Review: Philips 25W Vintage T30 Filament & 4.3W Classic Design LED Globes

In an earlier review of the IKEA Lunnom Filament LED Globe, I noted the increase in popularity of vintage mood lighting and the number of incandescent “carbon-filament lookalike” globes that had flooded the market just as incandescent bulbs were being phased out. LED globes are the way of the future, so the trend seemed rather regrettable.

Despite this, I still wanted to get my hands on a carbon-filament lookalike globe while I still could. Luckily for me, strolling into a Bunnings store, it seemed that some of the older stock was being cleared out for AU$2-3 a globe which was so cheap that I decided to pick up both a filament type globe and a newer LED-filament type globe for comparison.

Philips 25W Vintage T30 Filament Globe

The first globe to be tested is this Philips Vintage T30 25W B22 globe, rated for 220-240V. The globe claims a lifetime averaging 1500 hours – a little more than the average incandescent that would normally be rated for 1000 hours. But nowhere near the rating for LED globes.

Its colour temperature is claimed to be 2700k, which is just a normal warm white, and also claims to be able to withstand 8000 switching cycles. It comes up to full brightness immediately and is suitable for dimming (owing to the incandescent technology).

The globe is Made in China, but distributed in Australia and New Zealand.

With a claimed output of 60 lumens, this globe achieves just 2.4 lumens per watt.

The T30 style globe resembles a long “sausage”. The filament runs in a U shape down to the tip and back. The base is a nice golden brassy colour, and the envelope seems to be “metallised” internally to make it appear even warmer.

Because it works like a good-old incandescent lamp, it features a traditional BC base that is identical to what you would find on regular incandescent globes, rather than the more plastic variation on modern LED retrofit globes.

Looking up-close, a very thin coiled filament is supported by a number of support wires along the length.

Lighting it up, it produces a quite warm and full glow, which is not particularly bright but good for decorative effects. If you’re used to LED globes, the light actually seems “fuller” – this is probably due to the continuous nature of the spectrum from incandescent globes. The long filament does vibrate somewhat sympathetically with the mains frequency resulting in some “flicker” when viewed at just the right angle to the glass.

As a simple bulb, it works across a wide range of voltages, although the output very much depends on the input voltage. It looks like the globe is about 22W at 220V and 25W at 240V, which means the labelling is practically spot-on (as Australia was a 240V country). The one good thing is that the power factor is unity – it’s basically a resistor after all.

Philips 4.3W Classic Design LED Globe

This was probably an earlier LED replacement globe which has now been deleted from the catalog. With a rating of 4.3W, it might sound weak, but it delivers a whopping 470 lumens – 7.8x more light than the 25W globe above and achieving a very respectable 109 lumens per watt efficacy. The globe claims to have the same 2700k colour temperature, but is not dimmable and usable within 170V-240V. It claims a lifetime of up to 15,000 hours – ten times longer than the filament globe above

The main attraction of this particular globe is that the whole globe is in a traditional GLS A-series bulb envelope, so it will fit into every regular fixture. The glass outer envelope serves the function of insulation, but is not evacuated as it doesn’t need to be. It seems that all four LED filaments are paralleled in this design.

The BC base is a plastic-insulator type, with the driver integrated into the base.

Even the outer glass envelope is marked just like a regular filament globe as well, which is pretty cool. It is Made in China as well, with an Australian RCM.

The globe starts off with close to 4.6W consumption, quickly warming up and settling around 4.4W – so very close to the claimed 4.3W rating.

The globe is found to be workable and regulated down to about 120V. Below there, the globe flashes and cycles between about 60V and 80V, and runs at reduced brightness at about 80-120V. Power factor is a rather “ordinary” 0.55-0.58 within the main 220-240V working range.

Oddly enough, the colour of the light coming out of the globe feels somewhat cooler despite having the same claimed CCT. It’s also a bit bright to look at directly, and the look of LED filaments are not quite as delicate as the thin filaments. I can still see why there is an allure to the regular filament globes.


The Philips Vintage filament globe is representative of the “decorative” filament globe. It achieves just 2.4lm/W, producing 60 lumens which is pretty dim. It has disadvantages of being fragile due to the filament and a limited lifetime. However, it does have an allure of having the delicate filament visible, a warmer full spectrum light and a brightness that does make it rather pleasant to look at.

