As a VoIP, modem and telephone enthusiast, I’ve reviewed a number of different Analog Telephone Adapters (ATAs) in the past including the Linksys PAP2T, Zyxel Prestige 2302R, Cisco SPA112 and Grandstream HT702. These are standalone devices which let you connect old fashioned telephones and place calls over VoIP, most commonly using the SIP protocol to a Voice Service Provider (VSP). Of course, this was supposed to be an “interim” solution to allow for those with older telephones to keep using them in a way they were familiar with. Other users might prefer to use softphones (i.e. an app) running on a computer or a phone, or a physical IP-telephone which is a telephone set built to work with an IP network directly.
VoIP has had a bad reputation amongst some consumers because of a number of problems. These problems can arise because of network connectivity issues – such as firewall and NAT traversal problems which can cause one-way audio or no audio at all due to port-mapping issues or dropping of media packets, or sudden drop-outs in audio mid-call. Intermittent connectivity can also result in problems maintaining registration and receiving incoming calls. The choice of codecs and packetization can affect latency and perceived call quality, with packet loss, jitter buffer changes and delays degrading user experience. Compatibility issues with soft-switches and end-user devices can result in unexpected problems with call features. An ATA is also subject to problems with getting the right volume to the end user device, having a 2-wire interface that’s prone to echo, getting dial-tones and call-progress tones that match user expectations, getting correct dial-plans to correctly receive and send dialled numbers, correctly recognizing DTMF and data tones and forwarding them and generating the right ringing voltages and cadences to ring end devices. While SIP is a “standard”, interoperability isn’t always ensured especially with more sophisticated calling features, and thus configuration of SIP devices can be quite daunting especially if you are to ensure it operates correctly. Troubleshooting problems isn’t always as straightforward as one might like, although often the problem is not with the VoIP devices themselves but in the way they interoperate, or intervening routers/firewalls process data streams.
Regardless, a reader contacted me about wanting to know more about the Grandstream HT802 – one of their latest two-line ATAs to reach the market, at a very affordable price of AU$57. They wanted to know what was inside, and what I thought of the product. As I wasn’t in the market to buy one myself, they purchased the ATA for me and submitted it for a review challenge test.
As with the HT702, Grandstream ATAs all come in a glossy white box which is rather anonymous, save for a label on one side that details what is contained inside.
Flipping the front out and lifting the box open, we are greeted with the contents.
A GNU leaflet is on top, with an internal cardboard spacer holding the ATA unit wrapped in a foam-bag, and the power supply and Ethernet cable in their own plastic bags in the other compartment. Note that there is no manual, quick start guide or other leaflets, just like the HT702, although the Cisco SPA112 still comes with a sizeable quick start guide. I suppose, it’s 2016 and whoever is intending to use the device probably has a computer and internet connection, so it’s excusable. Full documentation can be found from their website. The unit is covered by a one-year warranty, which isn’t particularly long especially for a networking device which is expected to be relied upon and running 24/7, however, when it comes to ATA’s, it seems to be the most common warranty on offer.
The first thing I noticed was the included Ethernet cable, which is about 1.5m long, has at CAT5e rating, is made by Rapid Conn and has all pairs wired though. That’s a good result to see, as some cheaper networking equipment have often “sub-standard” cable (e.g. two-pairs) which could cause problems if used with other devices.
The next thing I noticed was the power supply. As this is often one of the more failure–prone parts of the system, it was interesting to take a closer look. This particular power adapter is made by ShenZhen Sunlight Electronic Technology Co. Ltd. Unfortunately, as I haven’t had any prior experience with their products, I’m not able to comment on its reliability just by looking at it, but it does have a VI efficiency rating and appropriate Regulatory Compliance Mark for the Australian market. Whether this same power adapter is used for the HT802 in different markets is not known.
Compared with the Cisco SPA112 which came with a Phihong power supply, this power supply is a lot smaller and doesn’t obstruct adjacent plugs which is a great feature. Another key difference is that this adapter has a 5V 1A rating, compared to the 5V 2A rating of the SPA112, suggesting the ATA also uses less power. This is further compounded by the fact the Phihong only reaches an efficiency mark level of V, which is one step less than this adapter. The HT702 came with a Mass Power adapter with an efficiency level of V and a 12v 0.5A rating which is more similar in power rating, but still has a lower efficiency level.
