LED COB Lights - June 2017

In 2017 a new type of LED lights - with the power regulator integrated onto the LED board hit the market. They were cheap and would automatically reduce the light output to limit the temperature to about 100C.

The problem is getting rid of the heat.

COB lighting info

Energy Use, Automotive Pollution and Whatnot

Energy usage was measured using the watts up? standard power meter from Electronic Educational Devices INC http://www.doubleed.com The accuracy is listed as a fraction of the measured value: +/-3% +/-2 digits for loads >10W and +/-5% +/-3 digits for loads <10W.

As of 2007 I started using a UPM EM100 Energy Meter and it seems to be much more accurate at lower power levels.

Power Meters

Belkin Insight

Bullfrog Power provides these to their customers and, as of August 2012, this is the most accurate power meter I have seen. It is the most precise (reading to 0.1W resolution) and works well in cases where the EM100 and Watts Up? fail to give reasonable results.

UPM EM100

This model is readily available in Canada, seems much more accurate (and consistant) than the Watts Up? at low power levels. However, very strange readings were obtained that made me doubt the accuracy of this meter. For instance the BergHOFF Induction cooktop registered 42W (outrageous and highly unlikely) when off, but the Belkin meter registered 1W (quite likely given the EU standby power draw requirements).

Watts Up?

I used one of the Watts Up? power meters when I first started power metering around 2000. It had issues measuring anything < 10W accurately.

Digital "SMART" flexnet isA2 iConA Hydro Meter

In Kitchener/Waterloo these meters seem to be configured to report the real nominal 240V current after displaying LST053 and the KWh used after displaying LST003. Multiply the current by 240V to get the power draw in watts.

Also there is a set of 8 bars below the numbers which function like the old rotating dial. The number of bars per hour is the Watts/hour. Or if you count to get bars/second and then multiple by 3600 that will give Watts. Counting seconds/bar is useful at very low power draws; but the digital readout (mentioned above) is much faster and more accurate.

Analog Hydro Meter

An excellent way to measure the power use of some things, and find vampires is to just use the hydro meter on your house. You need to time how long it takes to do a full, or partial, revolution of the dial.
P = (3600 * dial revolutions * Kh) / revolution_time_seconds
Kh will be indicated on the meter. For low power measurements I typically count how long it takes to do 0.1 or 0.2 of a revolution as that'll be 3 minutes! Note that the meter does become inaccurate at about 15W and less. For low power my meter doesn't seem to turn at all.

I've used the hydro meter to measure the power use of things like the furnace fan, oven and to verify the impact of my 3 biggest vampires (microwave oven, central vacume transformer, power bar for computer and it's extras).

Note - I've seen a brand name P4 computer draw <80W while all other P4's I've seen are >100W. That implies that the efficiency of the power supply is a lot better. From web sites (Wikipedia) efficiency is typically 75%, peaks at about 1/2 the power rating and can be as high as the low 90's.

Belkin Insight Power Meter

Here are some initial tests - comparing the Belkin Insight meter to the EM100.

LED Tubes for T8 Ballasts

As of November 2015 the big box stores have started carrying LED based T8 tubes to replace fluorscent ones.
But this is a complex upgrade as the measurements below attest. The T8 ballasts are constant current power supplies and designed to work with 400V fluorescent tubes. Having the ballast work at a lower voltage (about 45V) and/or lower power is reducing the efficiency of the ballast. Not below and you'll see that going from one tube to two connected to the ballast gives you 2x the light but only a 50% increase in power draw.

Technically one should be able to put multiple Globe LED tubes in series for more light - using one ballast - up to the power limit of the ballast (typically 2 x 32W or 74W). Some ballasts I've used are rated for exactly 2 bulbs while others can take 1 or 2 or 3 and some will work with 17W tubes up to 32W tubes while others are only rated for 25W or 32W tubes. Power supplies (ballast) will drop in efficiency as the power output drops. Note that there are special ballasts made with different "ballast factor" - basically they put out more or less current in order to increase (HO or High Output) or decrease the amount of light emitted and power used.

I have heard of, but not seen for sale, T8 LED tubes that are designed to run off of 120Vac. These Globe tubes will be quickly destroyed if they are connected to 120Vac. There are also LED tubes which are designed to work with an external power supply (like a ballast) that is designed to power LEDs.

