Vac Meter 2, discrete solid state. 2015.

The front panel, using retired 1980s obsolete university gear.
Two more pictures at bottom of page.

In 2015 I wanted to improve the RF performance of my AM kitchen radio so I
needed a Vac meter with wider bandwidth than the 2013 meter which gave
1.5Hz to 300kHz.

The aluminium case, meter, 3 pole 12 position rotary wafer were found in
excellent condition within some old 1980 test gear rescued from a rubbish bin
at ANU.

Contents of this page:-
Sheets 1 to 8 are schematics of my 2015 Vac meter giving 12 Vac ranges for
0.5Hz to 5MHz :-
0.0Vrms to 1.0mV, 3.2mV, 10mV, 32mV, 100mV, 0.32V,
1.00V, 3.2V, 10.0V, 32V, 100V, 320V.

SHEET 1 :- Basic block diagram of whole unit.
SHEET 2 :- Rotary wafer switch Sw2A,B,C.
SHEET 3 :- Amp 1, gain = x10.00, +20dB.
SHEET 4 :- Amp 2, gain = x10.00, +20dB.
SHEET 5 :- Emitter follower buffers for CRO, F meter, etc.
SHEET 6 :- Amp 3, gain = 1.0, +/- 0.0dB, meter driver.
SHEET 6A :- Amp 3 basic circuit + explanations of GNFB + rectifiers.
Meter dial :- Image for customized analog meter dial to reduce errors.
SHEET 7 :- Power supply, +/-15Vdc regulated.
Fig 1 :- Passive 1:1 or 10:1 R divider probe for CRO or Vac meter.
Fig 2 :- Passive 10:1 capacitance divider probe for CRO or Vac meter.
Fig 3 :- Active probe for CRO or Vac meter.
SHEET 9 :-  Switched R divider for Vac range calibration.
SHEET 9A :- Three useful attenuator switches for calibration.
SHEET 8 :- Band-pass filtering for reducing noise.

SHEET 1 is the overall picture of main element layout.
Many R are not numbered and 3 amp schematics have been reduced to symbols.
The bypassing of Vdc rails to 0V and chassis case floor and including LC filters
prevents RF instability.

There are three cascaded amplifiers with a huge total amount of open loop
gain exceeding 100,000. Each amp has local or GNFB, and is isolated to prevent
any oscillations. The +/- Vdc rails are well grounded to a very low Z common path
of aluminium floor of the box. I used many 2uF polyester caps rated for 250V and in
white plastic boxes. Xc 2uF = 0.08r at 1MHz. L3 to L6 = 40uH chokes offer low R
between amp rails to maintain regulated +/-15Vdc. The 40uH + 2uF form filter
networks to ensure any Vac above 10kHz at +/-Vdc rails cannot find its way to
the next amp to cause oscillations. Each choke has 5 turns of 0.5mm Cu
insulated wire taken from Cat-5 cable and wound through 2mm bore of ferrite
tube 20mm long, to make a toroid choke.

Protection against excessive Vac applied is shown after Sheet-2 Amp1 and
Sheet-3 Amp2 schematics below.

Manual Vac range selection is easy for 12 Vac ranges from 1mVac to 320Vac.
After turn on, the unit takes 12 seconds for Vdc rails to fully stabilize. I always
try to select a higher Vac range than the Vac I think may be present.

Measuring above 320Vrms could be a problem if you don't know the peak
Vac with a non sine wave. Most Vac in audio amps are sine waves, square
waves, or triangular, but pulse waves and noise Vac may exceed 1,000V easily.
To measure Vac between 100V and 1,000V requires a resistance divider rated
for Vac and Vdc peak levels. Tube amps create peak Vac + Vdc above 3,000V.
A resistance divider shown on page for 2013 meter will withstand 4,000Vdc for
5 minutes.

If the meter reads below 0.1 x full swing, better accuracy is gained by switching
to a lower Vac range which will swing the needle higher for easy accurate reading.
If the meter reads full scale, switch up to higher Vac ranges until the meter settles
above 0.1 x full swing.
The Vac range labels tell you which dial to read, and practice makes perfect !

SHEET 2 Input switching for Vac meter, 2015.
The Vac ranges are approximately 10dB apart. There are three scales on the
dial plate of a 100mm wide analog meter. (more below).

For non standard R values, you MUST use only 1% metal film in series or
parallel to get correct R within 1% or you get errors exceeding 1%.
Every R value shown allows for all combined loading by other R around it
during use.

Trim caps C2 to C8 could be high V rated with adjust screws for C = 3pF to 8pF.
I used turns of insulated wire from Cat-5 cable wound around 1mm solid copper
poles 15mm long soldered to contacts of switch. Turns of wire are adjusted for
flat sine wave response to 6MHz.

Cin of the meter is determined largely by the rotary wafer switch with
unavoidable C < 20pF.

SHEET 3. Amp 1, gain x 10.
Protection for Amp-1. Accidental HV input damage is limited by UF1004
clamping diodes d1-d6 UF4007 across Q1 gate to source, and from source to 0V.
Amp 1 is used for V ranges 0 - 100mV. The max gate V swing = +/- 2.1Vpk.
If 500V is applied to V ranges 1-5, it is applied across R3 470r, 1/4W, current
exceeds 1A so R3 rapidly burns open. R3 needs only 23mA for 1/4W. A 50mA
fuse could be fitted to limit heat in R3 to 1.2W, but it would still fuse open after
some time. The fuse and holder must be placed for easy replacement and not
increase Cin.

SHEET 4. AMP 2, 2015.
Amp-2 input has Q1 source follower input buffer with Rin = 2M2 and R3 + C1,
C2 form LF pole 0.09Hz. The 2M2 does not cause significant loading of 1k0
network around Sw2C positions 1-5, or for SW2C positions 6 -12. But R10 on
Sheet 2 needs to be trimmed carefully to get correct Vac division.

Q1 2SK369 source has CCS dc feed from Q2 PN100 for high open loop gain.
The Q1 follower isolates the input of following gain amp with Q3 to Q8, preventing
instability. R5 200r is prevents oscillations above 10MHz. Amp Bandwidth is from
0.25Hz to 6MHz.

Protection for Amp-2.
When using the Vac ranges 1-5, Amp-2 is fed by Vac from output of Amp-1 via
switch Sw2C and its resistance divider.
The highest normal Vac level into Amp-2 = 10mVac. But during turn on/off, and
during gross overload of Amp-1, maximum possible Amp-2 input is about +/- 7Vpk.
Therefore the high Vpk swing at Amp-2 input is limited to +/- 2Vpk by UF4007
diodes arranged for least increase of input C.

