Tube tester and curve tracer
Ken's Triode Tube Tester 2016
Some images of plate curves produced by the device under development.
The above image is an early test using a 12AX7, as the B+ capacitor discharges at 10 volt intervals the bias is stepped (in the case above 0v, -.5V, -1V, -1.5V, -2V, -2.5V, .3V) and current readings taken, the entire set of curves are produced in one capacitor discharge. Improvements have been made since this trace was taken, I shall do some more and post them soon.
A grid stopper cleaned things up a lot, the data is collected 100us after setting the grid voltage. BTW The B+ discharge capacitor is 110uF.
19 September 2014. A used but quite good EL34. The maximum negative grid bias can be a limitation with power tubes. 243 data points taken by the tester and used to produce this chart in open office. This seems to be a good test of power tubes, it is easy to spot weak tubes, It will be noted some data points were collected at greater than the tubes maximum dissipation rating, the measurements are taken so fast the temperature rise in the anode is minimal, as stated previously the entire test uses the equivalent power of less than 5 watts for one second. The 0 to 300V (x axis) scale, the 0 to 300mA (y axis) and the bias voltage legend to the right of the chart are generated automatically by Open Office or Excel. Simply import the data and create a scatter graph.
Testing has been carried out with these tubes so far, 6L6, 6V6, 6F6, EL84 and a EL34. An plug in adapter was made up to test the EL84 as only one octal socket is wired up for power tubes.
Primary Objectives of this project:
1/ To build a simple to use device that would enable non technical people to test the most common tubes used in the type of tube amplifiers used by musicians.
2/ To provide some idea of the tubes remaining life and to provide some tube matching information, it is assumed the tube works.
3/ Have the device powered from 12 volts DC.
4/ The device must work in a stand alone mode, i.e. no phone, PC or tablet required.
5/ If possible using a Bluetooth connection export data to a PC, Bluetooth proposed and not hard wiring for safety reasons.
6/ Have no continuous high voltages present, the device is to use a capacitor designed for use in a small xenon camera flash, it is charged just prior to a test and discharged during the test.
7/ Require no special software to import data to PC. Bluetooth serial port with ASCI pre formatted for easy import to any spreadsheet.
8/ Use triode connection of pentodes and beam tetrodes, if a tube won't perform in triode connection it cannot perform in pentode connection.
9/ Use a simple, select tube and press the start button to test and display on the inbuilt LCD transconductance, plate resistance and the percentage of new tube trans-conductance. Further pressing of the start button to display, amplification factor, anode current, bias voltage, test voltage and the raw data readings. A Bluetooth broadcast of basic test result to occur.
10/ Within the limits of the available B+ voltage (300V) and bias voltage (-25V) the device will export a set of triode curves, this output only sends formatted data to Bluetooth nothing on LCD.
11/ Be electronically robust and withstand a anode to cathode short when the start/test button is pushed.
12/ Be able to test tubes at maximum dissipation while limiting the absolute power used per test to < 5 joules.
The project so far:
The image above is of the prototype unit with an EL34 plugged into it.
How it works:
A 6V6 is plugged in and selected with the Chicken Head Knob.
The 6V6 pre-programed test points are transconductance measured at 250V on the screen and plate (triode connected) with test points at -12V and -13V bias (center value -12.5V).
The anode resistance is measured at 240V and 260V, it is the slope of this line, 250V is the center value, if you look at a 6V6 triode curves you will see the line is quite straight between these points.
The slope of this line, using ohms law can be used to determines the plate resistance at that point.
The Start / Next button is pressed:
After about 10 seconds for a 6V6 the following appears.
The LCD displays:
The transconductance as measured.
The plate resistance in K ohms.
A percentage of transconductance as compared to the RCA book value.
Note: If another tube type fitted it will read the tube parameters at the above test points, the percentage value may be meaningless as it is derived from the specification of the selected tube.
The Start / Next button is pressed again:
The amplification factor is displayed.
The Start / Next button is pressed again:
The current at 260V -12.5 volts bias is displayed. It can be seen this data was taken at a dissipation of 11.46 watts so the tube is working at a reasonable level.