The best thing about the power supply? It doesn’t use any barrel jack plugs. Instead, they’ve gone with the standardized micro-USB B connector, meaning that should the power supply fail, swapping in another power supply will not require any big effort at all. It also means that it’s probably possible to run the ATA off a USB power bank, say in an emergency or remote usage situation.
The ATA itself is a square-shaped design with rounded corners, of a similar size to the SPA112 and PAP2T. The top cover features a “water ripple” style design, with a central glossy window section featuring four light-pipes for LED status indications of power, internet, line 1 and line 2.
From the side, the unit has a ribbed edge all the way around, although from the front, this is purely cosmetic.
On both left and right sides, the ribs lead to vents which help cool the ATA’s internals.
On the rear, colour coded plastic jacks are used, with green jacks for the two FXS lines, and a blue jack for Ethernet. Ethernet activity lights are provided on the jack itself. A reset button hole is provided, and the microUSB-B power inlet as well. The rear colouration differs slightly from earlier photos of the HT802 used by resellers.
Slots for wall mounting in all four orientations are provided, as well as rubberized feet. On the whole, in terms of volume, this unit has a similar volume to the HT702 with the HT702 probably being slightly smaller. The HT702 had the advantage of being able to stand vertically or be used horizontally, whereas this unit is more like the SPA112 in being designed to sit on a desk horizontally or mount on a wall.
Even though I’ve only just got my hands on it, the first step is to void the warranty. Below the warranty label seal on the underside is a single Philips screw that holds the unit together. By piercing the label and removing the screw, the unit can be broken apart, although with much difficulty as the case clips on rather tightly. It makes for a robust feeling unit, even without the screw.
We can see that the unit’s top cover has an interesting design. Due to the use of light-pipes to bring the bright-blue LED indications from the PCB to the top of the unit, it seems the designers paid a lot of attention to the cosmetic issue of light leakage. As a result, each light pipe has black plastic moulded around it to prevent “interference” from adjacent LEDs, but the whole unit also has a plastic “wall” which sits around most of the PCB and “tucks in” near the Ethernet jacks as well. This feature stops light leaks through the side air vents, but also serves as additional insulation to prevent stray wires or dust from getting far into the circuitry. It’s not a bad idea, but one potential downside is that it will affect the free flow of air inside the unit, which might increase operational temperatures and consequently affect lifetime. This isn’t likely to be a big issue though, as the maximum power available is 5W, and the actual power used is quite a bit less (closer to 2-3W in normal operation).
Looking inside, we see that the PCB is a little smaller than the case itself and has a fair amount of empty space. It would have been very viable to make the ATA quite a bit smaller, although it may have compromised its aesthetics slightly. The PCB is V1.5A and has all the components laid out rather neatly with generous spacing. It is clear when compared with the HT702 and the SPA112 that this design is more integrated. It’s good to see that the line protection polyfuses (F1-F4) that were in the HT702 (and not in the SPA112) are also present in this design, as this is a relatively inexpensive way to ensure that faults with any telephone device connected to the unit don’t cause hazardous situations and provide a little more protection for the line codec. However, it seems that this PCB omits many diodes which were used as reverse polarity protection in the HT702, instead, relying on the USB power supply to be intrinsically “keyed” to prevent reverse polarity connection. This is appropriate, as long as you don’t have any mis-wired USB ports, and it probably avoids any voltage-drop problems. As usual, there’s a row of through-hole pads for test/debug.
Rather interestingly, the design has just a single electrolytic capacitor from Teapo. This particular brand isn’t particularly special, but their capacitors are quite acceptable. It seems to have had a black paint applied to the top which I could scratch off with a fingernail. There are two additional places for electrolytic capacitors to be fitted, but they were omitted. One of them seems to be for bypassing the USB power input, and another seems to be bypassing a regulator output – their omission is probably not ideal, since the USB cable inductance could cause issues, and the regulator might not be as happy with the smaller MLCC ceramic capacitor installed. However, it doesn’t seem to cause any problems, and reducing the number of electrolytics generally means improved lifetime/reliability as they are one of the more temperature-sensitive components.