For Vac ranges 6-12, Amp-1 is not used. DUT input is fed from outputs SW2C
R dividers to Amp 2 Q1 gate. The dividers all have 3M0 plus smaller R, with the
largest small R = 98k, for V range 0 - 0.32Vac, position 6c. If 3,000V is accidentally
applied to input, maximum Vac output possibly applied from SW2C at position
6c = 100Vac, but this is limited to +/-2Vpk by UF4007 around Q1 gate and
source to 0V. Maximum Iac in 3M0 with 3,000V applied = 1.0mA, so heat in
3M0 = 3W, and with 2 x 1M5 in series each 0.5W rated, the R will only fuse
if the high Vac is maintained for some time, which is very unlikely.
Use of 3 x 1M0 each 1W would be better, but then I'd have more clutter in a
small space.

Q1 source drives Q3+5 bases which are non-inverting input to the gain amp.
The Q3+Q5 are in two parallel differential amps (LTP) with PNP and NPN
bjts to give complementary action and best HF response.
I have idle Idc = 5.6mA in Q3,4,5,6 for high gm and high gain.
Differential gain is > 50 with collector loads of less than 2k2. Q7+Q8 have higher
gain in a complementary pair in common emitter mode.

Amp 2 open loop gain > 12,000 at 500Hz, but reduced to just 10.00 with
62dB GNFB. Q7+Q8 collector outputs are loaded by NFB network R21+R21.
The bottom of R21 is connected to 0V via C8+C9 each 8,200uF in series.
These are 10Vdc rated electrolytics which each need 7.5Vdc by divider with
R16+R16, each 10k0. So the effective C from R21 to 0V = 4,100uF, and the
R22 300r + 4,100uF set an LF amp pole = 0.13Hz. The arrangement gives
excellent Vdc stability.

To prevent inevitable RF oscillations with "uncompensated" high gain amps,
the open loop gain is reduced with C7 trim-cap 9-35p in series with VR1 1k0.
With VR1+C7correctly adjusted, there is no sign of oscillation over 6MHz.

Amp 2 Q7+Q8 collector output is isolated from other stages with 100r to inputs
of monitoring buffers on Sheet 5 and to Amp-3 input on Sheet 6.
The following amp stages have some shunt C which are likely to cause RF
oscillations. Series R between 100r and 220r are used at input or output to
prevent RF oscillations.

THD and noise is negligible.

SHEET 5..Emitter follower buffers. 2015.
Here are two simple emitter followers directly connected to output of Amp 2.
These allow 2 external devices to be connected to the Vac meter such as
Frequency meter, CRO, or alternative Vac meter. Such devices cannot affect
the working of Amps1,2,3.

Protection. I have FR3004 diodes after output caps C1-4 to +/-15Vdc rails.
Q1+Q2 can only be fused with accidental application of HV to the output
terminals. Excessive Iac or Idc current in R4 or R7 from an external HV source
will fuse them open.

HF F2 > 5MHz, and F1 is determined by C1+R5, 5uF + 330k and loading of
a CRO or other Vac meter etc in parallel.
If a CRO Rin = 1M0, R = 240k and -3dB F1 pole = 0.13Hz.

SHEET 6 Meter Amp 3.
Amp 3 is almost identical to amp 2 but without unnecessary input emitter
follower stage because the previous Amp 2 has low output resistance < 150r.

Amp 3 operation is DIFFICULT to understand.
SHEET 6A. Basic action in Amp 3.
Amp-3 on SHEET 6A is drawn here more simply with triangle symbol used for
amplifier and the two + / - input ports have Rin approx 50k, and the output at
Vo is a collector current source. R&C numbers on 6A are same as for SHEET 6
Amp-3 schematic.

The metering function depends on a basic principle :-

Idc flow from charged C in a full wave rectifier circuit = 0.707 x Ia rms flow in
Vrms source.

Alternating current at output of Amp-3 is applied to a diode bridge after to
produce dc flow in the meter coil R // VR3+R17. The Vac flow in the meter R is
reduced to negligible levels with shunt C18+C19 so only DC is applied to the meter.

The action is much like a full wave PSU rectifier where an AC source charges
a C after a diode bridge, and the resulting Vdc without ripple voltage is applied
to an R load. The C18 value must high at 470uF ripple Vac at low F is very low,
and to prevent meter needle wobble exceeding + / - 10% at 0.5Hz.
There is negligible meter wobble at 5Hz. Amp-3 output Iac flows through NFB
network which includes R15, diode bridge with 1N5711, C18+19 and R16.

For meter full swing, Vac across R16 680r = 100mVrms, so Iac = 0.14706mArms.

Idc in VR3+R17 // meter R = 0.14706mAdc.

Meter R = 1k0 and the adjusted total of VR3+R17 = 3.125k, so total R load for
DC flow = 680r, hence Idc = 0.1Vdc / 680r = 0.14706mAdc. The adjustment of
VR3 is fairly sensitive and you could use 1k0 trim pot + 2k7.

Amp-3 output is from Q5+Q6 collectors which are a virtual current source with
Ro > 50k0. If the output produces + 0.147 mA pk for 0.1Vrms at R16, and if open
loop gain = 10,000, then Vac difference between input ports = 10uV.
The transconductance of the amp is transformed by voltage gain to be about 14A / V.
The Vac across R16 680r is made linear to input Vac by GNFB, 80dB max at 500Hz.
Thus current flow from Q5+6 is controlled accurately by GNFB. Vac across R16 is
almost identical in wave shape to the input Vac at Amp-3 input.

The GNFB Vac at R16 contains THD in current flow with diodes and rectifier.
This is amplified to prevent its creation, so that Amp-3 output voltage is varied
to do whatever is needed to reduce THD at R16. This ensures the Idc flow to
meter is linearly proportional to the Iac rms flow in R16, and that the Vdc applied
to meter tells us the True Vrms value for any Vac input wave form.
With sine waves at both input ports, the wave at collector output appears like a
square wave with verticals = +/- 0.5V approx, top and and bottom horizontals
are curved up and down.

The relationship between Vrms, Voltage Root mean square and Vdc is explained
further at Fig 5, 1/2 way down page at 2013 Vac Meter.

