Originally Posted by Malouff
whitis I am sorry but currently I don't own a CECB yet.
Neither do I.
Originally Posted by Malouff
Are you wanting me to identify what guide/image fits into what category?
The idea is that members who have different converters and have used them could post one message per box (or family of boxes) with information about guide behavior that you can't see from the screen capture.
Answering these questions can depend on understanding the overview of PSIP transmission in my previous post and a general understanding of algorithms which can be used to store PSIP data.
- Does the box show: current program only, now/next one channel at a time, now/next grid (all channels), or a multihour/multichannel grid.
- How many days/hours of PSIP information does the box appear to store? Take into consideration that TV stations may not transmit a full 16 days, so we can just get a minimum bound based on the highest number across all channels. Limits on how far you can scroll might be an indication.
- Does it store for all channels or just the current channel? Some boxes might be able to give, for example, 12 hours of information but only on the current channel and forget that data when you tune to a new channel.
- Is there evidence of non-volitile memory (flash/battery backed) for program guide information?
- Does it appear to acquire PSIP information for all channels when it is turned off for an hour?
- Does it appear to acquire PSIP information for all channels after having been unplugged and then left turned off for an hour (power failure recovery).
- Does it appear to acquire new PSIP information when you channel surf and are not using the program guide.
- Does it appear to acquire full available PSIP information while you are channel surfing , if you spend at least 1 minute per channel? If you spend significantly less than 1 minute per channel does it appear to be populating data if you scroll 16 days into the future?
If you tune through a channel that sends the full 16 days in say 6 seconds, you might get the next 3-6 hours of programming plus 30 hours at some random time in the future.
- Does it appear to acquire PSIP information when autoscanning for new channels?
- What is the maximum amount of PSIP data sent by any station in the area where your are tested, if you have an indication of this from other equipment of from the station itself
- Does the box retain stored program guide information if unplugged for 10 seconds?
- Can you scroll through the program guide channels without the box tuning to another channel? I.E. can you peek at the program guide and still listen to your current program? Does the box interrupt your program and scan through one or more other channels to acquire data when you load the guide every time, only when it needs info it doesn't have, or never.
- If the box requires tuning to another channel momentarily to get the information, and does so with your permission, does it return you to your original channel?
- Does the box appear to store EIT-0 for all channels you have visited, without using program guide.
- Likewise for EIT-1
- Likewise for EIT-2
- Likewise for EIT-3
- Likewise for EIT-4+, how far out?
- How much detail does it give on each progam (current, next, 1-3 hours out, 3-6 hours out, and further out? I.E. is it throwing out data for programs that don't come on for a while to save space?
- Does the box crash occassionally? Any indication may be EPG related?
- Are there performance problems in the general use of the box that appear related to program guide information? If the box has a way to turn off the program guide, this can help compare.
- Will the box show you as much data as it appears to remember or is the user interface more limited in what it will display? Example: retaining 3-6 hours (EIT-0/EIT-1) but only showing now/next.
- Does the box have a function you can invoke to initiate population of program data for all channels? I.E. push a button and it scans all the channels while you take a potty break. Does it offer quick (1 minute/EIT0/EIT1, 3-6 hours data), medium (3 minutes EIT0 ... EOT3, 9-12 hours data), and full modes (1 hour, 16 days data or max supported)?
- If the box is left off for an extended period of time (days), does it turn itself on periodically to collect program guide information so it will be ready when you do use it?
- Can you select a program on the grid and view program details without changing the channel?
- Does it have a sleep timer that will cause it to go into standby? Some people leave the TV on all day and when they go to bed, thus the box never has a chance to load program data.
- Do you have knowledge of how many hours/days data the local stations transmits? When they update the info? You might have some or all of this based on using a digital TV with full program guide, by calling the station, or because you work there.
- Any quirks? Notable good/bad features?
- Summary question: Does the box opportunistically pick up as much information as it can, subject to a limit of not storing data too far into the future (but at least 3-6 hours), and retain that information? How many hours/days by how many channels (one/all)?
