There are a huge variety of radios sold under the “Baofeng” (officially, Fujian [Nan’an] Baofeng Electronics Co.) name. Some are probably official private label models, some are outright fakes, and some fall somewhere in between. To confuse things further, Baofeng originally didn’t bother to register their brand trademark in the US. Someone else did some “trademark squatting” by registering it. While this was playing out at the US Patent and Trademark Office (USPTO), Baofeng branded their radios as “Pofung“. Between the Baofeng / Pofung confusion, official private label models and the fakes, it is hard to tell if a particular “Baofeng” radio is legit or not.

To complicate things even further, at least two companies claim to be “the” official United States distributor of Baofeng radios:

https://baofengtech.com (who also refer to themselves as BTECH) state that “We have personally dealt with hundreds of counterfeit listings and have had them removed from marketplaces. Unfortunately, most sites will not prohibit the counterfeiter from creating another listing. The most effective means to stopping counterfeits is to buy direct from a BTECH authorized distributor.”

https://baofengradio.us claims “Baofeng Radio US is the Authorized Distributor of Baofeng products in the United States. All products ship from within the United States.”

In addition to those two, there are a large number of other sellers on Amazon, eBay, etc. who are also selling “Baofeng” radios.

The manufacturer isn’t helping the situation. Their Baofeng Official Website Announcement says “While Baofeng has worldwide distributors and resellers, products are all produced by Baofeng. The concept that radio not sold by one of the distributors is counterfeit is factually inaccurate.” Aside from that statement appearing to claim that anything marked with a “Baofeng” label is genuine, it does not provide a list of “worldwide distributors and resellers” so anyone could represent that they’re a factory-authorized distributor or reseller and the purchaser has no way of telling if that is true or not. To confuse matters further, there are many “Certificate of Authorization” letters allegedly issued by the manufacturer listing dozens (if not hundreds) of entities that are “authorized”. These letters seem to come in a a surprising variety of styles, and every one I’ve examined is expired. You can do a Google image search for yourself by clicking here.

My first experience with this situation was ordering a package of two “Baofeng UV-82HP” radios from Amazon in November 2019. This was a “too good to be true” deal with 2 radios, 4 batteries, 2 coiled combo speaker/microphones, 2 earpieces and two whip antennas. When I received them, there was no FCC ID number on/in the radio (required for them to be legally sold in the US) and after contacting the seller I was given the alleged FCC ID for these radios, which turned out to be for a different model entirely. I returned then to Amazon for a refund, stating that they were counterfeit. I also left a review saying the radios were counterfeit. Amazon approved and published my review and it showed up among several other reviews also stating that the radios were counterfeit. Apparently the seller complained because my reviews and the other reviews mentioning “counterfeit”, “fake”, etc. were removed without explanation. That product listing is still active on Amazon, three and a half years later.

I eventually ordered replacement radios on Amazon from seller “BaoFeng Tech”. This is the same seller as https://baofengtech.com, also known as BTECH. In fact, their Amazon pages show both “Sold by BaoFeng Tech” and “Visit the BTECH Store”. They came with the expected FCC IDs inside and performed as expected. Since then I’ve been making sure to only order products that are sold by BTECH.

You can probably get the same radio from other sellers, particularly if you’re outside the US. But I decided to stop “rolling the dice” and always order from BaoFeng Tech / BTECH.

BTECH themselves are not helping the situation with their multiple brandings. Also, while their radios have customized battery compartment labels with their website listed and they have custom packaging distinct from any other seller I’ve seen, all of their BL-8 batteries had been mis-labeled (along with everyone else’s) as 2800mAh capacity instead of the actual 1700mAh capacity. This continued for many years until they introduced a new version of the BL-8 pack which includes a built-in USB-C charging port in the pack. That pack is correctly labeled as 18mAh. Unfortunately they didn’t change the part number, so a “BL-8” battery might or not might have USB charging depending on when it was made and who sold it. It makes me want to jump up and down and yell “You’re not helping!” regarding confusing naming.

Having said all that, BTECH does seem to be the most reliable supplier in terms of the product being exactly what it is represented to be.

This is something I worked on quite a few years ago, and I’m finally getting around to posting it.

A long time ago I ran across a fan fiction story titled The Black Company in Middle Earth by a user named McJunker. You almost certainly know Middle Earth as the setting for J. R. R. Tolkien’s The Lord of the Rings, but you may not have heard of Glen Cook’s The Black Company novels. That is something you’ll probably want to rectify, as McJunker explains. The Black Company in Middle Earth is a full novel in its own right – if it were to be published as a paperback, it would be close to 200 pages long.

Anyway, I contacted the author because I loved the story. I was working on my copy editing and document formatting skills at the time because I was getting ready to have a paper published in a professional journal. I asked if I could “publish” his story in a somewhat more polished form, as a PDF. He loved the idea, and off and on over the next year or so I worked on this project. In the meantime the author entered military service, served in Afghanistan and came back home. How time flies…

Anyway, once McJunker returned I showed him the result and he loved it. But I never got around to posting it for some reason. So here it is: The Black Company in Middle Earth (PDF edition). I hope you find it as enjoyable to read as I did. If for some reason you’re prefer reading it online, you can read it here: The Black Company in Middle Earth (HTML edition). I’m partial to the version with my copy edits, but if you want to read the original, feel free to check it out via the link at the top of this post.

You passed away a month ago, but I’ve been so torn up about it that it took me this long to put thoughts to paper. Plus, I’m still in disbelief – you were such a positive whirlwind of energy that I have a hard time believing you were taken from us so soon.

