This is only a preview of the March 2020 issue of Silicon Chip. You can view 37 of the 112 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to ""True valve sound" Guitar Overdrive & Distortion Pedal":
Items relevant to "Programmable Thermal Control with a Peltier":
Items relevant to "Building Subwoofers for our new “Bookshelf” Speakers":
Purchase a printed copy of this issue for $10.00. |
Want to “probe” a veHicle’s EHT?
You’ll need this
by Dr Hugo Holden
1000:1 AC EHT Probe
for Ignition Systems
It’s surprisingly tough to measure the actual output voltage of an
automotive (or aircraft/boat) ignition system. You can’t use a standard
high-voltage probe because the voltages involved are way too high; they
can exceed 50kV! Nor can you use a standard EHT probe because these
are designed for DC use and will severely distort a fast-rising (or falling)
AC waveform. This simple design is the answer.
P
eak voltages from the ignition
coil secondary windings are typically in the range of 10-30kV but
can be higher – and can exceed 50kV
in some circumstances.
These high voltages occur for a very
brief time across a spark plug’s terminals before spark ionisation, or under
any test condition when the spark plug
is not connected.
This ‘open-circuit’ coil secondary
voltage value is an important ignition
system parameter.
The rate that the voltage increases
with time is another important parameter. A fast rise time to the spark ionisation voltage is thought to be beneficial
90
Silicon Chip
in overcoming the ohmic resistance of
fouled spark plugs, because less energy
is dissipated due to a shorter time interval before spark ionisation.
Also, a certain voltage threshold is
always required to initiate spark ionisation (the spark’s early phase, known
as phase one).
This voltage depends on the spark
plug’s gap and the composition of the
gases and the gas pressure and temperature between the gap.
However, during the spark’s burn
time (phase two), the spark plasma
has a low impedance, and the spark
gap voltage is relatively low – just 30V
with some aviation spark plugs and
Australia’s electronics magazine
around 1000V for a typical automotive spark plug.
By comparison, in free air (ie, outside the cylinder and not under pressure, a typical automotive spark plug
has a gap voltage of around 600V.
To measure the high initial prespark ionisation peak voltage or the
open-circuit output voltage of the
spark generating system, you need a
special probe with a flat frequency response also having the ability to avoid
corona discharge, which is a big problem with potentials over 30kV.
Making the measurement
Ideally, we want to use an oscillosiliconchip.com.au
passed their maximum ratings. Worse,
the probe tips do not easily interface
with insulated spark plug connectors,
which are the best way to link up circuits running at these high voltages.
Also, the probe needs to have a total load resistance of at least 50MΩ,
so there is little loading of the system
being tested. This equates to 1kΩ/V
for a 50kV test. A 200MΩ load is feasible, yielding 4000Ω/V, however, the
higher the resistance, the more lowpass filtering effects occur due to distributed capacitance. High-frequency
compensation, therefore, becomes a
little more difficult.
A high series resistance value leads
to a low-pass filter effect, because even
just 1pF of stray capacitance results
in a significant low-pass filter being
formed. For example, with 100MΩ and
1pF, the filter created roll-off (-3dB)
point is only 1.6KHz.
scope to capture these spark events.
So we need to scale down the typical
30kV open-circuit voltage down to say
30V (ie, dividing it by a factor of 1000)
and feed it into the typical 1MΩ//15pF
input impedance of a scope. This
needs to be done while maintaining
a broad frequency response, so that
the recorded waveform maintains its
original shape.
We also need to make sure that the
oscilloscope (and user!) is not at risk
of damage from these high voltages.
While inexpensive high-voltage or
“EHT Probes” are generally available
(eg, to measure CRT anode voltages),
they are meant for measuring static
DC voltages. We published a design
to build a low-cost EHT probe in the
April 2010 issue (siliconchip.com.au/
Article/121). That design is capable of
measuring up to about 25kV.
But this type of EHT probe gives
very low false readings on fast risetime waveforms; the rise times of ignition system secondaries are in the
microsecond range, and the high-order Fourier components can be in the
100kHz to 1MHz range.
