Category Archives: Bits and pieces

Keysight parts division: Clearly not profitable

Keysight 34401-88304 rear bezel in original packaging

Half width Agilent / Keysight instruments made up until the late 2000’s had a plastic back bezel (34401-88304) which the rear bumper attaches to. They are frequently broken in electronic labs because the rear bumper often slides off the end of the desk. Someone then grabs the instrument to pull it forward, not knowing it is caught on the back edge of the desk, destroying this part in the process. After the 100th time trying to glue mine back together – it was time to get a replacement.

They are frequently seen for sale on eBay for $100 or more, but the part’s true cost from Keysight is only around $12 – if you can be bothered with the hassle of ordering it directly from keysight – Which brings me to my point here.

Where U.S. Customers can order with ease directly from the website – getting this part in the UK involved 12 emails, 4 phone calls, to people in 3 different countries, and hours of both mine and Keysight’s time. In the end I was charged just £14 (including VAT and shipping). Ridiculous. Keysight likely spent £1000 of their own time getting me this part.

I am not faulting the service here – Given the rigmarole I am astonished at how little this cost! It even comes in a ludicrously oversized box:

I ended up having to collect in person from Mayfair post office –  who are never thrilled to have to hold large items – resulting in the spectacle of me carrying this large geeky box through the crowded streets of Central London (not that a single person would have recognised that logo).

Yeah… just the shipping alone would have cost £14.

In the end a rare example of dealing with a big brand like this, without punishing my credit card in the process. I also have some Keithley parts to order this month. That is not going to be quite so cheap.

Restoring a thrashed Plextor PX-40TSi SCSI CD-ROM Drive

Recently I purchased a Plextor PX-40TSi SCSI CD-ROM drive from eBay. It was described as being in “excellent” condition and unusually it came with all of the retail packaging, manual & accessories which was re-assuring. This particular one was very reasonably priced considering that these drives, despite not having made for some 20 years now, remain stubbornly expensive – commonly fetching $50 second hand.

This is because they are commonly found in older non-PC industrial, professional audio, electronic/RF and medical equipment which is still in use and valuable, and as a result of this people still buy up these drives to fit or replace existing units. Compounding the problem there are annoying people like myself buying them to connect to ancient computers for no good reason.

This of course is all nothing compared to Plextor SCSI CD-RW Drives. The PX-W4012TS (40/10/40) still fetches an astonishing $500 price tag. A quick bit of research reveals these may be being used for vintage Akai DPS units, but that doesn’t quite seem to justify that kind of price tag. If you know anything more – please drop a comment!

My Agilent 16702B is an example of an older piece of equipment which uses a SCSI CD-ROM drive. SE drives like this one tend to fetch a higher price than newer LVD drives too for some reason.

When I unboxed this unit, I quickly noticed the plastic around and on the eject button was noticeably worn. This drive has clearly had a lot of use. I plugged it in, pressed eject and the tray came out with a horrendous grinding noise. It was also apparent that it was full of dust, dirt and even quite a lot of hair. Yuck!

Upon inserting a CD, it spun up with a terrible screeching noise, and down, and up, and down, and so on, but never managed to read a disc. I complained profusely to the seller, who immediately refunded all of my money, leaving me with this piece of junk of a drive.

With the tragic 2019-nCov outbreak in China likely to leave me without the PCBs for my next project for quite a while – this has left tech-time a little aimless to say the least. What more can one do than spend some of that time to see if I can bring this thing back to life while we wait and hope for the best.

Problem 1: Drive full of crap

A quick blast with compressed air sorted that out.

Problem 2: Can’t read any discs

My first thought was to dab the lens of the laser with a cotton tip soaked in isopropyl alcohol. Astonishingly this fixed the problem. This trick sure as heck hasn’t fixed any other faulty optical drive I’ve had over the years. With the drive now demonstrated to be functional, we can now push on and tackle the rest of the issues.

Problem 3: Grinding noise when ejecting the tray

This was a nice easy one to fix. I just pulled all of the gears out, cleaned out the copious volumes of dust and other crap, re-fitted them, replenishing the plastic grease that would have been there originally.