The very first time I met you was when you were working at Stan Veit’s Computer Mart of New York. I was working for a company a few blocks away, designing computers. That first time I just ran in there to get some random part I needed, rather than waiting for mail order. Somehow we ended up talking. After that I’d tell my work that I urgently needed a part, just to go to CMNY to talk to you some more. Dave Levine showed me how to program the Computalker to say ‘things’ to you after I’d left.

While we were both in our own orbits, we passed each other often enough to become close friends. I have a series of vivid memories, though I can’t put them in any sort of definite time sequence other than by relating them to other events in my life, looking at concert ticket stubs, and so on. After we reconnected, I told you a number of these – some you remembered, some you could even pin down to a specific time and place, and others you didn’t remember at all. As you pointed out, we were partying pretty hard together during some of those times.

Stan’s CMNY closed and he ended up at Synchro Sound. Shortly after that, my project was shut down and I was looking for work. I asked Stan for a job and he was kind enough to set me up with a job there. You were working in that shoe company office in the Empire State Building and I’d commute back to Manhattan and meet up with you.

Eventually I lost track of you. I would run into various friends of yours at the strangest places – in a diner in the middle of nowhere in New Jersey, at the DMV, and so on. I’d always ask “Have you heard from Sue?” and the answer was always “No”, sometimes with a “If you do hear from her, let me know.”

I had pretty much given up finding out what ever happened to you when by chance I came across Stan’s obituary at legacy.com and saw that you had posted there. Knowing your married name I was able to find you with a web search. I discovered you were happily married and were doing what you loved. I decided I didn’t need to interject myself into your new life just to say “hello” and reminisce.

I don’t know why it took me so long, but 10 years after I located you, I decided to reach out and contact you via email. That was in October 2020. You were surprised but very glad to hear from me, and we had a brief but intense exchange of long emails filling each other in on what we’d both been doing since we saw each other last. I promised I’d come visit, but somehow never managed to find the time.

At the beginning of December 2021 you posted that you were in the hospital. I figured it wouldn’t slow you down at all – you were a force of nature, after all. With increasing disbelief I saw your less and less frequent posts, until your husband David announced that you’d passed away on December 30th. I was in shock – how could someone so full of life pass away so soon? It just wasn’t fair!

As I said above, it has taken a month for me to write this, and I’m in tears even now. One thing I do know is that wherever you are now, you’re definitely livening up the place with the joy you spread so freely to all around you. The lights here on Earth may have dimmed, but the heavens gained a new star when you joined them.

You go by Suzie Kerr Wright these days, but to me you’ll always be Sue (and when I need to distinguish between you and anyone else named Sue, “Sue with blue hair and roller skates”, or “Blue Sue” for short).

Farewell, Sue. I love you and I miss you more than you can imagine.

(AKA “How to gum up the inside of your $7000 printer with labels”)

NOTE: This is how I print LTO labels, provided to you for informational use. I am not responsible if you have a label come off inside your printer or for any other damage following this procedure may cause – you perform these steps at your own risk!

If you have an LTO tape library or autoloader with a media robot, you know that you need to label all of your LTO tapes with a unique barcode which both identifies the tape and tells the library what generation of media it is. You can purchase pre-printed labels such as the HP Q2013A which contains 110 labels with a random starting sequence number (you might get WZT653 through WZT752), for the incredibly low price (hah!) of only $92 (list price, street price is between $52 and $100). If you want a custom start sequence on your labels, you will find out that they are VERY expensive – once you find a web site that sells them, you’ll likely end up with a “Request for a Quote” button, and like they say “if you have to ask, you can’t afford it”.

Fortunately there is a very inexpensive solution to this problem. You will need 3 things:

• OL173WX labels in “Standard White Matte” from https://www.onlinelabels.com – if you buy the 250 sheet package for $39.87, you’ll get 8000 labels (32 per sheet) which should last you a lifetime, and still costs less than half of what a package of 100 preprinted HP labels costs.
• Free LTO label generating utility at https://tapelabel.de.
• An Epson SureColor P10000 printer (or any other printer that exhibits the problem described below).

Using the free LTO label generating utility, create a sheet of labels with your desired settings. I use Library Type: IBM, Tape Type: LTO6, Label Type: OL173, Color Scheme: full colors. I specify the 3-character alphabetic prefix I want and the starting sequence number (usually 0 for the first set of labels). Leave all of the other options at their defaults. Cick the “Generate my labels” button and open the PDF file it generates. Make sure it looks like the labels you were expecting. Be sure to check the “fine print” – the small “L6” (or whatever generation you are using) on the right side of the label.

Now, if you go and print these on your P10000, you’ll get wonderful-looking labels, EXCEPT that the barcode is cut off on the bottom of the page. The P10000 doesn’t want to print close to the trailing edge of a sheet of paper, and just discards the remaining part of the image without telling you. You need a trick or two:

