Wednesday, June 28, 2017

Amazon Wand adventure timeline

Amazon Dash Wand With Alexa


21 June 2017 21:51

Thought I do one last thing before bed, so tried to add the Amazon Wand I just got and of course like so much lately it is just not that easy. I opened up the app and found it had updated and wants ALL my info all over again. Now it is just setting here with what I'm guessing is a busy indicator.

22:27

OK of course just after I posted the above it came back. BTW I was referring to the the Alexa app.Turns out the Alexa app is not used to setup the wand despite it having Alexa built in. I dug out the instructions and it says go to a web page on your phone. No idea why because the web page just says use the dash button setup section of the Amazon app to set up. Opening the Amazon app up it of course wanted to update but at least this time it remembered who I am. Got it to connect on second try.  Looking at the tutorial I see it puts items directly into you shopping cart  NOT your shopping list like I could swear the ad I saw said. You can not even move the items from your cart to a list. I also tried scanning 3 random codes. A field guide, a gallon of bottled water and  dog food. It got the first one OK but then started saying "I'm having trouble". It usually says this when it can not reach the Amazon servers. I scanned the first item again and failed again. So all in all it is about as useful as the Echo remote. But at least it was free after rebate and maybe they will sort the bugs out down the road. If I can get it to connect again tomorrow I'll see what kind of luck I have adding random stuff. I'm afraid most of the stuff I'd scan though Amazon is not going to carry.

27 June 2017 16:00

Well been trying on and off but mostly getting "I'm having trouble right now". Today seemed like the perfect thing to try as I needed to reorder dog treats I usually order from Amazon. I used the wand to scan the bar code and it sounded like it worked so. Left it till later to check. Then later I remembered I had not placed my order and I went into Amazon but the cart item counter in the upper corner showed no items. So I went and scanned again and again it sounded like it worked but nothing showing. So I found the treats in my wishlist of recurring items and added to my cart. Then when I went to check out I find this.

So it appears they made some changes since I first tried it. On the plus side it appears they fixed the bit of not adding stuff they do not carry since it did find my frozen yogurt. Oddly it appears speaking into the wand now using the Alexa command to "add to shopping list" is now screwed up. (I could not tell the scan worked so tried speaking it.) Also despite me regular ordering these treats, it appears they could not figure out what treats I wanted to order because if I click on one of the "Choose Top Chews" buttons I get this.

BTW the UPC for the Cadet treats is 768303013119 and the UPC for the Top Chews bag is 31400066341 so a dup UPC is not it.

Hopefully they keep working on this and make it go to the shopping list instead of the cart or this staging area (which seems like the worst option). It would also be nice if the wand could say added [name of item] to your list / cart when is scans something so you know it worked. Saying "I'm having trouble" for any error and nothing when it "works" is very old school Unix style which appears in a lot of places but is just wrong in an interface like this. Especially as often as my Echo often mis hears me. That was my whole point of wanting this wand to eliminate mis hearings and get more detailed entries on my shopping list by using the scanner. Also trying to trigger an IFTTT event seems to just get you the "I'm having trouble" response so it does not seem to even be usable as an iffy Alexa remote now. But frankly that is no biggie. I have 4 Echos I can talk to already. I just need the scanner bit to be useful.

Update 7/6/2017

I think I might have sorted something. I think the Wi-Fi just takes a bit to get going. That makes the need for it telling me what it added even stronger. A post in the user's group got me to go look at mine's connection history. It looks like when you wake it up it stays up for around 4 minutes after you last release the button. Which is good for battery life. But if your Wi-Fi signal is weak or, like mine heavily encrypted, it might just be taking too long to connect. I know it is not unusual for it to take a couple seconds for one of my phones to connect. As you can see from this log pic my Wand appears to often connect to the Wi-Fi but not send anything.

