* * *

From DATA to SIMULATION (Part One)

2.1a Computer-Aided Measure - Introduction

Collecting data in tables is a tiresome work, even if it's still common nowadays in many famous lab's, so don' t be afraid.... but.... why don' using a computer to do it??.

Nowadays many computer-aided programs for data collection exists, even focused on tubes. Obviously the cost discorage the DIYers. Strictly speaking the cost is maily due to the software part that besides collecting data in a very quick and precise way allows for data analysys.

Obviusly we're interested in spending the minimum and get the result in the fastest way so the IDEA is:

use ready made low cost part

bulding a simple hardware part to collect data

using other (common) programs for data analysys.

As I think we're not interested in speed, nor in up-to date hardware I chose to use my old PC to collect data. This permitted me to use out-of fashion (=low price) data acquisition card.
Normally the only problem with this card is that tey're "slow" as they can make 2-3 data collection each second (!) .Actually this speed is more than we necessary need.
So using this card my old and forgotten PC has became alive: now it's a nice measuring station.

BASIC REQUIREMENTS OF THE A/D - D/A CARD
When I asked my supplier the prices for old Ad/Da card for IBM Pc I realyze that if I had'nt speed problems, for a small amount of money I could afford a "luxury" one.You can have a look at it in my supplier site http://www.micromed.it/catalogoc.htm The code of my card is "SH1".

The SH1 card has the following characteristics.

4 A/D 10 bit converters (1024 values ): input range 0-2V, 0-20V.
2 D/A 8 bit converters (256 values ):output +/-5V
8 TTL I/O programmable in group of 4

Many of these spec's actually outperform my needs.Even with that, this card cost me about 120klire (60USD) bundled with user's manual & evaluation/development software.

Looking at the block diagram of the card I decided to modify the ADC inputs to equalize the ranges.

It was easy: all you have to do is to replace the resitor networks in the input of each ADC with a 100k+100k partition network. Besides I suggesti you to buy metal film with 1% tolerance.
The resistor network is easy to find. You've only to follow the path from the input connector to the TL084.

With this change each input od the ADC is now rated 0-4V range (10bits).

2.2 Doing It ! (Hardware)

Here comes the block diagram. First of all:
NOTE THAT TO AVOID OPTOCOUPLER'S COST, THE GROUND OF THE PC AND THE MEASURMENT HARDWARE ARE IN COMMON. THIS COULD BE DANGEROUS AND CAUSE SHOCK AZARDS TO YOU OR YOUR EQUIPMENT BECAUSE OF THE HIGH TENSION PRESENCE IN THE H.T. SUPPLY SO ...

PLEASE BEWARE.

COMMENTS TO THE CIRCUIT:

As you can see Ia Anode current is "estimate" through the voltage drop across a 5ohm resistor. This simplifies considerably the circuit but introduces a little systematic errors for high currents (vg near zero).For example vith 50ma the voltage drops is 0.25V so:
Vg(actual)=Vg(measured)+0.25V.
I've found that this error is tolerable in many situations so I did'nt try to eliminated it even via software.

The availability of TTL outputs allow us to drive "protection" relays. This realys switch off in case maximum rating are reached or each time no measure is done.

As I couldn't find a High voltage OPAmp to generate the entire 0...-50V Vg range, I used a TTL output to divide the range in two: 0...-6V e -6...-50V (relais "vg range sel."). The first range is useful for signal tubes, the second for power ones. Beside As i had a AD converter more than actully needed I decided to "specialize" one AD for each range of Vg, thus increasing precision.

1.3 Interface+Supplies

The availability of many ADC allow us to use a AD converter "specialised" for each measure. This simplifies considerably the software. Infact all the work could be done actually with only one ADC !!!.

Please note that even if the Vg and Va generators are not so precise in consequence of the combined effect of :

number of steps (256 steps on 300V voltage means that each step is 1.2 V large)
nonlinearities of the circuit

the Va will be measured with good precision, because the AD has a 12bit conversion = 1024 values. For example in my prototipe the scale limit for Va is set to 400V thus the precision is 0.4volt.

As you'll see I've used only low cost IC's (TL81) so the amount of money (excluding power supply) won't reach 70klire (35USD) .

1.3.1 DAC interface - Va, Vg suppliers

Here comes the detailed schema

SUPPLY SECTION

DESCRIPTION

DAC interfaces are composed of level shifters (IC1,IC2) whose aim is to map the +/-5V range at DAC outputs in the range 0..... 6/8V.

This voltage is used to pilot the Va or Vg source.
Note that the AT supplier although very simple has a good linearity. The fet used as serial regulator has a 500-600Vds.

The vg source is divided in two according to the ranges.
The first range (Vg Low) 0/-6v is directly obtained from IC3. This allows good precision at low voltages ( 256 steps in the range 0-6V are about 0.025V large!).
This situation permits very good precision when investigating bias point of low signal tubes.
The second range (Vg High) -6/-60V is made using a Video Driver BJT from scrap (the BJT that drive the Cathode Tubes of TV sets).
Vghigh source has a lower linearity compeared to Vglow, but we'll see in the software section that this'll be overcome using a lookup table.Anyway this source has about 0.5V step, more than enogh for our porpousals.

1.3.2 The ADC Interface

The ADc interfaces are made of IC whose job is to raise input impendance and, where necessary, to amplify the voltages.

MEASURE INTERFACE SECTION

The choice between different amplification factors should be made as to guarantee enough number of counts for the ADC and, at the same time, allowing for a meaningful "dinamic range" of measure.

In details:

Va: IC4 act as a simpler decuopler while the partition network in the input reduces of 1/100 the voltage measured.
Ia: IC5 amplifies of around 40 times the voltage drop across the 5ohm resistor so that a 20ma current (0.1V drop) originate about 4v.
VgL: the low range of Vg (0/-6V) is divided for 1/10 then amplified by IC6 of 3 times.
VgH: the high range of Vg (-6/-60V) is divided for 1/100 decoupled by IC7 then amplified for about 4 times by IC7.

1.3.3 Tuning & Tips

A) Offset setting: you need a DVM connected to each opamp outputs (pin 6). The adujust the multi-turns 5k trimmers to obtain a voltage reading as closa as 0v as possible.

B)Verifying ampliofication factors: You need a reference voltage in the range of 0-4,5V (use a common battery) connected to a 10K potentiometer. Choose a reference voltage to avoid saturation of the amplifier (i.e if the gain iis10 and output of the amplifier should be 4V max choose a value below 0.4V) Apply the reference voltage at the input of each stage disconnecting the partition network (when present). Then set the gain trimmer to obtain the right amplification factor.

Easy isn'it?

Back to summary