP itself.... I think I will hack it (if possible)
Ciao
Andrea
Well, close. Not the DAC from tentlabs, as this is a diy project for me and not a kit build (i do a bit of both now and then).
More based on the old dac G.Tent worked on, together with some other people. And one of those other people made a modern version with the pll in software (a nos design).
It's going to be a mix of both as i'm going to use the old dac's output stage (i have the pcb nearly finished) for two pcm63. And CS8614 and pmd100 before that. And a pll in a pic16f628. So i need a VXO and that i'll probably source that from Tent again. And it requires a small i2c dac to drive the VXO.
Biggest job will be the programing of the pic with the pll and the i2c stuff, the 8614 has many registers. That's the plan and i'm actually busy with the schematic. See how it goes.
I'm still interested in your excursions in the teac! 🙄
I just received the Teac, time to think how to hack it....
I was thinking about keeping the hacked Teac isolated from the DAC.
This means that I need a double isolated channel, one for SPDIF and one for the clock injection.
SPDIF is easy, I just need a Toslink receiver.
Clock is harder, mainly because of the higher frequency.
Toslink seems not up to the task (max 6 Mbit), have to look if pulse trafos are ok.
I also have some ultra high-speed optocouplers PC410 but I'm not sure about their permformance at 11Mhz.
Ideas/opinions? Is galvanic isolation worth the trouble?
Ciao
Andrea
I was thinking about keeping the hacked Teac isolated from the DAC.
This means that I need a double isolated channel, one for SPDIF and one for the clock injection.
SPDIF is easy, I just need a Toslink receiver.
Clock is harder, mainly because of the higher frequency.
Toslink seems not up to the task (max 6 Mbit), have to look if pulse trafos are ok.
I also have some ultra high-speed optocouplers PC410 but I'm not sure about their permformance at 11Mhz.
Ideas/opinions? Is galvanic isolation worth the trouble?
Ciao
Andrea
I opened the WAP and gave a look inside.
Tried playing 44.1 kHz material and oscilloscope.
11.2 Mhz clock can be found on U60 (pins 3-4, but also others)
BCK can be probed on R426 near the DAC and is 2.8MHz
I probed both signals together and got a stable trace, so they are in sync.
Oddly when I try the same test between 11.2MHz clock and SPDIF (probing the across the LED) out I can't get a stable oscilloscope trace so it looks like the SPDIF output is not in sync with the master clock.
I'm afraid this can spoil my plans (SPDIF out to DAC board with XO and clock feeding back to the WAP).. or am I worrying too much?
Ciao
Andrea
Tried playing 44.1 kHz material and oscilloscope.
11.2 Mhz clock can be found on U60 (pins 3-4, but also others)
BCK can be probed on R426 near the DAC and is 2.8MHz
I probed both signals together and got a stable trace, so they are in sync.
Oddly when I try the same test between 11.2MHz clock and SPDIF (probing the across the LED) out I can't get a stable oscilloscope trace so it looks like the SPDIF output is not in sync with the master clock.
I'm afraid this can spoil my plans (SPDIF out to DAC board with XO and clock feeding back to the WAP).. or am I worrying too much?
Ciao
Andrea
Last edited:
Made another test with a CD player.
Used bitclock (master clock was under the board) and spdif out and they were in sync.
On the WAP I could clearly see that there was a sort of "beating" between the traces.
Yo avoid bit loss the (averaged) SPDIF and BCK frequencies must be the same, so I guess it will still be possible to reclock the incoming SPDIF-recovered IIS data with the same clock used to "slave" the WAP...
Surely time will tell but educated guesses are welcome ;-)
Ciao
Andrea
Used bitclock (master clock was under the board) and spdif out and they were in sync.
On the WAP I could clearly see that there was a sort of "beating" between the traces.
Yo avoid bit loss the (averaged) SPDIF and BCK frequencies must be the same, so I guess it will still be possible to reclock the incoming SPDIF-recovered IIS data with the same clock used to "slave" the WAP...
Surely time will tell but educated guesses are welcome ;-)
Ciao
Andrea
Clock injection - Part 1 (study)
As you may have noticed the Wap uses two different crystals to stream 44.1Khz or 48 kHz data.
The crystals are located near a hex inverter (74HC04) so at first I thought that the oscillator was based on logic gates with some trick to switch between crystals.
Investigating better I found a different beast: a colpitts oscillator based on a bjt ; the logic gates serve for xtal select and clock buffering.
The schematic I tracked shows how it works: the first inverter (between pins 5 and 6) polarizes one of the 2 diode strings (based on double diode BAV99).
The conducting diode string becomes the return path to gnd for the selected xtal (the other is left floating), the oscillator produces a sine of the desired freq and the logic gate on the right (pins 1 and 2) amplifies and squares the waveform. Not shown a second gate (pins 3 and 4) buffers the output (pin 3 connected to pin 2)
The problem now is: how can I perform clock injection without losing the double frequency capability?
Pin 1 would be a good candidate if using a single frequency (lifting it from the PCB) but I'm open to suggestions....
Ciao
Andrea
As you may have noticed the Wap uses two different crystals to stream 44.1Khz or 48 kHz data.
The crystals are located near a hex inverter (74HC04) so at first I thought that the oscillator was based on logic gates with some trick to switch between crystals.
Investigating better I found a different beast: a colpitts oscillator based on a bjt ; the logic gates serve for xtal select and clock buffering.
The schematic I tracked shows how it works: the first inverter (between pins 5 and 6) polarizes one of the 2 diode strings (based on double diode BAV99).
The conducting diode string becomes the return path to gnd for the selected xtal (the other is left floating), the oscillator produces a sine of the desired freq and the logic gate on the right (pins 1 and 2) amplifies and squares the waveform. Not shown a second gate (pins 3 and 4) buffers the output (pin 3 connected to pin 2)
The problem now is: how can I perform clock injection without losing the double frequency capability?
Pin 1 would be a good candidate if using a single frequency (lifting it from the PCB) but I'm open to suggestions....
Ciao
Andrea
Attachments
You've been busy..
Are there two inverters polarizing the diodes (one at a time)?
So why connect your clocksignal to the output, using the output of both inverters to select one of your clock signals. And forget the other stuff.
Interesting
Are there two inverters polarizing the diodes (one at a time)?
So why connect your clocksignal to the output, using the output of both inverters to select one of your clock signals. And forget the other stuff.
Interesting
- Status
- Not open for further replies.