I'll have to do some testing to see if that lines up with failures to scan. Maybe asking the time first and then scanning should be recommended use. After all it is not "on" all the time. When you press the button it has to power up from power saving mode, connect to an access point and get an IP before it can think about sending your request to the server. One would expect it to be doing all that in parallel with scanning or recording audio to send to the server. And that it would hold onto the data till the network is ready for it to be sent to the server. But maybe it cannot for some reason and attempts to send before the network is ready? Again from my logs it would appear that it often times out before it can talk to the server since about half the time 0 bytes are transferred.

So if you press the button and ask the time, that ought take long enough to give it time get connected before it sends that audio to the server. Then you can do the thing you really want to do since it appears to stay connected to your network for about 4 minutes after your last request. If it says it cannot tell you the time for reason X then that just means it is not ready yet. Just ask again and it should work. Though if fails 3 times you might have a real issue. I did a couple quick tests. Seems like if I pick it up and scan it works about 1/2 the time. If I ask it the time it seems to always work. Though I've only tried this a couple times so far. In the logs it looks like this
First instance I scanned which did not work but I could see it had connected to the network so I went back over and asked it the time which worked fine of course.
The second instance I scanned the same UPC code that did not work last time and it worked this time (though I could not tell at the time that it did). I then asked for the time which worked and scanned again. The second time I heard the ding that I believe is supposed to tell you it worked.
The curious bit is the earlier transferred a LOT more data though it did less. It was connected longer due to the time between me scanning and then going back ask the time.
After a wait I tried to repeat the second test, scan / ask time / scan, and indeed only second scan and asking the time worked.

Again odd though I did the exact same thing yet it sent almost double the data. Especially odd since the first scan failed.
I also grabbed the connection info before it disconnected to see the signal quality.

At 99% (-46 DB) you are not going to get much better so I think we can rule that out. I have a second Wand coming to do some comparison tests to confirm I do not have an iffy one but from what I'm seeing in the user group I think this is just how it is working right now.

Posted to Amazon customer reviews 7/8/2017 to give users work around to connection issues.

From testing I've done with 2 Wands on enterprise class APs with network monitors and such, there seems to be a race condition with the scanner in that it seems to try and send to the server before the network interface has powered back up. Or maybe just times out waiting for the network to come up. Hard to say which from the outside. Either way I'm seeing it connect to the network and transfer 0 bytes. I'm guessing QA tested it on an insecure WiFi connection which is quicker to connect. Note low end, weak or loaded WiFi might be even slower to connect and see more issues than I'm seeing. BTW If your WiFi is insecure (as in your Wand is not having trouble you best go check) stop reading this and secure it NOW!

The work around is:
If you ask it the time and then scan it seems to work most of the time. Asking the time seems to give the Wand time to finish connecting to your WiFi. I'm seeing about a 50% failure rate if I try and scan something as my first action. If your WiFi is slow to connect you might get an error asking the time. If so ask again.

However:
If it fails twice you probably have a real problem. As in your on your network or the Amazon servers. I was just trying to confirm a couple things while writing this and I can see the Wands connecting to my network and sending data but still getting "having trouble right now" for everything I try. Yet my Echos are able to talk to IFTTT and add to my Amazon shopping list just fine. I can only conclude the problem is on their end.

Hopefully Amazon will fix this plus have scanned items go to your shopping list instead of adding to cached bit of your cart like many have already posted. Note scanned items were going straight to the cart when I first got mine but now they seem to be going to a cache that only shows when you view your cart. As in the cart icon in the upper right corner will still show 0 items in it so it might not look like the scan added anything. Once to view your cart you will see and button to select an item like the one scanned even if the item is one you order often. I'm guessing this is to address people complaining about which seller's item got put int their cart. Note too, when I first got mine it seemed to just ignore anything scanned Amazon did not sell but now it appears to add to cache.

Dear Amazon,
If you read the above, you should also have it tell me what it added when it works too. I noticed this AM when a UPC is not found it now asks what it is, which is better than the ding if "worked" and nothing if didn't which was all but useless. Best fix the race condition first though as it sounds like people are already returning them as bad. Hopefully the above work around helps well enough for people to hold out for an update. I have to believe an update is in the works. From the number of people I see posting about their issues and the fixes looking so simple, you have to have too much invested in this to just have people not use them.