Problem 4: Screeching noise during operation

This was going to be a little more difficult to fix. From the sound of it this was likely the bearings in the spindle motor.

Upon removing the spindle motor assembly – a manual spin reveals that the bearings aren’t sounding very healthy. This being a Plextor drive it was fitted with two good quality deep-groove ball bearings as we would expect, but after likely more than a decade of daily punishment even those were completely knackered. Not once in my life have I seen an optical drive so thrashed as to require new bearings!

There are quite a few spindle motors for CD-ROM drives for sale on eBay, but none like this one. Bummer. We’ll have to try hammer those bearings out without destroying the motor in the process.

Without any other obvious ways of doing this, I just went straight at it with an automatic centre punch. After a good dozen or so actuations the spindle separated without any damage.

The bearings used here are 3x7x3mm in size which I was able to purchase couple of replacements of at a cost of just £2.50 each – free shipping.

Fitting the new bearings was easy – I just dabbed a tiny bit of superglue on the brass casing – as it appears there was something gluing the previous bearings in possibly? The spring between the bearings is perhaps there to tension them correctly so perhaps this isn’t necessary.

The process for re-fitting the spindle was similar, pressing from the disc side with a hex spacer this time. This took quite a bit of force.

Conclusion

After a couple of hours of repair, I was surprised to be able to return this very tired old drive to full operation – with it now working as good as it would have done new!

Unfortunately my eBay purchased bearings aren’t amazing quality so the drive is not quite as quiet as I was hoping it’d be, but it is certainly a hell of a lot better. I think I might chuck these out and try find some better ones at some point.

P.S.

The problems I have shown here were very trivial to fix but this isn’t necessarily always going to be the case. If you have any experience of repairing optical drives – I would be really interested to hear from you what sort of issues they develop, and indeed how to address them!

A look at a vintage Sony CD Burner

Recently I was looking around on eBay for an Adaptec SlimSCSI PCMCIA card – for an escapade unrelated to this post, there was of course one there for sale, for a reasonable price, but it included this lump of a CD Burner which I really didn’t want.

I was tempted to just throw it straight in the trash… but…

This happens to be one of the very earliest commercially available 2x (two-spin) units, which likely had a list price of thousands of dollars. I remember hearing about these things in the early to mid 1990s but I never actually saw one with my own eyes.

I knew someone who worked for a company who owned one – he would sneak into the room it lived in after hours to create his own music CDs. He was eventually caught, and sacked. Unfortunately his severance didn’t include the CD Burner.

I didn’t get access to anything like this until around 1999, and that would have been a much more modern 8x4x24 IDE unit which sold at a price affordable to the average person.

Anyway… let’s take a look at this beast.

The first noticeable feature is that it is a “Caddy” drive. These were common during the very early days of CD-ROMs because software CDs in particular cost a lot of money, and most people didn’t have a way to copy them, so paying extra for a caddy was worthwhile to protect the investment. Such CD-ROMs were loaded into a caddy, and left there, reducing the chances of damage to practically zero.

It was common for cheaper CDs to be swapped between caddies as the caddies themselves weren’t especially cheap either. In the end CD-ROMs got cheaper and all the caddy swapping just irritated people. Having a CD Burner eliminated the problem caddies were designed to solve entirely, making it a rather interesting feature for this unit.

Another interesting feature is the SCSI interface. This is because it is an external unit – early CD burners like this were typically standalone because they were insanely expensive, and moved from computer to computer as required. A PCMCIA type attachment  as I have shown would have been a popular choice as the unit could be connected to any desired laptop without having to purchase and fit a SCSI adapter to each.

The reason for the choice of SCSI over the more common IDE interface, is that IDE never had a standardised external connection mechanism (although it was possible at a pinch). There was no point in defining one because it was not hot-pluggable, and indeed the consequences of disconnecting drives during operation were generally dire. It was also an un-terminated bus, meaning it would be clobbered by signal reflections if cables were over 30CM or so. Adding additional connectors required for an external interface would have exacerbated the problem significantly.

