All posts by Ed Tittel

Full-time freelance writer, researcher and occasional expert witness, I specialize in Windows operating systems, information security, markup languages, and Web development tools and environments. I blog for numerous Websites, still write (or revise) the occasional book, and write lots of articles, white papers, tech briefs, and so forth.

Safely Eject Gets Weird When Swapping Drives

Go figure: I had to grab some or all of 7 different SATA drives to complete file transfers from the old production PC to the new. Mostly, this meant waiting for Teracopy (a bulk transfer utility I highly recommend). It aced the job of moving files from my old drives to the new 12TB Ironwolf. Along the way I learned something interesting — namely, that Safely Eject gets weird when swapping drives. At least it gets weird when using dual-drive caddies. Let me explain…

How Safely Eject Gets Weird When Swapping Drives

Safely Eject appears in the System Tray (aka Notification Area) of the taskbar as a teeny-tiny USB Flash Drive icon, as you can see in the lead-in graphic. Turns out it’s a quirk of the chipsets used to bridge multiple SATA drives through a single USB connection that coming back after an eject doesn’t always work.

Here’s what happened on my 5800X Flo6 production rig and the ThinkPad P16 Gen 1 Mobile Workstation. No drives in the caddy showed up after ejection, then disconnect/reconnect of the USB cable. As Copilot says “After eject, the bridge may not reinitialize that port properly.” A reboot typically fixes such problems, but that got vexing given that I had to work my way through 7 drives in all.

Suggested Mitigations

Courtesy of Copilot, I worked my way through a couple of possible mitigations. Turns out the ASMedia driver wasn’t in use, so it wasn’t a possible culprit, either.

The right technique involved a multi-step combination of commands and physical actions:

Step 1: Unmount the drive to be removed from the caddy using the mountvol <drive-letter> /p command (e.g. mountvol e: /p)

Step 2: Turn off the power on the caddy. For one caddy that meant using the power switch, for the other it meant unplugging the power input from its brick.

Step 3: Wait 5-10 seconds for the device to reset completely.

Step 4: Power the caddy back on, possibly with one or two new drives inserted, after removing one or two old ones. Wait for those drives to get initialized, then show up in File Explorer. Proceed.

This worked properly on both the P16 and the Flo6 PCs. Safely eject is fine for single-drive devices (of which I have more than a dozen). But I now know that using the mountvol command, plus cycling the power around drive swaps, is the right way to keep my dual-drive devices working as they should.

Here in Windows-World, the path to proper device function has its occasional twists and turns. For my dual-drive caddies, this particular turn is worth making…

Facebooklinkedin
Facebooklinkedin

SFF Copilot+ PCs Remain Scarce

Last year, I really got my hopes up when ARM announced a developer’s kit version of a small form-factor (SFF) PC with Snapdragon X. Alas, it never materialized. I’ve also read about plans from numerous vendors to offer mini but mighty PCs of this type, including Geekom and ASUS. But right now, SFF Copilot+ PCs remain scarce. In fact, I can find only two for sale presently: the Lenovo ThinkCentre Neo50q Tiny (Snapdragon) and the MSI Cubi NUC AI+ 2MG (Intel).

Why Do SFF Copilot+ PCs Remain Scarce?

I speculated about this when ARM announced, delayed and then canceled its own SFF SnapdragonX based model. Looking at the enclosures for the 2 aforementioned models, I’m still leaning that way. Mostly, in fact, it has to do with support for only 1 or 2 USB-C ports (typically 40Gbps/Thunderbolt 4/USB4).

Thus, there’s a nearly inescapable add-on expense when buying such a unit — namely, a TB4 or TB5 dock, most of which cost from US$350 to over $500. That’s a big bump in cost, cubic volume, and complexity for PCs designed to be affordable and, above all, compact.

I’m Interested, But Is the Market?

Because of the need for more ports, power, connections and displays, I have to believe the general marketplace finds current Copilot+ SFF PCs unappealing. It’s one thing to get useful capability in a compact and reasonably powerful package. It’s another thing entirely to have to turn around and spend from 44 to 63 percent of the purchase cost on a TB4/5 dock to get all the ports and connectivity modern office workers need.

If these OEMs build such SFFs, will buyers come? Initial excitement and plans said “Yes.” Subsequent actual product offerings, options, and limited choices say “Maybe” at best. Too bad: I like the category and what it brings to the desktop. I may be in a (small) minority, though…

Note Added 9/24: Oops! Wrong One..