Root mean square, rms,
is also defined better than I could at

Basic units need to be understood.
The Watt is a current flow = 1 Coulomb per second
= 6.2415 x (10 to power of 18 ) electrons, which is called 1.0 Joule.
This is the number of electrons in a 1 Farad capacitor charged to 1 Volt.
This is the number of excess electrons above what would exist if there is no
measurable Vdc across the capacitor. 
Where you have 1 Volt applied to 1 Ohm, I = V / R = 1 Amp.
For 1 second, the work done is 1 Joule, or 1Coulomb per second.
So where you have 1 Amp of current flowing,
there are 6.2415 x 10 to power of 18 electrons flowing per second.

Electrical power is done at a rate per second measured in Watts :-
Power, Watts = ( V x I ), or ( V squared / R ) or ( I squared x R ).
The power generates heat in a resistance, causes motion in an electric motor,
removes heat in a refrigerator, creates sound in air or water.

Electricity bills have power units in Kilowatt Hours, kWh.
My typical winter bill is 16kWh per day. I day = 24hrs, so each hour the average
power = 16 / 24 = 666.6Watt-hours. It means 666.6W is average drawn each hour. 
Current = P / V = 666.6 / 240V = 2.78 Amps rms. 667Watts is about equal to
1/3 a 2kW rated room heater, equal to 0.89Horse power, and about 30 times the
average power I generate within myself when not doing much, and is about 5
times the power I generate when riding a bicycle 30km across town to have coffee.
Modern civilization is an extremely energy hungry beast compared to 1717 before
the industrial revolution where the vast majority of all ppl were poor, and survived
by producing 591Watt-hours each day.
A sine wave alternating flow of current must have peak +/- Vac = 1.414V to
provide the same heating power in R as 1.0Vdc. The sine wave V and I can
be expressed in terms of Root mean square which equates the Vac and Iac
as equivalent to Vdc and Idc which will generate the same power in an R,
called RL, Resistance Load.

The peak Vac and peak Iac may vary greatly for any electric flow wave form,
but whatever these V & I values may be, the Vrms and Irms can be measured
using the meter I describe here, and in all meters giving "True Vrms" so the
question in your mind, "What is electricity?" need not ruin your day.

It can be proven mathematically that some simple Vac waves of +/-1V peak at
any constant frequency have Vrms values according to a simple table :-

Square wave
, 1.0Vrms = Peak Vac / 1,

Sine wave
, 0.707Vrms = Peak Vac / sq.rt 2,

Triangular wave
, 0.577Vrms = Peak Vac / sq.rt 3.

Vac wave-forms we measure have have very different shapes and may be
usually measured in Vpk, Vpk-pk, or Vrms.
Pink noise signals used for testing speakers sounds like a rumbly big waterfall,
and on the CRO it looks like a very blurry display because of the constant
randomly varying amplitude, frequency and phase. If we measure pink noise
Vac as Vrms, the meter may show slow Vrms changes due to very low F within
the noise causing meter needle to wobble.

The Vrms voltage measurement of Vac will be found to generate the same
heating in a load R as would the same applied DC Vac or Iac waves may
be a series of regularly repeating pulses of varying lengths of time, and may
be seen as a stationary wave on a CRO because of the repeating triggering
time of the CRO. The peak value of Vac or Iac change could be many times
the Vrms value.
So peak Vac measurements alone do not tell us how much continuous power
that wave will deliver to a load, only the maximum peak current and power.

Engineers find it useful know the Vpk and Ipk as well as the True Vrms and Irms.
If we can see Vpk for a wave on CRO, we can calculate the peak Iac for a given
load R. We can estimate average Iac from the wave shape and its duration as a
fraction of total time for 1 wave, and work out the power liberated in the R load
where that Iac exists.

The Iac flow in a transformer winding feeding a diode rectifier and can be viewed
on a dual trace CRO using both channels in differential mode across a 10r0 in
series with winding end and input to diodes before the reservoir C, if one is used.

Amp-3 open loop gain = 12,000 maximum, reduced to very close to 1.000
between Vac input and top of R16 which feeds the GNFB input port of the amp.
The 82dB of NFB ensures the Vdc applied to the meter remains directly
proportional to the input Vrms, so the meter may be calibrated to read Vrms,
and accuracy is good down to less than 0.1mVrms.

Amp-3 output is from high Z current source of Q5+Q6 collectors. During voltage
measurement, the wave form between collectors and 0V looks like a basic square
wave with curved arches instead of straight horizontals. It looks baffling until you
realize the amp is doing all it has to to make the Vac wave form across R16 and
at at NFB port very close to Vac at input.

The analog meter used for this project was made in Australia before 1985
when we still made good things. However, although the mechanical quality remains
excellent, the needle movement was not linear to the applied Vdc and errors of up
to 15% at low readings and 8% at middle of scale were found.

I have a section of re-calibration of analog meters below......

In Sheet 6 Amp 3, biasing, dc stability, meter LF pole are dependent on R14 82k,
and C14+16 136uF ( 2 x 68uF NP ). This part of GNFB network seemed to work
better at VLF than for the network in Amp-2. The only disadvantage is that any
noise below 0.2Hz generated in R14 82k is amplified by open loop gain and not
corrected fully by NFB. I saw some very slight CRO trace bounce at low Vac levels.
In Amp-3 meter amp, this does not cause any visible meter needle wobble while
reading Vac for any range. Rin at Q2+4 bases is about 50k so R&C network
= 82k//50k + 136uF = 30k + 136uF so F1 = 0.0376Hz in theory.
This is below C+R couplings elsewhere so meter gives F1 at 0.5Hz.

The 82k is a GNFB path from Q5+Q6 collectors to Q2+Q4 bases to maintain stable
Vdc operation without excessive Vdc offset at output.

Base currents to npn Q2 and pnp Q4 flow in opposite directions and are each
approx 0.25mAdc. Only the difference between base currents flow in 82k, about
0.003mAdc. Across 82k, Vdc < 0.24V. It is ideal to operate Amp 3 with Vo close
to 0Vdc, so VR1 is adjusted so Vdc at output < +/- 10mVdc. With VR1 set, there
is enough voltage gain at DC to keep output of the amp close enough 0Vdc under
all conditions.