Cheap EBay Boost and Buck Converters - Jan 2019

Here is the EBay buck convert specs for the 10A model. It has two small heatsinks on the circuit board sides and looks pretty beefy.

Input voltage: DC 10V-32V.
Output voltage: DC12V-35V(adjustable).
Output current: 10A (MAX).
Input current ：16A (MAX) (more than 10A please strengthen heatsink).
Output power: natural cooling 100W (MAX), strengthening cooling 150W (MAX) real power.
Conversion efficiency: 94% (when Input 19V 2.5A Output 16V, Reference).
Output ripple: 2% (MAX).
Working temperature: industrial (- 40 ° c to +85 °c) ( environmental temperature, more than 40 degrees, please
reduce power to use, or - enhance cooling).
Full-Load temperature: 45 degrees.
No-load current: 25mA typical.
Voltage regulation: ± 0.5%.
Load regulation: ± 0.5%.
Dynamic response speed: 200uS 5%.
Short circuit protection: No (Please installs the fuse or protection circuit at input parts).
Input Reverse protection: No (please comply with a reverse protection or connect a diode at input parts.)

The measurements below were done with a single meter - moving it around to measure both voltages and currents.

Intel i5 8th gen Power Test - July 2018

Below is power draw running Prime 95 to test an i5-8400. With 5 or 6 cores in use the CPU would trip the "core power limited exceeded" and reduce power draw (lower number in brackets) after about 5 seconds of running all cores at full boost speed. HWinfo64.exe reported 68W CPU power draw with 4 cores in use and 81.5W when 6 cores were in use before it tripped.

The ISK 300 case has a very low efficiency power supply

Intel i5 Power Test

For this test I tried 3 different power supplies and unplugged the video card to see the effect on power draw.

System setup common to all tests: Asus P8Z77-V LX motherboard, Intel i5-2320 3.0?GHz CPU, 2 of 4G DDR3 sticks, DVD, chassis fan (aprox 4W by itself with Antec EA380 power supply)

The video card used in 1/2 of the tests is Asus HD7770-DC-1GD5

Video Card Power Draw

This first test uses an Asus M2N SE Plus motherboard with 2G RAM, Antec 80Plus EA380 power supply, LE1640 CPU and 80G SATA HD

Fan Speed and Heat

P4 Fan Speed, July 2010

There are articles on changing the 3-wire P4 fans to run at various speeds. The trick is to run them not on the 12V, but to use 5V or 7V (diff between 5V and 12V). Note that if 7V is used - the ground reference is gone and the BIOS will not be able to measure the fan speed.

To control the speed for some of the measurements, a resistor was put in series with the 12V supply (1W 10%). Here is a comparison:

Voltage	Speed	Resistance	Temp CPU, chassis
2.8GHz Celeron D
12V (stock)	2650 rpm		49 / 36C 40/24C at startup
7V	?? rpm
5V	1850 rpm		59 / 38C
2.8GHz Celeron D Resistor Fan Speed Reduction
12V	1600 rpm	34 ohm	55 / 41C
12V	1760 rpm	25 ohm	54 / 42C
12V	1875 rpm	20 ohm	53 / 41C
2.4GHz P4 Resistor Fan Speed Reduction
12V	1900 rpm	20 ohm	38 / 37C

P3 Fan Speed

Many of my Pentium III computers are quite noisy and so I experimented with adding a resistor in series with the 12V line to slow down the fan. I tried resistors from 40 ohms up and was not happy with the noise reduction till I hit 160 ohms.

AMD vs Intel - Oct 2017

I am considering a new home computer and did some [rought] performance measurements and [importantly] power draw. We want a computer that doesn't use much power and is small and is ok for gaming like Minecraft.

Conclusion. The Ryzen3 is excellent value for money and, based upon online tests it's performance is similar to the i3-8100 with a GTX 1030 GPU. But we don't need that performance and are not willing to have 2x the power draw all the time.

Trying various hard drives to check their power draw

AMD AM3 CPU Power Use - June 2012

Test computer is Asus M4A785T-M motherboard with 4G of DDR3, one SATA HD and Antec 80Plus Bronze power supply

CPU Power Use - Sept 2008

I've seen a lot of confusing numbers from computer systems - Dell and IBM computers drawing around 60W compared to 100W for other P4 computers. I initially assumed that this was due to differences in power supply efficiency - but this wasn't testable (non-standard power supplies).

