machine – Experimental Engineering

Posted on March 15, 2018March 15, 2018 by The Engineer — Leave a comment

Raspberry Pi 3 Model B+ Initial Tests & Benchmarks

Yesterday, the Raspberry Pi community got a nice surprise – a new Pi! This one has some improved features over the previous RPi 3 Model B:

Improved CPU – 64-Bit 1.4GHz Quad-Core BCM2837B0
Improved WiFi – Dual Band 802.11b/g/n/ac. This is now under a shield on the top of the board.
Improved Ethernet – The USB/Ethernet IC has been replaced with a LAN7515, supporting gigabit ethernet. The backhaul is still over USB2 though, so this would max out at about 300Mbit/s
PoE Support – There’s a new 4-pin header, and a matching HAT for power over ethernet support.

The USB/LAN Controller is now a BGA package, supporting gigabit ethernet. The USB connections are still USB2 though, limiting total bandwidth. This shouldn’t be much of an issue though, since anything over the 100Mbit connection we’ve had previously is an improvement.

The CPU now has a metal heatspreader on top of the die, no doubt to help with cooling under heavy loads. As far as I know, it’s still the same silicon under the hood though. The WiFi radio is under the shielding can to the top left, with the PCB trace antenna down the left edge of the board.

The power supplies are handled on this new Pi by the MaxLinear MxL7704, from what I can tell from MaxLinear’s page, it seems to be somewhat of a collaborative effort to find something that would do the best job, since they apparently worked with the Foundation to get this one right. This IC apparently includes four synchronous step-down buck regulators that provide system, memory, I/O and core power from 1.5A to 4A. An on-board 100mA LDO provides clean 1.5V to 3.6V power for analog sub-systems. This PMIC utilizes a conditional sequencing state machine that is flexible enough to meet the requirements of virtually any processor.

The bottom of the PCB has the Elpida 1GB RAM package, which is LPDDR2, along with the MicroSD slot.

A quick benchmark running Raspbian Lite & a SanDisk Ultra 32GB Class 10 SD card gives some nice results:

Raspberry Pi Benchmark Test
Author: AikonCWD
Version: 3.0

temp=45.1'C
arm_freq=1400
core_freq=400
sdram_freq=500
gpu_freq=300
sd_clock=50.000 MHz

Running InternetSpeed test...
Ping: 45.278 ms
Download: 151.50 Mbit/s
Upload: 9.52 Mbit/s

Running CPU test...
 total time: 11.3003s
 min: 4.48ms
 avg: 4.51ms
 max: 44.50ms
temp=56.4'C

Running THREADS test...
 total time: 10.2161s
 min: 3.94ms
 avg: 4.08ms
 max: 21.49ms
temp=59.6'C

Running MEMORY test...
Operations performed: 3145728 (2418384.67 ops/sec)
3072.00 MB transferred (2361.70 MB/sec)
 total time: 1.3008s
 min: 0.00ms
 avg: 0.00ms
 max: 9.99ms
temp=60.7'C

Running HDPARM test...
 Timing buffered disk reads:  66 MB in  3.01 seconds =  21.91 MB/sec
temp=51.5'C

Running DD WRITE test...
536870912 bytes (537 MB, 512 MiB) copied, 34.6011 s, 15.5 MB/s
temp=46.7'C

Running DD READ test...
536870912 bytes (537 MB, 512 MiB) copied, 23.5404 s, 22.8 MB/s
temp=45.6'C

AikonCWD's rpi-benchmark completed!

Posted on August 17, 2017August 16, 2017 by The Engineer — 2 Comments

EpEver MT50 Control Panel Teardown

Here’s the MT50 controller from EpEver, that interfaces with it’s Tracer MPPT solar charge controllers, and gives access to more programming options on the charge controllers, without the need for a laptop. The display is a large dot-matrix unit, with built in backlight. Above is the display on the default page, showing power information for the entire system.

The rear plastic cover is held in place by 4 machine screws, which thread into brass inserts in the plastic frame – nice high quality touch on the design here, no cheap self tapping plastic screws. Both power & data arrive via an Ethernet cable, but the communication here is RS-485, and not compatible with Ethernet! The PCB is pretty sparse, with comms & power on the left, LCD connection in the centre, and the microcontroller on the right.

On the left of the board is the RS0485 transceiver, and a small voltage regulator. There’s also a spot for a DC barrel jack, which isn’t included in this model for local power supply.

The other side of the board holds the main microcontroller which communicates with the charge controller. This is a STM32F051K8 from ST Microelectronics. With a 48MHz ARM Cortex M0 core, and up to 64K of flash, this is a pretty powerful MCU that has very little to do in this application.