High-voltage compensated probes
which can handle 40kV are available,
but they are hard to find and expensive. Also, on some common ignition
system tests, they could be pushed
My probe design
Fig.1 and the photo at left show my
probe design. I’m using a spark plug
as a feed-in element, by trimming the
metal part away. Bramite (similar to
Garolite) was used as insulating material along with PVC tubing, and parts
of the assembly are glued with Torr
Seal from Varian Vacuum Technologies (a white epoxy resin which is also
EHT CONNECTION
Fig.1: this somewhat simplified diagram shows the
main part of the EHT probe. It consists of a high-power
resistor immersed in oil within a section of PVC pipe,
and a parallel brass rod which forms a distributed
compensation capacitor. The high-power resistor forms a
voltage divider in combination with the smaller resistors
below, while the distributed capacitance also forms a
divider in combination with the 1300pF capacitor. This
can be made up of two or more lower-value capacitors in
parallel.
SPARK PLUG WITH OUTER
METAL REMOVED
TRANSFORMER OIL
BRASS ROD
FORMS
COMPENSATION
CAPACITOR
SPACER
50M 50kV 15.5W 1%
RESISTOR
BNC-BNC COAXIAL
CABLE, 1.5m LONG
(RG179 COAX,
C = 95pF)
(OHMITE MOX96025005FVE)
SCOPE OR DSO
16mm ID PVC TUBE
& JOINERS
SCOPE INPUT
CIRCUIT
SPACER
FILLER PLUG
SC
56k
3W
1300pF
R.cal
1M
A
ZD1
75V
1M
15pF
K K
A
ZD2
75V
20 1 9
siliconchip.com.au
Australia’s electronics magazine
March 2020 91
Fig.2: the upper trace shows the 100V peak-to-peak square wave I’m applying
to my prototype while the lower trace shows the resulting 100mV peak-topeak waveform at the output. You can see from its shape, with no apparent
undershoot or overshoot, that the probe is correctly compensated.
an excellent insulator).
The input capacitance of the probe
is a little lower, at about 2pF, compared to a spark plug which is typically around 8-10pF. The typical
output capacitance of an automotive
ignition coil is around 50pF, and the
HT wiring contributes another 10pF
or thereabouts.
As shown in the diagram and the
photo, the main body of the probe is
made from PVC pipe. This is filled
with oil and houses the 50kV resistor.
Without the dielectric oil, the corona
discharge becomes very difficult at
peak voltages over 30kV. The oil solves
this problem.
The main compensation capacitor
is a brass rod which runs alongside
this oil-filled tube, acting as a highfrequency coupling capacitor distributed along the length of the resistor
by proximity.
It’s connected directly to the lowvoltage end of the 50kV resistor and
supported by the upper insulating
plate.
It must be mounted parallel with the
50kV resistor and centred 30mm from
the middle of the PVC pipe for correct
operation. That means there will be
around 18mm from the edge of the rod
to the edge of the pipe, depending on
the exact outer diameter of the pipe.
92
Silicon Chip
This dimension is critical for correct operation.
There are effectively three resistors
in parallel at the bottom of the divider:
56kΩ, 1MΩ and the 1MΩ input impedance of the scope. These combine
with the 50MΩ resistor to provide the
1000:1 division ratio at DC and low
frequencies.
At higher frequencies, the compensation capacitor and 1300pF of capacitance form a capacitive voltage divider
with a similar ratio, in parallel with
the resistive divider.
The 75V zener diodes were added
just in case any corona discharge occurs accidentally, which could harm
the oscilloscope input amplifiers.
Enlarge the central holes in the
22mm and 44mm discs so that the body
of the spark plug will fit through both.
Now make a hole in the middle of
one of the PVC end caps for the spark
plug body to pass through, plus a small
hole in the other end cap for the resistor lead, as well as a larger one, to suit
the filler plug.
Use the end cap with two holes as a
template to trace them out in the middle of the brass sheet, which will later
be bent into a bracket and attached to
this end cap.
Glue the 22mm and 44mm discs
together, and glue the PVC endcap to
the bottom of the 44mm disc. Now
place one of the spacer discs over one
of the resistor leads and feed this lead
up through the PVC endcap and two
round plates. Cut this lead short, then
solder it to the tip of the spark plug.
Next, pull the resistor back down so
that the spark plug is reaching down
inside the PVC endcap and seal around
the spark plug using the Torr Seal
epoxy, so that it is oil-tight.
Spread some epoxy all around the
edge of the spacer and then slide the
PVC pipe over the resistor. Spread
a generous amount of epoxy around
the end of the pipe, then push it into
the end cap firmly. Allow the epoxy
to set, with the pipe’ right-way-up’ so
that the upper spacer is resting on top
of the resistor body.
Place the second spacer over the
remaining resistor lead, spread some
epoxy all around its edge and push
it up into the pipe as far as it will go.