As you can see in the lead-in graphic, I’d originally landed on the Lenovo ThinkCentre M75q as a Copilot+ SFF PC. Closer examination researching this topic yesterday showed me, it’s not. It lacks sufficient NPU oomph to qualify as such. But that’s when I discovered that indeed Lenovo DOES have an SFF Copilot+ PC — namely the ThinkCentre Neo 50 (Snapdragon) Tiny now mentioned at the head of this story. I guess I got my minis mixed somehow. Glad to fix it, though…

Facebooklinkedin
Facebooklinkedin

Jabra 75 Headset Driver Disappearance

One of my favorite things about Windows made the scene in 1995, with the debut of Windows 95. It’s called Plug and Play (PnP) and it lets the OS detect and configure hardware devices automatically. It has made using Windows peripherals much, much easier than it was in the old days before PnP came along. This strongly contributed to a sense of shock and abandonment the other day, when a Jabra 75 headset driver disappearance delayed an attorney phone call for troubleshooting. Sigh: let me explain…

What Caused Jabra 75 Headset Driver Disappearance?

USB devices, particularly audio ones, involve a whole series of nested drivers. This runs the gamut from the audio device itself, to the USB hierarchy, to the Intel Smart Sound driver (which works between the device and the USB port to manage specific audio formats and functions). Some recent update to Windows 11 on my Lenovo ThinkPad P16 Mobile Workstation broke this complex chain somewhere in the middle.

How did I know? PnP quit working. Normally, I just plug the USB cable into a USB-A port on the P16, and audio switches over from its built-in speakers and microphone to the Jabra 75. I use it most especially for video calls, where clients need to see me as well as hear me (e.g. speaking engagements, webinars, depositions, and so on). When plugging-in resulted in “no play,” I knew I had a problem. Fortunately, it was easy to fix.

Plug and No Play Means Reinstall Driver

I visited the Jabra website, where I had to figure out what I needed was a new copy of the Jabra Direct software. With that downloaded and installed, and a fresh reboot to make sure everything registered properly, play immediately followed plug-in when I inserted the USB connector from the headset base to the P16. Problem solved!

But it took me a few precious moments to figure this all out, and then to fix it. In the meantime, I switched over to my production desktop where the driver was already installed and working properly (it’s the new build, so it got a new driver after the offening update, apparently). The client couldn’t see me (no camera on that rig) but we did complete the call. They hired me for an expert engagement, too, so I guess it didn’t go too badly.

Here in Windows-World, one must always be ready to adapt and overcome. So that’s what I did.

 

Facebooklinkedin
Facebooklinkedin

Closing Thoughts ON Tecra A60-M2

Well, it’s been an interesting and pleasantly surprising ride. On August 18 — just about a month ago — Dynabook sent me a snazzy new Tecra A60M2, one of its Intel-based Copilot+ PCs. Initially, I was a bit put off by the unit’s mid-range CPU (Core Ultra 5 225U) and its modest RAM endowment (16 GB). If buying for myself, I usually go for i7 or i9 class CPUs, and 32 or 64 GB of RAM. But as my 30-dal eval period is ending, I find myself unexpectedly and favoraby impressed with this PC. Thus, here are my closing thoughs on Tecra A60-M2, as I get ready to box it up and ship it back.

Weighing Closing Thoughts on Tecra A60M2: Pros & Cons

All in all, the A60-M2 handled everything I threw at it, and was more than able to cope with my daily regimen of email, web surf and search, heavy Copilot use, and intense workouts for Word, PowerPoint and Excel. I was expecting some occasional fits and delays, but encountered nothing but smooth, reliable performance. This is an excellent business machine for productivity/office workers, if under-powered for developers, data scientists or AI model makers or testers.

PROs

The display is clear and legible, if a bit washed out in direct sun. The IR camera and fingerprint scanner are the best I’ve ever used: incredibly fast, and only occasionally required repositioning of face or finger. Aside from a handful of Dynabook utilities, mostly useful not nugatory, there’s zero bloatware on this laptop.

Despite its modest RAM config (a single 16GB DDR5-5600 SODIMM, in fact, with room for one more) it is user-ugradeable to 64 GB. That’s a big plus against many other Copilot+ PCs I’ve seen lately, all with soldered RAM. It’s Wi-Fi 7 wireless and TB4 USB-C ports (x2) do their jobs nicely. That said, USB-C attached NVMe is middle-of-the-road, too (somewhat above UASP, but not 2x or better).