The front of the PCB has the ENIG contacts of the front panel buttons, and the LCD backlight assembly. There’s nothing else under the plastic backlight spreader either.

The front case holds the LCD module in place with glue, and the rubber buttons are placed underneath, which is heat staked in place.

The LCD is a YC1420840CS6 from eCen in China. Couldn’t find much out about this specific LCD.

Posted on April 1, 2017April 24, 2017 by The Engineer — Leave a comment

Website Hosting Updates!

Over the past few weeks, the host I’ve been with for over 3 years, OVH, announced a rather large price increase of 20% because of Brexit – the current universal excuse to squeeze the customer for more cash. This change has sent the price of my dedicated server solution with them to over £45 a month. Doing some napkin-calculation gave me £18 a month in extra power to run a small server locally. So I’ve decided to bring the hosting solution back to my local network & run from my domestic internet link, which at 200Mbit/s DL & 20Mbit/s UL should be plenty fast enough to handle the modest levels of traffic I usually get.

Obviously, some hardware was required for this, so I obtained this beauty cheap on eBay:

HP MicroServer Gen 8 — HP Proliant MicroServer Gen 8

This is a Gen 8 HP Proliant Microserver, very small & quiet, perfect for the job. This came with 4GB of RAM installed from the factory, and a Celeron G1610T running at 2.3GHz. Both are a little limited, so some upgrades will be made to the system.

4 SATA drive bays are located behind the magnetically-locked front door, there’s a 250GB boot disk in here along with a pair of 500GB disks in RAID1 to handle the website files & databases. For my online file hosting site, the server has a backend NFS link direct to Volantis – my 28TB storage server. This arrangement keeps the large file storage side of things off the web server disks & on a NAS, where it should be.

First thing is a RAM upgrade to the full supported capacity of 16GB. This being a Proliant server machine, doesn’t take anything of a standard flavour, it’s requirements are DDR3-10600E or DDR3-12800E (the E in here being ECC). This memory is both eye-wateringly expensive & difficult to find anywhere in stock. It’s much cheaper & easier to find the ECC Registered variety, but alas this isn’t compatible.

Over the past 48 hours or so, I’ve been migrating everything over to the new baby server, with a couple of associated teething problems, but everything seems to have gone well so far. The remaining job to get everything running as it should is an external mail relay – sending any kind of email from a dynamic IP / domestic ISP usually gets it spam binned by the big providers instantly, regardless of it actually being spam or not – more to come on that setup & configuring postfix to use an external SMTP relay server soon!

If anyone does find something weird going on with the blog, do let me know via the contact page or comments!

Posted on December 5, 2016March 27, 2017 by The Engineer — Leave a comment

OpenVPN Server Speed Tweaks

I’ve been running my own VPN so I can access my home-based servers from anywhere with an internet connection (not to mention, in this day & age of Government snooping – personal privacy & increased security).

I’m on a pretty quick connection from Virgin Media here in the UK, currently the fastest they offer:

To do these tests, I used the closest test server to my VPN host machine, in this case Paris. This keeps the variables to a minimum. Testing without the VPN connection gave me this:

I did expect a lower general speed to a server further away, this will have much to do with my ISP’s traffic management, network congestion, etc. So I now have a baseline to test my VPN throughput against.
The problem I’ve noticed with OpenVPN stock configs are that the connections are painfully slow – running over UDP on the usual port of 1194 the throughput was pretty pathetic:

I did some reading on the subject, the first possible solution being to change the send/receive buffers so they’re set to a specific value, rather than letting the system handle them. I also added options to get the server to push these values to the clients, this saving me the trouble of having to reissue all the client configurations.

sndbuf 393216
rcvbuf 393216

push "sndbuf 393216"
push "rcvbuf 393216"

Unfortunately just this option didn’t work as well as I’d like, downstream speeds jumped to 25Mb/s. In the stock config, the tunnel MTU & MSSFIX settings aren’t bothered with, some adjustment to set the tunnel MTU to lower than the host link MTU (in my case the standard 1500) prevents packet fragmentation, MSSFIX let’s the client TCP sessions know to limit the packet sizes it sends so that after OpenVPN has done the encryption & encapsulation, the packets do not exceed the set size. This also helps prevent packet fragmentation.

tun-mtu 1400 
mssfix 1360

After adjusting these settings, the download throughput over the VPN link has shot up to 136Mb/s. Upload throughput hasn’t changed as this is limited by my connection to Virgin Media. Some more tweaking is no doubt possible to increase speeds even further, but this is fine for me at the moment.

Posted on February 8, 2016 by The Engineer — Leave a comment

De La Rue Coin Counting Machine Followup – PSU Oddness

I did a little more digging into the PSU circuitry of the small coin counting machine, and it’s even more strange than I thought!
The part I originally thought was a transformer on the PSU board is in fact a DC-DC converter module!