SCSI on the other hand with hot-swapping built in from the get-go, and with its integral termination mechanism was ideal for this kind of application.

Time to take it apart…

As can be seen from the outside, this is a big unit. We can see it has its own built-in switching power supply down the right hand side. Also as expected a standard form factor 5 1/4″ drive with an interesting bespoke mounting solution. Sony likely also sold this as an internal drive (CDU920S).

Next step is to pull the drive out, and open its top cover.

We are immediately presented with the main PCB, which has a lot of components on it.

And – yup – on the other side as we would expect there is a ridiculous number of chips on it. A modern CD/DVD burner only has two or three chips at most.

Lastly the caddy and optical mechanism. Quite a bit of extra stuff there to deal with that caddy!

Burning a CD with it

I went to the trouble of installing Adaptec Easy CD Creator 4 on an old PC to see if I could make it work. After hours of frustration I gave up.  I think it has some hardware issues – as when a disc is inserted it just makes a lot of spinning up / spinning down / clunking noises, doing very little else.

A quick look at the Keithley 2380 DC Electronic Load

There’s plenty of information about these floating around the internet – so this is not a review, but my first impressions. Let’s unbox and take a look at this 2380-120-60:

If you’re expecting something the usual quality of a Keithley instrument, say, a Source Meter – prepare to be disappointed.  In terms of external build quality – I actually found it slightly worse than another Chinese designed & manufactured instrument I own which cost one tenth of the money.

The first issue I noticed is that the tabs on the rear bumper don’t line up with the holes in the metalwork. I’m sure they did in the CAD drawing but unfortunately due to sloppy molding the only way I could get the rear bumper to stay attached is by cutting them down to size a little. Not the kind of thing one expects when purchasing an instrument from a Tier 1 vendor.

Anyway – let’s have a look inside:

The cover isn’t very well formed either – The position of the bends wrapping down to the underside mean it has an extremely tight fit on the inner chassis, suffice to say I had quite a job getting it off. Eventually after a bit of bashing and brute force it succumbed. On my aforementioned £200 Chinese cheapie, that cover glided straight off!

Immediately we can see what the deal is here. That top PCB is 100% identical to the one found in the BK Precision 8601 which its self is designed by ITECH. I’m not going to go to the trouble of dismantling it but from what I can see that bottom PCB is looking pretty damn similar to the 8601 too.

Don’t believe me? Take a look at this:

So… this probably isn’t a Keithley instrument, but some kind of re-branded deal.

A little disappointing – but not a surprise. Given the relentless pressure of Chinese competition on these instruments, I guess western players need to cut every possible cost, and it appears that in this case, the design of the instrument has been outsourced entirely.

There is some good news – some of the gripes in this video i.e. grinding part numbers off chips, and not-particularly-well insulated mains cables – all fixed in this unit.

2380 Rear – Identical to the BK Precision 8601 minus the remote control port

Serviceability?

The uncomfortable feeling I now have – is that this is not a serviceable unit i.e. we can’t buy parts from Keithley like we can for their other products. This is a problem we often get with re-branded products – the marketer cannot provide parts because they don’t manage the life-cycle of the product.

If anyone has gone to the trouble of inquiring with Keithley on this matter I would certainly like to hear from you.

There certainly doesn’t seem to be a downloadable service manual on the Tektronix website.

The good news is perhaps – if parts aren’t obtainable from Keithley, we may be able to get them through the BK Precision / ITECH route instead.

It appears that only the keypad and outer chassis (that’s “shazzy” – if you’re from Australia) differ from the 8601. VFD appears to be identical, and I suspect it’s even running the same firmware too – it does report the same startup message “BIOS Ver1.10” seen on the 8601.

Conclusion

Despite being less than impressed with the physical build – the specification is of course significantly better than my existing £200 unit (which barely has any written specifications at all).

Given the identical spec to the BK Precision 8601 (understandable given it has the same internals) – I’m not quite sure where the value is for that extra £450 asking price.