No PCBs are used. I built Amp1, 2, and 3 on separate pre-drilled boards about
about 120mm long x 85mm wide with all tracks and terminals between parts using
short lengths of 1mm dia solid copper, formed unto a U using long nose pliers,
pushed through two holes of board, then ends folded flat under the board.
The layout of bjts and R parts are copied from schematics as I show them.
It is always easy to know where you are during later service work. Leads of
R or bjts are surface soldered to wire tracks, and most C are under board
with leads up through holes to tracks. With all larger C under the boards and
wire tracks, there is no clutter in the way of measuring Vac on boards.

Only practice makes a nice looking board. It will not be as neat as a PCB,
but the circuitry complete HF and LF stability. After you have done about 10
boards you should become skilled and make reliable boards very unlikely to
ever develop dry joints even if the unit is dropped to a hard floor. Each board
is mounted off the metal case floor on 4 x 16mm dia x 35mm long timber dowel
spacers at each corner and fastened with 4 gauge x 16mm long c/s cupboard
hinge wood screws down through board and up through case floor.
This allows easy removal of boards, plus gives a short path for 2uF caps from
Vdc rails to case. Such C are shown on SHEET 7 for PSU and necessary for
RF stability.

NOISE could be a huge problem if you build this Vac meter. To test for noise
generated by the 3 amplifiers, the input must be shorted to 0V using an RCA
male plug with short wire shunt. I tried valiantly to build the unit without a steel
sheet shield around Vac range switch and Amp. Noise was only low enough f
or me when putVac range switch and Amp1 inside an additional steel sheet
metal box inside the Aluminium case.
Then I found the equivalent noise at most sensitive Vac range 0-1mVrms < 10uV,
quite OK considering bandwidth of Vac meter is 0.4Hz to 6MHz. The regulated
rails of PSU help keep very low frequencies so low that a CRO used to monitor
Vac does not show any trace movement when set to DC.

SHEET 7. PSU, earthing, feeds to three amp rails.
Sheet 7 PSU for 2015 has 7815 and 7915 regulators for +/-15Vdc rails for the 3 amps.
For noise free Vdc output I found C6+C8 470u+u47, and C12+C14 10,000u+u47 were
needed. The Sheet 7 AND Sheet 1 arrangements gave me the lowest noise at all F
when viewing the output of buffers on Sheet 5, with input shorted to 0V and with most
sensitive Vac range selected. With amplifiers used to measure Vac < 0.3Hz, Vdc rails
must have very low LF noise, and only regulation removes the very low F noise generated
by random variations in mains levels into the PSU.

The output resistance from regulators appears to be < 1r0 and low enough to prevent
Vdc rails moving at VLF.

The 0V rail connects to aluminium case for low frequencies near output of PSU
via R5 180r, and with low value C16 0.1uF.
The 0V rail of input RCA socket plus other points on 0V rail are bypassed to Al case
 floor with 2uF. The 0V rail is is a solid 1.2mm dia Cu with total length about 350mm long.
There are several 2uF to 0V from points along 0V rail length to prevent the rail being
a tapped inductance with rising Z at HF. The whole total arrangement works fine with
the metal casing and with shielded+LC filtered IEC mains chassis plug.

Sheet 1 shows additional chokes and caps used on each amp board to ensure
Vdc rails remained free of noise or possible RF oscillations. The chokes are ferrite
tubes 20mm long, 6mm oa dia, with 2mm bore dia. I have 6 turns of 0.5mm Cu dia wire,
polythene insulated, from Cat-5 cable, to make a tall toroid coil and which gave
40uH at 1MHz so XL = 251r.
This is far more inductance than a 100mm long piece of 1.0mm dia wire which has
L = 0.17uH, and XL = 1r06 at 1MHz.
See the calculator for wire inductance at
Rw is < 0.01r. I might assume the Idc flow does not cause significant lowering of
ferrite choke reactance The arrangement of 40uH plus 2uF gives a low pass filter
with pole approx 18kHz. At 400kHz, 40uH + 2uF give XL = 100r, and XC = 0.2r,
so attenuation = 0.2/100 = x0.002 = -54dB. HF in one amp rail cannot find its way
to another amp rail to cause RF oscillations. The exact route and cause of RF
oscillations in this instrument or any other electronic gear may be difficult to
forecast or analyse or cure so its best to try to isolate each rail for each amp,
and but have the common 0V rail bypassed with 2uF several times along its
length to a very low reactance such the aluminium case floor.
This is the meter dial for an unknown but better brand of analog multi-meter.
Many people will struggle to read it because its so complex and you can see
why DMM have become so popular since 1985. The thick black curve arching
across the dial is not black, really is a mirror so that the image of the needle
should be hidden behind the needle which means you are looking at meter at 90
degrees and you read the meter correctly - a correction of "parallax error".
A quick Google of "parallax error meter" will bring up countless analog meter

Many analog meters have a linearly drawn dial scale of typically 0 - 100.
The one I have needs 0.1Vdc for full scale at 100. But many will be found
to be inaccurate if checked with Vdc = 5mV, 10mV, 25mV, 50mV, 75mV, 100mV.

I found my meter gave 7% error at 50, and 15% at 10, so I thought I needed
to draw a new scale for dial plate. But did I really need to re-calibrate the dial?
I then thought I better measure Vdc applied to meter by Amp 3 by applying a
number of known accurate Vac inputs using the 0 - 1.0 Vac range setting.
I used 1kHz from my low THD oscillator.

First thing needed is a know 1.00Vrms applied to meter input in 1V range,
and making sure Amp 2 was producing a measured 100mVrms at input to
meter Amp 3, and that Amp 3 then produced whatever Vdc was needed for
full swing of meter which needs the meter installed, and adjustment of VR3
seen below meter in SHEET 6 Amp 3. Amp 2 gain needs to gave gain close
to 10.00, but within +/-5% is OK, and Amp 3 gain is 1.0, and VR3 compensates
for any errors, and for varying Vdc needed for full meter swing, slightly
different to the nominal amount in spec sheets.

The best way to produce a number of accurate Vac is to use just ONE
reference  Vac then divide it with a switched attenuator use the same
brand of resistors of equal value and 1% tolerance, and low enough value
to allow loads down to 100k be connected without change to each Vac at
each switch position.To check my meters I found a suitable aluminium box
210mm x 100mm x100mm long, and installed this schematic.....

SHEET 9. Three useful attenuator switches :-
The schematic shows 3 rotary wafer switches each 1pole x 12 position,
all made before 1980, and 50mm dia types. S2 + S3 are for testing ranges
of Vac or Vdc in -10dB steps and S1 is for meter dial plate calibration.

