Addition to the user manual for the AVR Transistor tester

K.-H. Kübbeler
is only valid for the test version 0.94k

Introduction:
Based to the Software of Markus Frejek I had started to modify the software.
At the beginning the reason for this was a problem, which I had with the programming of EEprom.
Because writing to the flash memory was without problems, I believed that it is the quickest and
best way to get a run capable tester, if I put the texts and parameters away from the EEprom and put
them in the flash memory instead. Beginning from version 0.92k you can select with option
USE_EEPROM, if EEprom should be used or not. By analysing the software I had an additional
idea, which I have implemented as a test:
In order to display the voltage, values are needed in mV and not in steps of the ADC. This was done
by original software for every value which is displayed. On the other side the original function
ReadADC reads the ADC Value twenty times, adds every value and divide the sum by twenty. The
resolution of the result is again 5000mV/1023 (same as ADC). If I read the ADC value instead of
20 times now 22 times, build the sum and double the sum and divide by nine, then my maximum
value is 22*2*1023/9=5001, what matches nearly perfect to the wanted mV resolution. But with
this idea real work started.
Now all “if statements” in the program must be adapted to the new resolution. So I learned more
and more about the software of Markus Frejek. My actual version of the function ReadADC reads
the ADC value 45 times, but adds only the last 44 values and divides the result by 9. Because I had
the initial ambition to put the software of the reduced version (without resistor and capacitor
measurement) in 4K flash, I did modifications which were not really necessary. So I replaced the
wait loops by calls to a new written assembler routine, which uses only 66 bytes of flash, every call
need only one instruction, but serves a total range from 10µs to 5s in steps of 1 2 3 4 5 10. The
routine includes the Watch Dog Reset for all calls above 100ms.
Wait calls with interim value such as 8ms need two calls (5ms and 3ms). I don't know any
implementation, which is more economical if you use many wait calls in your program. The wait
calls matched the exactly delay time if the lowest time wait call does. Only the 100ms wait time
calls are 1µs longer if your clock is 1MHz because of the additional Watch Dog Reset. The calls
uses no registers, only the Stack Pointers for the return addresses in the RAM (at most 28 Byte
stack space in current release) is used. The software version for 8MHz clock needs less memory
than the 1MHz version, because the LCD-functions can then use this calls too. In this case the
additional calls 1µs,2µs,3µs,4µs and 5µs are possible.

1
 New features and modifications

1. Measuring of resistors are upgraded, so that potentiometers can be connected. The display
format for this measurement is x-□-y-□-z in the first row, where x, y and z can be any of
the pin numbers 1-3. The second row shows the value of the resistor which is connected to
pin x and pin y followed by the value of the resistor connected to pin y and pin z. If you
don't want to compute the total value yourself, disconnect pin y and start measurement
again. If the potentiometer is adjusted to one of its ends, the Transistor tester cannot differ
the middle pin and the end pin!

2. Capacitors are displayed with its symbol in the format:

Pin number - symbol - Pin number.

3. The measurements of capacitor values are done by measurement of load time. The original
software did this with a program loop, which reads the corresponding digital input pin until
switch occurred and count the loop cycles. This has the handicap that the resolution is
limited by the total time consumption of one loop cycle. I have replaced this loop for little
capacitor values (about <50µF) by a technical feature of the AVR that the counter can save
its counter value by a external event. The counter can operate at full clock rate (1MHz or
8MHz). The external event can be build by the output of the comparator. The comparator
can operate with any ADC input pin and Band cap reference. So I discharge the capacitor,
prepare the comparator to the proper pin input, start the counter at 0 and start charging of the
capacitor with the 470kΩ resistor. Now I check in a program loop, if the counter flags
signals a overflow event or a input capture (external) event. I count the overflow events until
I detect the input capture event. In this case I stop the counter and check if I must count a
additional overflow, because the counter can't be stopped by the input capture event. The
input capture counter and the overflow counter built together the total time, from which I
subtract a experimental find out constant to eliminate the measurement offset. I don't know,
if this constant must be adapted to other boards or processors. Capacitors with bigger values
as about 50µF are measured in an previous test. This is done by up to 500 load pulses with a
length of 10ms, which is done with the 680Ω resistor. Load pulses are repeated until the load
voltage reached more than 300mV, measured without any load current. The value of the
capacity is then computed from the count of load pulses and the reached load voltage from a
table. The table contains the factors to get the capacity in nF units from load time and the
reached voltage with a spacing of 25mV. Interim value of voltage will be interpolated.
As a result of the lower load voltage the measurement time is much faster, because this
advantage works also on discharging.
Furthermore a diode, which is parallel connected to the capacitor don’t disturb the
measurement in most cases, because the flux voltage of most diodes is not reached.

4. Capacitors are measured only in three combinations: Pin 1 (-) and Pin 3 (+), Pin 1 and Pin 2,
plus Pin 2 and Pin 3. The measurement with the other polarity are omitted.