If I’ve misunderstood anything here – please correct me.

Retrofitting a Bosco IXO cordless screwdriver with a USB-C charger connector

In this age of USB-C – I’m beginning to tire of things which require Micro USB to charge.

Today’s annoyance: My Bosch IXO. At the time of writing neither my model or the current one charge with USB-C. No doubt people at Bosch are beginning to ask “Do we need to take this new USB seriously?” – but that doesn’t help us much right now.

So let’s open it up.

Damnit. It’s all on one PCB, and there’s a nasty piece of rock hard epoxy behind the micro USB connector. This rules out making a ‘Retrofit’ PCB – which would have been a cool project.

After a good blasting with my Leister Hot Jet S – the old connector and the epoxy is gone. Small bummer that I tore those unused pads from where the connector was, but we don’t need those anymore. I’ve also got a couple of wires there to lead to the new connector

Now – what are we gonna put in its place?

I bought a few Micro-USB to USB-C adapters off eBay and tore the plastic case off one of them. They are compact and have everything we need.

We now have the new connector soldered down. This was quite tricky. I found that the easiest way was to position it by tying it down with some copper wire before soldering. The shell of the USB-C connector is soldered down to the ground pad on the PCB previously used by the Micro USB connector.

Wiring is simple. With those red/black wires I previously soldered on connected to the Micro USB GND/VBus connections. The required 5.1K resistor is already on the adapter PCB.

For added strength that copper wire also goes through the PCB.

You can also see a tiny piece of Kapton tape under the connector. This was to stop the shield of the USB-C connector from shorting the pads used by the old micro-USB connector. By not outright removing all of those pads I made life rather difficult for myself – too late now.

And finally some new epoxy to ensure we’ve got a nice strong bond to the PCB.

Last job is to grab some needle files to increase the aperture in the case to the new larger size needed by the USB-C connector.

In the end it turns out we have something a lot tougher than the original arrangement, because of the longer length of the USB-C connector it now protrudes from the case meaning we can make a nice snug fit for it, significantly lowering the chances of damage by yanking of the charging cable.

That extra length of the connector also means we don’t have the annoying problem of plugs which aren’t the same shape as the supplied charger not fitting in either.

Conclusion

Unfortunately this doesn’t get our IXO all the way into the world of USB-C. PD aware “true” USB-C chargers like for example the Apple 18W iPhone charger, or USB-C Macbook chargers refuse to charge it.

To get full USB-C compatibility we would need an extra chip to negotiate the charge voltage. There are quite a few of these on the market now but we would have to make a custom PCB due to space limitations in the IXO. This is good enough for me – anything which has a USB-A connector on it will work just fine.

This all took me about 2 hours. Not an easy mod unfortunately. But if you have the skill and patience – well worth attempting.

Fixing a Sony KDL-40W5810 with a “14 blink” error code

My dinosaur TV back in action after repair – reassuring to see that little has changed in the world since it last powered on

A little while back I bought this TV cheaply off Gumtree – probably one of the dumber things I’ve done, as it had an array of small but annoying problems, but recently it upped the game and failed to turn on, giving the above described blink code (which shows as 13 blinks, then followed by 14 blinks continuously).

Frustratingly there doesn’t appear to be any official triage documentation for this model online. A bit of googling around offers a few possible causes ranging from mainboard failure, LCD panel failure (?) and TCON (timing controller) failure.

But which is it? I first checked all the voltages on the power supply, all good, looked briefly at the TCON, couldn’t see any obvious issues there. I also took a look at the backlight – the mainboard wasn’t even attempting to power it on, so not likely the issue, but just to be double sure I powered it up manually by applying +3.3V to the BACKLIGHT_ON signal on the power supply. Sure enough it came on and none of the error signals were asserted.

Now I know the problem is either the mainboard or the TCON. I took a stab replacing the mainboard – hitting up eBay for a cheap second hand part. Upon installing it, I now get an 8 blink error code? More googling reveals an issue with the audio amplifier.

Ah. I see, so someone has torn the connector from the PCB when removing it from the donor TV. We can see those small traces have been torn the pads the connector was soldered to.