All 3 old switches use contact 12 to feed a rotating disc which is the switch
pole which can point contact 12 and to 11 other contacts. So 12 Vac are
possible including 0Vac. Most modern 12 position wafer switches have
separate pole connection which allows 12 different Vac above 0V.

Consider S1 first, for meter calibration.
S3 uses R1 to R11= 9 x 270r and 2 x 135r metal film x 1% x 0.5W arranged
so that when 10.000 Vrms is applied to input, you can get 11 output voltages
of 10.0V to 1.0V in 1.00V steps, with the smallest Vac being 0.5V and 0V.

To make a new calibrated dial :-
1. Make sure work area is clean, and free of any iron particles
( from drilling, filing etc)

2. Remove meter from its mounts, remove perspex front cover.
Measure size of new dial to equal existing for top, and 2 sides.

3. Cut white cardboard template to be exactly equal to existing dial plate.

4. Adjust zero adjust screw to center position.

5. Slide template behind needle and fix with masking tape at top and two sides.

6. The needle must be free to move without touching template.
Cardboard must be flat.

7. Mount meter vertically on unit temporarily without front cover.

8. With meter turned on, and with Vac range at 0-10V, and with RCA input grounded
for low noise, and with no Vdc present across meter, the needle position is drawn
in pencil behind the needle near end of needle and at bottom of template.
Use a finely sharpened HB pencil.

9. Connect meter input to S3 output with setting at lowest 0.05V position.
10. Connect low THD sine wave 400 to 1,000Hz from signal gene with
low Z output less than 600r to S3 input.

11. Turn on signal gene and adjust level at S3 input to 1.000Vrms using a
known reference meter, I use my Fluke 117.
(Two other DMM give similar readings for 1.00V, with less than +/- 0.4% difference.
But if in real doubt, then build a reference signal generator with guaranteed output level
( Maybe not so easy ). See 100Hz gene

12. Turn up S3 to give full swing of meter. The Vdc across meter should be
close to the nominal Vdc needed for full swing of meter, in my case, close to 0.1Vdc.

13. Mark needle position '100' behind needle in template at end and at bottom
of template.

13. Switch S3 down one position and mark behind needle for '90', and all
subsequent positions down to '5'. Check all 3 times, and an hour later. use enough
pencil marking to ensure they appear well enough when later scanned in black
and white.

14. Remove the meter from its mountings, remove template with care.

15. The template is scanned to make a preview scan, then the smaller dial
plate area scanned at 300dpi, black and white. Make sure the outline of dial
plate is accurate, and shows up with vertical and horizontal boundaries.
I've been using ArcSoft PhotoStudio 2000 since about 2002, and a Cannon
scanner from 2001, both still working better than many others.

16. Save the scanned image as "meter-dial-1-2016" and as .bmp in your
relevant "Test gear" folder which is a sub-folder of your larger "Audio Technical" folder.
( I have hundreds of files in many folders in Audio Tech and I need to be able
to find them later easily.)

The scan size may be quite large but may be reduced to get an image to fill
about 1/2 height a PC screen in MS Paint when "1x size" is used.
Thin lines may be drawn in black over feint/thick/untidy drawn markings.
Save the image as a monochrome .bmp, and that should replace .bmp with
grey pixels.

17. Open IrfanView, open .bmp, and increase canvas size enough to plot
position of needle bearing center. Save in original folder.

18. Re-open image in MS Paint with larger canvas size.

Now comes the real work of drawing up a dial worthy of printing.
The size of image on screen will be much larger than the template,
and the x2, x6, x8 function will be needed to create a credible dial plate.

19. Draw vertical line down from 1/2 way across the meter plate lower boundary.

20. Draw at least 4 lines through marked needle positions below lower boundary
of template and to intersect vertical line. 2 radii each side of center line will do.
You should find an "average intersection point CP" on center line, then draw a
horizontal through vertical, and remove mess of other lines. Distance from CP to
"end of needle marks should be the same, within +/- 2mm.

21. Using a ruler on PC screen, measure from CP to plot scale baseline
intersections for the 2 Vac and single dB scales. Plot curve baseline positions along
radii at relevant distances from CP with small cross using a "dot".

22. The dots can be joined to give a multi-faceted baseline curve, with minimum
line width fill in line so at line size steps there is a thicker line, but not more than
2 pixels wide, or high.

23. Tidy up curves without losing essential voltage positions.

24. Between 1 and 2, divide distance in 1/2, and that will be 1.5. plot a dot
near curve baseline The divide each 0.5 into 5 parts
with 4 radii lines so distance between each looks equal.
Thicken up the lines, make them say 50mm long at 1 and 2, 40mm long at 0.5,
and 30mm at each 0.1 position.
Measure and trim line lengths using a ruler. The process here is interpolation
with negligible errors because we know the needle will be at 1 and 2, and at
1.5 if we adjusted Vac input.
23. The process is continued for 0.0 to 0.5, and 0.5 to 1, and then for 2 to 3
and so on for the whole 0 - 100 scale. I took 8 hours to get that looking
acceptable, that's 4.8 minutes per fine line for scale division.

24. The scale 0-32 was plotted by reading off voltage from 0-100 scale and
drawing radii and 0.1 divisions.

25. The dBV scale is far from linear because its based on logarithmic increase
of Vac, but scale marks are drawn from voltages on 0-100 scale, aided by
calculation for 1dB reductions of voltage. 

26. Lettering for scales is typed in at whatever size is needed to get dial
to look right.

Perhaps you can find a meter scale drawing app which automatically can
draw a scale to suit the markings from a template.
But I bet you can only find one to make a linear dial which we do not want;
we want THIS meter to tell us what IT measures, which may be different to
the next meter along.

SHEET 6B. Meter Dial plate.
27. This is the image similar to what I finally ended up with in MS Paint.
It is saved as .gif, and then printed. The dial size on paper will be too large
because the printer tries to fill the A4 page.
A measurement of length is taken, and size ratio to real template length
calculated. The overall image size in IrfanView ( or some other imaging program )
is adjusted by the ratio, and is saved, and then this printed sheet should show the
dial details much smaller, but the same size as the template, and this can be
confirmed if template is laid over image.

28. The image is trimmed to outside boundaries with scissors, and taped to
existing dial plate, and the meter used to check accuracy with varied Vac from S3.