The Start / Next button is pressed again:
The B+ voltage at the end of the test is displayed, it should be 240V for the above test, this is read by the unit at the end of the test.
The Start / Next button is pressed again:
This screen displays raw data as read by the ADC to determine the above results.
The 12 bit ADC has a practical range of zero to a little over 4000, I like to see all values over 1500 to obtain reasonable resolution.
The Start / Next button is pressed again:
Back to the first results screen.
The Stop button is pressed:
Back to the menu tube select screen.
-------------------------------------------------------------------
Bluetooth
At the end of each test a small Bluetooth module broadcasts a result.
A 6V6 was plugged in about 1/2 hour ago and the start button was pressed 11 times. Any software that can view a bluetooth serial port at 9600 baud can receive the data.
Below is the information as received by my laptop the numbers at the right are for user convenience.
mA/V=4.37 mu=10 k=2.3 mA=41.9 @260V -12.5V bias #2
---
mA/V=4.35 mu=10 k=2.3 mA=42.0 @260V -12.5V bias #3
---
mA/V=4.37 mu=10 k=2.3 mA=41.9 @260V -12.5V bias #4
---
mA/V=4.39 mu=10 k=2.3 mA=42.0 @260V -12.5V bias #5
---
mA/V=4.37 mu=10 k=2.3 mA=42.0 @260V -12.5V bias #6
---
mA/V=4.39 mu=10 k=2.3 mA=41.9 @260V -12.5V bias #7
---
mA/V=4.37 mu=10 k=2.3 mA=42.0 @260V -12.5V bias #8
---
mA/V=4.37 mu=10 k=2.3 mA=42.0 @260V -12.5V bias #9
---
mA/V=4.37 mu=10 k=2.3 mA=41.9 @260V -12.5V bias #10
---
mA/V=4.39 mu=10 k=2.3 mA=41.9 @260V -12.5V bias #11
---
mA/V=4.37 mu=10 k=2.3 mA=42.0 @260V -12.5V bias #12
---
It can be seen the result changed a little from test to test, the mA/V maximum was 4.39 mA/V and the minimum 4.37 mA/V, less than 0.5 %.
The mA at 260V -12.5V bias varied between 41.9 mA and 42 mA about 0.25%.
No Bluetooth broadcast of raw data is provided.
--------------------------------------------------------------
Plate Curves.
Leaving the 6V6 in place from the above tests. The menu item below was selected.
The start button was pushed and about 15 seconds later a block of data was received via the Bluetooth it was pasted into Open Office and a scatter graph produced, see below.
It can be seen the tube current exceeded the span limit of the menu selection, there are some options:
1/ Delete the faulty data points, on this range the graph can be stretched out a long way in the x y axis with good resolution, very handy at times.
2/ Run the test again with a different menu selection as shown below.
This is menu selection # 13 used for the above graph, it has a span of 350 mA and different grid voltage spacings.
This particular setting allows currents up to 360mA, as you can see the tube types mentioned are only a guide, you could test a lot of tubes with this setting provided an adapter plug was made to suit.
--------------------------------------------------
Menu Selections
This is designed for easy testing of guitar amp tubes.
Select the tube with the Chicken Head Knob and press start.
8 tests availiable (results on Liquid crystal display)
#1 12AT7 ECC81
#2 12AU7 ECC82
#3 12AX7 ECC83
#4 6V6
#5 6L6
#6 EL34 6CA7
#7 octal A power tube with suitable wiring plugged into the octal socket. The actual test points suit a EL84
#8 not allocated at this point in time
8 plate curve tests availiable (results Bluetooth only)
#9 12AT7 ECC81 or similar tubes
#10 12AU7 ECC82 or similar tubes
#11 12AX7 ECC83 or similar tubes
#12 6V6 or similar tubes
#13 6L6 or similar tubes
#14 EL34 or similar tubes
#15 not allocated at this point in time
#16 not allocated
---------------------------------------------
Calibration
If the stop button is held while the power is applied a 16 selection calibration menu appears.