So when I did get some computers for which I could swap the power suplies (and had low power use) I verified that it wasn't 80Plus efficient power supplies which caused the lower power use, but that it was the CPU.

Computer Power Use - April 2014

I decided to evaluate the Celeron CPUs as a way to reduce power and cost of computer purchases. But first I had to test the effect on performance for Quartus compiles (1, 2, 4 or 8 simultaneous compiles). Performance for Windows is worse as it's all I/O bound on the Samba file server.

Computer Power Use - March 2016

This is a quick test with some 6th generation Intel CPUs. The power supply is Antec MT-350 80Plus, motherboard SuperMicro X11SSZ-TLN4F

CPU	Off	In BIOS	In Linux	running MPrime	ECE 327 FPGA Compile Time (seconds) 1x, 2x, 4x, 8x
i3-6100 3.7GHz	8.6W	39.0W	26.0W	?W	-
i5-6600 3.3GHz (3.9GHz max)	8.6W	42.6W	35.8W	?W	-
i7-7600 3.4GHz (4.0GHz max)	8.6W	41.6W	25.8W	?W	-

Computer Power Use Summary

Modern PCI-Ex video cards seem to draw 20W more than old PCI video cards (inefficient power supply).

CPU Power Use - Sept 2008

I've seen a lot of confusing numbers from computer systems - Dell and IBM computers drawing around 60W compared to 100W for other P4 computers. I initially assumed that this was due to differences in power supply efficiency - but this wasn't testable (non-standard power supplies).

So when I did get some computers for which I could swap the power suplies (and had low power use) I verified that it wasn't 80Plus efficient power supplies which caused the lower power use, but that it was the CPU.

Computer Power Use Summary

Modern PCI-Ex video cards seem to draw 20W more than old PCI video cards (inefficient power supply).

80Plus power supplies such as the Antec EarthWatts and, at least, 2007 vintage Antec TruPower Tri power supplies offer 20% to 25% reduction in power use when the computer is on. On computers which are on 24x7 this means a payback on upgrading is just over a year at $0.10/kWh.

Intel Core 2 Duo microprocessors are power hogs. AMD Athlon 64 and 64x2 microprocessors use signif. less power at the same level of performance. Typical AMD systems I have use 60W under normal use; while Intel Core processors are around 120W - assuming inefficient power supplies.

Proper choice of computer motherboards can make a signif. difference. I've measured fancy "Delux" motherboards with WIFI which draw 20W more than other motherboards and this is signif. if one is aiming for a sub 60W power draw (which is easily achieved).

Car - Gasoline Consumption - MaxiTrip 1.2 TP100

Using a MaxiTrip 1.2 TP 100 I measured the following gasoline consumption on a 2005 Toyota Echo (in Aug 2011):
cold idle: 1.8L/h, 20C ambient, 1250 rpm
warm idle: 0.5 to 0.6L/h, 26C ambient, 650 rpm, coolant 84C

I hoped to have a flat road on a windless day to measure the fuel consuption at 70km/h in 3rd, 4th and 5th gear to see the effect of the gear ratio. However, it was very difficult to hold a steady fuel consumption - so much so that I gave up. Holding a steady speed wasn't the issue - but the gas consuption bounced between 2.4 and 4 L/100km while holding an apparently steady speed. Perhaps by logging the data and doing some averageing it would be possible.

I would like to remove the power steering pump and see the effect upon fuel consumption as it's apparently signif. However, that first requires removing the serpentine belt and tensioning the belts appears difficult.

Toyota Echo Power Steering Delete - August 2014

Note that our Echo is running oversized tires from our previous car 185/70R14 (76.0" circumference, 24.2" diameter). The stock sizes are 175/65R14 (72.1" circumference, 23.0" diameter) and 185/60R15 (74.5" circumference, 23.7" diameter). Our car shipped with the 185/60R15 but was sold to us with 175/64R14 tires. Tire size calculations are from http://tire-size-conversion.com/tire-size-calculator/

Comments online indicate a 7 to 10% or 2 to 3mpg increase for a BMW E30 and an increase in the idle fuel consumption of 0.07 gpm (from 0.19gpm to 0.12 gpm). Our car is about 0.55 L/h or 0.14 gph (US gallons per hour) with power steering and 0.5 L/hour without for a 10% reduction of fuel consumption at hot idle. Summer highway driving (trip to and around the east coast of Canada) shows a 5% reduction in fuel consumption.

