Hi Russ,

I have an issue with my new Buffalo II DAC that I hope You could help me resolve.
I did extended testing in order to identify the source of the issue.

Description of the issue:

After "cold" starting the Buffalo II board there is a 2-3 minutes "warm up" (meaning getting the right temperature) period of the board. During that period consistent SPDIF lock drop-outs occur (about 5-6). When warmed up no more (not even one) lock dropouts occur on the same SPDIF source.

The following factors have been tested and seem to have no influence on the issue:

- SPDIF source jitter quality.
If the device is warmed I could not produce any drop-out even with low quality SPDIF sources (supposedly as result of the high quality jitter reduction of ESS), but after cold start during the warm-up period even extremely precise SPDIF sources produce drop-outs (e.g. mine RME Digi 96/8 board, that provides very low jitter: < 1 ns)

- SPDIF source sampling frequency and quantization bit depth.
It does not matter whether it is 44.1KHz/16bit or 96KHz/24bit or even 192Khz/24bit during warm-up period droup-outs occur. After warmed-up (no change in the source) no drop outs.

- SPDIF source level: Consumer or TTL level (through isolation transformer circuit with opened Buffalo II SPDIF switch) SPDIF .
Even feeding the SPDIF source into the Buffalo II through an isolation transformer circuit (providing TTL level and galvanic isolation) has no influence on the drop-out behavior. During warm up consistent drop-outs (even with different sources) after warmed up no drop outs with any tried source.

- Buffalo II Power supply.
The behavior is the same with LCBPS and shunt regulated power supplies (Even DC voltage was tried from 5.0 up to 5.5 Volts but has no influence).

- Cabling.
During all test I used a fine grade (75 Ohm) cable for connecting the SPDIF Source and the Buffalo. (Though with even lower quality cabling produced the same results.)

- Operation temperature of the device providing the SPDIF input.
Theoretically it could be assumed that the SPDIF source device has some influence on the provided SPDIF signal according to e.g. operation time/temperature. It has been also been tested, no influence.

- No ESS learning time.
If the device is turned of after getting warmed up and turned back on after a few seconds there is no drop-outs so it seems that ESS does not need relevant time (e.g. even seconds) for the identification of the SPDIF source format. (Theoretically it could be assumed that ESS Hyperstream DAC's Time Domain Jitter Elimination needs some learning time)

(All examined factors have been tested also in combination as I referred in some of the previous points)

Conclusion (on empirical basis, taking into consideration all tested factors)

It seems that the drop-out behavior is very consistent and coming from the Buffulo II (or the ESS 9018).

None of these device's (Buffallo II, ESS 9018) specification contains any operating temperature limitation that could explain this behavior (On the contrary, I assume that both devices should be fully functional on normal "room" temperature e.g. 25°C ).

On one hand this raises my concern that we should not exclude the possibility of having some soldering defect on the Buffullo II board (the few input soldering was double checked) or a faulty ESS 9018 chip which could lead to instability of the Buffalo II operation even though at the monent it seems that after some warm-up time it works fine. On the other hand I have very limited experience with Buffallo II and ESS 9018 so I might miss some relevant factor or correlation.

I am sure that You are very experienced in the accepted behavior of Bufullo II and ESS 9018 because of product prototyping and assembling/testing many Bufullo IIs.

Please give me your opinion about this issue/behavior, I would really appreciate if You have any solution ideas or suggestion regarding to further problem identification.
(Needless to say It would be a very relevant information If You say this is not the normal behavior of Buffalo II).

Thanks for Your help.

Gyula

Edited by user Sunday, July 18, 2010 6:56:21 AM(UTC)  | Reason: Not specified

Hi Russ,

Thanks for Your fast Response.

Regarding to Your reflections:

- The only thing I can think of is that the XO freq may change slightly over temp change, but it is very unlikely that this should cause any issues.
I agree, it is extremely unlikely that a Crystec Oscillator with a specification of 0°C to +70°C / ±20 ppm change the freq so much that it causes a lock dropout.

- I would need to see some pictures of your setup to be able to offer more advice.
This is the picture of the Buffullo II (with temporary boxing for testing). I also made a 7 minute video of the "cold" start of Buffalo II showing the lock dropout behavior. The recorded behavior is very typical. After cold start at 0:37, 1:00, 1:30 and 2:10 there are lock dropouts, then it is gone (the video is 7 minutes long though it could be 2 hours, because after this certain 2-3 minutes time of "warm up" it becomes rock stable).