Onto the “heart” of the ATA. In this case, the ATA is powered by the DSPGroup DVF9918 chipset, an upgraded version of the DSPGroup DVFD8187 used in both the HT702 and SPA112 ATAs. This chipset features a dual-core 400Mhz CPU (rather than the 220Mhz CPU + 120Mhz DSP in the older one). It also features a hardware crypto engine, which should improve performance for encrypted (TLS/SRTP) calls and provisioning and secure boot which should ensure firmware integrity which is probably quite important seeing the IoT device botnets as of late. It has a QSPI interface for faster interface with firmware flash, and support for up to 6-lines (only two available in this model). The codec support of the chipset is claimed to be “G.711, G.722, G.723.1, G.729A/B, iLBC, G.722.2 and more”. Seeing a better chipset in a product at a similar price-point to the others is a very nice perk, as it’s likely to be better supported, more flexible, and more performant. Just having this chipset has reduced the chip-count and crystal-count as well, which reduces costs for the manufacturer.
The chipset is paired to an Etrontech EM63A165TS 16Mx16 (32MiB) 166Mhz SDRAM. The firmware is contained within a GigaDevice 25Q128CSIG 128Mbit (16MiB) QSPI NOR Flash chip. Ethernet 10/100Mbit/s physical layer is provided by a Microchip KSZ8081, as in the HT702.
A big improvement and integration has been the use of a Silicon Labs Si32260-FM2 ProSLIC as the phone line interface. Compared with previous Silicon Labs devices, the one chip now handles two lines, performing all functions necessary to handle the lines while reducing power consumption and parts count. Specific features for this chipset include ultra-low power consumption, patented low-power ringing which may have been the enabling factor in making this unit operate off the restricted power available from a 5V/1A USB-style charger. It also claims wideband voice support, although this is probably not so applicable as many phones may not have enough frequency response to take advantage of it.
The underside features a few diodes, transistors and capacitors. There are two areas where silkscreen paint is generously applied.
This matches up with the wall-mounting brackets on the rear, so that if any screw were to have made it through the casing, there would be at least two more layers of insulation – silkscreen print and solder resist. This shows that they have paid much attention to insulation and maximising user safety – after all, these units could be hooked up to long internal house-wiring instead of being directly connected to a device, so could be vulnerable to high voltages – e.g. lightning impulses and static discharges.
Since this unit was given to me for review, I decided to also take a look at the power adapter. Normally such an investigation ends in tears with a destroyed case and lots of effort in trying to crack it open, however, in this case, it turns out there’s one screw (not a security sort) and a little bit of squeezing between us and the circuit-board inside.
The screw in question is underneath the label, in one of the corners. I’m not sure such an arrangement is as sturdy or safe as needed, but at least we can look inside. The rear of the plug pins has been covered internally with a plastic shield so that the soldered wire joints are not going to make accidental contact with the PCB. The PCB is wedged in place with some foam-rubber pads and sits in rails in the casing.
The board can easily be slid out of the case, which reveals a PCB with copious amounts of silicone goop applied to secure wire-ends, however poorly done, especially on the live/neutral wires making them virtually ineffective.
The unit has a fusible resistor as the main protective/inrush-limiting element. Rather disappointingly, this unit has many positions for the addition of surge protective MOVs, interference suppressing inductors and capacitors which are all not fitted. This likely means that the supply radiates more radio noise than strictly necessary, and may have a significant noise and ripple in its power output. It would also be more vulnerable to mains transients and surges than a properly “fitted” power supply. I wouldn’t recommend using this supply with other USB devices for that reason. The transformer does appear to have adequate insulation from a visual inspection, but this cannot be confirmed without a destructive teardown.
The power supply appears to be a Vifipower design, Model R60, and features a spark gap (very wide, probably not effective) between live and the secondary negative/shell connector. Why this has been put in is not certain. The primary to secondary isolation appears appropriate. The soldering looks good with shiny joints and smooth texture (slight mottling due to lead-free solder). The unit appears to be controlled by a 6-pin IC, which could not be identified.
The biggest disappointment in the power supply is the use of many “unknown” capacitors. These come from KT and Sancon, and are not widely known. These could prove to be a potential source of failure in the future, especially after long operating hours – but I suppose they will easily last the first year, since the adapter is fairly efficient and doesn’t heat up too much (measured casing temperature of 30 degrees in a 21 degree room).