P.S. I'd gladly pay $100 for one of these that would reliable send the UPC to IFTTT or similar so I could rig a kitchen inventory.

Update: 7/26/2017 another glitch

This time it is dumping Alexa commands into the cart cache
Bet they are starting to wish they started with Alexa and added dash instead of the other way around about now.

Update: 8/7/2017 Some good some bad

The Wands seem to take at least 2 shots before they work now. Still seeing a lot of "what time is it" entries in the cart cache too but on the plus side it now says "added" when it does add something to you cart.

Update: 8/18/2017 Just huh?

Lately it almost always says "I'm having trouble right now" but seems to most get the item into you cart cache or do the action. It is also adding commands to the cart cache. Note the "turn off light bar" command showing below actually did turn off the light bar before saying "I'm having trouble right now". This works find for me but will be a problem for those expecting to ask it question like an Echo. Also from the few times it seemed to scan an item successfully, the "added" response appears to have been removed.

Update: 10/30/2017 Same old

Still getting the can't do right now error almost every time I try and use either Wand. About half the time I get something in my cart.  Not see much posted in the Echo groups lately so maybe most have just given up on them. Did see something saying the bar code scanner in the Show actually works. Not really feeling like spending $250 to  see if it does. Maybe if they blow them out for Black Friday.

Tuesday, June 13, 2017

Glucose meters, WTF?

Enhanced version of my original Facebook post

Been using a True Test meter just to keep an eye on things every week or so since mid 2013. This is because I read at the high end of OK on the yearly physical blood panel that you fast for. Doc was worried my reading might be higher when not fasting but they do not seem to be. Pretty steady readings. (Excluding off diet days ) Should note I took a reading right before the blood panel each year to check meter's accuracy and the reading would always be real close.
Anyway recently I ordered what I thought were the right replacement strips but they are for a newer True meter model. The packaging looks almost the same. After some research I discovered my type strips had been pulled from the market because they did not meet new accuracy regulations. (FDA says they now have to be within 15% instead of the previous 20%) So I did more research to see what new meter I should get. I finally settled on the FreeStyle Lite which Consumer Reports, among others, said was more accurate and had highest repeatably. When I tried it though I was shocked how high my readings were. OK not call the doc high but but the chart here shows how things got a lot less steady on May 19th. Doc had me on some short term inflammation meds known to raise BP and glucose though so I just checked it every day or 2 till after I was off the meds. But the readings were still higher than I was used to after the meds ended on the 22nd. So I bought test solutions to check the meter. The readings with the test solutions varied widely too. High solution was 346 to 281 and low was from 65 to 43. You might notice my readings for Jun 5th are 120 and 7 minutes later 111. Which seems off but is within the 15% guideline. The test solutions however are not even close.

So I just ordered two more meters. The other top rated Bayer Contour Next and the one that matches the strips I got by mistake, a True Track. The test solutions are good for a couple more months so when the new meters get here I'll test all 3 side by side with the same drop of high and low solutions to see just how close they are. I only wish I still had a few strips for my old meter to compare with.

How I did testing

The new meters came with test solutions so I had 4 to try. I also found that maybe I did not shake the solutions enough in the tests above. This time I gave each 20 flips before putting 5 drops in a spoon. I then sampled each with meter in rotation, just in case testing some how tainted the solution. I did 2 tests with each meter in each solution. I would have liked to do more but the Bayer only comes with 10 strips and did not want to buy a bunch more if the meter is useless.

Each vial of test strips is labeled with what to expect when you test with. Here is what these test strips should report. I've highlighted the solutions I had to test with.