29. I found all was well when I measured, scale had errors < 1% at all meter positions

Tests in several Vac ranges gave less than 1% error of reading, and was better
than all other analog meters I have used or made.

Amp-3 in Sheet 6 has bandwidth of 0.2Hz to 6MHz with GNFB.
Meter F response gives readings -3dB at 0.5Hz and 6MHz.
Checking other Vac ranges.

In SHEET 9 switched attenuators, S3 is a Vac attenuator with the same -10dB Vac
steps as I have in my two analog meters. With S1 set for highest input resistance to S3,
ie, position 2, then the 100k has negligible loading effect on most signal generator
source output resistances which are usually 600r or lower.
But my switched attenuators do not have compensation C across each of R13 to R23
so that the stray circuit capacitances do not create huge errors with F above say 50kHz.
With such C in place, the input impedance to S3 becomes mainly capacitance above
say 100kHz, and if C was say 22pF across R13 68k input C, at 1MHz the Xc 22pF = 7k3,
very much below 100k which exists at say 1kHz, where 22pF = 7M3, and negligible.

So the attenuator has limited use for wide bandwidth.

For all operation of any circuit above say 50kHz, circuit impedance and circuit operation
may be drastically altered by connection of any meter or oscilloscope probe.
Good signal generators which work above 50kHz should have Rout = 50r, so that whatever
they connect to does not affect the source input Vac.

With source resistance of 600r feeding S2 input set at position 2 for highest input
resistance to S3, Rout at S3 position 6 = 1k0. So source resistance 600r is loaded
by 100k, and its Vac is not affected by 100k. But whatever connects to S3 output at
pos 6 has Rout 1k0, and the Vo at pos 6 is less likely to be affected by high F.
The price paid is that there is 1/100 or -40dB Vac reduction.

Below position 6, there 6 more Vac available with lessening Rout.
In Voltmeter 1 I used old type rotary 12 position switches giving 11 Vac above 0V,
with highest Vac being 100Vrms and +320Vrms is read by change of probe cable
to an extra input terminal into which up to 320Vrms is OK.

But in This Voltmeter 2, the input switch has 12 possible Vac with modern rotary
switch I have a separate pole terminal so up to 320Vrms is measurable by using
the range switch without change of input probe lead to separate input terminal.

A sine wave for 320Vrms is + /- +/- 453.3Vpk. This is dangerous territory for any
technician, and may challenge RCA leads or other coaxial cables. If you must play
around with more than 100Vrms anywhere, make sure you know what you are doing
before and during and after.

Some DIYers might just draw up a dial template with pencil and ruler, and then go
over it with black ink pen and ruler, and all including interpolated fine division marks
each fraction of a Volt. The template can be erased to remove pencil lines, leaving
only ink marks. This is easy, but most ppl end up looking at a mess they should have
done on a PC in a drawing program.

Voltmeter probes and cables.

The simplest voltmeter probes have 2 x 1 meter long read and black cables, well
insulated against a maximum of 3,000Vpk and with very flexible multi strand wire,
with shrouded 4mm plugs one end and plastic probe handles with 2mm dia pointed
metal probes for DUT end. All DMM are sold with these leads which I found to
be safe enough until constant use fatigues wires and cracks the plastic insulation.
Beware the old meter probe which gets very bitey if measuring +600Vdc.
These cables are very prone to high RF pick up and hum, but are OK for everything
from DC to 1kHz where DUT circuit R < 10k0, and signals are > 10mVac.

Many DMM can only read Vac down to 10Hz and up to only 1kHz if the DUT has
high circuit resistance so that DMM input C shunts signals above 2kHz which may
be the -3dB F2. Most DMM have high Rin usually > 5M0.

For measuring 0.5Hz to 6MHz for low level signals down to 1mVac, the meter
cabling should resemble good probe leads used for a CRO, ie, use coax cable
with good shielding and low shunt C. But the best coax cable has 33pF per meter,
and if meter Cin = 32pF like many CRO then minimum C shunt when probing
= 65pF with a 1M probe lead. Many are 1.5M. so C shunt = 82pF at least.

To get -3dB F2 = 6MHz, and if Cshunt = 82pF, DUT circuit Z should be 320r.
The best coax locally available which will last years without breaking the inner
wires is is RG58CU. It has 67pF per metre, and at least a metre is needed for
most general work to reach between DUT and Vac meter or CRO.
With 1M cable and CRO, total C = 100pF, and DUT circuit impedance should
be no more than 250r to get -3dB F2 = 6MHz.
For measuring an anode circuit of EF86 where R = 100k, and probe Cin = 100pF,
F2 = 16kHz; the act of measuring reduces the working HF Vac, and may cause
a power amp circuit with NFB to oscillate badly at HF. So high shunt C and low
probe input resistance needs to be avoided like the plague. 

For most audio work with circuit R < 10k0, RG58CU has good shielding and
F2 = 160kHz.

Standard coaxial cable properties are listed at
The cable with lowest C is RG79A with 10pF per foot, or 33pF per meter.

To avoid the high capacitance of coax cable, a CRO probe with a resistance
divider with capacitor compensation allowing two switchable output levels may
be used.

Most switchable CRO probes have a 1:1 ratio where there is no R divider
in the circuit. At the CRO, Rin = 1M0 in parallel with 32pF. If the CRO cable
has C = 67pF, the total 1:1 probe has Zin = 1M0 // 100pF.
When 10:1 ratio is selected, a 9M0 is switched in series with cable output.
A trimmer C = 4-20pF shunts the 9M0 and is adjusted for 1/9 of the Cin to
cable and CRO. This means the probe input C = 10pF plus any C between
probe tip to probe case if used. Some probes have along probe 50mm long
which will allow too much RF entry for low level work.
Good probes have a short probe end and metal case extending out to shield
9M0 and which can be connected to DUT 0V rail or case with a short lead.
So many 10:1 probes have 10:1 Cin about 15pF.
Notice that the HF cut off is much higher for low probe Cin.
If a Vac meter input circuit is set up the same as a CRO, many standard CRO
probes may be used for measuring Vac.
My Vac meter described here has Rin = 3M0 which can be switched down to 1M0.
Cin = 20pF so my Vac meter can be used with many available CRO probes of
1:1 or 10:1.

Many DMM and other Vac meters have high Rin < 5M0, and high Cin of perhaps
1,000pF and CRO probes are NOT suitable.