This was causing the 8 blink error code, as those small traces lead to an ADC which tests for DC offsets from the audio amplifier. This check ensures that we don’t end up with smoke pouring out of the speakers in the case of a blown transistor in the audio amplifier. In this case there was no DC offset, the error was just the ADC inputs drifting all over the place because they weren’t connected to anything.

JST PH header bodged in place of the missing original header

Above you can see I’ve dickied a JST PH connector in place of the missing original – not quite the same as what was there but close enough, also taking care to reconnect those small traces. The original harness mates with it OK – just doesn’t latch.

So I switched it back on with the new mainboard… 8 blink error is gone, and now we’re back to 14 blinks. FFS. The good news is that the eBay seller fully refunded me when I complained of the damage, and I’ve still got the replacement mainboard, but, my TV still ain’t workin’.

So, let’s look at that TCON then.

The TCON is under the aluminum shield circled in red. It converts the serialised picture data from the mainboard to the parallel signals required to drive the LCD panel.

I had purposely not gone down the path of replacing this because it’s not an easy part to find and I had read on other sites that if you disconnect the LVDS cable (the large black one which interlinks it to the mainboard), then power on the TV and it remains powered on (without a picture), then you definitely know that the TCON is at fault.

This was not I observed. In my case I get the 14 blink code regardless of whether or not the LVDS cable was connected. Additionally I had also spent a bit of time checking all of the supply voltages on the board, everything looked OK, hence attempting the mainboard first.

It turns out that bit of advice was rubbish in the case of my TV, because the 14 blink error code is seen both in the case of a TCON which fails self test, and when the TCON is disconnected.

I know this, because I found another (brand new) TCON very cheaply at a local TV service company, mis-labeled as “for Samsung” – knowing this is a rare and pricey part, I wasn’t going to be asking any questions. Upon installing it, my TV is now working again.

The old TCON PCB

I’m picking that few faulty TCONs can easily be repaired, because this board subject to more thermal stress than anything else in the whole unit. After years of thermal cycling it’s likely that this one has died as a result of cracked BGA solder joints.

If you are keen to repair a TCON with this kind of fault – a good starting point would be to re-flow or re-ball, perhaps even replace the gamma processor and its memory (the three BGA chips on the left). The TCON itself (silver topped chip in the centre) is less likely to be an issue due to cooler running temperature.

Conclusion

Success in the end, with a lot of time spent on a worthless item. Perhaps a small bonus that the largely needlessly replaced mainboard fixed a couple of other minor faults. I’d have been a lot more successful if I’d never bought the damn thing in the first place.

Does the Dell Latitude M.2 WWAN socket have the SATA interface on it?

Recently while trying to answer this myself, I found a lot of discussion on this subject, but no definitive answer.

M.2 B Key sockets have a variety of possible modes as defined by the NGFF specification:

The second to last mode “RFU” (Reserved for future use) is defined by Dell as:

I am unsure what HCA stands for, but have seen it printed on the silkscreen of some of my earlier latitude models. Clearly it is some kind of proprietary device that requires one PCIe lane. It may stand for Host Channel Adapter – Implying some kind of non-NVMe PCIe SSD (i.e. has an Option ROM).

None of this tell us anything about what Dell have actually implemented on thier WWAN slots, neither does their tech support, or any of their documentation. I found myself eyeballing the traces extending from the socket, but even this was inconclusive as a lot of them are fed up from vias under the socket.

It’s given that it’ll have SSIC/USB 2.0 because almost all WWAN cards use those interfaces, but what of the others?

Fortunately there’s no need for any further conjecture because the schematics for these models are floating around the internet, so let’s answer that question definitely.

The above diagram covers models 7280, 7380, 7480, 7490 and likely others too. So there we have it. The interfaces on the WWAN socket are:

  • USB 2.0
  • SSIC (Chip-to-Chip USB 3.0)
  • PCIe (Permanently disabled in BIOS)

No SATA.