5. Unfortunately the measurement of capacities had not the expected accuracy with respect to
different AVR's, what I know from a single user's response. As a reason I assumed the
difference of the internal 1,3V reference voltage, which was used with the comparator.
Therefore the actual release can use a table with the theoretical dependency of the load time
in respect to the comparator voltage. The table is spaced in 50mV steps and will be

2
 interpolated according to the actual reference voltage. This function uses some of the limited
flash memory. I hope that the reason for the different measurement values is found. Now
the reference voltage is read by every power on if this function is selected with the option
WITH_AUTO_REF. I noticed that the reference voltage is permanently somewhat to low, so
that you can choose an offset with the Makefile option REF_KORR. The measured
reference voltage will then be corrected (added) by your value (mV units).

6. Discharging of capacitors is changed. If the voltage is below 1300mV the capacitor is

shortened by the output pins of the connected ADC port (Port C). I believe that this is legal
because every output port has a built in resistance of about 20Ω. The data sheet Figure 149
(page 258) shows curves up to 2V. Of course I can not guaranty, that no damage can occur.
I have tested the function with a 15mF Capacitor many times and I have never noticed any
problem. The current should be below the specified limit of 40mA. Damage can occur if you
don't discharge a (high voltage) capacitor before connecting it to your tester.

7. The format of displaying diodes is replaced by an diode symbol surrounded with the pin
numbers. This should be known by every user with any first language. Unfortunately I see
no way to show different symbols for special diodes like breakdown diode. Please notice the
pin numbers instead. If the outside pin numbers of two diodes are identical, it can be a two-
in-one LED, a breakdown diode or something else. Notice the flux voltages!

8. The measurement of single diodes is supplemented by a capacity measurement in inverse

direction. Probably is is possible to select a diode for different purposes. I have measured
values between some pF up to 5nF (base emitter diode of a BU508A Transistor). A
experimental parallel connection of a diode and a capacitor with 330µF was detected
correctly. If the capacitor has greater value, only the capacitor is detected. Of course this
measurement can also be done for diodes of bipolar transistors if you connect the base and
only one of the other terminals (collector or emitter).

9. To save program place, the format of displaying pin numbers of bipolar transistors is
changed to the form EBC=xyz . Where EBC means Emitter, Base and Collector and xyz
means the sequence of the corresponding pin numbers.

10. The measurement for bipolar transistors is also done with common collector (emitter
follower). The hFE value of High Power transistors a more acceptable if measured by this
way. In this kind of connection there is no risk to overload the base, even if the base is
supplied with the 680Ω resistor. In case of Darlington transistor the voltage over the base
resistor is too little, so this measurement is done also with the 470kΩ base resistor (release
0.92k). The normal measurement with common emitter is also done as before and the
higher value of hFE is displayed. Because Darlington transistors can have very high hFE
values (>20000), the display layout is changed from hFE= to B= to save space.

11. The differences of reference voltage of ATmega8, ATmega88, ATmega168 and ATmega328
are applied as noted in the data sheet. You can also use the feature WITH_AUTO_REF with
a Makefile option, in this case the reference Voltage is read out with ADC. The factor to
transform the measured load time to capacity is dependent of the reference voltage. If the
curves in the data sheet are not only for one example, reading of actual reference voltage
should not be required, but I'm in doubt about it.

3
12. The measurement of big capacitor values is expanded to values up to 100mF. Because I
don't have such big capacitors, I could not yet test this. Capacity of Gold capacitors could
not measured correctly by this method (too quickly).

13. Use of the AVR with 8MHz crystal is added to the Makefile. The correct fuses for the
different ATmega's is selected with the „make fuses-crystal“ call. If you prefer the operation
with 1MHz, the clock divide by 8 fuse is selected automatically for the ATmega88 line. The
ATmega8 must be connected with a 1MHz crystal, if you wish to use the 1MHz crystal
mode (no clock divide).

14. Beginning from version 0.92k the program can be configured with Makefile options. You
can select a supported language, if you wish to use measurement of resistors or capacities, if
you want a serial output, if the battery voltage is displayed at the beginning, if you wish to
use the self test function and if the program should use the EEprom. Additionally you can
select your clock frequency (1MHz or 8MHz). You can also select which processor you have
installed (m8, m48, m88, m168 and m328) and which programmers (avrisp2) you have
connected if you use avrdude as programmer interface (call: make upload). The avrdude
program checks signature before any load of program data is done. If the correct processor is
not found, program terminates with a error message. You can select a clock generation with
internal RC generator (make fuses) and a clock generation with a external crystal (make
fuses-crystal). You can use the crystal version only if you have installed a 8MHz crystal
between pin 9 and pin 10 of your ATmega. If you wish to use an 1MHz clock operation with
crystal, you can do this. If you have a ATmega88 ATmega168 or ATmega328, the clock
divide fuse is selected automatically by Makefile. Only for the ATmega8 you must install
the 1MHz crystal if you wish to use a 1MHz clock operation with a crystal. Linux-user
should be able to get an ready transistor tester with 4 steps:

• Edit the Makefile, select your processor type, your programmer and options
• call „make“
• connect the ISP-plug and call „make upload“
• if necessary, call „make fuses“ or call „ make fuses-crystal“ and disconnect the ISP-
plug

15. Beginning from version 0.92k the ReadADC function adds 4 to the sum before dividing by
9 (round up to next integer).

16. Beginning from version 0.93k you can select the special option AUTOSCALE_ADC for
the ReadADC function, so that not only the 5V reference is used for measurements,
additionally the internal reference (1.1V for ATmega168) will be used if the input voltage
enables that (Input voltage must be lower than 1V). Especially if the software switch the
reference from 5V to 1.1V, the software must wait more than 5ms until the selected
reference is stable. The reason for this is the external installed 100nF capacitor at the AREF
pin (21) of the ATmega. Switch back to 5V is much quicker (300µs).
The necessary additional wait time results in a significant greater measurement period for
capacitors with big values, because very often must be switched between 5V and 1.1V
reference. The resolution of ReadADC function is always mV!

4
 17. Version 0.94k has a new option in the Makefile „NO_AREF_CAP“, which will reduce the
wait time between switching the ADC reference between 1.1V and 5V. You can use this
option only if the 100nF capacitor connected to the AREF pin (21) is removed or
replaced with a 1nF capacitor. I have not noticed any degradation of measurements if this
capacitor is removed. This option does have effect only, if the AUTOSCALE_ADC option is
selected too. I have installed a 1nF capacitor after all.

18. Version 0.93k changes the measurement of resistors. The measurement of voltage of the
directly switched pin (ADC port) is removed, only the Pin which is connected across a
resistor to the power (VCC or GND) is measured. The potential drop of the direct
connected pin is computed with respect to the known internal resistance of the switched
port.
Four different measurements are made:
Low-Pin connected to GND and High-Pin connected across 680Ω to VCC
Low-Pin connected to GND and High-Pin connected across 470kΩ to VCC
Low-Pin connected across 680Ω to GND and High-Pin connected to VCC
Low-Pin connected across 470kΩ to GND and High-Pin connected to VCC
Corresponding to the measurement results, two of the measurements are selected (680Ω or
470kΩ pair) and the results are averaged. If the AUTOSCALE_ADC option is selected and
the voltage of one result is below 0,99V, an weighted average is build with factor 4 for this
value, the other value is weighted with factor 1.
In table B2 you can see the benefit (measurements around 22kΩ).

19. In version 0.93k have I added fill characters to the different texts to get the same length for
every implemented language. In version 0.94k I have removed this fill characters again,
because I had noticed, that there is no space left in EEprom for the additional data required
to implement the option LCD_CYRILLIC. To make this option selectable again, the overrun
area of the table RLtab was additionally removed.

20. By optimizing the program in version 0.94k it is possible to select the self test function
together with all other options without AUTOSCALE_ADC for a ATmega8. Therefore the
output of capacity value and the output of resistor value now uses a common function.

21. The computation of the current amplification factor for bipolar transistors is in version 0.94k
done with „long int“ (32 bit) , but result is limited to 65535 (match 16 bit). The overrun
problem was reason for the implausible results of previous versions (BC516 measurements).

I think, that optimising of measurement results is nearly finished. Some more response from users
would be helpful.
I hope, that I have not forgotten any important item and will introduce a last item, for which I will
spend a separate chapter.

5
 Self test Function
Beginning with release 0.9k I have implemented a self test function. Usage is very simple. If you
have installed test terminal with clamps, put all clamps together to a piece of uninsulated wire and
press the start button. The program notice the shorten probes and start the self test function. After
finishing the self test the transistor tester will continue with normal measurement, if no equipment is
connected, the program will end with „part unknown or damaged“. The unhappy side of the self
test function is that the 8K flash is used near the limit. The length of the ATmega8 version 0.9k is
about 8000 bytes. The length of the ATmega88 version is with 8122 byte very near at the limit. Use
of EEprom (with option USE_EEPROM) is one of the only possibility to save memory. Some
functions like ReadADC are already implemented in assembler syntax.
The separate steps of the self test function is generally displayed on row 1 of the LCD display with
the letter T followed by the step number. Every step is repeated 8 times, before the program
continues with the next step. In every step only measurement results are displayed, no error analysis
are done, you must interpret the results yourself.
At this place I will give you an additional important hint. Never do a measurement with connected
ISP plug!
The ISP interface influences the measurement.

Here are the Test steps:

1. Measurement of the 1.3V (or 1.1V) reference Voltage (Band gap Reference). In row 1 the
text „Ref=“ and the measured Voltage in mV is displayed. The second row shows the
resulting factors for capacity measurement.

