Teardown: Generic CX-FU02+ USB SoftV92 Data Fax HSF Modem

My tech shopping in Hong Kong continues with more goodies. One of the things I was on the lookout for was the last of the dial-up modem. Rather unsurprisingly, for such an advanced country, it’s almost impossible to find any new modems at all. The used modems I found were all locked up in shops which didn’t open in the two days I visited.

Unboxing

I found one new modem buried in the expansion-card-box pile of one particular shop. It was contained within a cardboard box which had a gradated lime-green package similar to the TP-Link packaging, but was not branded as such. It was generic, unbranded and completely devoid of any model numbers on the outside. Due to weight reduction measures, the package didn’t make it with me – just the goodies inside.

Everything could fit in a small-ish zip-lock bag, namely the modem, a driver mini 8cm CD, a phone cable and the receipt. I ended up haggling and got the modem for HK$140, or about AU$23.33.

Given the lack of choice (this being the only one), it comes in a fairly small “soap on rope” form factor which is bus powered (more convenient than some older modems). Two LEDs are provided to signal status, on a glossy black plastic case which feels a bit thin and flimsy.

Cost reduction seems to dictate the inclusion of just one phone line port. I suppose the socket for a phone is rarely used anyway, and it simplifies DAA design. No audio inputs/outputs are provided, because it seems that it’s all handled digitally in the driver.

There is absolutely no branding or model number on the rear. The unit is held shut with two screws, one screw covered by a warranty label that was (messily) removed.

Teardown

I was secretly hoping that I would get something that had an Agere chipset in it, but sadly, that was not to be. Instead, this one contained a more predictable result.

The modem is built around a Conexant CX06836-11 CXHSF-USB chipset. The chip itself has a date code of Week 12 of 2002, and almost looks like it was salvaged from another device and recycled with bent pins on the left side, an amateurish soldering job and left over glue or flux on the top of the package. It could be new-old-stock soldered by hand though. The line interface also looks rather simplified, however, instead of opting for an all-silicon DAA, it seems they used a size-reduced transformer of the sort you might find in laptop modem cards. The single plastic phone socket is askew, and the USB wires are soldered and glued down with hot-melt instead of using a connector. The soldering of everything looks a bit hand-done.

The underside does have a few components, and it seems that this modem has a PCB identification code of QBA41-C, design dated 12th October 2013 and PCB made 16th February 2017. A very new product, with a very crappy design. Interestingly, the phone socket doesn’t even have its fourth pin soldered, which explains the poor mechanical support and tendency to shift in its place. It would probably break given enough plug insertion/removal cycles.

I suppose this is what happens at the end of a technology’s lifecycle. Everything becomes integrated into a single chip where possible, the price is reduced to absolute minimum, the size of the product and power consumption is cut back to bare minimum, the functionality might be somewhat reduced and then … build quality is sacrificed. Even then, this modem probably works similarly to an average PCI soft-modem.

Installation

The modem was not quite plug and play. Windows 7 x64 did not have any drivers already available through Windows Update, and thus using the disc was necessary.

It identified as USB HSF Modem with VID of 0572 and PID of 1300. This already speaks of the Conexant heritage, and its similarities to the Conexant HSF/HSFi PCI modems.

The provided driver package has a SUSBUTyK.inf file which covers the modem. The drivers are otherwise default Conexant drivers, of which the CD covers Windows 98 SE through to Windows 10 64-bit, as well as Linux through Linuxant.

The driver identifies it as USB SoftV92 Data Fax Modem.

It is, however, unsigned. I’m not sure if this holds true for the Windows 8 and above packages, as this could preclude them from being installed on later versions of Windows.

The drivers are indeed Conexant provided, with a version number of 7.63.0.50 and date of 15th March 2007. The ati strings reported by the modem are as follows:

ati1
255
OK
ati2
OK
ati3
SoftK56V_B2.1_V7.63.00.50
OK
ati4
USB SoftV92 Data Fax Modem
OK
ati5
009
OK
ati6
SoftK56 
CModem Version 12
Rksample Version 342
OK
ati7
255
OK
ati8
Feb 19 2007 # 17:53:52
OK
ati9
AUSTRALIA
OK

A few test calls proved that the modem worked, however, the calls sounded identical to the HSFi PCI modems. The performance did seem to result in lower link rates and more difficulties achieving V.90 connections over my “calibrated” VoIP termination. This may be a consequence of the DAA set-up which appears somewhat simplified.

Conclusion

You can still buy a dial-up modem, and it’s not that expensive. However, it’s built like some rookie’s project out of salvaged spare parts and it seems to be partly simplified to reduce costs. Its performance is nothing unusual, if anything, slightly worse. The modem has officially come to its logical end …

Aside

I did also buy a TP-Link Archer C7 router to replace my ageing TP-Link WR-740N (x2) and WD MyNet N750 (x2) as the latter started to have issues where the wireless would play up and require a reset. Needless to say, I got it for about AU$85, which is much cheaper than locally, and it works quite well on the default firmware despite the interface being D-Link-ified with large shiny colourful buttons and slow response. It’s not quite feature complete, but it’s good enough for me. The additional coverage and removable antennas work as a bonus. I guess I can’t really stress it out with my shitty 9Mbit/s down and 1Mbit/s up connection.