The Philips Classic Design LED globe is an LED retrofit that tries to replace the filament globe. With a very efficient 109lm/W, it is 45.5x more efficient and can even provide general illumination with a much longer lifetime. The light quality, however, does feel somewhat cooler and harsher and lacks the delicate look with the thick LED filaments. The brightness makes it somewhat less pleasant to stare at.

I guess it is a worthwhile trade-off to save money and the environment to go LED, but filaments do still have the upper hand of looking cooler and more “retro”.

Bonus: Older Filament Bulbs

Now that most incandescent globes are outlawed and the only sort that are available are halogen retrofits, I was pleased to find my old collection of incandescent globes which consists of a range of pearl/frosted, clear, round, fancy round, candle-shaped, reflector spot, etc types.

I got a few of the clear globes to take a few macro shots …

An unbranded 100W globe – short coiled thick filament with two intermediate supports.

Looking closer at the filament – it’s a coil of coils. Below is another globe – a cheap variety-shop 60W globe with five insulated supports!

It seems the filament on this globe is made of a single strand that is coiled.

Finally, going down to a 25W globe, we can see that the filament has stretched from a number of years of use, with more stretching around the supports – this filament is now more vulnerable to failure by shock and by developing “thin” spots. But as expected, lower wattages mean thinner filaments.

Posted in Lighting | Tagged , , , | Leave a comment

Quick Review: Strontium Nitro 16Gb Class 10 UHS-I MicroSDHC Card

If you like to tinker with single-board computers, then you probably have a bunch of Raspberry Pi (or equivalent) boards all needing a microSD card to boot from. Often, for simple projects, the card doesn’t have to be anything too special – as long as it’s reasonably fast and large enough, it will do. As I haven’t purchased any new microSD style cards in a while, I was running low with a few of my older boards sitting around with no card to use.

Having resisted the urge to pick up a few cards, I saw an online bargain posting for a three-pack of Strontium Nitro 16Gb microSDHC cards with adapter for AU$10.80 including shipping which I couldn’t turn down. At that price (assuming the adapter cost nothing), I was paying AU$3.60 for each 16Gb card, or about 22.5c/Gb. That’s not the cheapest flash memory has gone for – recently we’ve seen SSDs at the 1Tb mark drop to 19.9c/Gb, but the overhead of getting several cards also increases costs slightly.

The Product

The product comes packaged in a retail hanging card, in the standard Strontium yellow-red-white colour scheme. The package claims 433X speed, up to 65MB/s (read) speeds with lower write speeds. The items are packed within a clear plastic bubble, which can be seen from the outside. The cards are labelled “horizontally” in a striped-colour design which bears some similarity to Sandisk and Samsung offerings. The packet has the Australian RCM mark on the rear, indicating locally approved stock, with a product number of SRN16GTFU13C1A and claims to be Made in Taiwan. The packet is perforated for easy opening, although I didn’t find this to be effective.

The plastic tray can be removed from the cardboard.

The adapter is very plain, but functional, only having the brand printing on the front and no specifics on the rear.

The cards all contain the same etching on the rear, which claims to be Made in Taiwan.

The card is pre-formatted with a volume label of STRONTIUM with 15,632,171,008 bytes total capacity (14.5GiB).

Further analysis shows that the card is partitioned with an MBR, containing a partition that starts at sector 8192 (4kiB) into the card.

The relevant card identification is as follows:

CID: 8454464d53000000202c5d002a01181
CSD: 400e00325b59000074877f800a400091

Performance Testing

Unfortunately, due to the change in my system, my benchmarking system is no longer consistent with the old SD card database. Instead, tests are being performed using the Transcend RDF8 USB 3.0 reader connected to my new workstation. Testing was performed with the card in its factory format.

HDTune Pro

On a fresh, unwritten card, the read speeds averaged 91.2MB/s. However, once fully written, the true sequential read performance is revealed to be about 68.5MB/s which is slightly above the package claim of 65MB/s.

Writing to the card achieved an average rate of 10.8MB/s, barely reaching the Class 10 requirement.

Random access I/O figures showed relatively consistent read service times, but write service times did occasionally encounter longer delays of up to 351ms.