CONs

Three of the four case surfaces excluding the top deck (made of magesium alloy) are plastic composites. The unit is strong and relatively rigid — it meets MIL-STD-810H durability requirements– but the top deck flexes quite a bit, and the upper surface is a fingerprint magnet. The 4-cell Li-ion polymer battery is rated at 60W, but the unit is power-hungry enough to give it a typical 6-8 hour life for normal workday tasks. I’m spoiled by double-digit SnapDragon X values of 11 or more hours, I admit. It also charges back up to 40-50% in 30-40 minutes using the co-supplied 65W USB-C charger. That goes a bit faster from a TB4/TB5 dock.

I have some others cons to report, but they’re more nitpicky than serious:

  • Touchpad is positioned left-of-center on the lower deck, and takes some getting used to. Works fine (and accurately), tho…
  • The display is Full HD (1920×1080 only) with no 2K or 4K upgrade options available.
  • Keyboard layout is idiosyncratic enough that a long-time Lenovo ThinkPad user like myself had to adjust touch typing techniques to get fully up to speed.
  • Device weighs in at 4.1 lbs (1.86kg) which makes it noticeably bigger and heavier than other Copilot+ models I’ve used, especially those of the SnapDragon X persuasion.

One more thing, this Dynabook model’s MSRP is $1,249 as configured (the 512GB PCIe Gen4 NVMe is the only major item I didn’t mention earlier). This runs from $250-639 more expensive than Lenovo’s ThinkBook and IdeaPad Snapdragon X models. Buyers must want the bigger screen and higher durability to make that extra cost worthwhile. But it’s a durable, hard-working, surprisingly capable beast, and sure to provide its users with years of computing satisfaction.

Facebooklinkedin
Facebooklinkedin

So Long i7 Skylake

It’s been a long, long ride. It must’ve been right around 8 years ago that I first brought up my Asrock Z170-based, i7 Skylake PC. Today, I made the decision to lay it to its final rest. I’ll decommission it in hours. Indeed, as I bid “So long, i7 Skylake,” I’m saving the drives and ordering a 12GB SATA drive to which I’ll back everything up before wiping and sending them onto Goodwill (or elsewhere). It’s safe to say this has retirement been a long time coming.

In yesterday’s blog post, I had planned to bring out the heavy guns next. That is, I unplugged all the drives from the i7 Skylake, and inserted a new, blank NVMe from which to boot. At first, it looked  like it wasn’t going to work. I even got past providing a license key. But then, the newer NVMe wouldn’t run even when I provided the right Intel RST driver version. When I switched back to the old drive, the PC would no longer boot, not even to the UEFI screen. Something serious quit working along the way, and I no longer want to spend any more time on this. It’s over…

What Does So Long i7 Skylake Mean?

Because I’m moving the unit out of my office entirely, it means I can set up a couple of other machines.  I’d set them aside during the “parallel runnings” phase as I was using the Ryzen 7 5800X build as my primary production desktop, but running the i7 Skylake just in case I needed something from that PC. Now, I’m going to content myself with snapshots of its drives for a while, and put the old beast to rest.

As I was looking inside the case today, I see it was reasonably equipped for a machine of its day:

  • 4 physical cores/8 threads (vintage 2015)
  • 32 GB DDR4 RAM
  • NVIDIA 3070 Ti GPU (originally a 1080)
  • TONS of storage (11 drives totaling 16TB)

But it’s been getting flakier and flakier over the past 3-4 months. So yesterday, when I couldn’t get the machine to handle a clean install of Windows 10 Pro, I decided “enough is enough.” I have plenty more PCs to work with, so I don’t need to nurse this one along anymore. If somebody in central Texas wants to come by and pick it up, they can have it, gratis. (I’ll be taking the 3070 Ti, but the rest of the build, including 6 drives with about 8 TB of total capacity — wiped, of course — will come along for the ride. Contact me if you’re interested.) Otherwise, it too is headed for a Goodwill drop-off.

I’m sad to see it go, but I’m definitely glad the X5800 build is working out quite well. Hopefully, it will see me into the next decade. But here in Windows-World, the strangest things can (and often do) happen. Stay tuned: I’ll keep you posted.