Strip Ranges
Contour Next EZFreestyle LiteTrue Track
lowhighlowhighlowhigh
Level 13746
Level 2111138
Low3060
High248372
True Control 084114
True Control 1175237
True Control 2361489

The disappointing results

Here are the readings I got. The ones in range of what they should be are in green

Readings
Contour Next EZFreestyle LiteTrue Track
Freestyle low3680140
Freestyle low (2)3849130
Contour Next lvl 212468267
Contour Next lvl 2 (2)12973264
Freestyle high182283Hi (>600)
Freestyle high (2)183296Hi (>600)
True Control 17998230
True Control 1 (2)73103211
So I just do not know what to make of this. Interestingly other than one reading they all are in range when tested with the same brand solution but only the Bayer was in range with someone else's solution. So my first thought would be the Bayer though it is readings seem low. I guess the thing to do is get more Bayer strips and check my blood with all 3 meters for several sessions. Given the above results I'm not sure if any of these meters tells me anything about my glucose levels. I'll update this again after I get a few readings.

One last thing. As I mentioned above the FreeStyle Lite Blood Glucose Test Strips are the most expensive at $74.50 for 100 and it appears Amazon has stopped carrying them. The TRUEtrack Blood Glucose Test Strips come in at 100 for $20 and Contour-Next Contour Test Strips come in at around $21 for 100.

Update 6/17/2017: First blood test. I got a drop of blood on my finger large enough to do 2 rounds of tests with the meters. First round

Second round

 To me the interesting bit is even though the Bayer and TrueTrack meters vary 8.41% and 12.26% respectively but they both vary in the same direction which would tend to make me think there might actually be that much variance in the blood sample itself. Let's look at the data again.

Readings
Contour Next EZFreestyle LiteTrue TrackAverageStandard DeviationMeanMin from max
Freestyle low368014085.3352.2080.00388.89%
Freestyle low (2)384913072.3350.2449.00342.11%
Round deviation5.56%63.27%7.69%
Avg for meter / average for test53.07%18.18%71.25%
Avg for meter / min for test2.78%79.17%275.00%
Contour Next lvl 212468267153.00102.62124.00392.65%
Contour Next lvl 2 (2)12973264155.3398.19129.00361.64%
Round deviation4.03%7.35%1.14%
Avg for meter / average for test17.95%54.27%72.22%
Avg for meter / min for test86.03%3.68%290.44%
Freestyle high182283600355.00218.10283.00329.67%
Freestyle high (2)183296600359.67215.67296.00327.87%
Round deviation0.55%4.59%0.00%
Avg for meter / average for test48.93%18.98%67.91%
Avg for meter / min for test0.27%59.07%229.67%
True Control 17998230135.6782.2598.00291.14%
True Control 1 (2)73103211129.0072.58103.00289.04%
Round deviation8.22%5.10%9.00%
Avg for meter / average for test42.57%24.06%66.62%
Avg for meter / min for test4.11%37.67%202.05%
Blood test 1 sample 1107143106118.6721.08107.00134.91%
Blood test 1 sample 2116135119123.3310.21119.00116.38%
Round deviation8.41%5.93%12.26%
Avg for meter / average for test7.85%14.88%7.02%
Avg for meter / min for test5.19%31.13%6.13%
Avg for meter / max for test6.64%12.59%5.94%
Average Round deviation (repeatability)5.35%17.25%6.02%
Average of averages25.03%26.07%57.00%

Repeatability wise Bayer seems to be best by a bit and best all around. Especially when you consider second place for  Repeatability is TrueTack and over all is Freestyle.

Something else I just noticed it the strip I just received for the Bayer meter have different ranges than the first ones that came with the meter.


Lastly I realized I had not said anything about the kits the meters come with.  Here are pics of them. The Bayer and TrueTrack came with test solutions and the Freestyle did not. I bought those separately.

The Freestyle

The Bayer 

And the TrueTrack

The Bayer kit seems not thought out to me. Especially given the meter seems the most accurate so far. Personally of the 3 lancet devices I prefer the TrueTrack one. It is easier to use and seems to get a larger sample for the same setting.

UPDATE: 8/5/2017 Added 6 more days worth of tests and some more color coding.