2. Comparing of the 680Ω resistors.

In row 1 the cryptic text „L1+2- .3- 2+.“ is shown. Meaning of this is as follows: The L is
the symbol for Low meaning the 680Ω resistors. The 1+ stand for: resistor at pin 1 is
connected to VCC (+). The following 2- means: A GND (-) connected resistor at pin 2.
The result of this measurement is displayed in row 2 at the first place.
In row 1 follows now a „ .3-“ which means, that the first connection of measurement 1
continues(.), but that the GND (-) connected resistor at pin 3 is now in action. The result is
displayed in the middle place of row 2.
The last measurement of this test „2+.“ means that now the resistor at pin 2 is connected to
VCC (+) and the GND connection of measurement 2 continues (.). The result of
measurement is displayed at the last place of LCR row 2.
Please remember, that the resolution of the ADC is about 4.88mV!
All these combinations with respect to the internal resistance of the pins should result to:
5001 / (18+680+680+22) * (18+680) = 2493 .

3. Comparing of the 470kΩ resistors.

Now the display shows in row 1 „H1+2- .3- 2+.“. The same procedure as done in step 2 is
repeated with the 470kΩ resistors (symbol H).
Result should be nearly 5001 / (18 + 470000 + 470000 + 22) * (18 + 470000) = 2500 for all
combinations.

6
 4. In this step nothing is measured, but the order is displayed „ isolate Probe“,
which means that it is time to separate the probes (release from wire).

5. This step tests the capability of GND (-) connected 470kΩ resistors (H) to pull the test pins
to GND . Row 1 shows the text „RH-“ .
Row 2 should display zero for all three pins.

6. This step tests the capability of VCC (+) connected 470kΩ resistors (H) to pull the test pins
to VCC (+). Row 1 shows the text „RH+“.
The best value for this three measurements is 5001.
Great differences from the best value for test 5 and 6 are errors such as isolation problem,
flux material or damaged port.

7. Measuring of internal resistance of pin output switched to the GND signal.

The text in the 1st LCD row is Ri_Lo = (mV). In the second row of the LCD three voltages
were displayed. The internal resistance of the port C outputs switched to GND (-) are
measured with the current of to VCC (+) switched 680Ω resistors. Only the three pins of the
ADC port are measured, the resistor port B (PB0,PB2 and PB4) can not be measured
without hardware modification. Is is assumed that the port resistance of the different ports
are nearly identical. To get the resistor values, you must divide the displayed mV values by
about 7 (see test 8).

8. Measuring of internal resistance of port outputs switched to the VCC (+)signal.

The needed current is generated with to GND connected 680Ω resistors .
The text in the 1st LCD row is Ri_Hi= (mV). In the second row of the LCD three voltages
are displayed (in difference to VCC). It are the same measurements as those in test 7 to the
other side.
With the following steps you can get the resistance:
To get the voltage of the 680Ω resistor: 5001 – (result of test 7) – (result of test 8)
To get the current build: (voltage of 680Ω resistor) / 680
Then you can get both resistor values by dividing the voltage (result of test 7 or 8) by the
current.

9. A 50Hz rectangle signal is generated on Pin 2 and the same signal in opposite direction on
Pin 3. Pin 1 is switched to GND . The current is limited with 680Ω resistors. This test is
repeated 8 times with 5 seconds period each.
You can check the time of the wait calls, if you have an oscilloscope or frequency counter.
If you don't use the crystal clock version, the result may be inexactly.
A exactly clock frequency and wait time are important for measurement of capacity values.

At the end of test function the text „Auto Test End“ is shown in row 1 and the version number of
software is shown in row 2. Then the program continues with the normal measurement task.
Beginning from version 0.93k, a test step is not further repeated, if the start key is pressed. If
you leave the key pressed, every test is executed only once. You can configure self test only
together with all other options for a ATmega168 (ATmega328 untested) . For the ATmega8 you
must at least omit the option AUTOSCALE_ADC because of the limited flash memory.

7
 Appendix A
Result of self test
Software Version 0.94k

Microcontroller 1. Result 2. Result 3. Result

Mega8 @ 8MHz Test 1 Band gap Ref 1237
Reference Voltage
should be 1298mV RHfakt. 753 RLfakt. 4887
Signature 1E 93 07
WITH_AUTO_REF Test 2 RL1+ RL2- RL1+ RL3- RL2+ RL3-
comparison 680Ω 2488 2488 2484
best value: 2493
Test 3 RH1+ RH2- RH1+ RH3- RH2+ RH3-
comparison 470kΩ 2493 2493 2493
best value: 2500
isolate probe
Test 5 RH1- RH2- RH3-
470kΩ (Isolation) 0 0 0
best value: 0
Test 6 RH1+ RH2+ RH3+
470kΩ (Isolation) 4995 4995 4995
best value: 5001
Test 7 TP1- RL1+ TP2- RL2+ TP3- RL3+
Pin resistance Low 132 132 137
Test 8 TP2+ RL1- TP2- RL2+ TP3- TP3+
Pin resistance High 151 151 151
(VCC - voltage)=