 

Facebooklinkedin
Facebooklinkedin

Win10 Boot Follies Galore

Here’s my situation. I’m still running the old i7Skylake with its 2015 vintage Asrock Z170 motherboard. That machine is running Windows 10 Enterprise. Thus, it’s not eligible for the ESU (Extended Security Updates) offer from MS to keep that machine alive for another year. “No problem,” I thought, “I’ll just clean install Windows 10 Pro, and take up the offer that way…” Instead, I’m dealing with Win10 boot follies galore, unable to boot to USB media to replace the current Windows image. It’s been heartbreaking…

Describing Win10 Boot Follies Galore

Copilot agrees that something is hinky with the UEFI on the Z170, and it’s preventing the PC from booting off a USB flash drive. So far, I’ve:

  • Turned off Fast Boot in UEFI, and discovered that toggling CSM (the compatibility support module that supports both MBR and GPT partitioning schemes) kills UEFI completely for UEFI version P7.60. Turns out that’s a known gotcha.
  • Built rescue and install disks on 8GB media to avoid FAT32 issues (using the usually reliable Media Creation Tool, and the still more dependable Macrium Rescue Media Builder)
  • Run those UFDs from USB 2.0 ports, on the off-chance that USB 3.x isn’t working for boot

So far, nothing has worked to install a different Windows 10 version on this PC. But I have a plan…

Bring Out the Heavy Guns

When all else fails while installing Windows, I’ve observed that disconnecting all non-boot drives, and replacing the boot media with a completely blank drive will sometimes work. I’ve got a 1TB Crucial T705 NVMe that I’ll prep in that way, and give it a try. IMO, it has a good chance of getting me over this hump.

I won’t have time to do this until the weekend. Stay tuned: I’ll follow up on Monday with a report on that experience. I’ve been bit on the hindquarters many times in Windows-World, but this bite kind of stings…

Facebooklinkedin
Facebooklinkedin

Windows 11 Does Multi-Monitors Right

I’ve been living the life of a native Windows 11 user on my production desktop — the newer, Ryzen 7 X5800 based PC — for a couple of months now. One thing I’ve noticed is that Windows 11 handles multiple monitors with great alacrity and aplomb. IMO, Windows 11 does multi-monitors right, in several important ways. Let me elaborate…

Why Say: Windows 11 Does Multi-Monitors Right

To begin with, I didn’t need to engage in Display settings shenanigans on my dual-monitor desktop setup. At all. Indeed, Windows 11 correctly detected and arranged both of my displays during its initial bootup. I didn’t have to do anything to get them working in extended mode (to use left- and right-hand devices as a single logical desktop with proper mouse tracking all the way across). In Windows 10, this requires manual intervention and setup.

It gets better. Updating NVIDIA graphics drivers is likewise much better behaved. On Windows 10, post-install, all windows moved onto the primary display. Remote desktop sessions appeared in full-screen windows that needed to be maximized to show a top-center control bar. On Windows 11, windows reappear where they resided prior to the update, and RDP sessions remain as previously set.

GPU driver stability also seems to be improved. On Windows 10, I often had to deal with “display blink” — a phenomenon that caused either my left- or right-hand display to blink on and off every few seconds. So far, I’ve been through three NVIDIA update cycles on Windows 11 with nary a sign of such misbehavior.

All’s Well…

It may still be too early to proclaim success or improvement, but I’m doing those things anyway. So far, display handling in Windows 11 is better behaved and more predictable than it was on Windows 10. Should things change, I’ll be sure to report on issues I encounter. But so far, it’s been a refreshing breeze to work with multiple monitors on Windows 11. Fingers crossed that things keep going in the same way for the foreseeable future!

Facebooklinkedin
Facebooklinkedin

Pondering NVMe Performance Premiums

I admit it: I’m something of an SSD nut. I’ve been fascinated with these solid-state alternatives to spinning media since they first appeared in the mSATA days. Indeed, I acquired my first SSD in 2008, in the form of a SATA based Intel X25-M. I jumped on the mSATA bandwagon early in 2011, about two months after Intel dropped its M.2 SSD 310. Over the past decade and more, I’ve spent far too much time wondering if the latest performance boost is worth the typical doubling in cost over previous generations that vendors exact for riding on the bleeding edge. Here’s what I think…

When Pondering Performance Premiums, Don’t Forget Price

I bought a very nicely priced Samsung 990 EVO Plus 4TB Gen4 NVMe to include in my current desktop build earlier this summer. It set me back just over US$200. I was just reading about a new offering from Lexar — the NM1090 Pro 4TB — that costs “just $360” in the words of that story’s title. The difference is what got me thinking about today’s musings.

The thing about leading edge hardware is the whole device chain. That means it’s not just the drive itself in this case, but the slot into which it plugs, the bus upon which it rides, and the motherboard that houses all the pieces and their connections. Here at Chez Tittel, I have perhaps one or two systems — both laptops — that could use the Gen5 high-speed capabilities the Lexar drive can deliver.  I own neither of them (one is on loan from Lenovo, the other from Dynabook).