Gasoline contains 33 kWh/US gallon. One hp is 745W and gasoline engines are typically 20% efficient. That calculates to be 2.2hp for 1L/hour consumption. Given the change in fuel consumption the power draw of the power steering is about 0.05 L/h or 0.11 hp (hot idle) and 0.28 L/h or 0.62 hp (highway average). Highway fuel consumption is around 5.5L/100km and so at 100 km/h that equates to 12.1 hp.

Data Power-De-Steering

2014: 6538km 340.6L  5.21 L/100km power steering pulled winter tires

 2013 172.7L 3341km 5.44  Stock, Majestic tires assumed 5% smaller by tire size calc.

 2012 239.7L  4721km  5.33L/100km  Stock, Majestic tires assumed 5% smaller by tire size calc.

 2011 145.3L  2828km  5.41L/100km  Stock, Majestic tires assumed 5% smaller by tire size calc.

Effects of De-Power Steering

I've cut the power steering belt and found the steering to be "different". Low speed steering requires more force; but I can parallel park and back into the driveway easily enough. The big change is the centering force on the steering wheel when maintaining a constant turn - but you get used to that. Three point turns are the hardest.

After removing power steering fluid the effect was: ??

Links For Deleting Power Steering

• EcoModder Power Steering Delete - aprox 2mpg improvement
• Toyota Echo power steering delete comments Comments that since the PS pump turns easily at low speeds that it is unlikely that there are fuel savings of note.

(Un)Powered Steering

Various people have mentioned that removing the power steering pump reduces fuel consumption at idle and for in-town driving. When a car is equiped with power steering the gear ratio is modified so that, with power steering, the wheels would turn more. This means that removing the power steering (by removing the belt) will result in steering that takes much more effort at low speeds. This is easy to verify - when driving on in a safe place, turn off the engine and see how much force it takes to steer. In particular note the steering effort at 5km/h or 3mph.

There are options to remove the power steering.

• The steering rack can be changed; but it's often very hard to find ones as very few cars are sold without power steering.
• The power steering pump may be replaced with an electric one. Many people building EVs do this Jeep EV conversion using the Toyota MR2 electric pump. Here is MR2 Electric Power Steering

Oversized Tires

With our 1997 Escort I went with oversized tires to lower the engine RPM - as well as get cheaper tires. So instead of 185/65R14 I bought 185/70R14. 7 years later when the Escort had a failing transmission, was 12 years old, and not worth fixing the tires moved to our 2005 Toyota Echo.

Here is a comparison of tires for the Echo. Note that 175/65R14 is stock, 185/60R15 is an option (which our car came with - but the seller swapped out the 15" rims, put in 14" ones with stock no-season tires.

Note that sizes below are from http://www.1010tires.com/tiresizecalculator.asp
I assume that the calculated sizes are for summer tires.

Speedometer Calibration Measurements

In an attempt to calibrate the speedo in the '05 Echo we ran tests using a GPS:
P185/70R14 tires, at 100km/h, the GPS indicates ~98.5 km/h or 1.5% lower speed than the speedo!
P175/65R14 tires, at 100km/h, the GPS indicates ~92 km/h for

Our Cars Over the Decades

2016 Kia Soul EV Performance

Here is trip data from long drives.

We use the FLO charging App but PlugShare has really good maps.
ABRP (phone App) is ABetterRoutePlanner.com which is an exceptional route planner. Just be sure to set the temperature as range loss in the winter is signif.

Winter 2019

0C weather, car in "eco mode" with snow tires, 18kWh/100km driving at < 80km/h, 23 kWh/100km at 110 km/h highway.
Range: 125 km but do not go < 15 km range and charging > 90% gets very very slow. Charging ~124A at 380V when charge is < 70%, but only 20A around 90%
Electronics draws ~0.7 kW & climate control on minimum about 0.7 kW or 2kW if cabin is kept hot

Car Milage

2009 Nissan Versa 1.6L? 1.8L? automatic transmission, driven 335km in Vancouver May 2010. 29.324L of gas used in mostly highway driving. That gives 8.8L/100km when the car is rated at about 8 city and 6 highway. This car had a nice low idle and generally worked well, but the milage sucks!