But…

Just because we know what interfaces are there, we still don’t know what kind of peripherals will actually work. For example – if a socket can accept WWAN-PCIe – SSD-PCIe should also be no problem as the required connections are all there, however those devices will have their configuration pins tied differently, allowing the BIOS to determine exactly what’s attached.

This allows manufacturers (for example) to allow PCIe WWAN cards, but disallow PCIe SSDs. On top of that, just because an interface is there, it doesn’t mean that it’s actually enabled.

This turns out this is the case for my Dell Latitude. While the PCIe is there, it cannot be used because the port on the root complex is disabled in the BIOS, and there’s no way to enable it (without hacking the BIOS). I was able to confirm this by testing out a variety of B-Key PCIe devices, none of which were detected (even when strapped as WWAN-PCIe).

A dick move by dell, but given how rare WWAN cards requiring PCIe are, they had no reason to enable it, and my experience from owning previous models is that they’re pretty good at tying up loose ends like this.

Short of hacking the BIOS, or building a new type of USB 3.0 card from scratch, only WWAN cards will work in the WWAN slot.

Yours may differ

Older Dell models did have SATA+PCIe in the WWAN socket, but this is not the case for more recent units. The story will inevitably be different for other manufacturers.

Those Thunderbolt to PCIe boards which shouldn’t exist, but do

If like me you happen to have a Thunderbolt 3 enabled laptop, and are finding yourself wanting to attach arbitrary PCIe peripherals to it, you may have found yourself looking for some kind of adapter board.

You may also be aware that such items are strictly forbidden under the Intel Thunderbolt licensing programme. This is because thunderbolt chips are only sold for use on certified products. As a device like this is only part of a product, it could never be certified.

But that doesn’t technically stop it from existing. In theory this problem should go away with the upcoming release of USB4, but that may be some time away.

You may have seen one of these on either eBay or AliExpress:

Markings:
ART150728
21A-10030
AR SP BPD PCIe_Rev_2p1

But what is it? How can it exist? It appears to have the same dark green hue and silk-screen font as an Intel reference board (of which I’ve got quite few). I’d be comfortable to say they did indeed make this.

Quite how they’ve ended up for sale is an interesting question. What’s even more interesting was the transaction its self. I bought this off AliExpress, paid in U.S. Dollar, apparently ship from China in 20 days… but…

It arrived the next day! Shipped from the Amazon UK warehouse?!

Yup. This thing came from the Amazon UK warehouse. What the hell!

This certainly adds the the intrigue of who exactly it is that is selling them, not to mention how. The punchy price tag tells us they haven’t got an unlimited supply of them.

Sellers generally advertise them as for SSD use only. This is more because it only supplies 3.3V to the attached peripheral than anything else. Most PCIe cards also require 12V which this board does not supply. That doesn’t stop us from providing the 12V separately.

Plugging it in

The thunderbolt info dialog describes it as an “LT-LINK Node Lite” – a similar name to the AKiTiO Node Lite. We also see the warning there about a graphics device which “may not function properly” telling us that the firmware on this board was for some kind of eGPU enclosure.

Using it

To test it out I’ve decided to throw a rather curly scenario at it –

I’ll be attaching it to my ExSys PCI chassis. Fitted will be an Intel PRO/100B – a 24 year old conventional PCI adapter, which doesn’t support 64-bit DMA (nor does it have 64-bit drivers). Fortunately the source code for its driver is in the Windows DDK, so I’ve compiled it for 64-bit Windows for this test…

This test will also add another couple of PCI-to-PCI bridges onto the existing arrangement in the Thunderbolt chain.

The chassis is connected to the Thunderbolt adapter via the PCIe card option.

Yeeeha! Got PCI-to-PCI bridges? I quite frankly have no idea which of those is where, or doing what.

Even with a long chain of bridges,  a transition to conventional PCI, and the rather legacy nature of what I’ve attached – it works, and passes traffic.

So there we have it. You can indeed use peripherals other than SSDs on these!