Make sure that the resistor is fully
wedged between the two spacers so
Construction
Start by using a 16mm hole saw to
cut two round pieces of Bramite. Place
the Bramite sheet on a sheet of scrap
timber which is firmly supported at
either end, so that you drill won’t go
into anything critical while doing this.
The resistor leads can pass through the
central guide holes.
Cut three larger discs from the Bramite using much larger hole saws; one
around 22mm in diameter, one around
44mm and one around 64mm. (Tip:
you can buy a hole saw set which will
have most of the required sizes).
Australia’s electronics magazine
Fig.3: Fourier theory says that a
square wave can be formed from an
infinite number of sinewaves with
different amplitudes and phases.
The higher-frequency sinewave
components have lower and lower
amplitudes as the frequency increases.
This means that you can tell whether
the frequency response of a device
is flat by feeding a square wave into
its input and looking at the resulting
shape at the output.
siliconchip.com.au
Screw the plug into the bung to seal
it up and clean up any oil that squirts
out around the edges. Do it up tight
so it won’t accidentally come loose;
that could be messy! It’s a good idea
to silicone around and over the bung
as insurance against oil leaking out.
By the way, if you can’t get a proper oil filler bung, you could consider
drilling and tapping the end cap to
accept a regular screw thread, but if
you’re going to take that approach, it
may be necessary to thicken the end
cap material by gluing one or more
PVC discs inside it, to give enough
‘meat’ for the screw to form a good seal.
Final assembly
Fig.5: the output voltage of an unloaded ignition coil being driven by a Tung-Sol
EI-4 capacitor-discharge ignition (CDI) system, captured using the probe described
here. No sparks or corona discharges are occurring, resulting in an extremely
high peak voltage of -40kV, which matches well to the expected peak of 39.6kV as
determined by the coil turns ratio and primary voltage. After the initial discharge,
the residual coil magnetic field energy and energy stored in the coil’s distributed
capacity decays away in an oscillatory manner, due to the self-resonance of the
ignition coil.
it won’t move later.
It’s also a good idea to push some
epoxy into the hole surrounding the
resistor lead, if you can get in there.
Up-end the whole assembly, resting
it on two equally tall objects on either
side, so that it sits vertically, until the
epoxy on the second spacer has set.
CORRECT
COMPENSATION
UNDER
COMPENSATED
OVER
COMPENSATED
SC
20 1 9
Fig.4: compare your calibration waveform to the three shapes shown here.
If it looks nice and square, like the
one at the top, you’re finished. If it’s
rounded (under-compensated), reduce
the value of the 1300pF capacitor. If
it has overshoot (over-compensated),
increase the value of that capacitor
(eg, by adding a low-value ceramic
capacitor in parallel).
siliconchip.com.au
Bend the remaining resistor lead so
that it will pass through the small hole
in the end cap that you drilled earlier,
once the end cap is fitted onto the end
of the pipe. It should be long enough to
reach through the cap; if not, extend it
by soldering on some stiff wire. Glue
on the end cap using more epoxy, and
also seal around the wire exit.
Now is also a good time to coat
the inside of the hole you made for
the bung with epoxy and press it in.
Make sure it will be oil-tight when the
epoxy sets.
Now up-end the assembly, again
resting it on a couple of blocks and
let the epoxy set. The next step is to
pour a little transformer oil into the
oil filler hole. Wait a few minutes and
check that you don’t have any oil leaking out anywhere. If you do, you will
need to drain it, clean it up and apply
some more epoxy to seal the leak areas. Then try filling it with oil again.
If it looks good, add a bit more oil,
then a bit more, then start pouring it
in slowly until the pipe is almost full
of oil. Wait a while for any air bubbles
to surface, then add a little oil until it’s
just about full. Leave a small air bubble
inside to allow for thermal expansion.
Australia’s electronics magazine
Drill the holes you marked earlier
in the brass sheet and bend it to form
a bracket to support the PVC pipe (see
photo). Also, drill a hole to fit the BNC
socket next to the pipe. Make sure
the resistor wire end exiting the pipe
won’t touch this, as the bracket will
be Earthed.
I glued a 50mm wide sheet of brass
foil around the bottom of the tube so
that I could solder it to the bracket;
however, you could also use a section
of large diameter brass tube or come
up with some other arrangement to
attach the bottom of the tube to the
supporting bracket.