On the whole, it seems that this ATA features an upgraded design using the latest highly-integrated chips. It offers better performance specs as compared with older devices, while also being less complex and using less electrolytic capacitors which is a common cause of failure. The unit still retains the line-protection features seen in the HT702, and has paid much attention to insulation and user safety. However, the power supply was a slight let-down, with many components bypassed to reduce cost and the use of rather unknown electrolytic capacitors. At least, it would be easy to replace when the time comes. That being said, DSPGroup seems to have a clean-sweep when it comes to ATA chipsets, as the major VoIP equipment players all seem to be utilizing their chipsets in their products.
Administration and Configuration
If you’ve used an ATA before, or a Grandstream ATA specifically, you would feel rather at home with the HT802. When powered up and connected to a network, it will use DHCP to grab an IP. You can check your DHCP server or scan the network to find it, otherwise, just pick up a phone connected to either port and dial **** and press * to cycle through the settings to have the ATA read them back to you using its voice synthesis.
The login page already gives you the feeling of what configuration is like – Grandstream ATA’s have a fairly simple administrative interface, which looks a little “from the 90s”. The default login is “admin” and can be changed.
My unit came with Firmware 220.127.116.11, and at the time I started the review, I didn’t find any non-beta firmware available. However, mid-way through the review, they finalized 18.104.22.168 and are currently beta-testing 22.214.171.124, so it seems that this device is under active development and support. Judging on the firmware page, it seems the other HT8XX products are based on a similar “base”. Unfortunately, I’m not too sure about the firmware quality, as the present 126.96.36.199 has a change-log warning that factory reset via web UI is not possible and is a known issue.
The firmware update process is rather simple, and only requires navigating to the right area and selecting the file.
After you click upload firmware, nothing seems to happen as the firmware is slowly uploaded to the ATA and flashed. This process must not be interrupted, otherwise damage to the firmware may occur. After a minute, the page will change to confirm successful loading.
The main page is the “Status” page which allows you to see the status of the system at a glance, including the state of both lines, call features, firmware version and integrity, provisioning status, and networking information. A difference from the HT702 is the provision of the GR909 Test Page.
This allows you to run a variety of tests on the two lines to determine their status. These types of tests are like those available to telecommunications personnel testing on line-cards, but in most consumer cases, isn’t likely going to see much use. However, if you are remote from the ATA, or you have a long line connected to it, checking the results could give you an idea of what’s wrong.
In my case, I just ran the tests for the sake of it, and it all came out fine. The results are, however, are presented in a way that is quite cryptic without knowing some telecommunications terms – e.g. Vrg = Voltage Ring-to-Ground, Vtg = Voltage Tip-to-Ground, Vtr = Voltage Tip-to-Ring, Rrg = Resistance Ring-to-Ground, Rtg = Resistance Tip-to-Ground, Rtr = Resistance Tip to Ring, REN = Ringer Equivalence Number.
The next page along is the “Basic Settings” page. This page allows you to set the user password, whether Telnet server is enabled, the network IP settings, time-zone, language and reset settings. Note that with the user password, a limited access to ATA pages is provided, unlike with the admin password which is set on the next page.
The “Advanced Settings” allows you to configure the admin password, QoS tagging/VLAN tagging, STUN server, network connectivity, auto-update/provisioning, TLS, TR-069 settings, ring cadence, call progress tones, keypad features, NTP server, remote logging, backup/restore configuration and update firmware. The tones are set by default for the US market, and thus, for Australia are not correct. Trying to get settings which sound “authentic” is not really possible, as only two frequencies are catered for and our dial-tone is a mixture of three frequencies. At this stage, a characteristic Grandstream configuration “deficiency” begins to show – namely that the configuration pages are rather long “laundry lists” of options which aren’t always sorted in any logical order. This can make configuration daunting compared to a more sensible menu structure.
Each port has its own configuration page, which is the longest configuration page of all. I’ve uploaded the image above, so you can “go through” the settings by looking at it, but it is quite feature rich and should have most of the necessary settings available. Interestingly, register expiration time is configured in minutes (whereas other ATAs can be configured in seconds), but a register-before-expiration feature is provided to ensure the registration does not lapse. Various NAT-handling features are provided, with provision for dial-plan, auto-dial, call duration limit. DTMF, fax mode, line gain, VAD, echo canceller and jitter buffer settings are provided, and codec priority choices are provided. Instead of configuring p-time directly, we can configure the number of voice-frames-per-TX (packet). Support for SRTP is provided as well, for encrypted communications.