Many of these questions relating to scavanging are less important if a box stores at least 4 days of program info (when you turn it off daily) and thus can always display two days. You lose up to 24 hours due to the fact that a station may update its EIT table once a day and another 24 hours based on when you last turned the unit off. Thus if the station sends 72 hours and updates at noon and you turn the unit on at five minutes till, you are already getting only 48 hours and 16 hours later you are only getting 32 hours plus what could be scavanged.
This list could be used like the list of closed caption questions was used on the captioning thread. The information gives information that can be used to determine how useful the program guide will be and how intrusive it is, in the form of delays and the need to tune away from current channel to acquire data, to a particular set of viewing habbits.
I will illustrate with a hypothetical test of a Zenith style now/next converter box in a particular location.
The idea would be to unplug the unit (not merely turn off) for a few days (as many as it appears to have data) to deplete the memory so you can see when it gets populated. Note that if box doesn't have flash or a battery backup, unplugging for 10 seconds may suffice.
Then carefully experiment with the unit. If a DVR or VCR is available, set it to record output of box while experimenting. Suppose we have the following channels (zipcode 22903):
16-1, 19-1, 19-2, 27-1, 27-2, 29-1, 29-2, 29-3, 41-1, 41-2, 41-3.
Now, we should really look up on TVfool which physical channels are which since we have one company broadcasting three different networks but it turns out they are using separate frequencies. I will assume a now/next zenith type box that lets you scroll through channels without actually tuning to another channel but may tell you there is no data. I am guessing, here based on the unit doing the best that it can, storing EIT-0 and EIT-1 only for each channel. I.E. it is storing a little more info than it shows. Do not change channels except when part of the experiment to avoid acquiring extra data. Also, an unexpected result at any stage of this test could affect the assumptions in following tests; you need to know enough to adapt.
- Set channel to channel 16-1.
- Unplug unit to deplete memory
- Plug unit back in and immediately turn on
- Hope we are still on 16-1
- pull up program guide. Scroll through channels without tuning. Should see program listings for 16-1 only. If we have program data, this may be an indication of flash/battery backup (either of which is good but could wear out). Rest assumes we only saw 16-1. Exit program guide.
- CH+ to 19-1,
- pull up program guide. Scroll through channels without tuning. We expect to see listings for 19-1. We should see 19-2, and 19.3; if not, we have a bug (failure to store other virtual channels on same physical channel. We should also see listings for 16-1, if not we have a bug (failure to remember other physical channels). We will not see listings for 27-*, 29-*, and 41-*. If we do, the unit a second tuner for guide data (not likely), is telepathic, or we screwed up.
- Turn unit off for 3 minutes (but keep plugged in), then turn back on and pull up program guide. (Are we still on 19-1? This tells us if we have last channel memory as long as power is applied.)
- We should see now/next guide data for all channels. If not, we have a bug (failure to scan channel data when first turned off).
- Stay tuned to 19-1, for just over 1 hour (assume programs are less than 1 hour) with the box on so it can't scan.
- Pull up program guide and scroll through channels without tuning. We should still have now/next information for all channels. If not, we have a bug (failure to store EIT-0 and EIT-1 for each channel, storing ONLY the now/next shows).
- Stay tuned to 19-1 for just over 6 more hours. Note that this test is similar to watching a movie or a couple of consecutive shows on one channel without changing channels. Data will become depleted due to expiration in roughly 3 to 6 hours. We give it 6 in case we started at a best case time.
- Pull up and scroll through program guide without tuning. We expect to see no program listings for any channel but 19-1, 19-2, and 19-3. If we see more, we have a feature: storage of EIT-2 and EIT-3 and we should also check to see if it stores more than 2 and 3.
- Unplug unit for 10 seconds (more if non-volatile), plug back in, and leave turned off for 5 minutes. When we turn on and look at program guide, we should have now/next data for all channels. If not, we have a bug (failure to scan after power failure).
Note time in both local time and UTC when doing tests at the start of the test and after each long delay. EIT blocks start at hours divisible by 3 in UTC, not local time. Thus the time you do the test dictates how much usable data is in EIT-0 and EIT-1.
These tests require some understanding, not just following a recipe because you have to adapt the tests based on what you learn along the way.
For some, it may be helpful to use UTC and draw a chart on a piece of graph paper with virtual stations, grouped by physical channel, across
the top and time in UTC down the side in 1 hour or half hour blocks.