• Get a sheet of 8.5 x 11 plain paper and make a mark exactly 3″ from the long end.
• Line up the 2 sheets of paper (OL173 label and plain) with the plain paper on the BACK of the OL173, so the plain sheet sticks out 3″ (use the mark from the first step) beyond the OL173.
• Use a a small piece of Scotch tape (maybe 2″ long) to tape the two sheets together where they overlap on the back. Put the tape in the center of the overlap after making sure the two sheets are aligned. DON’T overdo the tape – one piece, just enough to hold the sheets together without skewing.
• Load the paper into the printer with the OL173 face up and the blank sheet at the trailing edge (in other words, labels first). Have the printer load the paper as “Cut sheet / Plain paper”.
• Go to the Adobe Acrobat Print dialog and select the P10000. Click on Properties to open the print driver menu. Make sure Media Type is set to Plain Paper, Color is set to Color, and Print Quality is set to speed (any of the higher-quality modes put too much ink on the paper, causing the labels to wrinkle). Make sure Paper Source is set to Sheet. For Size, select “User Defined” and then click on the “User Defined…” button and select Paper Width of 8.5″ and a Paper Height of 14″. Click OK to exit that menu and OK again to exit the printer menu.
• Back in Acrobat’s print dialog box, make sure you have the “Actual size” button selected. Set the Orientation button to Portrait. The print preview window should show you “Document: 8.5 x 11.0in” and “8.5 x 14.0 Inches” for the paper size. The labels should be up at the top of the preview box, not centered.
• Click “Print”. The printer should begin printing the labels from the top edge and work its way down to the bottom edge of the label sheet, correctly printing all 32 labels. It should not print anything on the plain paper. When the printer asks you to eject the paper, press the OK button and remove the printed page from the printer. Check the print for proper alignment on the labels – there shouldn’t be any bleed from the bottom of one label’s barcode to the color boxes on the next label. Bleeding downward generally indicates that you pushed the labels too far into the printer during the paper loading sequence. Left-to-right misalignment is a user error in not lining up the edge of the labels with the edge marker on the printer roll feed housing, but there is a fair amount of left / right tolerance on the labels. As long as the print isn’t off the edge of a label, you should be OK.

Remove the plain backing paper and tape from the finished labels and let the labels sit somewhere to dry out before stacking them. You can re-use the plain backing paper and its tape a few times, as long as the paper is not crinkled or torn and the tape still sticks down securely. When in doubt, replace the paper and tape – that is a lot cheaper than fishing bits of paper and tape out of the printer.

NOTE: I suppose it is possible for a label to come off and get stuck inside the printer (which could turn into a VERY expensive service call). However, the paper hold-down in this printer is mainly vacuum from the back side – as the printer is designed to print on expensive, delicate papers with fragile coatings, it does make an effort to not touch the front side of the paper.

Updated 1-Feb-2022 to add:

I have used this procedure to print labels on several different P10000 printers and there seems to be a bit of a manufacturing variation in where exactly the paper edge is in relation to the first printable area. On the second P10000 I tried, the labels were consistently a bit low on the page, which resulted in a small portion of the barcode bleeding over onto a subsequent label. Out of a half-dozen or so libraries I’ve used the labels in, one would occasionally misread barcodes. It was apparently “seeing” the tiny bit of the previous label’s barcode. I was unable to adjust the starting point on the sheet with the label generation program because the printer just can’t print that close to the leading edge of the paper. I ‘solved’ the problem by using a black marker to cross out the tiny bit of barcode from the previous label. That caused the barcode reader to read the correct barcode further down the label. I think this is a barcode reader alignment issue on that one particular library. This is just something to be aware of when printing the labels.

Executive summary: Utterly unsuitable for the task at hand. NOT recommended. Decent (probably) hardware let down by mediocre firmware.

I purchased a Netgear Business Essentials WiFi 6 AX3600 Dual Band Wall/Ceiling Mount, PoE Powered, Local Management access point for evaluation. I have several dozen older (non-AX) access points spread across a half dozen or so locations that I am looking to upgrade to AX3600. Based on a quick look at the specs, the Netgear WAX218 seemed as though it could be a potential candidate as a replacement model.

Before I go any further, I should mention that this is not a full review with lots of nice graphs showing throughput vs. number of clients, etc. – I didn’t get that far before giving up in disgust and boxing the unit up to return for a full refund.

In my opinion, a “Business” networking product should be able to be configured and deployed remotely, with on-site workers simply running the network cable (if there wasn’t one already), mounting the access point, and reporting the installation as complete so support can remotely configure and test the unit. The WAX218 fails to meet even this expectation. Initial configuration requires an on-site system with WiFi to connect to the special WAX218XXXXXX-CONFIG-ONLY SSID. The password is printed on the bottom of the unit, along with the actual SSID (replacing the X’s in the above example). After entering basic information such as the new admin password, the SSID and passphrase for the first WiFi network, etc. the WAX218 stores those and reboots – which takes an unusually long time. The next thing I did was update the 1.0.1.0 firmware that shipped with the unit to the latest 2.0.1.0 firmware found on Netgear’s support site. After uploading the firmware, the WAX218 goes catatonic for 10 minutes (!), displaying a countdown timer from 600 seconds until 0 seconds when it will resume communicating. This is accompanied with dire warnings not to interrupt the power, close the web session, etc.

With those preliminaries out of the way, I attempted to continue the configuration by connecting to the WAX218’s IP address over the hardwired Ethernet. After accepting my connection, it performed a redirect to https://routerlogin.com which gave me the message “You may not be connected to your Router’s WiFi network. To access routerlogin.com, your device must be connected to your Router’s WiFi network. Check your current connection and try again.” I then tried connecting via WiFi to the SSID I had defined in the initial configuration and got the same error. It appears the only way you can access the web management interface is via the dedicated WAX218XXXXXX-CONFIG-ONLY SSID. That pretty much rules out any remote management of these devices.

Next, I used the web interface to attempt to configure the timezone. The WAX218 allowed me to specify the offset from GMT, but required me to provide a specific start and end date for Daylight Saving Time (DST). It does not even support rules like “second Sunday in March”, so you need to remember to update the config on every WAX218 annually. What makes this particularly galling is that the WAX218 is running Linux as its underlying operating system. Linux comes with a very complete set of timezone rules for the whole world. Another annoying point is that when I clicked “Save” after making these changes, the WAX218 reported it was rebooting. This was a matter of just a few seconds, but config changes that don’t affect user connections should NOT affect user connections by rebooting the device.