CrystalDiskMark seemed somewhat more complimentary about the card’s performance, with sequential reads scoring 80.42MB/s and writes at 11.90MB/s. Small block 4k accesses do take a significant hit – around 3.6MB/s read and 0.6MB/s write. This level of performance is quite unremarkable.


The ATTO benchmark shows the card reaching full I/O performance by 64KB, which is not a bad result, suggesting potentially smaller flash page sizes. Smaller accesses do suffer significant penalties, as is normally the case with flash memory. The write speeds peaked at 11.7MB/s, with read speeds peaking about 76.9MB/s.


H2testw results are relatively consistent with the previous, indicating write speeds just compliant with Class 10 requirements and read speeds of 73.6MB/s which exceed the claimed 65MB/s. No data integrity problems were encountered operating with the default format.


The Strontium Nitro 16Gb Class 10 UHS-I MicroSDHC cards aren’t anything special, which is exactly what I expected given the price. The write speeds barely meet the Class 10 standard, with the read speed being quite close to (or even above, depending on the benchmark) the 65MB/s advertised, so it is an honest offering. Unfortunately, when compared with some of the other Class 10 competition which can push write speeds up to 20MB/s or even higher with read speeds closing in on 90MB/s, these cards do feel a little slow by comparison. For the price, it was still worthwhile.

Posted in Computing, Flash Memory | Tagged , , , , | Leave a comment

Note: Multiple VLAN Operation on Realtek RTL8111D NIC (& Others)

A while ago, I wrote about the wonders of having multiple VLANs at home, operating in a mixed “hybrid” access model. While it was great, I made a comment that simultaneous VLAN operation might not be possible under Windows depending on the NIC involved. While I had a late-model Intel NIC on my main desktop which happily supports Intel Advanced Networking Services, allowing me to participate in multiple VLANs at the same time through a number of virtual adapters, I noticed some other NICs only had one VLAN ID entry within their advanced driver properties.

Owing to a new project, I suddenly had to re-commission my former desktop. The Gigabyte 890FXA-UD7 uses a dual GbE LAN solution based around the Realtek RTL8111D solution. The problem was that the driver property pages had no VLAN entry!

I checked the usual culprits – namely updating the driver to the latest version on Realtek’s website. Unfortunately, this didn’t change the situation at all.

According to Gigabyte’s website as well as Realtek’s product page, the solution supports VLAN tagging – so what gives?

Realtek Ethernet Diagnostic Utility

The answer to the VLAN problem lay in an unexpected place – the Realtek Ethernet Diagnostic Utility. The tool can be downloaded on the same page as the drivers for the NIC.

I’ve never bothered to download or install this, as a “diagnostic utility” is usually used to diagnose issues with the network adapter, drivers or cables – none of which I’ve ever really had any issues with.

As it turns out, the diagnostic utility is both a diagnostic utility and a tool for configuring virtual interfaces and VLANs.

This can be done through the VLAN page, which allows you to add a new VLAN to be member to (which automatically installs the appropriate driver and creates a new virtual interface). You can also change the MAC address of the virtual adapter in case it’s necessary to avoid conflicts. For the most part, it works as expected, but adding VLANs will normally unbind TCP/IP from the main adapter, resulting an inability to access the untagged network. The simple fix is just to open the properties for the adapter and rebind the protocol – that way you can access both tagged and untagged networks simultaneously. The functionality is equivalent to Intel ANS-based VLAN.

It’s also interesting to see that the adapter is capable of reporting cable length fairly accurately (actual length was 5m) – this can be accessed in the BIOS as well.

The Green Ethernet page also lets you see the power savings achieved from Energy Efficient Ethernet operation. It was rather disappointing to see that the maximum energy saving for this chipset is 16%, and in the case of my 5m cable, just 14% was being saved. I guess something is better than nothing though.


As a result, if you have a Realtek adapter, you might not need to miss out on the fun of multiple VLAN operation under Windows. The deception is in the fact that to configure this, you need to use the Diagnostic Utility, which isn’t an intuitive place to go looking for such a feature. Once configured, the untagged network is not accessible by default due to unbinding of TCP/IP from the host adapter – you can safely rebind it and thus achieve both tagged and untagged operation simultaneously on the same adapter.

So I guess in the case of the RTL8111D, it’s not bad news at all to have a Realtek.

Posted in Computing, Telecommunications | Tagged , | 1 Comment