Silencing a Cisco 2951 (for where ever silence may be required)

A little while back I wrote a post on silencing a Cisco 2911 for home use. You can read about that here. I accept that it’s unlikely anyone would be using 2951 at home however you may find yourself in a situation where you need to reduce the noise level of this router. In my 2911 guide I went to the extent of rebuilding the power supply as a 12V DC version to reduce heat output. We’re not going to be bothering with that here, instead we’ll be looking solely at the fans. The first task is to remove the four original fans. Remove all the screws you can see on the fan module, and release the plastic clips holding the face plate.

Extracting the contacts from the mating connector

Next step is to unpick the contacts from the receptacle I personally used the Molex Mini-Fit Jr extraction tool (11-03-0044) – which is the correct tool for this, however you may find it more convenient to just cut the wires off the old fans and splice them onto the new fans.

Stock 2951 fans removed

In my case I’ve used the same gold-plated Mini-Fit Jr contacts (42815-0012) as were used on the original fans. You may not feel the need to bother with those. You can read more about these connectors here. Just in case it isn’t obvious – the pinout for that connector is as follows:

  • 1 : GND (Fans 4, 3)
  • 2 : +12V (Fans 4, 3)
  • 3 : PWM 3
  • 4 : TACH 3
  • 5 : +12V (Fans 1, 2)
  • 6 : GND (Fans 1, 2)
  • 7 : Module presence strap
  • 8 : PWM 4
  • 9 : TACH 4
  • 10 : PWM 2
  • 11 : TACH 2
  • 12 : PWM 1
  • 13 : TACH 1
  • 14 : Module presence strap

The next step is to re-install connections for fans 1 and 2 only, as we’re going to blanking up the holes for fans 3 and 4. If you’re not replacing the contacts entirely like me, you won’t need to be bothering with this.

Connector from bottom
Connector from top

As previously mentioned, connections for fans 3 and 4 are omitted, except we’ve strapped the tach signal from fans 1 and 2 to them to keep the software satisfied that all 4 are present. Now we have to assemble the new fan module. I’ve used two Delta AFB0612VHC fans. I don’t personally see the need to go splashing out on expensive “ultra silent” fans from the PC modding scene. You can if you like but they’re not going be much (if any) quieter than these.

The fan module with only two fans installed. It is important that these positions have fans fitted as this is the side with the power supply and CPU.

I have also blanked up the holes for fans 3 and 4, to ensure airflow remains consistent. Next step is to re-assemble the fan module, and plug it back into the router – making darn sure you’ve wired it properly first. Now, the obligatory ‘show env’ to check that everything’s OK:

Router>show env
SYSTEM POWER SUPPLY STATUS
==========================
 Internal Power Supply  Type: AC
 Internal Power Supply  12V Output Status: Normal

 External Redundant Power Supply is absent or powered off

SYSTEM FAN STATUS
=================
 Fan 1 OK, Low speed setting
 Fan 2 OK, Low speed setting
 Fan 3 OK, Low speed setting
 Fan 4 OK, Low speed setting

SYSTEM TEMPERATURE STATUS
=========================
 Intake Left temperature: 21 Celsius, Normal
 Intake Right temperature: 20 Celsius, Normal
 Exhaust Left temperature: 34 Celsius, Normal
 Exhaust Right temperature: 25 Celsius, Normal
 CPU temperature: 59 Celsius, Normal
 Power Supply Unit temperature: 49 Celsius, Normal

REAL TIME CLOCK BATTERY STATUS
==============================
 Battery OK (checked at power up)

SYSTEM POWER
===============
 Motherboard Components Power consumption = 55.3 W
 Total System Power consumption is: 55.3 W

 Environmental information last updated 00:00:04 ago

Router>

Looks good.

Caveat Emptor

As I stated in my previous 2911 guide, when we make modifications like this, we’ve got to consider the consequences. For the 2951 you’ll not be able to use it with any service modules. If you need to install service modules, I would recommend using 4 of these fans.

How much quieter is it?

To save you from asking – it now makes about as much noise as a middle of the range ATX power supply under a favourable conditions (i.e. small load, and room temp of 20 degrees Celsius.