I then enabled the SSH server on the WAX218 and logged into it via SSH to see if there was any sort of a decent menu system. There isn’t – it is a very deep tree of commands, none of which can be abbreviated. Other than checking to make sure the timezone setup was equally brain-dead via the CLI and a few other things (including doing a reset to factory defaults before packing it up for a return and refund) I did not experiment with the CLI in depth. There did not appear to be any supported way to get to a Linux shell prompt to investigate if there was a timezone file there at all. Frankly, Netgear would have been better off using a character cell based browser like Lynx to access the WAX218 web server. At least that way the device would be remotely manageable.

I did examine the T-bar / ceiling mount bracket included with the WAX218. It was a flimsy appearing piece of plastic, not something I’d be comfortable to support a 1.75 pound (yes, the WAX218 is huge and heavy) 8 feet or more over my head.

15 years ago, on June 28th 2006, Alaska’s Senator Ted Stevens said that “… the Internet is … a series of tubes”. While widely ridiculed at the time for failing to understand the nature of the services his Senate committee was in charge of regulating, I figured I’d take a look back on the occasion of the quote’s 15th anniversary. The following text is intended as humor – please don’t tell me that I don’t “get it” either.

While the Internet is indeed “a series of tubes”, very few people have actually seen these tubes and generally think (if they think about them at all) that they are as mythical as unicorns. In reality, the tubes are quite shy and generally burrow deep underground, where they are neither seen nor heard as they happily carry the Internet traffic of billions of users. Despite being solitary by nature, sometimes the tubes are forced by humans into larger groups for ease of herding. Even in captivity, the tubes do reproduce by fission, generally as one or more new, thinner tubes branching off of the parent tube. An example can be seen in the rectangular box near the foreground of the image.

As I mentioned, the tubes are generally solitary and out of sight, so most people don’t know what they look like. So I am pleased to show you this image of a large group of tubes at one of the main places humans have forced them together. This may be the only time people will ever see such a large gathering of tubes.

This image was taken in October, 2019 at 111 Eighth Avenue, NYC (AKA the Googleplex), although this photo was taken in a public corridor. The exact location is the 5th floor near the 9th Avenue end of the building. And yes, I really did take this picture nearly two years ago with the specific intent of making this post today.

This specific post and image are licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license. Contact the author here if you would like to discuss alternative licensing and usage.

If you’ve come here just looking for the Datatracker DT-5 manual, you can download it here if you want to skip the long walk down Memory Lane.

A little history

Long before there was Ethernet, computers were connected to terminals, modems, etc. via a huge number of different possible cabling systems. Even earlier on, mechanical teleprinters were connected to switching systems such as TELEX and TWX. One of the most popular connection types, particularly in later years as 20mA type connections were phased out, was called “RS-232”. The “RS” in there stands for “Recommended Standard” – in other words, “Wouldn’t it be a wonderful world if everything worked this way?” dreaming. The original version of this standard was published in May, 1960 and the most popular version, RS-232-C, was published in August, 1969. RS-232-D was published in 1986, RS-232-E in 1991 and RS-232-F in 1997. To be honest, I’d never heard of the -E or -F revisions until I was doing research for this article.

Wikipedia describes it thusly: “Because the standard did not foresee the requirements of devices such as computers, printers, test instruments, POS terminals, and so on, designers implementing an RS-232 compatible interface on their equipment often interpreted the standard idiosyncratically. The resulting common problems were non-standard pin assignment of circuits on connectors, and incorrect or missing control signals. The lack of adherence to the standards produced a thriving industry of breakout boxes, patch boxes, test equipment, books, and other aids for the connection of disparate equipment. A common deviation from the standard was to drive the signals at a reduced voltage. Some manufacturers therefore built transmitters that supplied +5V and −5V and labeled them as ‘RS-232 compatible’.” In other words, it was pretty much the Wild West out there.

Note that I disagree with one aspect of the Wikipedia article – +/-5V is a perfectly valid signalling level for RS-232-C as the standard specifies minimum voltages of +/-3V to +/-15V. In practice, “classic” RS-232 drivers normally operated from +/-12V as those were common power supply voltages in older computers. The incredibly popular MAX232 series (datasheet PDF) generates RS-232 levels of between +/-5V and +/-8V and I’ve never heard of interoperability issues with those parts.

Being the Wild West, there was a need for hired gunslingers to sort out problematic cabling. I was one of those gunslingers – for example, in the mid 1980’s a major New York bank flew me first class from the US to Venezuela (I have some interesting stories about Venezuela, but I’ll save those for another time) and set me up in a hotel suite for a week, just to install a bisync protocol converter on their IBM 4381 mainframe. And yes, they paid for my flight home, too!

The average breakout box of that era had a few LEDs to indicate signals and jumper wires to manually patch signals from one side of the box to the other. This TENMA 72-440 is a typical breakout box:

Enter the Datatracker

That type of breakout box is fine if you’re solving minor incompatibilities, but if you’re in a foreign country 2000+ miles away from home, you want the most complete breakout box possible. In my case, that was the Datatracker DT from Datatran Corporation. This was a box with 100 (!) LEDs, displaying a high / low signal for each of the 25 pins on each side of the breakout box, as well as being able to supply positive and negative voltages to force any pin on either side to a specific state.

This is the product listing from the Specialized Products Fall ’95 product catalog, showing the final iteration of the Datatracker, the DT-5. Note that $239 in 1995 dollars is over $400 today:

As I mentioned, there have been several iterations of the Datatracker, with the DT-5 being the final one.

