Verfasste Forenbeiträge
-
Beiträge
-
@solidcore I am currently running a 902 as an access point connected to my home router, and a 902 as a client connected to my music server:
Home router -> wired ethernet on port configured as a dedicated audio subnet -> 902 running OpenWRT and configured as an access point with audio WiFi network -> WiFi -> 902 running OpenWRT and configured as a client -> throttle cable -> Taiko network stack -> Taiko Olympus
On my 902 AP, I removed DHCP and all firewalls and configured it as a low power access point (1mW/3dbM). On the 902 client, I connected to that AP SSID, again with a low power radio setting (1mW/3dbM). For me, there is no conflict with my other WiFi networks.
This morning I finally converted all my 902’s to 3.3V so I was able to do some direct A/B testing with the Ian Canada battery supply and UcConditioner.
For me, a 902 client running OpenWRT was a clear step up from a 902 client running the original TP link firmware. I’m attributing this to the lower radio power of 1mW vs 10mW on the „low power“ setting on the original TP link firmware, but that is a guess on my part.
With my 902 running OpenWRT as an access point, there was also a clear step up when going to a USB battery pack and then again to the Ian Canada 3.3V supply. The 902 access point in all configurations was better than my home WiFi (Ubiquiti mesh network). Alas, what is upstream of the 902 client still matters here.
My OpenWRT configurations are in the links above, but happy to discuss/share here if of active interest
Best holiday wishes to everyone, and all health and joy in the New Year!
I too have had success with a 902 as an Access Point and 902 as a client device. I recently updated both to OpenWRT to get lower power radios as well. Here are links to the recipes I used for OpenWRT on both the client and AP:
OpenWRT AP config https://www.whatsbestforum.com/threads/review-the-best-yet-most-affordable-network-switch-tplink-wr902ac.41708/post-1099127
OpenWRT install on Mac https://www.whatsbestforum.com/threads/review-the-best-yet-most-affordable-network-switch-tplink-wr902ac.41708/post-1099130
OpenWRT Client config https://www.whatsbestforum.com/threads/review-the-best-yet-most-affordable-network-switch-tplink-wr902ac.41708/post-1099148
Next for me is getting the 902 AP on 3.3V (shared supply with my 902 client, while I wait for my second Ian Canada LiFEPO4 supply to arrive). That is an experiment for the next couple days
Has anyone compared the TL-WR902AC to the TL-WR802N? The later seems to be a simpler version (no USB, 2.4GHz only, etc)
Very interesting Matthias.
I remember when many of us were experimenting with ferrites on BNC cables between the Chord upscaler and the Chord DAVE DAC, there was a cut over point where the improvement max’ed out. At the time, I attributed it to the tradeoff between BNC length (longer was poorer) and ferrites count (more was better). The physical space required for ferrites seemed to drive the tradeoff.
The v3 cable is interesting because we can do more windings on the torroids as well, not just add ferrites
Of course, the most intriguing part of Eric’s work (for me) has been his ability to measure common mode noise and correlate that to SQ improvements. If I had the equipment, would be facinating to play with other cable filter configurations and see what frequency range my system is most sensitive to, and filter even more aggressively there.
Wonderful to have a measurable hypothesis that correlates with the empirical differences we hear. Alas, I no longer have access to lab equipment at work these days, so difficult for me to get a proper measurement setup going here.
Indeed, that would be a quick experiment. Has anyone done this stacking experiment with 2 v3’s (or a v2 + v3)?
I was thinking of a 10m ethernet cable, and doubling the windings on the torroids, but it would be much easier to A/B 1 vs 2 throttle cables (if that is audible, then time to think about a v4)
Got the last of my v3 parts in the mail today and assembled my cable. Significant benefit here, even with an Ian Canada powered/conditioned TP link wifi extender going to the Taiko Audio network stack.
With the WiFi extender I did not hear a difference between a generic Cat6a cable and my previous reference ethernet cable, but the difference with the v3 throttle cable is very clear.
Makes me wonder if more ferrite filtering is better in particular frequency ranges…is there such a thing as over filtering with common mode noise?