The Contour is still the one closest to the average of all 3 meters and seems the most repeatable

Readings
Contour Next EZFreestyle LiteTrue TrackAverageStandard DeviationMeanMin from max
Freestyle low368014085.3352.2080.00388.89%
Freestyle low (2)384913072.3350.2449.00342.11%
Round deviation5.56%63.27%7.69%
Avg for meter / average for test53.07%18.18%71.25%
Avg for meter / min for test2.78%79.17%275.00%
Contour Next lvl 212468267153.00102.62124.00392.65%
Contour Next lvl 2 (2)12973264155.3398.19129.00361.64%
Round deviation4.03%7.35%1.14%
Avg for meter / average for test17.95%54.27%72.22%
Avg for meter / min for test86.03%3.68%290.44%
Freestyle high182283600355.00218.10283.00329.67%
Freestyle high (2)183296600359.67215.67296.00327.87%
Round deviation0.55%4.59%0.00%
Avg for meter / average for test48.93%18.98%67.91%
Avg for meter / min for test0.27%59.07%229.67%
True Control 17998230135.6782.2598.00291.14%
True Control 1 (2)73103211129.0072.58103.00289.04%
Round deviation8.22%5.10%9.00%
Avg for meter / average for test42.57%24.06%66.62%
Avg for meter / min for test4.11%37.67%202.05%
Blood test 1 sample 1107143106118.6721.08107.00134.91%
Blood test 1 sample 2116135119123.3310.21119.00116.38%
Round deviation8.41%5.93%12.26%
Avg for meter / average for test7.85%14.88%7.02%
Avg for meter / min for test5.19%31.13%6.13%
Avg for meter / max for test6.64%12.59%5.94%
6/24/2017107125104112.0011.36107.00120.19%
6/24/2017108123123118.008.66123.00113.89%
Round deviation0.93%1.63%18.27%
Avg for meter / average for test6.52%7.83%1.30%
Avg for meter / min for test3.37%19.23%9.13%
Avg for meter / max for test6.00%7.20%1.20%
6/25/2017110113105109.334.04110.00107.62%
6/25/2017105124123117.3310.69123.00118.10%
Round deviation4.76%9.73%17.14%
Avg for meter / average for test5.15%4.56%0.59%
Avg for meter / min for test2.38%12.86%8.57%
Avg for meter / max for test4.70%4.17%0.54%
7/1/2017113132102115.6715.18113.00129.41%
7/1/201710910497103.336.03104.00112.37%
Round deviation3.67%26.92%5.15%
Avg for meter / average for test1.37%7.76%9.13%
Avg for meter / min for test14.43%21.65%2.58%
Avg for meter / max for test1.14%6.44%7.58%
7/2/2017112112109111.001.73112.00102.75%
7/2/2017109122106112.338.50109.00115.09%
Round deviation2.75%8.93%2.83%
Avg for meter / average for test1.04%4.78%3.73%
Avg for meter / min for test4.25%10.38%1.42%
Avg for meter / max for test0.96%4.37%3.42%
7/9/201799106104103.003.61104.00107.07%
7/9/201710210397100.673.21102.00106.19%
Round deviation3.03%2.91%7.22%
Avg for meter / average for test1.31%2.62%1.31%
Avg for meter / min for test3.61%7.73%3.61%
Avg for meter / max for test1.26%2.52%1.26%
7/16/2017104103101102.671.53103.00102.97%
8/4/2017107101109105.674.16107.00107.92%
8/4/2017107100121109.3310.69107.00121.00%
Round deviation0.00%1.00%11.01%
Avg for meter / average for test0.47%6.51%6.98%
Avg for meter / min for test7.00%0.50%15.00%
Avg for meter / max for test0.41%5.79%6.20%
Average Round deviation (repeatability)3.81%12.49%8.34%
Average of averages11.75%16.92%47.51%
LegendReading in range for test solutionLowest value of 3
Calculated valueHighest of 3
Reading with 5 of average
The pics of the new tests