The Datatracker DT-3

This is my first Datatracker, purchased around 1984. It has a 104636 Rev. D circuit board, copyright 1982. The distinguishing features of this unit are:

• No external battery access – the case must be opened to access the batteries.
• Dual batteries, one for providing +V (logic 0) and another for providing -V (logic 1) signal levels.
• No snap-on protective cover and no markings on the outside back of the case
• The carrying case is long and skinny and contains 4 clip leads and a M/F ribbon cable. One side of the case interior is rubberized to prevent damage to the case from the test pins.
• Identified on the internal serial number sticker as a model DT-3

In the above picture you can see that both the left and right battery connections have both a + and a – pin, and there are 4 switches for LED Common – left battery to DB-25 pins 1 and/or 7 and right battery to DB-25 pins 1 and/or 7. To me, this is a major advantage of this version of the Datatracker over the subseqent DT-4 and DT-5 variants. It is possible to generate both a high and a low signal at the same time with this version, a feature that was lost in newer models. It is rare to need both + and – levels and not be able to “borrow” one from another RS-232 signal in the cable, but when you need it, you really need it.

The DT-3 has a plain back cover, lacking both battery access and a listing of common RS-232 signals.

The skinny pale yellow wire wrap wire you can see in the above picture is a repair done by me – the batteries are held in place with double-sided foam tape, and I was overly aggressive when prying the lower battery free to replace it in the mid 1980’s. I damaged one of the printed circuit board traces and repaired it with some wire I had on hand, taking care to route it away from the batteries so it wouldn’t be pulled free during a future battery replacement.

This carrying case holds the Datatracker in the main compartment and the clip leads and DB-25 M/F ribbon cable in the smaller compartment.

Despite the faceplate of this Datatracker being labeled “Datatracker” with no version, we see from this sticker that it is indeed a DT-3 model. By 1984 Datatran had apparently produced over 30,000 of these units.

The Datatracker DT-4 (labeled as a DT-5)

This next Datatracker is a bit of an oddity – The front panel says DT-5 but the circuit board is 111-0008-1 Rev. D and is clearly labeled DT-4, copyright 1988. Compared to the DT-3 above, this unit has:

• Single externally-accessible battery compartment.
• Snap-on protective cover for the front of the unit to protect the pins and carrying case.
• The carrying case is shorter and wider.
• The LEDs are slightly larger than the ones on the DT-3. I believe they are also somewhat brighter at a given signal level than the ones on the DT-3.
• A current-limiting resistor was added in the battery circuit to protect against user-configured short circuits.
• In addition to the 4 clip leads, a ribbon cable with a DB-25M connector at one end and both DB-25 M and F connectors at the other end and a second ribbon cable with a DB-25F connector at one end and both DB-25 M and F connectors at the other are provided.
• The lower switch assembly is now 5-position instead of 6-position, reflecting the single battery.
• Some of the more common pin uses are labeled on the faceplate.

You can see the “DT-5” on the lower right of the faceplate. The LED’s aren’t really all lit – it is just the way they responded to the camera flash.

The back panel now has a list of common RS-232 signals and an access panel for replacing the (single) battery.

Despite what the faceplate says, the printed circuit board is clearly labeled “DT-4”. The single resistor directly above where the battery wires connect to the printed circuit board is to provide current limiting if the user accidentally jumpers the battery terminals together.

This shows that this particular unit was manufactured in the eighth week of 1989, and a total of over 64,000 units have been produced to date.

The inside of the snap-on protective cover contains an abbreviated version of the “8-Step Interfacing Method” from the manual. In the manual, it is the entirety of Chapter 4 (20 pages, beginning on page 43 of the PDF linked at the top of this post). Note the typo of “lable” instead of “label” near the bottom.

The Datatracker DT-5

The last Datatracker in this article is labeled DT-5 both on the case and the printed circuit board, which is 111-0049 Rev. A, copyright 1990. In most aspects it is identical to the DT-4. It comes with the same accessories (carrying case, 4 clip leads and two 3-connector DB-25 ribbon cables).

The only externally-visible difference is that the DIP switches are now labeled 1 through 25 instead of two sets of 1 through 10 and the one set of 1 to 5. That’s a nice touch, but hardly a major change.

The back panel has a similar list of common signals as the DT-4. However, references to the Bell 208A modem have been removed and the title is now “EIA-232-D/CCITT V.24 SIGNAL CHART”, showing the exact standard numbers and reflecting the changed standard from RS-232-C to RS-232-D.

The internal construction appears nearly identical to the DT-4, except that the battery leads have been threaded through the right-hand DB-25 pins, presumably to keep the wires in place when the case is assembled.

This unit came in a black leatherette carrying case. While a bit more subdued than the case the DT-4 arrived in, the foam inside it has degraded and covered everything with a thin layer of sticky dark grey dust. You can see some of this inside the battery compartment in the next picture.

This unit was manufactured in the 9th week of 1994, with over 84,000 units produced so far. This is likely the last version produced, as the product is not listed in Specialized Equipment catalogs after the Fall ’95 edition.

The snap-on cover contains the similar text as the cover for the DT-4. The “lable” typo has been fixed and there are other minor changes such as saying the test cables are “Y” cables.

Conclusion

I hope you’ve enjoyed this walk down Memory Lane. While newer interfaces such as USB and Ethernet have replaced RS-232 in the majority of uses, some commonly used devices such as Cisco routers and switches still have an RS-232 console interface. Every now and then I need to break out one of my Datatrackers to solve a thorny interface problem. The DT-3 is still my preferred device, 35+ years on. I definitely got my money’s worth out of it.