Getting There from Here Is More Than a Ride

Taking full advantage of the leading edge means a leading edge rig in which to house a leading edge drive. Right now, building such a system would cost me over US$2,000 (maybe even over $3K if I want to max out on the GPU side, too).

My strategy is to hang back one generation when building, because I can get reasonable performance for half the cost of buying into whatever the state of the art might be at purchase time. Of course, that means my build will be obsolete a bit sooner, but gosh: saving 50% lets me buy in more often at a lower overall cost. I happen to think that’s the right way to go.

Here in Windows-World, buyers can do as they like. I’d rather stretch my dollars a bit further, and use them more sparingly. But then, I’m not a gamer, nor do I run many applications where my productivity is diminished because I can’t operate at max bandwidth. What’s your take?

Facebooklinkedin
Facebooklinkedin

Careful With App Defaults

I’ve always got PCs coming and going here at Chez Tittel. That means laptops, mostly, with an occasional All-in-One or SFF desktop to keep things interesting. Further, that means setting up Windows to work the way I want. Too bad, then, that Edge isn’t inclined to make way for other Web-focused defaults (e.g. Chrome). Even when I pick Chrome as the default browser, Edge keeps popping up for other stuff. The lead-in graphic shows why I’ve learned to be careful with App defaults. Indeed, it shows Edge remains the default for .mhtml files (MIME encapsulation of aggregate HTML documents) even after Chrome gets the default pick.

Why Be Careful With App Defaults?

Elsewise, items opened in other MS apps will open in Edge rather than Chrome.  In this case, Outlook handles a lot of .mhtml files for links and Web items embedded in email messages. Those still go to Edge. The right way to do this is to click Settings > Apps > Default Apps > Google Chrome.  On that page, make sure Chrome gets associated to all file types you want it to handle.

In my case, I just made sure everything on the foregoing Settings page that used to show Edge as the default now shows Chrome. A hopefully obvious corollary to this exercise is that setting Chrome as the default browser in Windows 11, is not sufficient to make it take over for all Web-related items. Hopefully, that’s now fixed!

Just another happy day in Windows-World, living on the edge (horrible pun not just intended, but enjoyed).

Facebooklinkedin
Facebooklinkedin

Interminable Adobe Acrobat Updates

I’m feeling a little less happy about switching from Nitro Pro to Adobe Acrobat on my new 5800X production build PC. The former costs about US$110 for each new version, the latter about US$20 a month. Right now, it feels like I’m paying more and getting less. But I confess: it’s the interminable Adobe Acrobat updates that bother me most. Let me explain…

Why Say: Interminable Adobe Acrobat Updates?

I just went through my second update cycle on this program. I noticed it took a while for the last update. Today, I was actually working on something else, and not really watching closely. But I found myself thinking “Hasn’t it been about an hour now?” when the UAC prompt to run the Acrobat Installer finally hit the 5800X screen this afternoon.

I’m guessing, but it took as long as 90 minutes for the whole update cycle to complete on this PC. To make matters worse, Acrobat is subject to a “known issue” in the winget packages repo because its update process doesn’t signal completion as it should. Indeed, Copilot says:

Winget starts the next package update before Acrobat finishes, triggering a 1618 error (“another installation is already in progress”). This isn’t just a timing hiccup—it’s a flaw in how Acrobat’s installer communicates with Windows Installer, and it breaks the expected transactional flow of winget upgrade --all.

Is It Really “Less for More?”

Copilot also mentions that “[t]he in-app update is notoriously sluggish,” which IMO is understating things a bit. Painfully slow, is more how I’d put it. It explains that it’s old fashioned installer Enhanced Security settings (Sandboxing, AppContainer and Protected View) slow things down, but that its “monolithic update model” and “telemetry and plugin checks” also contribute to seemingly glacial update processing (my exaggeration, not Copilot’s).

That said, Acrobat has handled my PDFs with aplomb, and lets me edit, mark up and sign such documents with ease. Had I know it was such a crawl to update, I might have rethought my switchover. But I’m in for a year’s worth now, having signed up and paid the annual fee. I’ll be sure to keep this in mind when the next cycle comes around.

Here in Windows-World, as I’m fond of reciting: “It’s always something!” Today, it’s painfully slow Acrobat updates. What will it be tomorrow?

Facebooklinkedin
Facebooklinkedin