Compared with the HT702, some of the settings names have changed slightly and support for pulse dialling and high ring power has been lost (which would only really be useful for the oldest telephones). Compared with the SPA112, it seems both are highly configurable although there are differences in the options provided (e.g. SPA112 allows for T.38 repetitions to be configured).
The big feature difference probably boils down to the support for the more modern and computationally intensive Opus codec, which is supposed to offer better voice quality at any bitrate. Unfortunately, support for Opus is still somewhat limited amongst most VSPs, and is really most useful when making IP-to-IP calls. It is good to see that the HT802 supports virtually all popular VoIP codecs – something the SPA112 falls short on (no support for iLBC, G.723.1 and Opus).
By default, the HT802 also has a Telnet interface enabled, although I didn’t use it to configure the device. It also supports remote provisioning in various ways – XML configuration, DHCP options, 3CX auto-provision and TR-069. As I don’t use any of these features, I cannot really comment on their effectiveness.
The settings back-up and restore are text or XML files containing lists of parameters and values, which is rather cryptic. I did not test settings restore, although it doesn’t seem any of the FXS line parameters are directly readable from the text files.
P694=97 P730=0 P855=0 P695=0 P731=1 P856=200 P570=20 P571=20 P732=60 [...] <?xml version="1.0" encoding="UTF-8"?> <!-- HT802 XML Provisioning Configuration --> <gs_provision version="1"> <mac>[REDACTED]</mac> <config version="2"> <P694>97</P694> <P730>0</P730> <P855>0</P855> <P695>0</P695> <P731>1</P731> <P856>200</P856> <P570>20</P570> <P571>20</P571> <P732>60</P732> <P857>200</P857> <P4593>2</P4593> <P572>1</P572> [...]
Voice and Data Calls
Placing regular voice calls over the ATA proved to be straightforward. Without any tweaking, the call quality was quite acceptable at defaults, however, it seems it is a little too loud on its output. Using cordless DECT phones which “digitize” the signal again, noticeable distortion and crackling of the dial tone was heard, sounding like the dial-tone coming through a rainstorm on a tin roof. Echo was fairly well controlled. However, to get the most from the ATA, configuring it for Australian line impedance and tweaking the send/receive gain levels is necessary.
Testing with my local VSP using “within-VSP” calling showed that the HT802 had an interoperability issue where it could receive calls made using a softphone (CSipSimple) from another account on the same VSP with two way audio, but if the HT802 initiated the call, there would only be one way audio. A bit of troubleshooting later, it was determined this was because the VSP was using “reINVITE” to avoid media-relaying, and this means that once the call signalling has established a call, both ends are asked with an INVITE to “update” the call so it goes directly between the two parties and doesn’t involve the VSP. It seems that the HT802 seems to prefer UPDATE rather than INVITE, so setting the “Force Invite” feature fixes this problem and allows for reliable two-way audio. Of course, this is also dependent on all your “upstream infrastructure” (i.e. router, modem, firewall) being correctly configured (e.g. fixing RTP ports, port forwarding RTP ports and SIP ports to ensure direct IP-calls can be received).
It was at this point of testing that I proceed to grab out the data modems for some fax action. Unfortunately, some quirks were found. As Australia is an A-law country, the defaults of U-law are not appropriate. Configuring the codec list to prefer A-law over U-law, and then using passthrough mode and not T.38 resulted in call failures through my primary VSP.
The call failures were rather interesting, as the call proceeds normally until a fax tone is detected, and then silence occurs. From a packet capture, it seems that in passthrough mode with my VSP, the VSP’s gateway detects the fax tone and attempts to reinvite the HT802 in a more appropriate mode. The HT802 then somehow decides to choose U-law. This results in an asymmetric media condition, where the audio in one direction is in A-law, and the other is in U-law, and the HT802 then “becomes silent”.
In fact, in further testing, it seems the HT802 doesn’t like asymmetric media conditions – even if that means the ptime on one direction is 20ms, and on the other is 10ms. This always results in the HT802 “going silent”.
This is fixed, however, if you choose U-law as the first priority, and A-law as the second, which results in a proper fax pass-through call with my VSP, but then it would screw up all your voice calls. Even more curious is that this interoperability issue does not occur using my internal Asterisk + Digium Free Fax for Asterisk line. That line negotiates the codecs in the right order every time. As a result, your mileage may vary.