Mega168 @ 8MHz Test 1 Band gap Ref 1090mV

Reference voltage
should be 1102mV RHfakt. 865 RLfakt. 5649
Signature
1E 94 06 Test 2 RL1+ RL2- RL1+ RL3- RL2+ RL3-
comparison 680Ω 2493 2493 2493
Test 3 RH1+ RH2- RH1+ RH3- RH2+ RH3-
comparison 470kΩ 2497 2498 2498
isolate probe
Test 5 RH1- RH2- RH3-
470kΩ (Isolation) 0 0 0
Test 6 RH1+ RH2+ RH3+
470kΩ (Isolation) 4998 4998 4998
Test 7 TP1- RL1+ TP2- RL2+ TP3- RL3+
Pin resistance Low 131 132 132
Test 8 TP2+ RL1- TP2- RL2+ TP3- TP3+
Pin resistance High 156 156 156
(VCC - Voltage)=

8
 Appendix B1
Results of resistor measurements
Software Version 0.92k (old)

resistor Result of Result of Result of

Mega8 @ 1MHz, Mega8 @ 8MHz, Mega88 @ 1MHz
original software Signature 1E 93 07 Signature 1E 93 0A
Signature 1E 93 07
60Ω 0.1% 61Ω 1.7% 62Ω 3.3% 60Ω 0%

120Ω 0.1% 120Ω 0% 122Ω 1.6% 120Ω 0%

240Ω 0.1% 240Ω 0% 240Ω 0% 239Ω 0.4%

340Ω 0.1% 339Ω 0.3% 339Ω 0.3% 340Ω 0%

680Ω 0.1% 677Ω 0.4% 677Ω 0.4% 679Ω 0.1%

1360Ω 0.1% 1357kΩ 0.2% 1352Ω 0.6% 1361Ω 0.1%

3.90kΩ 0.1% 3865Ω 0.9% 3859Ω 1.1% 3896Ω 0.1%

7.80kΩ 0.1% 7676Ω 1.6% 7652Ω 1.9% 7776Ω 0.3%

11.0kΩ 0.1% 10902Ω 0.9% 10.70kΩ 2.7% 10.91kΩ 0.8%

22.0kΩ 0.1% 21.1kΩ 4.1% 21.1kΩ 4.1% 21.1kΩ 4.1%

44.0kΩ 0.1% 43.1kΩ 2.0% 43.0kΩ 2.3% 43.0kΩ 2.3%

50kΩ 0.1% 49.2kΩ 1.6% 49.2kΩ 1.6% 49.2kΩ 1.6%

100kΩ 0.1% 99.0kΩ 1.0% 98.9kΩ 1.1% 98.9kΩ 1.1%

200kΩ 0.1% 196.8kΩ 1.6% 197.7kΩ 1.2% 198.6kΩ 0.7%

270kΩ 0.1% 266.3kΩ 1.4% 267.5kΩ 0.9% 268.6kΩ 0.5%

470kΩ 0.1% 77.46µF ? 467.1kΩ 0.6% 469.0kΩ 0.2%

940kΩ 0.1% 919.6kΩ 2.2% 935.7kΩ 0.5% 940.0kΩ 0%

1.00MΩ 0.1% 973.8kΩ 2.6% 990.8kΩ 0.9% 996.2kΩ 0.4%

2.00MΩ 0.1% 1922.0kΩ 3.9% 1.975MΩ 1.3% 1.996MΩ 0.2%

10MΩ 1% 8433.8kΩ 15.7% 9.574MΩ 4% 10.11MΩ 1%

additional: ---- 43.87MΩ 12% 52.94MΩ 6%

50MΩ 1%

9
 Appendix B2
Results of resistor measurements
Software Version 0.94k

Resistor Result of Result of Result of Result of

Mega8 @ 8MHz Mega8 @ 8MHz Mega168@8MHz Mega168@8MHz
Signature 1E 93 07 Signature 1E 93 07 Signature 94 06 Signature1E 94 06
AUTOSCALE_ADC AUTOSCALE_ADC