I have pretty good WiFi in my home (3 access points), so I wasn’t able to detect any signal attenuation (signal strength stayed „excellent“)
Originally my plan was to have the extender PCB mounted bare to the enclosure, but there wasn’t a convenient mounting point on the PCB. Instead I added standoffs to the lower plastic case. After that, I put the cover back on because the plastic tabs on the cover keep the extender PCB in place.
If you have a convenient way to mount the extender PCB direct to the case that would make things cleaner. Given all the experiments/hacks I did before getting my enclosure, this was the easiest (incremental) path for me.
As a quick update, I was able to do a test on my DAC cable between my audio router and my streamer, using 3 x Würth 74271142, 2 x Würth 74271142s , and 1 x Würth 74271111S (parts for the v3 throttle cable, while I wait for the rest of the parts)
The impact was definitely audible, but not a game changer. There is some benefit to be had here for sure (alas, cable is too stiff to wind around the torroids
I suspect 5-8x Wurth 74271633S could be beneficial here, but that experiment is for the future
Eric, for the v3 throttle cable, what was the rationale for the ferrites that you chose? The 74271633S pushes into higher frequency ranges, and there are other Wurth ferrites with higher impedance in that 70MHz range.
Alas, there are so many ferrites to chose from (and combinations with different impedance match and mismatch to consider), I’m curious about how amenable the filtering choices are to a data driven approach
Torsten, what I did was get the following from (US) Amazon:
https://www.amazon.com/dp/B0D5B82V6K
This connects to slave on-off control input (J5). When the switch is in the off position, a voltage here turns the unit on, and no voltage turns the unit off.
I then used a USB brick on a switched outlet to turn it off overnight for charging, and have it on during the day
Other elements for my build for the combo of the Ian Canada mini battery + Ian Canada conditioner + the WR902AC (since tracking these things down for a first build is a pain, hopefully there are equivalents from local options in Europe)
https://www.amazon.com/dp/B07PKKY8BX?ref=ppx_yo2ov_dt_b_fed_asin_title
https://www.amazon.com/dp/B0F9TP2GZ6?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1
https://www.amazon.com/dp/B0CZ6WKTN5?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1
https://www.amazon.com/dp/B09XWS2FTM?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1
https://www.amazon.com/dp/B0CCRXF1N7?ref=ppx_yo2ov_dt_b_fed_asin_title
For the enclosure, I have one USB C input for charging, one switched USB C input for turning on and off, and ethernet output. I purchased standoff screws from Ian for the mini battery PCB, conditioner PCB, and the WR902 (I drilled mounting holes in the bottom of the case when I had the PCB out)
I am currently waiting on delivery of parts to build a v3 throttle cable. Looking forward to hearing what impact that has with the WR902
As an aside, note that Ian let me know that a new Pro version of the LiFEPO4 mini battery adapter is getting close to release, that support faster USB charging and have some other improvements
Thank you Eric. I have a 2m DAC cable that I will be experimenting with (maybe 15cm diameter is the tightest you can wind it, so no easy looping)
From previous experiments, there was a big benefit here in reducing power impulses from the NIC side (lower power consumption by the SFP, more network isolation to reduce packet volumes, look ahead caching of files to concentrate traffic -> better the sound quality).
My presumption there is some induced noise on the power or ground plane at the transceivers, and the Taiko preference for direct attach networking is related to that, but that is all speculation
That being said, I clearly (and unambiguously) hear the benefit of of common-mode noise reductions using the techniques you’ve detailed here. These seem to be complementary (and perhaps competing) optimization levers
Like Jawed, I was very active with trying to optimize digital transport with Chord DACs back in the day (ferrites on BNC cables, optical transceivers for isolation, shielding experiments, power optimizations, etc) I am getting a sense of deja vu with these discussions and experiences with ethernet, which is very interesting (see https://www.head-fi.org/showcase/audiowise-opto%E2%80%A2dx-optical-isolation-bridge-for-dual-spdif.23757/reviews#review-22155 for some experiences from 2019…I had forgotten about those Chromecast experiments!)
(with apologies again for intruding with English on this forum)
Thank you for the response. I was able to track down the issue to a flakey solder connection causing the wifi extender to get into a reboot cycle (multi-meter was able to integrate through the intermittant glitches).
Since addressing that, performance with the Ian Canada mini supply and conditioner has been outstanding (thank you for all the research with this solution)