Most CRO probes have input voltage rating equal to the CRO, often 600V peak,
or 424Vrms sine waves, and for both switched positions of 10:1 or 1:1.

The 10:1 probe reduces DUT Vac to 1/10 at Vac meter or CRO.
My most sensitive Vac range is 0-1.0mVrms, so the 10:1 probe is not very useful
for DUT Vac < 10mV. Most analysis of circuits are done while measuring Vac > 10mV.
This is OK because the SNR will be better.

Fig 1. Switchable 1:1 or 10:1 Passive CRO probe.
This is a typical 10:1 switched CRO probe used with a generic  CRO input
circuit with generic Zin = 1M0 bypassed with 32pF.

I made a non switched 10:1 probe with shielded metal case made from tin
plated steel sheet from olive oil cans. It has the above schematic, but without Sw1.
The probe case is 21mm dia tube 100mm long, capped at both ends with smaller
tube 6mm dia projecting 25mm over the 1.2mm dia wire probe tip, with about
1mm insulation. This allows reaching to most DUT test points.

Shielding is better than for other manufactured switched probes. The output cable
is 1M of RG58CU with C = 67pF.
My CRO has Cin = 33pF, so total Cin to cable = 100pF. The C1 is adjusted for
about 11pF and total Cin to probe is about 15pF. C1 is adjusted for best square
wave for all F between 1kHz and 1MHz, using a flat sig gene Vac source with
Ro < 200r.

With a wide band CRO probe you may find LF noise such as hum or audio Vac
will interfere with low level Vac above 10kHz.

The alternative way to measure or view Vac is to adopt the principle of using
TWO frequency bands, one from DC to 10kHz, and the other from 10kHz to
above the limits of the Vac meter or CRO. The lower can be done with 1:1
probe with high Cin = 100pF, or 10:1 probe with Cin = 15pF. But as F rises,
the reactance of C reduces and can affect viewing waves or measurements
because of loading the DUT circuit Z.

A Capacitance Divider 10:1 probe the best for HF viewing or measurements.
This is an even simpler type of probe and without a switch or 9M0, but with
C1 set to give 10:1 ratio at say 1MHz. Without the 9M0, probe bandwidth is
reduced at LF so that F1 pole is at 1.6kHz so that noise below 160Hz is reduced
 -20dB below the HF level. 50Hz hum is reduced -30dB, and VLF trace movement
at 1.6Hz or meter wobble is reduced 40dB.

Fig 2. Capacitance divider probe.
C1 for above probe needs to have Vdc rating of 1,000Vdc if possible.
The circuit gives a fixed Vac ratio division above 10kHz if the Rin to a CRO or
Vac meter is 1M0, and total Cin is between 50pF and 150pF.
This extremely simple probe allows you to measure all Vac in an old radio
without much de-tuning of LC circuits or losses which will alter AVC bias in
AM radios where there can be low level input between 455kHz and 2.2MHz,
but also subject to LF noise in AVC circuit. This probe is very easily made as
a metal tube extension to an RCA male plug with a trim cap inside and access
hole to screw adjust. This allows the 10:1 ratio to be adjusted for 1MHz, and
you should find 10:1 ratio remains for all F between 10kHz and 10MHz.

To avoid high Cin and Vac level loss, an active probe may be used with
j-fet + bjt in a high Z input emitter follower.

Fig 3.
Active Probe.
One would hope this is better than any passive probe for Vac less than +/- 10Vpeak,
or 7Vrms sine waves. It can be "easily made" with small circuit board 18mm x 100mm
long which can slide into 20mm dia steel tube. Cin will be total of 10pF, with 5pF at Q1
gate plus 5pF to shielding.

The bandwidth should be 0.34Hz to above 6MHz. With input network = 1M0 // 10pF,
Zin at 1.59MHz = 10k0, so the probe is good for all audio work. You should be able
to measure a 1mV signal at DUT if the noise is low enough to permit it.
To exclude LF noise, F1 LF pole is raised by reducing C1 from 0.47uF to whatever
value you choose. If you want F1 = 10Hz, C1 = 0.016uF, and for 1kHz, C = 160pF.
Using C1 = 10pF, total input C = 15pF approx, so F1 = 10.6kHz, and good for
measuring low level RF without the amplitude reduction of a 10:1 probe.

Measuring Vac at any high resistance anode circuit at HF is affected by C of a probe.
The anode circuit of an EF86 may have RLa = 100k and shunt C between anode
and all other things = 10pF, giving F2 = 159kHz with no meter or CRO probe connected.
If the meter probe C = 10pF, then total shunt C = 20pF and F2 = 80kHz.
To avoid HF attenuation at anode, you can connect 1k0 between B+ rail and 100k
anode load, then measure Vac across the 1k0.
If probe Cin = 10pF, F2 pole is 15.9MHz, and if probe C = 100pF, F2 = 1.59MHz, so
the working F response at anode is not disturbed while you probe it. But this means
you have an effective 100:1 probe, so Vac at anode should be at least 1Vrms. 

Oscilloscope probes might be purchased.

In Sheet 2 above, I have Sw1 to switch in 1M5 from input to 0V to reduce max Rin
from 3M0 to 1M0 which is the same as both my CROs.
Trying to measure low Vac from an MC phono cartridge using a test record may be
difficult. A Denon MC 103DL has rated output of 0.4mV at 1kHz, with 0.04mV at 20Hz,
4mV at 20kHz - if RIAA reverse EQ has been applied for cutting grooves.

It is better to use a low noise j-fet phono preamp to increase all F from 20Hz to 20kHz
while equalizing relative F levels with RIAA EQ network. If the network is accurate,
and the recorded signal has had accurate reverse RIAA EQ, and the cartridge response
is flat, you should see a flat sine wave response from 20Hz to 10kHz with -3dB poles
just outside these Fo. Three things have to be correct before you can say the cartridge
is GOOD.
Deviations from the flat may tell you about a cartridge. Testing 3 or 4 different MC and
MM carts tells you more, and all will vary slightly, and possibly colour the sound like a
graphic equalizer using unknown random settings. My pages on preamps tells you more.
Making a phono preamp for testing is not difficult if a kit with op-amp (OPA2134PA ) and
NFB RIAA is used.
Noise can be a problem, and will test your abilities.

Many audio amps may have noise >2.5mVac with no signal present.
It is usually mains related harmonics of 50Hz, 100Hz, 150Hz and 200Hz, plus
diode switching pulses at 100Hz plus hiss or rumble from noisy input devices.
Using the low Vac ranges and a CRO, you can see the truth about amp noise.