Used Datatrackers for sale can be found on eBay from time to time. As of the time this article was written, the least expensive one is listed for $19.98 and the most expensive one is listed for $65.00. Despite these often being sold “as is”, they are pretty indestructible aside from physical damage. In theory it is possible to damage the LEDs with an overvoltage condition, but that is very rare. The assorted accessories are easily replaced for the most part if they are missing, although you should make sure the unit you’re considering buying has the snap-on cover to protect the unit when not in use. It is very uncommon to find one of these units complete with manual, which is why I have scanned the DT-5 manual and uploaded it here. Also, bitsavers.org is also hosting a reduced-resolution copy of the manual here.

To start, I want to acknowledge that Windows 10 is a controversial subject – many people really like it, many people use it because it is what came with their PC, and some people fall into the “over my dead body!” camp. I’m in the latter group, and if you’re a Windows 10 fan this article is probably not for you. With that out of the way…

I really like the look-and-feel of Windows 7. It took a while to grow on me, but I’d happily keep running it if it was still in support and ran on modern hardware. But it isn’t, and it doesn’t. There was (IMHO) collusion between Microsoft and Intel to agree to not release Windows 7 and 8.1 support for newer processors and chipsets (citation here and a more user-friendly / less doublespeak version here). This was purely of academic interest until September 2019 when I purchased a Dell Precision 3630 with an Intel Xeon E-2286G CPU, which had just been released a few months prior. At that point I discovered exactly how incompatible this new hardware was. I was stuck with Windows 10 and had to make the best of it, which started me on a long journey of discovery.

Before I get into the details, let me list a few reasons I dislike the Windows 10 “out of box” experience:

• It keeps changing out from under me, with features being added and removed and applications that stop working because they expect a particular release of Windows 10. A perfect example of this is the massive changes to the user interface in the 20H2 Windows 10 update. It continues the trend of “dumbing down” the visible portions of the system.
• There is no control over what updates it installs or when it installs them – if you’ve been running a complex mathematical calculation or graphics rendering for several weeks Windows 10 doesn’t care – it is going to update and reboot whenever it wants to.
• Even supposedly professional editions of Windows 10 like the “Windows 10 Pro for Workstations” that shipped with my Precision 3630 come with all sorts of unwanted things like Candy Crush, Cortana, Xbox, the Windows Store and more.
• Extremely intrusive telemetry and advertising – is it unreasonable to expect that after paying over $300 for an operating system that I not be spied on and have an advertisement-free environment? Apparently.
• The “Aero Glass” theme and UI effects were removed. I like the visual styling of Aero and don’t understand why it was removed. It certainly wasn’t for performance reasons as newer hardware has faster CPUs and better graphics. If Microsoft wanted to make changes just to be different, they could have retained the Aero theme as a user-selectable option.
• Desktop Gadgets – Gadgets were deprecated during the Windows 7 lifecycle due to unspecified software vulnerabilities. Gadgets do run in a common environment (the sidebar.exe process) and have relatively unrestricted access to the system. However, I find the gadgets useful and have been running the same set of gadgets for over 10 years.
• Metro UI – I’m lumping lots of unwanted stuff into this category – the new-style Start Menu, Windows Store and File Explorer to name a few. Again, change for change’s sake without an option to return to previous behavior.

Upon looking around, I discovered Windows 10 LTSC 2019 (Long Term Servicing Channel). This solves a number of the major issues from the above list, primarily by omitting unwanted [by me] features like games and the Windows Store, being frozen at the 1809 Windows 10 release, eliminating advertising and the more intrusive telemetry features, and allowing somewhat more control over the update process. It will also be supported by Microsoft for bugfixes and security updates until 2029.

Note that enough customers are apparently choosing LTSC that Microsoft announced that the next LTSC release (presumably LTSC 2021) will only be supported for 5 years instead of 10. Microsoft claims that this is because (quoted from the linked article) it provides “… a more consistent customer experience and better planning.” That sounds like classic doublespeak to me. In any event, I won’t have to worry about my LTSC 2019 installs until 2029. Yes, that means Microsoft will be supporting an older LTSC version for 3 years past the newer version’s end-of-life. Go figure…

LTSC sounds perfect, except that Microsoft doesn’t want you to be able to buy it. They go on and on about why you don’t want it. For example, “The Long Term Servicing Channel is not intended for deployment on most or all the PCs in an organization.” here, “The Long-Term Servicing Channel (LTSC) is designed for Windows 10 devices and use cases where the key requirement is that functionality and features don’t change over time.” here and so forth. Basically, if you’re not controlling an MRI machine or running a brewery (more on this below), Microsoft says “No LTSC for you!” and makes it extremely difficult to purchase. This completely ignores the fact that (from the Microsoft quote) “… the key requirement is that functionality and features don’t change over time.” That is exactly why I want to purchase it.

LTSC is not available from most distributors and if you ask Microsoft about it, they will claim (in my experience) that it doesn’t have a part number, can’t be ordered, and so on. That isn’t true – the part number is KW4-00190 and costs a little less than $300 when acquired through legitimate channels like CDW, Connection or Provantage. This is an upgrade license and you need to have an existing Windows 10 license, but even the “free Windows 10 upgrade from Windows 7” appears to count. Of course, now that you know the part number you’ll need to convince the distributor to actually sell it to you. There’s a bunch of paperwork that apparently needs to be manually processed at some points and most distributors don’t seem interested in going through the hassle unless you are purchasing a bunch of licenses (the minimum quantity I’ve found is 5). However, a number of people have succeeded in purchasing a single LTSC license plus 4 of the least expensive Microsoft licenses which are currently NK7-00031 (Microsoft Identity Manager – software assurance – 1 user CAL) for under $10 each. That gets you the minimum of 5 licenses to participate in volume licensing. Of course, if you need more than one LTSC license you can add them to your order instead of the “filler” CAL licenses.