Group each block of three hours (starting at UTC hours divisable by three) together and write down the EIT number. Draw heavy lines at physical channel boundaries and the 3 hour boundaries. Then, when you tune each station, you can put an X in the blocks for what data the converter appears to be able to access.
A box that doesn't at least store EIT-0 and EIT-1 for all channels visited and avail itself of all legal opportunities, but only legal opportunities, to acquire that information is really substandard and the program guide will be of limited usefullness and very intrusive. That is less than a megabyte of storage and thus is likely to have no effect on the marginal cost of the unit. Thus, any box which doesn't give you 3-6 (depending on time of day) hours is pretty shoddy engineering.
If we were to give a letter grade, it might look like this before adjustment for quirks:
A+ - TV Guide Online, 14+ days, grid, details
A - 14+ days PSIP, full grid and details
station transmission may limit actual performance
B - 7+ days, all channels, grid, details
station transmission may limit actual performance
B- - 48 hours+ usable data at all times (store 96 hours)
provided you turn the unit off once a day
C+ - 24 hours+ usable data at all times (store 72), all channels,
C - 3-6 hours, all channels, grid, opportunistic scavanging
The least that can reasonably be expected.
D+ - now/next, grid, details, opportunist scavanging, EIT-0/EIT-1 stored
D- - now/next, storing less than EIT-0/EIT-1 for all channels
F - current only
Originally Posted by Malouff
I don't think there are many places that have 50 channels with 4 subchannels
Regarding the memory look at mine and bdfox18doe posts
Well, TV fool shows that a tiny percentage of the population (0.5% or so) has 53 total channels (analog/digital) and that is roughly where I got the 50 number from. So that was used as the maximum density the airwaves seem to support. After the switch the maximum number of physical digital channels is expected to be 29. But the freed up spectrum will be auctioned off and it is possible that some of the buyers could actually be new networks or stations and estimates might be low.
Of the limited number of digital channels around here (assuming I don't pull in stations from neighboring areas), most have 2 or 3 subchannels.
29 (NBC): 3 channels
41 (PBS): 3 channels
19: 2 channels
16: 1 channels
27: 2 channels
Thus the average would appear to be 2 but that is deceiving because 19, 16, and 26 are really the same company that has only recently added CBS, ABC, and Faux. So, 29 and 41 are more indicative. So, I am going to assume 3 virtual channels (1 HDTV and 2 SDTV) iis the norm but stations might try to transmit 4 SDTV programs at certain times of the day. Actually could be worse, because you could have low refresh rate information like weather maps, stock tickers, etc. And even if something has the same show on 24 hours a day, it still takes up a fair amount of PSIP space.
Thus, I figured 4*50 was a good off the cuff upper estimate, but might not have actually been high enough. And you need to use an upper estimate, lest you confuse people with stuff like "you can store 8 days of program guide if you have 24 or fewer virtual channels and 4 days if you have 48 or fewer". You also complicate the software handling as it needs to prune far out data for some channels to make room for near future data for other channels and create a higher risk of software bugs in appliances you can't update. The simplest and most reliable way to store the data would be to store the RAW PSIP blocks in 1K blocks of memory. Pruning and packing information could slow the box down, give rise to fragmentation problems, and generally contribute to unreliability.
PSIP broadcasts today seem to be pretty substandard (72 hours recommended minimum), which also complicates real world testing a box. After the transition, there may be pressure to get the program guide up to at least a week, especially if we start seeing more DVRs.
So, I think 26MB dedicated PSIP memory is still a good estimate for a true full program guide, unless virtual channels share EIT blocks which would severely limit the quality of guide data.
Originally Posted by Malouff
This is just my assumption/speculation.
I have been told that those boxes with a Extended EPG do tend to slow the box down and reading the forums it does seem to fit.
They should not slow the guide down if you just store 128 EIT blocks in a rotating window for each channel. It is when you try to compress the data (making assumptions that could prove invalid in the future) that you slow down the box and potentially make it unreliable by having to do lots of move operations or dynamic memory free list searches. Just dumping a handful of 1K PSIP blocks into memory each second shouldn't have much effect.
I did look at the memory discussion on the TR40 thread but saw no hard data there.