- I have not seen this issue ever occur myself. I have seen lots of dropped SPDIF, but this always came down to the quality of the source.
I used my most precise SPDIF source for recording which is an RME Digi 96 card (which Ultra-low jitter SPDIF: < 1 ns) though it has no relevance regarding to the lockout behavior (I do not want to repeat myself, all circumstances are described in my previous entry). The RME Digi 96 was already outputting the SPDIF when I powered the Buffalo II.
(There were just a few other topics in the forum where like "The frequency of the dropouts often decreases the longer the unit is used" are reported, e.g.: Odd dropout, Audio dropout)


Finally I would like to come up with another suspicion which is based on the short (2-3 minutes) duration of the issue. The AVCC module is packed with relatively dense heat generating SMD parts (and working in hardly dissipating shunt mode) so this is a specific area where the temperature change during 2-3 minute can be relevant. (Though I do not have detailed knowledge and experience of the Buffalo II as it is only a "black box" for me so during problem identification I can only change the outside conditions.) What do You think?

If You have any solution ideas or suggestion regarding to further problem identification please let me know.

Thanks for Your help,

Gyula

Edited by user Friday, August 6, 2010 5:00:43 AM(UTC)  | Reason: Not specified

Hi Russ,

Thanks again for Your fast response (it is really something...).

Your concern about the SPDIF source is very logical. Intitially I got the same idea that is why it was one of the main factor to test. (Please look through my original entry again because all relevant information is there). After extended testing I came to the conclusion that the issue is completely independent from the SPDIF source belong with a lot of other factors except for the Buffalo board.
It can be seen in the video that after the 2-3 minutes Buffalo "warm up" there is no dropout which would be impossible if there is a concsistent jitter problem with the source. (If I had switched off the "warmed up" Buffalo and after a few seconds turned it back on with the same source, there would not be even one droput. Needless to say it is very unlikely that the SPDIF source and the Buffalo warms up "together" and it is the SPDIF source what is changing not the Bufallo, especially because the SPDIF source was started earlier. Additionally the behavior is the same with other SPDIF sources).

So in my opinion according to the tested factors it seems that the problem cames from my specific Buffalo board (meaning all included parts, board, ESS, soldering, etc.). Please correct me if I miss any relevant factor or correlation. You said that the issue can not be fundamental as You have built, tested and delivered hundreds of units without this problem. Taking both into consideration it seems that it is my Buffalo (or some of its parts) that produce a faulty behavior. I think I have done quiet a lot of tests that leads to a consistent conclusion. What is Your opinion?

Still If You have any solution ideas or suggestion regarding to further problem identification please let me know.

Thanks for Your help,

Gyula

Edited by user Sunday, July 25, 2010 12:56:13 AM(UTC)  | Reason: Not specified

Hi Gyula,

I really am not sure what is going on, as I can't reproduce it at all.

As AVR300 suggested, I would check the load you are putting on your transformer.

I am still thinking this over, but as for now, I don't have any solid leads for you.

You don't need to bet me anything ;-) I'm only telling you my experience of the BI beeing underpowered.

I just simply can't reproduce your results with any of my flock of Buffalo IIs :). So I am quite sure its not a systemic problem. It could be anything from implementation to a damaged or defective part.

Hi Russ,

I got and installed my new transformer(s).

As I mentioned I use temporary solutions for circuits and case (later on all things will be finalized and integrated, but It has no influence on test results...). In the previous videos DSD and SPDIF isolation transformer modules (and the direct User Level SPDIF input of BufalloII) were used.

I have done the same tests as in the previous videos, basically using the Shunt regulated Power Supply and additionally (once again) the Linear Power Supply (when I used direct user level SPDIF to BuffaloII and the 5V was free to use) to cross check. (Both Power Supply's stability was checked with oscilloscope.)

Unfortunately there is no change in the behavior. E.g. as in previus video SPDIF lock consistently and systematically can get dropped by using the small fan.

Regards,

Gyula

Toroid connected to the BufalloII's Power Supply


Temporary layout with DSD module (the two SPDIF inputs are not is use at this setup...)

Edited by user Sunday, August 15, 2010 3:37:07 PM(UTC)  | Reason: Not specified