60Ω 0.1% 60Ω 0% 61Ω 1.7% 60Ω 0% 60Ω 0%

120Ω 0.1% 120Ω 0% 120Ω 0% 120Ω 0% 120Ω 0%

240Ω 0.1% 240Ω 0% 240Ω 0% 240Ω 0% 239Ω 0.4%

340Ω 0.1% 339Ω 0.3% 339Ω 0.3% 340Ω 0% 340Ω 0%

680Ω 0.1% 678Ω 0.3% 678Ω 0.3% 681Ω 0.1% 679Ω 0.1%

1360Ω 0.1% 1359Ω 0.1% 1358Ω 0.1% 1359Ω 0.1% 1359Ω 0.1%

3.90kΩ 0.1% 3894Ω 0.2% 3895Ω 0.1% 3909Ω 0.2% 3903Ω 0.1%

7.80kΩ 0.1% 7813Ω 0.2% 7782Ω 0.2% 7828Ω 0.4% 7808Ω 0.1%

11.0kΩ 0.1% 11.04kΩ 0.4% 10.96kΩ 0.4% 11.08kΩ 0.7% 11.02kΩ 0.2%

22.0kΩ 0.1% 21.30kΩ 3.2% 21.60kΩ 1.8% 21.30kΩ 3.2% 21.7kΩ 1.4%

44.0kΩ 0.1% 43.10kΩ 2.0% 43.40kΩ 1.4% 43.30kΩ 1.6% 43.6kΩ 0.9%

50kΩ 0.1% 49.20kΩ 1.6% 49.30kΩ 1.4% 49.30kΩ 1.4% 49.60kΩ 0.8%

100kΩ 0.1% 98.90kΩ 1.0% 99.10kΩ 0.9% 99.10kΩ 0.9% 99.50kΩ 0.5%

200kΩ 0.1% 198.1kΩ 1.0% 198.1kΩ 1.0% 198.4kΩ 0.8% 198.4kΩ 0.8%

270kΩ 0.1% 267.7kΩ 0.8% 267.9kΩ 0.8% 267.9kΩ 0.8% 267.9kΩ 0.8%

470kΩ 0.1% 468.0kΩ 0.4% 468.0kΩ 0.4% 468.4kΩ 0.3% 468.6kΩ 0.3%

940kΩ 0.1% 938.6kΩ 0.2% 938.7kΩ 0.2% 940.8kΩ 0.1% 940.8kΩ 0.1%

1.00MΩ 0.1% 995.8kΩ 0.4% 995.3kΩ 0.5% 997.1kΩ 0.3% 998.0kΩ 0.2%

2.00MΩ 0.1% 1989kΩ 0.6% 1990kΩ 0.5% 1998kΩ 0.1% 1990kΩ 0.5%

10MΩ 1% 9894kΩ 1.1% 9857kΩ 1.4% 10.10MΩ 1.0% 9940kΩ 0.6%

additional: 48MΩ - 4% 47MΩ - 6% 53.40MΩ/ 7% 48.3MΩ – 4%

50MΩ 1% 51MΩ 50MΩ not detected 50.2MΩ

The resistor measurement of many multimeter’s ends at 20MΩ or 40MΩ!

10
 Appendix C1
results of capacity measurements
Software Version 0.92k (old)

Capacitor Result of Result of Result of Result of Result of

Multimeter Mega8 @1MHz Mega8 @8MHz Mega88 @1MHz Mega88 @8MHz
PeakTech signature 1E 93 07 signature 1E 93 07 signature 1E 93 0A signature 1E 93 0A
3315 Original software WITH_AUTO_REF

56pF 58pF ---- 58pF 0% (42pF with 28% 58pF 0%

a Diode)
110pF 114pF ---- 114pF 0% 111pF 3% 117pF 3%

220pF 225pF 0.27nF 20% 226pF 0% 222pF 0% 228pF 0%

1000pF 1034pF 1.18nF 14% 988pF 4% 966pF 7% 981pF 5%

3.3nF 3.47nF 3.90nF 12% 3287pF 5% 3223pF 7% 3230pF 7%

7500pF 7.23nF 8.43nF 16% 7241pF 0% 7079pF 3% 7092pF 2%

10nF 10.45nF 11.93nF 14% 10.13nF 3% 9935pF 5% 9922pF 5%

33nF 33.2nF 37.27nF 12% 31.97nF 4% 31.26nF 6% 31.36nF 6%

100nF 97nF 112.36nF 16% 97.02nF 0% 94.98nF 3% 94.93nF 3%

330nF 333nF 385.41nF 14% 333.2nF 0% 325.9nF 2% 325.7nF 2%

1µF 955nF 1.10µF 15% 954nF 0% 934nF 2% 934nF 2%

2.2µF 2.2µF 2.54µF 15% 2198nF 0% 2150nF 2% 2149nF 2%

22µF 21.94µF 25.61µF 16% 21.92µF 0% 21.41µF 3% 21.41µF 3%

47µF 47.5µF 55.29µF 16% 47.30µF 0% 46.70µF 1% 46.9µF 1%

100µF 97.5µF 112.94µF 15% 91.30µF 6% 91.7µF 6% 91.3µF 6%

220µF 229µF 272.69µF 19% 219.2µF 5% 219.2µF 5% 219.2µF 5%

1000µF 1071µF 1371.95µF 28% 1076µF 0% 1063µF 1% 1066µF 1%

2200µF 2.231mF 3308.58µF 48% 2302µF 3% 2288µF 3% 2309µF 3%

4700µF 4.75mF 6103.00µF 28% 5042µF 6% 5042µF 6% 4982µF 5%

14.1mF 14.4mF 5Ω ?? 15.13mF 5% 15.13mF 5% 15.31mF 6%

11
 Appendix C2
Results of capacity measurements
Software Version 0.94k

Capacitor Result of Result of Result of Result of Result of

Multimeter Mega8 @8MHz Mega8 @8MHz Mega168 @8MHz Mega168 @8MHz
PeakTech331 Signature 1E 93 07 Signature 1E 93 07 Signature 1E 94 06 Signature 1E 94 06
5 WITH_AUTO_REF WITH_AUTO_REF WITH_AUTO_REF
AUTOSCALE_ADC AUTOSCALE_ADC
REF_KORR=17
56pF 58pF 58pF 0% 57pF 2% 57pF 2% 58pF 0%