Once it has been secured, bend the
projecting resistor lead over (making sure it isn’t contacting any of the
metalwork), trim it and solder it to
the central pin of the BNC socket. If
you’ve used brass foil or a brass tube
at the base of the PVC pipe, as I did,
you will need to solder an insulated
wire to the resistor lead instead and
feed it through a hole in the supporting tube, then seal it up.
Now solder the few other electronic
components between the BNC ground
tab and the end of the power resistor lead, with the zener diodes wired
back-to-back across them. See the accompanying photo, which shows how
I arranged the components.
Try to leave the 1MΩ resistor and
100pF capacitor accessible, as you
may need to replace these with different components during calibration.
Now cut the brass rod so that it’s
just a bit too long to fit between the
top and bottom plates. As you can
see from the photo, I made a bracket
from a small brass plate and some
brass tubing. This had the advantage
of both holding the rod in place and
March 2020 93
Standard
20mm joiner
(approx.
25mm OD)
10mm
5mm diam
The view of
the base of
the probe
from the
“front” side
showing the
point-to-point
wiring, along
with the
BNC output
terminal
and...
20mm OD
electrical
conduit
20mm
cL
15mm
29mm
29mm
also providing a convenient place to
make the electrical connection.
However you do it, make sure the
rod is fixed in place and parallel with
the PVC pipe, with the dimensions
described above - the critical one being the 30mm from the centre of the
PVC pipe/resistor to the centre of the
brass rod.
I held the top of the brass rod in
place by inserting it into a blind (shallow) hole drilled in the inside face of
the top plate. I soldered the 5mm rod
to a length of 7mm diameter rod, to
make it easier to tap the bottom of the
rod for an M3 screw to make the electrical connection. I then soldered this
7mm rod to the bracket, as shown in
the photos. But there are other ways
of doing this.
Regardless, you will need to run
a wire from the bottom of the rod to
the bottom lead of the resistor in the
PVC pipe and solder or clamp it at
both ends.
Calibration
You should find that your probe
provides very close to a 1000:1 division ratio when connected to a device
with a 1MΩ input impedance. Note
that many DMMs have a higher input
94
Silicon Chip
impedance than this, at least when
measuring volts. If you want to use a
DMM for calibration and it has a 10MΩ
input impedance, clip a 1.1MΩ resistor
across the DMM’s leads for the tests.
For the first test, use a relatively
high voltage DC source such as a 48V
supply or a bench supply wound up
to maximum.
Measure the voltage across the supply outputs using your DMM and write
it down, then connect the probe tip to
the + supply and the output ground
to the – supply. Measure the voltage
at the BNC cable tip, keeping in mind
the above comments about input impedance.
You should get very close to
1/1000th of the voltage. For example,
if your test supply measures 48.4V, you
should get 48.4mV at the probe output.
If you get a higher value, you can
slightly reduce the value of the 1MΩ
resistor in the probe to compensate.
Similarly, if its output is low, slightly
increase the value of the 1MΩ resistor.
AC calibration is just as, if not more
critical than DC calibration. For this,
you need a function or pulse generator
capable of producing a 1kHz square
wave of similar.
Ideally, it should be able to deliver
Australia’s electronics magazine
50mm
7mm diam
cL
(Above): looking at the
underside of the probe. It’s
attached to an 80mm diameter
disc of Bramite or similar
insulation, which is
in turn mounted on
a much larger sheet
for working
stability. All holes
should be
countersunk.
...here’s the
view from
the opposite
side. Note
the brass rod
compensation
capacitor.
a square wave of around 100V peakto-peak. I used a Tektronix PG506 calibration generator.
If you only have a low-voltage pulse
generator, you should build our Precision Signal Amplifier from the October 2019 issue (siliconchip.com.au/
Article/12025). It’s a simple and relatively cheap device which can boost
the output of a function generator up
to about 30V peak-to-peak, just sufficient for this calibration procedure.
The AC calibration is set by the
1300pF (1200pF || 100pF) capacitor.
This forms a divider with the brass
rod, which acts as an HF coupling capacitor distributed along the length of
the resistor. Fig.2 shows my probe’s
square wave response with the probe
plugged into the input of a Tektronix
2465B scope.
The upper trace is the input voltage which is a near 1kHz, 100V peakto-peak square wave from the PG506
generator. The lower trace is the output
voltage which is close to 100mV peakto-peak. Without the compensation capacitor network consisting of the brass
rod and 1300pF capacitor, the output
waveform bears little resemblance to
the input waveform and looks more
like a sinewave.
siliconchip.com.au
I used sinewave testing to determine
that the probe has a flat response from
DC to over 1.5MHz. The highest frequency of interest in an automotive
ignition system is about 300kHz.
