2022-06-18

AK103E (CM6206-LX) sound card

Cheap (~$12, less if shipped directly from China) USB sound card with SPDIF input and output (my main reason to buy it).





Irregular PCB shape - I'm guessing 2 PCB pieces are made from one panel, this would save about 5 cm^2 of the PCB. Producent margin and labor cost must be pretty low if this is economically viable.


https://tehnoblog.org/c-media-cm6206-5-1-usb-audio-sound-card-line-in-line-out-bass-extension-low-frequency-response-upgrade/ has quite detailed description of similar sound card. While mine has slightly different PCB (SMD capacitors, slightly smaller, not rectangular, no headphone output pins on the PCB, no pads for EEPROM) it also had incorrect polarization of electrolytic capacitors at the outputs:
Older (?) PCB variant, with THT capacitors seems better overall, having more capabilities (headphone output pads, EEPROM pads). SMD capacitors are also placed near to audio jacks, so they are not as convenient when fixing their polarity, requiring soldering iron with thin tip.
Following linked blog I've also replaced C30 and C31 with higher values - 2 x 1 uF (1.2 uF actually measured), 0805 size (they still fit, even of original were 0603) as 85 Hz cut-off frequency with default 100 nF capacitors would be pretty ridiculous and confusing.
Software is just awful, with multiple layers of transparency that cannot be disabled. Also what is weird - CMedia shares reasonably detailed (33 pages) datasheet of this chip but does not seem to provide drivers, so don't throw up mini-CD included in the box. There are drivers for Windows 7, 8.1 and 10 while local distributor shared only driver for Windows 10 on webpage.

While this kind of software should be simple to use, it is quite unintuitive. I found few comments that SPDIF output is not working on this card and I'm not surprised, it took me maybe 30 minutes to figure it out. While this sound card could have up to three outputs (line out, headphones - not available on this PCB - and SPDIF) they cannot be switched using Windows mixer as it would always show two inactive outputs as "disconnected" (even though detecting disconnected SPDIF output would cost more than this whole card). Only one output can be active at the time, so you can't have Line Out and SPDIF playing same time. In "Xear" CMedia software two out of three outputs would be grayed. You cannot just click on Speaker/Headphone/SPDIF output icon to switch to this (even if same action works for audio inputs), you have to use context menu of the icon from currently selected output (for example to switch from SPDIF to any other output you would have to use "SPDIF-Out Settings", to switch from Line Out to SPDIF - "Speaker Settings").

Other issue: I've never seen other sound card where lowering volume down to 0 would not mute sound completely.

It is quite hard to unplug optical cable from SPDIF output (input is fine). I'm guessing output socket (or its solder joints) might fail early.


RMAA line out -> line in loopback results (after correcting capacitor polarity and swapping line-in capacitors to 2x 1u):
Summary
Frequency response (from 40 Hz to 15 kHz), dB
+0.25, -0.31
Good
Noise level, dB (A)
-87.1
Good
Dynamic range, dB (A)
87.2
Good
THD, %
0.00895
Very good
THD + Noise, dB (A)
-76.6
Average
IMD + Noise, %
0.036
Good
Stereo crosstalk, dB
-82.1
Very good
IMD at 10 kHz, %
0.016
Very good
General performance
 
Good
Frequency response
 
Left
Right
From 20 Hz to 20 kHz, dB
-1.68, +0.25
-1.56, +0.37
From 40 Hz to 15 kHz, dB
-0.31, +0.25
-0.19, +0.37
Noise level
 
Left
Right
RMS power, dB
-85.9
-85.9
RMS power (A-weighted), dB
-87.1
-87.1
Peak level, dB FS
-72.2
-71.5
DC offset, %
+0.8
+2.6
Dynamic range
 
Left
Right
Dynamic range, dB
+85.9
+86.0
Dynamic range (A-weighted), dB
+87.2
+87.3
DC offset, %
+0.80
+2.58

One other modification I've tried was adding 10u ceramic capacitor at the voltage regulator output. No noticeable difference was observed, except for channel separation increasing by 1 dB.

2022-06-16

GTV LD ZASL6W-30 LED power supply

Small (up to 6W) LED power supply, 12V/0.5A




Repair:
  • C5 (47u/25V) was replaced as it completely lost it properties, not even being recognized by component tester as capacitor
  • THX203H needed replacing
As this exact model costs $5, repairing it might be not worth the time. I thought this only capacitor failure at first, but driver chip was available at the local electronics store that I had to visit anyway for $0.75.
I'm not 100% sure if I've fixed everything as this power supply behaves somehow weird - it's output voltage is stable with 100 Ohm load (120 mA) but falls down to 9 V with 27 Ohm load (~300 mA).



2022-06-11

Asus VW223D monitor repair

 This is 22" 16:10 (1680x1050) LCD monitor from 2008. It has CCFL backlight (4 tubes), so power consumption is rather high (55W according to user manual). I suspect this might be power consumption with maximum brightness, as in my test it heats up only a little. While I'll probably won't keep it, I like warm colors of this kind of monitors comparing them to LED/blueish cheap LED-backlight panels.

Disassembly: panel is screwed up to back part, front frame is held by one screw and latches.

Only VGA input




Realtek RTD2554VH



Delta DAC-19M020 power supply, used also in other monitors. Apparently still available (as used products, so it might be hard to predict their expected life time) on aliexpress for less than $10.


When I've tried to turn it on, it only was blinking shortly in regular intervals - pretty typical sign of power supply malfunction or overloading. It turned out to be typical failure, with 2200u/10V capacitors visibly bulging and losing capacity. Other capacitors seemed to be fine when measuring with tester, but I've actually replaced all electrolytic capacitors except for the high voltage one. 


Other problem: not working ON/OFF microswitch (probably because it was the most used one). Metal part soldering was broken and also its latches were streched, so it was only loosely attached. I probably could search for some replacement, but soldering back microswitch metal part and securing it with some epoxy was faster and seems good enough.