110pF 114pF 114pF 0% 114pF 0% 113pF 1% 118pF 4%

220pF 225pF 225pF 0% 225pF 0% 223pF 1% 232pF 3%

1000pF 1034pF 987pF 5% 987pF 5% 968pF 6% 1007pF 3%

3.3nF 3.35nF 3294pF 2% 3287pF 2% 3214pF 4% 3324pF 1%

7500pF 7.23nF 7241pF 0% 7240pF 0% 7038pF 3% 7310pF 1%

10nF 10.45nF 10.06nF 4% 10.10nF 3% 9841pF 6% 10.20nF 2%

33nF 33.2nF 32.50nF 2% 32.50nF 2% 30.46nF 8% 32.69nF 2%

100nF 97nF 97.06nF 0% 96.87nF 0% 94.26nF 3% 97.71nF 1%

330nF 333nF 332.4nF 0% 332.0nF 0% 322.8nF 3% 334.8nF 1%

1µF 955nF 953.6nF 0% 952.4nF 0% 925.8nF 4% 960.3nF 1%

2.2µF 2.2µF 2195nF 0% 2195nF 0% 2134nF 3% 2214nF 1%

22µF 21.94µF 21.87µF 0% 21.92µF 0% 21.26µF 3% 22.07µF 1%

47µF 47.5µF 47.40µF 0% 47.49µF 0% 47.27µF 1% 47.31µF 1%

100µF 97.5µF 91.68µF 6% 91.54µF 6% 90.72µF 7% 91.68µF 6%

220µF 229µF 218.8µF 4% 220.9µF 4% 218.8µF 4% 216.6µF 5%

1000µF 1071µF 1076µF 1% 1076µF 0% 1069µF 0% 1062µF 1%

2200µF 2.231mF 2289µF 3% 2281µF 2% 2289µF 3% 2246µF 1%

4700µF 4.75mF 5032µF 6% 4999µF 5% 5032µF 6% 4923µF 4%

14.1mF 14.4mF 15.34mF 6% 15.29mF 6% 15.17µF 5% 14.99mF 4%

12
 Appendix D
Results of semiconductor tests
Software Version 094k

semiconductor Result of Result of Result of

Mega8@8MHz Mega168 @8MHz Mega168 @8MHz
signature 1E 93 07 signature 1E 93 07 signature 1E 94 06
WITH_AUTO_REF WITH_AUTO_REF
AUTOSCALE_ADC
1N4148 Diode, 721mV, 0pF Diode, 729mV, 0pF Diode, 725mV, 0pF