While doing research for this blog post, I found an eBay seller who says that they are an authorized Microsoft Partner and Reseller (although Microsoft doesn’t provide any way to actually verify that), selling “Microsoft Windows 10 Enterprise LTSC 2019 for 20 PCs – Volume MAK” for $339.96 with free shipping. That works out to $17 per PC, which is a fantastic deal if this is a legitimate license. I’ve redacted the seller’s name and location from this screen capture, just in case it isn’t:

Assuming that you’ve managed to successfully purchase the licenses from somewhere and get them enrolled properly at Microsoft, you can then visit the Volume Licensing Service Center to register them to machines and download the installation media. In my case, I downloaded SW_DVD5_WIN_ENT_LTSC_2019_64-bit_English_MLF_X21-96425.ISO. If you are not planning on installing for US English, be sure to download the appropriate installation ISO for your locale. IMPORTANT: Do NOT download the 17763.107.101029-1455.rs5_release_svc_refresh_CLIENT_LTSC_EVAL_x64FRE_en-us.iso from the Microsoft Evaluation Center – that will give you an auto-expiring trial copy of an older base release that can’t be activated with a license key.

Once you have your copy of LTSC 2019 installed and activated, you may want to install the free O&O ShutUp10 utility, available here. This utility lets you disable or enable many Windows “features” that are either not controllable using Windows 10 built-in methods, or which require Group Policies to be applied (which are beyond the scope of this article). ShutUp10 gives you a clean interface which describes each function and an easy on/off slider to change it, as well as batch action buttons for things like “Apply only recommended settings”.

Now that you have a working Windows 10 install, you can take a look at it and decide if you’re happy with the look and feel of Windows 10 or if you would prefer it to look and act more like Windows 7 did. You can stop here if you’re happy with the way the system behaves, or you can read on if you’d like an environment a lot closer to Windows 7. Yes, this really is Windows 10:

I’ll discuss each of the programs I’ve used to get the comfortable look and feel of Windows 7.

The most important program is definitely Classic Shell from here. This free program does most of the “heavy lifting” to restore the Windows 7 style user interface. It is a newer version of Classic Shell, since Classic Shell ceased development in December 17.

Next is Gadgets Revived from here. This free program restores Windows 7 style sidebar gadgets to the desktop. As a bonus, it seems to have fixed the memory leak that affected the official Microsoft sidebar in windows 7 SP1 and later. Most of the gadgets you see on my desktop above are from AddGadgets.com. These free gadgets work with Windows 10 and Gadgets revived. Because Microsoft dropped support for gadgets, these have not been updated in some time, but they still work perfectly. The weather gadget on my desktop is from Win7Gadgets here, since both the Weather Underground and Microsoft weather gadgets have been nonfunctional for years now.

To round things out, I’m using the paid Glass8 Aero Glass add-on from here. While this is paid software, it is very inexpensive, starting at €3 (on the “Donate” link). Once you have paid, you will be emailed a link and password to be able to generate your activation key(s). Installing an activation key will remove the desktop watermark and enable automatic downloading of updates from Microsoft. Each time the Windows kernel is rebuilt by Microsoft, the location offsets (“symbols”) that the add-on uses to hook into the display manager and it needs to fetch the updated symbols from Microsoft. It’s very technical inside, but is very easy to use. You will probably also want to download the GUI for adjusting Aero Glass settings from here.

One of the issues when trying to use an older laptop (such as my Dell Studio 1558) in the modern world is the lack of high-speed WiFi cards – even the Intel Centrino Ultimate N 6300 is limited to a theoretical 450Mbit/sec rate, as it only supports 802.11a/g/n and not any of the newer WiFi standards.

Obviously, newer WiFi cards are available but they are generally in the M.2 form factor and not the Mini PCIe that older notebooks tend to have. These newer cards also come with a newer type of antenna connector (MHF4 instead of UFL). This means that the cards are not usable in older PCs without an adapter. These adapters are readily available from many sellers. Unfortunately, in many notebooks they won’t fit in the space available for the WiFi card. This picture shows how much space the adapter takes up:

My Studio 1558 is one of the notebooks that doesn’t have enough room for that adapter. I thought I was out of luck until I discovered the ALFA Network AWPCIE-AX200U. This is a mini-size card that incorporates the Intel AX200 module, but with a Mini PCIe connector and older-style UFL antenna connectors. That card fits perfectly in my Studio 1558:

As you can see from that picture, even with the correct card things are a tight fit. You can also see that the 3rd antenna cable (grey, with a clear plastic cap on the end) is tucked away in the cable management channel. Unlike the N 6300 card, the AWPCIE-AX200U only needs 2 antennas. The card to the right is a DW5808 mobile broadband (LTE) card. Since the AWPCIE-AX200U is a WiFi 6 card, it operates on the traditional 2.4GHz and 5GHz WiFi bands and works perfectly with the antennas built into my Studio 1558. Other systems should generally be problem-free, but read the section on “things to note” below. WiFi 6E is the version of the spec that adds the 6GHz band. Intel has the AX210 card that adds support for the 6GHz band. But the 6GHz band (which actually extends up to 7.125GHz) is likely outside the range of older notebook antennas. So I don’t know if ALFA thinks it will be worth it to introduce a card based on the Intel AX210.

Here are the AWPCIE-AX200U and the AX200NGW side-by-side to show the difference in form factor:

Once I had the card installed, Windows detected the card correctly, but I decided to install the latest Intel drivers anyway. After that, I connected to my home WiFi network without any problem. Speed is unchanged as I’m running older Cisco Aironet 702i access points that don’t support the newer 802.11ac/ax standards required for higher speeds. At some point in the future I’ll replace them with some other type of device, but at the moment 802.11ax-capable access points (particularly ones with 10GbE LAN ports) are very expensive.