To avoid noise above or below the F band you wish to measure, a bandpass
filter (BPF) is connected between DUT and meter or CRO input. This may alter
DUT behaviour and lessen Vac you wish to measure so a simple alternative is to
place an active BPF between Amp-2 output and input to Amp-3 meter amp and
buffers for the CRO.

SHEET 8, BPF, 320Hz to 32kHz.
The bandwidth as shown is 320Hz to 32kHz and excludes most mains related
harmonics, diode noise, and RF noise.
If there is radio station RF pick which is converted to audio by DUT, you may
see the AF on CRO with BPF, without other signals present.

The BPF allows a clearer view of a test signals between 1kHz and 20kHz,
measure low level signals more easily without noise.
Noise in an audio amp may be 50 times higher with no GNFB connected than
when GNFB is connected. Sw1 allows BPF switched in or out so a comparison
may be made with BPF or without BPF.

Additional switching could be used to alter F1 and F2 by altering R&C values in filter.

Simple bjt emitter followers will produce low enough noise and THD and produce
BW wider than op-amps.

In my page on THD measurement I show use of an LC bridged T notch filter
to remove 1kHz from sample Vac from an audio amp output. This allows inspection
and measurement of THD and noise. At low level signals a switched hum filter
allows me to remove H below 320Hz. At very low levels with THD < 0.1%,
I use op-amps to amplify the THD signal x10 and I have a BPF to pass all HD
between 2kHz and 11kHz so THD of 1kHz may be seen and measured without
too much noise.
Where you have more than one F present in any Vac,
Total Vrms = Sq.root of the sum of Vrms squared of each F.
If you have 0.1Vrms 2kHz, and 0.033Vrms of 3kHz, total Vrms = sq.rt ( 0.01 + 0.0011 )
= 0.1054Vrms, so you can see how two Vac with 3:1 amplitude ratio make very little
difference to the Vrms measurement of the largest Vac. If you have 1.0Vrms of 1kHz,
and THD = 10% = 0.1Vrms, total Vrms = 1.005Vrms.

If DUT noise is high, you should try to eliminate it before making serious
measurements. I lost count of how many audio amps I had to fix or modify
before being able to measure them properly. If noise < 1mVrms, then measuring
a 10mVrms test signal is easy.
Most Vac meters will struggle to measure THD signals < 1mV. But a CRO is
useful for measuring below 10mV. I have taped a 1-10mvrms scale beside CRO
screen to allow measurement when using the most sensitive CRO Vac range.
My CROs also have useful switchable amps for x5 or x10.

Well shielded probe cable is essential for wide bandwidth Vac at low levels.
12mm of unshielded probe wire length may allow RF noise pick up to exceed
the signal level you want to measure. Probing a superhet radio near the input
stages may pick up the oscillator signal which obscures the wanted RF signal.
Magnetically induced pick up is not prevented by non ferrous cable shields or
metal boxes.

If your CRO has 15MHz bandwidth, then a high level of 20MHz oscillation
will be seen on the CRO as a wide blurry line which the CRO is unable to
display as a wave form. A 250MHz CRO would have no trouble displaying
any wave up to 250MHz, but usually only if DUT circuit resistance < 50r.
Most DIY audio enthusiasts will not have to deal with anything above 300kHz.
But unwanted oscillations up to 100MHz do occur in gear you have made or
you have to repair. When I first used a 2SK369 + triode for a cascode input
stage in MC amp, the circuit oscillated above 20MHz.
The measurements of audio Vac and Vdc seemed odd.
The circuit layout included unintended L and C elements forming RF resonant
LC networks. What appeared to be only an audio amp was also an RF oscillator.

Presence of high RF oscillations may be impossible to see on a CRO
but their presence may become obvious by just touching 0V points with a
short lead to the metal chassis/case. The DUT output should be connected
to an audio amp and speaker set for low levels. The touching of points along
the 0V rail with screw driver, or shunting of 0V points to chassis with short wires
should be always inaudible. But if you hear a click during a touch or shunt
procedure, it is because the RF ceases or starts which causes a rapid Vdc
change which is heard as an audible click. The act of measurement of Vo
from an wide BW signal amp will often start HF oscillations because the
added 100pF from probe lead causes 90 degrees phase shift at HF so NFB
becomes PFB and it oscillates. Usually, using a 220r added in series to probe
with 100pF prevents the phase shift at the amp, so no oscillations.
The probe F2 pole is 7.2MHz, allowing high enough F measurement,
but loading at 7.2MHz is only 308r. For Vdc measurement, a 47k resistor between
probe end and DUT prevents any shunt C affecting the circuit. 
All coax cabling or cables with a parallel pair of wires have properties not
easily understood because each conductor has inductance and there is
distributed capacitance between conductors along the cable length.
This means long lengths of cables act as "transmission lines" - and you need
to Google more about them because I don't have time to define and explain
everything. But short lengths of coax cable used for probe leads or audio
interconnect cables can be considered to have low inductance, and low
resistance, and simple shunt capacitance between inner wire and outer shielding.
Used carelessly, coax cable C can cause phase shift and oscillations at HF.

Coax cable bandwidth depends on the source impedance feeding the cable
input and the terminating impedance and cable length. Without providing
more info about coax cable properties, you might assume that the lower the
source R and termination R become, the wider the bandwidth.
Coax cable is designed for a "transmission line" and cable losses per 100
metres may be quoted but the properties are only valid when source and load
resistances are 50r or 75r, and you have cable lengths > 2 metres.
Coax cable data is not relevant to a DIY enthusiast trying to fix an old radio
with fairly high circuit Z throughout, and using the very minimum of test gear.

The BEST description of basic oscilloscope probe properties is at :-

There is much which may be applied to measuring Vac.

Aluminium top cover off, and steel cover off top left box with Amp 1, and input range switch.

Inside the Amp1 box with range switch. It is a bit messy, but is typical DIY wiring
with discrete parts, and final result is after many variations to overcome
many problems to get optimum results.

Back to Education and DIY directory.

Back to Index page.

Perhaps these tables are useful :-

Vac scale 0-100 Read off
Vac scale 0-32 Draw

Vrms scale 0-100.0
dB scale
-21 to +3

DUT Circuit R
F2  -3dB 100pF
F2  -3dB 15pF