But you don’t need a sinewave
sweep to check the frequency response; a single square wave test
will do the job much more easily and
quickly.
According to the Fourier theorem,
a square wave or rectangular wave is
composed of a fundamental frequency
and a plethora of harmonic frequencies, the higher-order ones being responsible for the rapid rise on the
leading edge of the waveform.
This is shown in the simplified diagram of Fig.3.
Therefore, if a square wave is passed
through the system, it is immediately
apparent from its shape at the output whether the frequency response
across a broad range of frequencies is
flat or not.
If the HF response is limited, the
fast rising and falling edges are rolled
off. If the rising and falling edges are
peaked, then the HF response is excessive. If the flat top of the wave has distortions or bends or tilts, then the medium frequency (MF) or LF responses
are abnormal.
Most oscilloscopes have a calibration output voltage which is a square
wave, so that the compensation capacitor on the 10:1 probe being used can
be set for a flat response. The procedure for calibrating this probe is much
the same, except that you may need
to replace the 100pF capacitor with
a higher or lower value to achieve
calibration.
Fig.4 shows what square waves look
like at the output of a probe which is
correctly compensated, under-compensated or over-compensated.
If your square wave looks like the
one in the middle, you need to reduce
the value of the 100pF capacitor (try
removing it entirely first).
If it looks like the one at the bottom,
then you need to increase the value of
the 100pF capacitor or connect another
low-value 100V capacitor in parallel.
As noted in the parts list, it’s best to
use NP0/C0G ceramic capacitors here
as they do not change in value with
temperature.
Otherwise, your probe’s calibration
could be different on cold and hot
days. They’re also extremely linear for
the best possible performance.
SC
siliconchip.com.au
Parts list – 1000:1 AC EHT Ignition Probe
1 spark plug
1 200mm length of 20mm outside diameter PVC conduit
2 PVC end caps to suit conduit
1 450mm x 225mm x 6mm (or similar) sheet of Bramite (#)
1 100 x 50mm sheet of 1mm thick brass plate
1 250mm-long, 5mm diameter brass rod
(or 1 200mm long, 5mm diameter rod and 1 50mm long, 7mm diameter rod)
1 1/8” NPT female bung and matching plug
1 50MΩ 50kV 15.5W 1% resistor (Ohmite MOX96025005FVE)
[Digi-key, Mouser]
1 1200pF 100V NP0/C0G ceramic capacitor [eg, Kemet C322C122J1G5TA]
1 100pF 100V NP0/C0G ceramic capacitor [eg, AVX SR151A101JAR]
1 56kΩ 1% 3W resistor [eg, Stackpole RSMF3JT56K0]
1 1MΩ 1% 0.25W resistor
2 75V 1W zener diodes
1 chassis-mount BNC socket
1 1.5m-long RG179 coaxial cable fitted with BNC plugs at each end
Various brass machine screws, washers and nuts
1 tube of Torr Seal epoxy resin
1 one-litre bottle or can of transformer oil
(#) Bramite is a material used as the backboard in meter boxes. It should be
available from electrical wholesalers.
It’s a request we’ve had many, many times in the past:
Can I buy back issues in PDF format?
Sorry, No. At long last . . . YES!
As you know, for some years subscribers of our online version (siliconchip.com.au)
have been able to search for and read articles from previous issues.
However many readers have asked us if we could make whole issues available.
Until now, that has been impossible -- the online version has only been practical
due to our printed edition production processes.
But that’s all about to change: following years of work, we’ve been able to produce
a digital version (in PDF files) containing all articles in any issue -- just as if you
had a printed copy in your hands.
The digital edition PDFs will be supplied on a quality USB drive, at least 32GB.
They will be recorded in five-year blocks (60 issues), covering:
n
n
n
November 1987 - December 1994
January 2000 - December 2004
January 2010 - December 2014
n
n
n
January 1995 - December 1999
January 2005 - December 2009
January 2015 - December 2019
Each five-year block is priced at just $100, and yes, current subscribers
receive the normal 10% discount. If you order the entire collection, the
6th block is FREE (ie, pay for five, the sixth is a bonus!).
All PDFs are high resolution (some early editions excepted) and the USB
Flash Drives are high quality metal USB3.0, so can be used over and over!
Want to know more? Full details at
siliconchip.com.au/shop/digital_pdfs
Australia’s electronics magazine
March 2020 95
|