I did fire up MetaGeek’s inSSIDer app to see if there were any high-speed access points within range. The version of inSSIDer I was using (2.1, the last version before compulsory registration and activation) didn’t know about the newer standards, which was no surprise as it is over 8 years old at this point. So I bit the bullet and downloaded version 5.4 and registered it. As you can see, it detected an 802.11ax network at a speed of 2.4Gbit/sec:

The AX200 module (and thus the AWPCIE-AX200U) supports Bluetooth 5.1. The Intel documentation does not specify whether the Bluetooth antenna is internal to the module or if one or both of the external antenna connectors are used. Regardless, the Bluetooth radio sends and receives a good strong signal – it was able to “see” the televisions of several of my neighbors, a few houses away and had no trouble pairing with my test device (a JBL FLIP 5 speaker). I had to use Windows Device Manager to disable the Bluetooth 365 module that was already in the computer so there would only be one active Bluetooth radio. The BT365 module is in a location that requires near-total disassembly to get to, so I ran this way for a week. Eventually I needed to disassemble the laptop for an unrelated reason and took the opportunity to remove the now-unneeded BT365 module. Note that the ALFA website and the AWPCIE-AX200U datasheet both say that the module supports Bluetooth 5.0, not 5.1. This appears to be a typo, as the Intel datasheet for the AX200 clearly states Bluetooth 5.1, which is confirmed by this Microsoft support article. Performing the steps in that article shows that the AWPCIE-AX200U is using LMP (Link Management Protocol) 10, which is Bluetooth 5.1:

There are a few things to note before you run out and buy one of these cards for yourself:

First, some computer manufacturers (HP for example) have a “whitelist” of supported WiFi cards, which are generally limited to a bunch of overpriced, near-obsolete cards. Putting any unapproved card into one of those computers will give you an unfriendly message such as “Unsupported wireless network device detected. System halted. Remove device and restart.” You may have some luck searching for things like “<your computer brand and model> BIOS mod” or “<your computer brand and model> whitelist hack”. Of course, flashing an unapproved BIOS is something you do at your own risk. When considering future purchases, you may want to “vote with your wallet” and avoid purchasing any systems that try to impose this sort of needless lock-in.

Second, there is good deal of variation in what pins are connected on notebook Mini PCIe connectors. The AWPCIE-AX200U does its best to avoid conflicts by setting all pins not essential for operation to “Not Connected”. It is possible that some incompatibilities could still arise. If you post a comment here I’ll try to help you. Another issue is that the Mini PCIe USB pins may not be connected in your computer. In that case, the card will still work fine as a WiFi adapter, but the Bluetooth portion will not be detected.

If you’re interested in purchasing one of these cards, I highly recommend Rokland Technologies. They’re an authorized Alfa distributor and ship quite rapidly. Be sure to order the AWPCIE-AX200U card and not the AWPCIE-AX200 (no “U” at the end). The non-U card has the newer MHF4 antenna connectors and you will need adapters (and a place to put them) if you want to re-use your existing antennas.

Updated 23-Apr-2021 to include a side-by-side comparison of the ALFA and Intel cards and their antenna connectors.

I have previously mentioned in passing, for example here and here, that the Dell Studio 1558 notebook computers are prone to overheating. I figured it was time to go into this in a bit more detail.

As I mentioned in the first post linked above, the configurations of some Studio 1558 computers (particularly ones with discrete graphics and quad-core CPUs) tended to overheat even when fresh from the factory. Dell addressed this with the A04/A05 BIOS which resolved the problem for most customers. However, people are still reporting the issue and selling used 1558 systems with comments like “The unit boots to BIOS, but a few minutes later it shuts off. No further testing has been done.” This is a different problem, caused by a combination of accumulation of lint in the fan and degradation of the heatsink compound over 10 years or so.

The fix is easy, once you get to the fan. Getting there is the hard part. Refer to the service manual (or DCSE if you have it) for detailed disassembly procedures. If you aren’t comfortable disassembling the system to that extent, most repair shops that fix laptop computers should be able to do this for you. I also offer this service, although I probably won’t be cost-effective for you unless you are having other services performed at the same time.

This picture shows the accumulation of lint that has built up over 10 years of use. The air exhaust through the cooling fins is completely blocked. The heatsink assembly was removed from the motherboard and the lint was carefully vacuumed out. Then the fan was removed and the whole heatsink assembly was put through an ultrasonic cleaner to remove any remaining lint. A new replacement fan was then installed.

Here you can see the hardened heatsink compound. It had hardened to the extent that there was almost no thermal conduction between the top of the CPU and the heatsink assembly. After carefully chipping away the larger chunks of material, the assembly was left to soak in solvent to remove the remaining compound. At the same time, the old thermal pads were removed from the part of the heatsink assembly that cools the HD5470 graphics chip and the 2 lower video RAM chips. The surfaces of the graphics and RAM chips were cleaned to remove the oils that had separated out of the old thermal pads.

The existing CPU chip (an i7-720QM) also had a large amount of hardened thermal compound on it. As this repair order also involved upgrading the CPU to an i7-740QM, no cleaning was done on this chip.

After installing the new CPU chip, all surfaces were cleaned with 2-part Arctic Silver cleaner / prep fluid. New thermal pads were cut from Arctic Silver thermal pad material (1mm for the 2 video RAM chips, .5mm for the graphics chip) and Arctic Silver heatsink compound was used on the CPU. After reassembly, the processor idle temperature is in the 49-51° C range and with intense CPU usage, reaches 76° C. The system has run under heavy load for more than 24 hours with no faults.