Here is the milage of our 3 cars. The Chevy Sprint we had for 12 years. The Ford Escort Wagoon, aka "Black Beast from Hell", we had from it's 5th to it's 12th year (transmission failure) and data from the 2005 Toyota Echo is lacking as we've only hhad the car since xmas 2009. Note that the tires on the Escort were about 3% oversized and that's been factored in to improve the milage 3%.

Here is a graph of the milage of various cars we've had in L per 100km and mpg (Imperal gallons)

Our Ford Escort had an automatic transmission. A friend with a 1999 Escort Wagon with a 5-speed reports the following milage:
I am currently getting 32 mpg with our Escort in the winter.
When we lived in the country, we were getting 41 mpg average annually.
Monthly averages ranged from 37mpg in winter to 44mpg in summer. Usually on a trip in the summer we would get close to 50mpg.
Monthly averages are ranging from 31mpg in winter to 42mpg in summer.

Emissions Testing

One of the better things to happen is that many places now require cars to be emission tested. One rough stat says that 10% of the cars produce 90% of the pollution because their emissions equipment is broken or because they're burning oil, etc. Please email me testing stats for your vehicle if you are in Ontario. Here are some stats:

Vehicle	Age	Engine	HC (ppm)	CO (test/max)	NO (test/max)	Dilution
2005 Toyota Echo Sedan (5 spd)	5	1.5L	5 / 66	0.01 / 0.37	2 / 505	??		14.21, 2568 rpm
			2005 Toyota Echo Sedan (5 spd)	6	1.5L	2 / 66	0.00 / 0.37	1 / 505	??		14.7, 1804 rpm
1 / 150	0.00 / 0.7	N.A.				??		13.8, Idle 1804 rpm
1990 Eagle Vista 4WD idle (full time 4WD)	11	2.0L	105 / 200	0.07 / 1.00	N.A.	14.97
1990 Eagle Vista 4WD 2500 rpm			55 / 200	0.4 / 1.00	N.A.	15.1
1991 Chev Sprint	10	1.0L	102 / 116	0.12 / 0.66	84 / 1380	15.5
	11		37 / 116	0.02 / 0.66	82 / 1380	14.02		2320 rpm
			75 / 200	0.05 / 1.00	N.A.	14		Idle 814 rpm
1992 Subaru Justy 4WD Bad Oxygen sensor	9	1.2L	0 / 106	0 / 0.59	1457 / 827	14.8
1992 Subaru Justy 4WD Good Oxygen sensor			6 / 106	0.01 / 0.59	139 / 827	15.01
	11		5 / 94	0.02 / 0.52	260 / 732	??	2539 rpm
			9 / 200	0.03 / 1.0	N.A.	??	813 rpm
1997 Ford Escort wagon (4 spd auto)	5.5	2.0L	18 / 86	0.14 / 0.48	184 / 656	14.9		1674 rpm
			14 / 200	0.01 / 1.00	N.A.	14.8		Idle 1032 rpm
	6.5	2.0L	5 / 76	0.01 / 0.42	238 / 580	13.11		1884 rpm
			18 / 200	0.09 / 1.00	N.A.	12.77		Idle 738 rpm
	7.5	2.0L	0 / 66	0.09 / 0.37	6 / 505	14.9		1877 rpm
			0 / 200	0.21 / 1.00	N.A.	15.2		Idle 921 rpm
	9.5	2.0L	5 / 68	0.01 / 0.38	5 / 526	14.9		1813 rpm Gas cap replaced
			5 / 200	0.02 / 1.00	N.A.	14.9		Idle 988 rpm
1999 Ford Escort wagon (5 spd)	4	2.0L	11 / 76	0.05 / 0.42	11 / 580	??		2753 rpm
			12 / 150	0.3 / 0.7	N.A.	??		Idle 768 rpm
1998 Chrysler Truck	3	5.3L	19 / 92	0.06 / 0.51	415 / 1269	15.1