1N4150 Diode, 678mV, 0pF Diode, 681mV, 0pF Diode, 682mV, 0pF

BA157 Diode,623mV, 17pF Diode, 631mV, 16pF Diode, 620mV, 15pF

BY398 Diode, 541mV, 0pF Diode, 553mV, 0pF Diode, 542mV, 0pF

1N4007 Diode, 654mV, 13pF Diode, 665mV, 9pF Diode, 658mV, 11pF

LED green Diode, 1954mV, 6pF Diode, 1970mV, 6pF Diode, 1951mV, 4pF

ZPD2,7 2xDi, 729mV, 2659mV 2xDi, 738mV, 2674mV 2xDi, 730mV, 2656mV

BU508A NPN, B=9, 613mV NPN, B=9, 621mV NPN, B=9, 615mV

BU508A B+E Diode, 613mV, 5201pF Diode, 621mV, 5285pF Diode, 611mV, 5344pF

BU508A B+C Diode, 595mV, 261pF Diode, 597mV, 267pF Diode, 591mV, 272pF

2N3055 NPN, B=21, 617mV NPN, B=21, 626mV NPN, B=21, 625mV

BC546B NPN, B=381,780mV NPN, B=376, 777mV NPN, B=387, 771mV

BC556B PNP, B=266, 790mV PNP, B=429, 787mV PNP, B=266, 790mV

BC639 NPN, B=180, 722mV NPN, B=180, 733mV NPN, B=188, 724mV

BC640 PNP, B=185, 716mV PNP, B=227, 725mV PNP, B=187, 719mV

AC128 (Ge.) PNP, B=68, 270mV PNP, B=64, 269mV PNP, B=66, 271mV

BC517 NPN, B=26996, 1419mV NPN, B=28220, 1413mV NPN, B=28250, 1404mV

BC516 PNP, B=65535, 1430mV PNP, B=65535, 1420mV PNP, B=65535, 1417mV

BS170 N-E-MOS,D,2616mV,66pF N-E-MOS,D,2562mV, 67pF N-E-MOS,D,2564mV, 68pF

J310 N-JFET N-JFET N-JFET

BRY55/200 Thyristor Thyristor Thyristor

IRFU120N N-E-MOS,D,4151mV, 922pF N-E-MOS,D,4156mV,894pF N-E-MOS,D,4153mV, 933pF

IRFU9024 P-E-MOS, D,3525mV, 960pF P-E-MOS, D,3525mV,926pF P-E-MOS, D,3534mV, 965pF

ZVP2106A P-E-MOS,D,3217mV, 115pF P-E-MOS,D,3220mV,114pF P-E-MOS,D,3217mV, 113pF

ZVNL120A N-E-MOS,D,1560mV, 140pF N-E-MOS,D,1535mV,138pF N-E-MOS,D,1535mV 138pF

13
 Appendix E
Pictures

Test version with unmounted 4x20 Display

Modifications: 8MHz Crystal (Pin9+10) and

Pull-up resistor (Pin12+13), better (Pin13 and VCC)

14
 Appendix F
Known errors
Software Version 0.94k

• The measurement results of little capacity values vary with the Pin combinations.
Combination 1:2 values are about 3pF less than the values of the other pin combinations
(1:3 and 2:3). This effect is equal on any tested AVR processor.
• Germanium Diodes (AC128) are not detected in all cases.
• Does program work correctly without the automatic shut off?

15
 Appendix T
to do list (not sorted)
Software Version 0.93k

• Add more and better documentation.

• By my tests I have noticed that the measured voltages of the internal Band cap reference is
lower than the data sheets let me expect. The reason is unknown. VCC?, ADC-error?
• Check if transistor tester could get better “interpolated” ADC values if additional noise is
added to the signal or to the ADC reference (see ATMEL document AVR121: Enhancing
ADC resolution by oversampling). If all items are identical, there can't be any enhancement
of resolution by oversampling. Can enough noise be generated with the ATmega counter?
How additional noised affects the upper and lower limit values?
Of course this method can not eliminate all of the ADC errors.
• Test of the effectivity of this method can be done by building a ramp input signal and
monitoring this signal. The ramp signal can be build by slowly charging a big capacitor with
the 470kΩ resistor. The growing of the voltage can then be monitored with the LCD display
in a special part of self test. The difference of the ReadADC function alternatives 44/9,
22*2/9 or 11*4/9 can be monitored too.
• Think about how we can get the real internal resistance of port B output (resistor swiching
port) instead of assuming, that ports are equal.
• Can discharging of capacitors be made more quickly, if the minus pin is additionally raised
with the 680Ω resistor to VCC (+)?
• Who is using the serial port? I did not test this function and even I don't know how.
• Can inductance be tested?
• How measurement results changes by variation of the supply voltage between 4,5V and
5V?
• Find better organisation of directory structure.
• Check if the tester can use floatingpoint representation of values. The risk of overflow is
lower. There is no need to use multiplication and division together to build a multiplication
with a non integer factor. But I don't know how much flash memory must be spend for the
library.
• Write User's guide for configuring the tester with the Makefile options and description of the
build chain.
• Development of a new board with crystal or crystal generator clock for ATmega.
• If the holding current of a thyristor can not be reached with the 680Ω resistor, is it harmless
to switch the cathode directly to GND and the anode directly to VCC for a very short time?
The current could reach more than 100ms. Will the port be damaged? What is with the
power supply (voltage regulator)?
Check the Port afterwards with self test function!
• Can voltage regulators be checked? (Input, Output, GND)
• Can optoelectronic couplers be checked?
• Can the transistor tester act as clock generator for AVR devices, which can't be programed
because of wrong fuse configuration?
• Test if 0,1Ω can be displayed with the AUTOSCALE_ADC option of ATmega168. In this
case probably the ESR measurement of electrolytical capacitors is possible.

16
Additional:
As you can see, the list of ideas and “to do” tasks is still rather long. My purpose is to make the
transistor tester more precise, faster and multifunctional. Probably I purchase a second board for
easier checking of different processors (mega8 and mega168).
I can not promise you, that the task list will ever be empty. Also I can not promise you, that my
software is free of errors. I don't give you any warrenty! Using my software is your own risk.
During my tests never a ATmega or other part was damaged. But I give you the promise that I never
will publish a release without source code!.
By now I develop my software with the GNU toolchain under a Linux (Ubuntu) operating system.
I had got a useful hint from another thread on www.mikrocontroller.net to use my programmer
(Diamex ALL-AVR ) without any problems. The same programmer was not able to program the
EEprom of my ATmega8's with the Windows driver. Now I can also use the original software of
Markus Frejek for my tester. I still hope, that my software become a official version. But I still wait
on answer to my email to 5volt and to linuxgeek from February 2012.

17