Here’s the device after desoldering it from the PCB. It turns out that instead of a transformer, it’s an inductor.

Underneath is the controller electronics, with an COB controller & the switching transistors are under a protective covering of silicone.

Driving this whole lot of PSU randomness is the mains transformer, with a secondary voltage of 35v.

The only reason I can think of that the manufacturer went to this much expense with the power supply is stability – a coin counting machine that miscounts due to power supply surges, sags & spikes wouldn’t be very much use. It’s not likely I’ll see anything similar again, unless I manage to get hold of something like medical grade equipment.

Posted on February 3, 2016February 3, 2016 by The Engineer — Leave a comment

De La Rue Coin Counting Machine

Here’s some teardown photos of an old De La Rue coin counter, used in businesses for rapid counting of change into large bags.

An overview of the whole mechanical system of the counter. Coins are loaded into the drum at the rear of the machine, which sorts them into a row for the rubber belt to pick up & run through the counter. The coin type to be sorted is selected by turning the control knobs on the right.
The control knobs adjust the width & height of the coin channel so only the correct sized coins will be counted.

The counter is driven by a basic AC induction motor, the motor power relay & reversing relay is on this PCB, along with the 5v switching supply for the main CPU board.
The SMPS on this board looks like a standard mains unit, but it’s got one big difference. Under the frame next to the main motor is a relatively large transformer, with a 35v output. This AC is fed into the SMPS section of the PSU board to be converted to 5v DC for the logic.
I’m not sure why it’s been done this way, and have never seen anything similar before.
The edge of the coin channel can be seen here, the black star wheel rotates when a coin passes & registers the count.

Here’s the main controller PCB, IC date codes put the unit to about 1995. The main CPU is a NEC UPD8049HC 8-bit micro, no flash or EEPROM on this old CPU, simply mask ROM. Coin readout is done on the 4 7-segment LED displays. Not much to this counter, it’s both electronically & mechanically simple.

Coin counting is done by the star wheel mentioned above, which drives the interrupter disc on this photo-gate. The solenoid locks the counter shaft to prevent over or under counting when a set number of coins is to be counted.

Under the frame, here on the left is the small induction motor, only 6W, 4-pole. The run cap for the motor is in the centre, and the 35v transformer is just visible behind it.

Main drive to the coin sorting mech is through rubber belts, and bevel gears drive the coin drum.

Posted on November 26, 2015July 13, 2016 by The Engineer — Leave a comment

nb Tanya Louise – Compressor Install

Compressed air is a rather useful power source, especially when all maintenance is done by the on board crew instead of by boatyards.

Screwfix had a good deal on a 50L 3.5CFM air compressor, to save space this has been permanently mounted in a free space & air will be piped to where it is needed from a central point.

Because of the total height of the machine, the compressor itself has been unbolted from the tank, a copper line connecting the two back together at a larger distance.

In one of the very few free spaces available, under a bunk. A pair of timbers has been screwed to the floor to support the tank.

The tank is strapped to the wooden supports with a pair of ratchet straps, the compressor itself can be seen just behind the tank. The copper line on the top of the tank is going back to be connected to the compressor outlet.

Compressor control remains on top of the tank, the pressure switch & relief valve centre. After an isolation valve, the feed splits, the regulator installed will be feeding the air horn with 20PSI, replacing the existing automotive-style 12v air pump. The currently open fitting will be routed to a quick connect on the bulkhead. This will be accessible from the front deck, an air hose can be fitted to get a supply anywhere on board.

More to come when the rest of the system gets installed!

73s for now.

The Shack

So, here is where all the action happens.

Main radio of course is housed on the left, it’s partially hidden under my currently over-populated breadboard.

All 3 monitors are linked to the same PC, using a pair of video cards. This is a very flexible system with so much screen real estate.

Main system power is provided by the pair of power supplies next to the radio – these are homebrew units using surplus switched mode PSU boards. Check my previous posts for more details.

The main power supply system. These two supplies are cross connected, giving a total DC amperage of 30A at 13.8v. There is also a link to a large 220Ah lead-acid battery bank (orange cable), to keep me on the air during power outages. This cable is getting upgraded to something more beefy shortly. The white cable is currently supplying power to my online radiation monitor.
The main high-current DC outputs are the Speakon connectors next to the meters. The top one is powering the radio directly, the bottom is linked through to my 12v distribution box for lower current loads, such as the oscilloscope, audio amplifiers, tools, etc.

Attached to the side of the desk is the radiation monitor itself.

