Factless

It turns out that the 52 inch DLP projection TV I bought made by Toshiba has a flaw in regards to the lamp included with it.

References:
* Toshiba 52HMX85 bulb has blown! (Digital Home Canada)
* Toshiba Investigates DLP Lamp Problems (Audioholics Editorials)
* Information for Owners of 2005 DLP® Televisions (Toshiba Canada Support)

The good news is that Toshiba has owned up to the issue and has extended the warranty for the bulb to two (2) years automatically for TV models affected by the problem. This is great as my bulb blew last night. I wasn't sure about the "dimming" I had experienced before it blew was my mind playing tricks on me or not because I haven't been watching a lot of television lately (into summer reruns).

I also found out that bulbs certainly seem to vary in price, from about $230 to nearly $300 Canadian in a couple of Canadian stores to as little as equivalent to $200 Cdn on eBay (incl. shipping) for the same Toshiba brand replacement bulb (D95-LMP - part no. 23311153).

My replacement bulb from Toshiba should be sent tomorrow, so I won't be without high definition television for too long. I suppose I could hook up my 27" RCA CRT (tube) standard definition TV is I get desperate, but I've been watching so little television I doubt I will bother.

Posted by mctaylor at 5:16 PM | Permalink | Comments (1)

It took longer than it should have (roughly two weeks), but in the end the problem I was having accessing Akamai hosted websites such as eBay Canada, Chapters-Indigo was resolved.

I had to use the BroadBandReport web forums about Cogeco, to report the problem to some dedicated Cogeco employees who eventually took a closer look at my problem (the traceroute ICMP checksum errors appear to have been a red herring), and in the end they found both a broken Akamai box, and a bug-like issue (I think means a misconfiguration or an obscure error).

So the problem was fixed, and I can buy obscure computing and RF gadgets at surplus or near-surplus prices again, and appease my bibliophile tendencies.

Posted by mctaylor at 5:07 PM | Permalink | Comments (0)

Lately I've been trying to design a 24V_dc @ up to 2 Amperes power supply. The power supply is for a surplus Datum (acquired by Symmetricom) LPRO Rubidium oscillator that I picked up off eBay.

I figured I should use a linear power supply with good voltage regulation, and able to meet the up to 1.45A needed during 15 minutes of initial warming up (literally, there is a heater for the Rubidium (vapor?) as I understand it). So I figured that I should use something better than just a 3-terminal 78xx or LM317 series voltage regulator with a pass transistor.

I quickly realized that the LM723 was the closest thing I could easily find to fit my needs, with better regulation than the 78xx series and since I would need a pass transistor anyhow, the ~100mA limitation of the chip itself is not a concern. I will actually use the LM723C which main difference is its rated operating temperature range.

I also added two suggestions to hopefully improve the voltage regulation, first I added a zener diode based pre-regulator, which drops the input voltage from the bridge rectifier and filter capacitor to a minimal amount needed for the voltage drop of the pass transistor and LM723 regulator, which is about 3V. So I selected a 28V zener - a 1N4751 with its own pass transistor to handle the necessary current for this pre-regulator zener diode which gives roughly 1V "extra".

The other thing I did was try to reduce the noise on the output of the LM723, by adding a circuit based upon an interesting article from Wenzel Associates entitled Finesse Voltage Regulator Noise! I don't remember how I found this article (maybe from the time-nuts mailing list), but I thought it was a gem. Maxim/Dallas Semi have a similar Application Note, AN3656.

The one thing I didn't add to the design was using an external shunt voltage reference, like the LM399 or cheaper LM329. It would of only required the 3/4-legged IC plus a FET as a constant current source (set with a resistor) - total cost of about $1-10 depending on which reference IC used. This would of reduces the output drift of the LM723 due to ambient temperature change, by an order of magnitude or two. I didn't add it because I could not figure out how to add the reference without messing up the non-inverting input when used with a PNP series pass transistor.

Although I didn't need it, I found the explanations for the fold-back current limiting confusing, as I could only find one equation to specify the value of two resistors...(R3, R4 in figure 6 of the National Semi LM723 datasheet)

Posted by mctaylor at 11:09 AM | Permalink | Comments (0)

Sigh, for the past two weeks I have been having intermittent problems with several major web sites, which all happen to be Akamai customers. Akamai offers "high performance content delivery" via their solution called EdgeSuite.

Unfortunately for companies like eBay Canada, Chapters-Indigo, CBC, Analog Devices, and Linear Technologies, it has not been working very well for me, a Cogeco cable modem based High-Speed Internet (Standard, Residential) customer.

So far it has cost Chapters-Indigo a few hundred dollars, and a undetermined amount for eBay Canada in lost sales.

The technical problem appears to be part of Cogeco's internal network is mangling packets.

[root@push mctaylor]# traceroute www.analog.com 490
traceroute: Warning: www.analog.com has multiple addresses; using 216.221.80.240
traceroute to a257.g.akamai.net (216.221.80.240), 30 hops max, 490 byte packets
1 10.70.128.1 (10.70.128.1) 7.854 ms 7.248 ms 5.741 ms
2 d226-12-85.home.cgocable.net (24.226.12.85) 13.037 ms 11.686 ms 14.307 ms
3 cgowave-0-114.cgocable.net (24.226.0.114) 16.738 ms 14.149 ms 14.032 ms
4 cgowave-0-109.cgocable.net (24.226.0.109) 12.848 ms 14.293 ms 13.436 ms
5 cgowave-0-198.cgocable.net (24.226.0.198) 14.271 ms 13.259 ms 13.748 ms Icmp checksum is wrong
6 d221-80-240.commercial.cgocable.net (216.221.80.240) 14.506 msIcmp checksum is wrong 12.903 msIcmp checksum is wrong 14.600 ms

Compared to accessing a popular web site, not located within Cogeco's network.

[root@push mctaylor]# traceroute www.google.ca 490
traceroute: Warning: www.google.ca has multiple addresses; using 72.14.205.103
traceroute to www.l.google.com (72.14.205.103), 30 hops max, 490 byte packets
1 10.70.128.1 (10.70.128.1) 6.430 ms 9.424 ms 6.591 ms
2 d226-12-85.home.cgocable.net (24.226.12.85) 12.487 ms 12.694 ms 14.357 ms
3 gw-google.torontointernetxchange.net (198.32.245.6) 12.834 ms 12.746 ms 11.978 ms
4 66.249.94.92 (66.249.94.92) 11.904 ms 13.540 ms 66.249.94.96 (66.249.94.96) 12.030 ms
5 72.14.232.66 (72.14.232.66) 23.527 ms 27.627 ms 72.14.236.142 (72.14.236.142) 26.070 ms
6 qb-in-f103.google.com (72.14.205.103) 14.245 ms 13.561 ms 13.535 ms

Note: The 10.70.128.1 is located within Cogeco's network, it is either a Cisco UBR (Universal Broadband Router) or an IP router.

Of course, I didn't know all this when I first reported the problem, initially for me the problem was that accessing popular sites was flakey, but that doesn't matter really. Now I cannot seem to get this information the Cogeco's internal technical staff. That is the real problem, not getting this more specific, and useful details to the right people.

Oh, well.

Posted by mctaylor at 10:44 AM | Permalink | Comments (0)