Efficient Lighting

Light Source	Power	Brightness Lumens	Lumens/Watt	Cost	Lifetime (hours)	Comments
T8 4' fluorscent tube	32W	2950	92	$4	20,000 hour life	instant start 5000K temperature is too blue/harsh
GE Standard CF (Compact Flourscent)	15W	700	47	$10.50	6,000 hour life	fairly long - doesn't fit all lamps
Soft white	40W	460	12	?	1,500
Long life	40W	420	11	?	3,000
Soft white	60W	840	14	?	1,000
Halogen	60W	900	15	?	3,000	www.lighting.philips.com/nam
Noma CF	10W	520	52	$3	8,000	very short, time delayed start, 2700k (rated for -15C or warmer)
Noma CF	13W	800	61	$2ea (6-pack)	8,000	very short, very narrow base
Ikea CF	11W	?	?	$5	life?	short, time delayed start, harsh color
Sylvania CF	30W	2000	67	10.00	6,000 hour life	fat and long
Sylvania CF	13W	800	62		8,000 hour life	very short and thin
Sylvania CF	9W	450	50	13.00 for 2-pack	8,000 hour life	small socket, starts at reduced brightness
Sylvania CF	4W	160	40	8.00	6,000 hour life	small socket, starts at reduced brightness
IKEA CF	5W	170	34	7.50	10,000 hour life	small socket, starts at reduced brightness
IKEA CF	7W	300	42	7.00	10,000 hour life	small socket, starts at reduced brightness
Home Hardware CF	25W	1500	60	??	10,000 hour life	short, but fat - fits many lamps
Commercial Elec. (Home Depot)CF	23W	1600	70	??	10,000 hour life	short, but fat - fits many lamps, instant start
Home Hardware CF	15W	900	60	$5	6,000 hour life	very short - fits all lamps (rated as 70W equivalent)
Philips Marathon CF	15W	900	60	$22 aprox	10,000 hour life	very short - fits all lamps
Philips Marathon CF	20W	1200	60	$25 aprox	10,000 hour life	short - fits many lamps
Pricemark [Sylvania?] spiral CF	15W (12W actual draw)	?	?	$8 aprox	10,000 hour life	short but fat,Starts quickly
Pricemark [Sylvania?] spiral CF	20W (14W actual draw), 23W	?	?	$8 aprox	10,000 hour life	short but fat,Starts quickly
incandescent filtered with a red mask ie emergency or car tail light	140	460	3	?
High Efficiency LED	NA	NA	80 (1998 vintage)	?	very long
High Efficiency OLED	NA	NA	50 (2000 vintage, achieved in production Dec 2005)	?	very long	Organic LED

CFL Bulb Type	Comment
Phillips Marathon	These bulbs have a noticable time delay to turn on and take about 30 seconds to achieve full brightness
Pricemark (Osram / Sylvania?)	They turn on very quickly at full brightness. I've heard that they don't last as long.

Flexnet 530-X Power Meter Information

This is various info I harvested in an attempt to get information about these meters.

Gov. of Can. info (partially quoted below): www.ic.gc.ca/pics/lm/electric/ae/1611r1.pdf The iSA2 meter with Flexnet 530-X is a solid state bidirectional meter approved for metering energy
Energy registered in the forward direction is preceded by the code LST003 prior to displaying the registered value on the display. Similarly energy registered in the reverse direction is preceded by the code LST004.
The iSA2 meter is equipped with a liquid crystal display (LCD). The LCD can be programmed to provide additional information. Information related to the additional display features can be found in the iConA technical manual. [can't find that online]
The iSA2 meter is also known as the iConA.
The raw energy calculation is carried out to a resolution of 1*10-9 Wh and is transmitted by digital serial interface over the Flat Flex cable (FFC) to the register display board, once per line cycle.
A test pulse signal is generated by the metrology chip, which is programmed according to the customer.s preferences, from 1 Wh/pulse to 32 Wh/pulse. This signal is routed to the infrared (IR) test pulse LED on the register display board.
The FlexNet radio can be configured to transmit readings in the following four resolutions: 1 kWh, 0.1 kWh, 0.01 kWh and 0.001 kWh.
The typical resolution used for transmitting on-air meter readings is 1 kWh.
Mfg site: http://www.sensus.com/web/usca/electric/product-line/electricity-metering/product/icon-a-residential-meter

Back to my home page