Under the radio is the core NAS of the network. It’s an array of 9 4TB disks, in RAID6, giving a total capacity after parity of 28TB. This provides storage & services to every other machine in the shack, the Raspberry Pi on top of the disk array is doing general network housekeeping & monitoring, also generating the graphs for the Radiation Monitor page. A Cisco 48-port switch is partially out of frame on the right, providing 100MB Ethernet to the devices that don’t require gigabit.

Posted on July 11, 2015July 11, 2015 by The Engineer — 1 Comment

Simple Dishwasher Repair

Earlier today, one of my neighbours put their dishwasher out for the scrap man. After asking if I could appropriate it in the interest of recycling the Ham Way™, I was told it wasn’t draining. The engineer called out to fix it had claimed it was beyond economical repair.

A quick test showed that indeed the drain pump wasn’t operating correctly – very poor pumping capacity & a horrid grinding noise.

Here is the drain pump on the bottom of the machine. Strangely for a dishwasher, everything underneath is very clean & free from corrosion.

On removing the securing screw & unlatching the pump from it’s bayonet mount, the impeller instantly tried to make a break for freedom – it has come off the splines of the rotor shaft.

In the past I’ve tried to remove these rotors manually – and totally destroyed the pump in the process. They are usually so well secure that replacement is the only option. This particular one must have vibrated off the shaft somehow.

This repair was easy – removing the rotor from the main pump body & gently drifting the impeller back onto the splines.

Here the pump is reassembled & ready for reinstallation.

On test the pump sounds normal, & works as expected.

Engineer 1 : 0 Throwaway Society

Posted on December 4, 2012March 27, 2017 by The Engineer — Leave a comment

Site Hosting

There have been quite a few updates to the hosting solution for this site, which is hosted locally in my house, from the above setup, in a small comms rack, to a new 22U half rack, with some hardware upgrades to come.

Core switch here has been removed, with the rest of the core network equipment. The site was kept online by a direct connection into the gateway to the intertubes.

New 22U rack, with the core switch, FC switch & management & monitoring server installed.

As I had no rack rails to start with, the servers were placed on the top of the rack to start off, here is the Dell PowerEdge 860 pfSense core router installed, with the initial switch wiring to get the internal core network back online. This machine load balances two connections for an aggregated bandwidth of 140MB/s downstream & 15MB/s upstream.
The tower server behind is the NAS unit that runs the backups of the main & auxiliary webservers.

Still with no rack kits, all the servers are placed on top of the rack, before final installation. This allows running of the network before the rest of the equipment was installed.

The main server & aux server are HP ProLiant DL380 G3 servers, with redundant network connections.

Still to arrive are the final rack kits for the servers & a set of HP BL20p Blade servers, which will be running the sites in the future.

Stay tuned for more updates as they happen!

Posted on November 29, 2012May 12, 2013 by The Engineer — 1 Comment

Brightwell Brightstar II BSL4 Dosing System

Here is an old chemical dosing system for industrial washing machines. These units are 4-pump models, with dual pumpheads. The motors are reversed to operate alternate pumps in the same head.

From 2006, this is a fairly old unit, and made in the UK.

Main controller PCB, with interface to the power electronics via the ribbon cable, an external serial port for programming to it’s left. Powered by an ST microcontroller. The LCD is below this board.

Main power supply, sense input & motor driver boards. The PSU outputs +5v, +12v & +24v. The inputs on the lower left connect to the washing machine & trigger the pumps via the programming on the CPU. The motors are driven by L6202 H-Bridge drivers from ST.

Motor & gearbox assembly on the back of the pumphead. These are 24v DC units with 80RPM gearboxes.

UPDATE:
As it seems to be difficult to find, here is the user manual for this unit:
[download id=”5557″]

Posted on August 26, 2010 by The Engineer — Leave a comment

Nokia 7110

Another phone from the mid 90s. This is the nokia 7110.

Here the slider is open showing the keypad.

Here the battery is removed, a Li-Ion unit.

The battery cell & protection circuit removed from the casing.

This is the rear of the PCB removed from the housing. Data & charging ports on the right hand side f the board.

Front of the PCB with the RF sections at the left hand side & the keypad contacts on the right.

Closeup of the RF sections of the board, big silver rectangular cans are VCO units.

Closeup of the top rear section of the PCB, with SIM cnnector, battery contacts, IR tranciever at the far left. Bottom centre is the external antenna connector.

The logic section of the board, Large chip is CPU, to right of that is the ROM storing the machine code. Other chips are unknown custom parts.

The Mic & the loudspeaker removed from it’s housing.

LCD from the front of the unit, SPI interfaced. Flex PCB also contains the power button, loudspeaker contacts & a temperature sensor.

The scroll wheel removed from the front housing.

Tiny vibration motor removed from the rear housing, alerts the user to a text or phone call.