Of course, if your VSP supports T.38, it’s pretty stupid not to use it for faxing, since it improves your fax success rates rather dramatically. In the case of the HT802, the T.38 implementation seems to be quite solid, reliably picking up the fax tone and sending out “well framed” packets. It was happy to work T.38 with both VSP and my local Asterisk box with no flaws.
As I wanted to better understand the match and quality of the line codec, I needed to do pass-through fax testing. Using FOLDS-B, it was determined that a Tx setting of -4dB and an Rx setting of -10dB was optimal (default is 0/-6). Testing of the frequency response and signal levels were based an Australian region modem (Netcomm Roadster II) and a lot of calls (about AU$4).
The result is actually quite good – although there is some high frequency roll-off visible, it is not as severe as the other ATAs I’ve previously used. A good result all-over.
When it came to data calling, using my digital VSP termination calling into a V.90 capable modem bank, I was able to achieve connect rates of 44kbit/s to 48kbit/s, implying a fairly good codec implementation. Given that SiLabs codecs are used in most of my “preferred” ATAs, this is no great surprise. Connections using “back-to-back” analog modems going through my local Asterisk server were able to achieve 31.2kbit/s symmetrically, which is a good result. However, in all cases, where the modems were attempting a V.8bis handshake early in the call, it seems the HT802’s “fax tone detection” kicks in even if switched off, and interrupts the stream just enough to make V.8bis stutter badly or fail entirely. A data connection can still be established, just without the benefit of the V.8bis negotiation. This is the same behaviour as seen in the HT702, although is not seen in the PAP2T or SPA112.
Looking at the network behaviour of the ATA, it seems like the incorrect frame checksum issues seen in the SPA112 and HT702 are not present in this ATA, which is good.
On the whole, once configured, the ATA was mostly frustration-free, although the lights on the top of the ATA are an interesting design choice. Where the HT702, SPA112 and PAP2T all have edge mounted lights which can be seen from a wide variety of angles, the top-mounted lights of the HT802 are mostly best seen from above. They are also fairly bright for their purpose, so in a dark room, can be quite distracting. I suppose the fix is simple – some black electrical tape to cover it up.
The ATA isn’t particularly weighty, so in the case of using stiff cables, it can often “slide” off the table or lift-off of it. It runs fairly cool, and according to a USB power meter, consumes about 300-400mA under most circumstances. The use of a USB port for power is a real boon, as it means it’s easy to find replacement power adapters and it could also be possible to travel with it and power it from a PC or USB power bank.
The use of American call progress tones is not really suited for Australians, although I suppose we (as people) can adapt to live with it. I’m still yet to find a perfect configuration, although this is somewhat close.
The HT802 is Grandstream’s latest ATA. Its fairly well priced, and offers a full complement of features. Specifically, it uses a more modern high-specification DSPGroup DVF99 series chipset which brings more performance, hardware crypto and secure-boot features to the unit. It also brings Opus support. These features alone make the HT702 and SPA112 look “old” by comparison, especially when you consider they still retail at a very similar price.
The HT802’s design is rather simple, but it still features the line-side polyfuse protection seen in the HT702 making it more robust. Attention to safety can be seen throughout the design, with electrical isolation everywhere. The use of a USB micro-B connector for power is also a good convenience, especially should the power supply need eventual replacement. The power supply was compact, not obstructing adjacent outlets as the SPA112’s does, which consumers would much appreciate.
Call quality was good in general, and data calls completed with a high completion rate. Frequency response was good, and the stability of the unit was unimpeachable.
On the downside, the power supply didn’t appear to have the same level of quality internally, with cost-saving component reductions which might compromise longevity and tolerance to transients. The configuration interface is still rather cumbersome and difficult to work with. The unit was “too loud” out of the box and firmware quirks with Fax pass-through in G.711a were experienced with my VSP, although T.38 worked flawlessly. However, it did seem that the fax tone detection interfered with V.8bis negotiation of data modem connections, just as it did in the HT702.
If you’re on the market to buy a VoIP ATA, there really aren’t many choices left. I suppose if you’re considering buying the older Grandstream units or the Cisco SPA112, the HT802 is noticeably superior. Call quality issues under heavy network loading/administration with the SPA112, the FCS packet errors with both SPA112/HT702, data call quality problems with the HT702 and T.38 framing quirks in the PAP2T were all absent from the HT802. I think it’s just become my favourite ATA … now that Cisco isn’t really paying any attention to this market.