If the load default values page is selected then save values to memory (EEPROM) is selected the unit will function quite well, the accuracy determined by the tolerance of the components used and firmware, this will normally produce better results than a calibrated tube tester produced during the heyday of vacuum tubes.
27 Oct 2014
About to redraw the board with a large number of refinements, it is so easy to add a lot of components and blow the cost/complexity out. Need to think about it for a while. What do I want/need.
07 October 2012
The PCB has been redrawn, and gerber files sent to the PCB house.
Modifications:
Two high-side FET switches that will safely handle a anode to cathode short when the 110uF cap is switched to the circuit under test, harder to make satisfactory was flash-over protection just after the FET switched on, early tests showed a 250mA fuse would go off like a flash lamp and blow the FET, the circuit can now handle the fast current rise time and the fuse is not required. Thank goodness for spice simulations, an actual test circuit was short circuited about 25 times with 350V on the low ESR capacitor, BTW 300V used for actual testing of tubes.
A firmware controlled capacitor discharge circuit, testing low current tubes like a 12AX7 was a bit slow.
Provision for a much higher negative bias voltage, the bias generator will be switched off while the ADC reads, capacitors and a firmware controlled regulator will hold the required bias voltage.
A firmware controlled linear heater supply. As a 5 bit DAC was available, 220mV steps are used, the series pass transistor is a TO-247 to lower the thermal resistance to the aluminum case. The heater current is now rated at >1.6 amps (KT88, 6550A) and is short circuit proof. A linear supply is used as it is very low noise. Series pass transistor dissipation will be < 9.2 W maximum. It seems there is a method of determining tube life by rerunning the tests at a lower heater voltage and comparing the results using said algorithm. I shall give this a try.
A socket fitted to suit an eBay Bluetooth module.
Changed the output socket to a 14 pin IDC for easy change of valve/tube sockets, IDC pins 1-2-3 -6.3V, 4-5-6 +6.3, 7-8 grid,9-10 cathode, 11-12 not used for extra HT isolation, 13-14 anode1 and anode 2.
1 November 2014
The new PCB has arrived and I have loaded all the components, the program has been extensively modified as many things on the PCB were modified and 96% of the program memory had been used in the original version, at the moment I have I have all the original features reinstated with a program memory usage of <70% there is now room to add some more features.
The new software adjustable heater supply works well and has a current limit of 2 amps.
The new negative bias generator easily generates 40 volts and the design will allow it to do twice that voltage with some component changes, as expected some interference is produced as it employes one inductor (the first one was inductor less), testing will confirm if it needs to be off when measurements are made, at the moment it has an unshielded coil and the B field is detectable. The E field should not be a problem.
The high side switches B+ work a treat, the over current protection has not been enabled as the special high voltage SOT-23 transistors have not arrived, this will only affect the testing if I plug in a shorted tube then repairs may have to be made.
5th November
Above is an image of the new board under test, the series pass transistor on the heatsink will be soldered directly on the PCB and thermally coupled via an aluminum block to the die-cast aluminum case. The ribbon cable exiting to the right of the image goes to the start and stop buttons and the potentiometer with the chicken head knob. Note extensive use of 2512 resistors. The plug in board at the top is for Bluetooth. None of the TO-220 devices get warm, they are used as are a small footprint and very economical, IRF840's are used for the B+ generator and for the two high-side switches. Still missing the high voltage SOT-23's for the short circuit protection so the shorted tube test has not been performed.
This is a view of the underside of the board, no i/o on the micro is expected to deliver over 1mA, heavy currents only flow in the power supply ground plane, all switching power supply's and unnecessary micro functions are off when the ADC reads.
A four layer board would be good, the two layer version seems fine, a little larger maybe as I did not want to keep placing tracks on the ground plane.
9 May 2015
This project is on hold due to high work commitments, development will eventually continue.
25 Feb 2016
Not forgotten simply delayed!
I will again change the micro again as they seem to grow in power and memory size by the day with no increase in cost. The cry from potential customers is ........ Just get it on the market!
6 November 2019
I'm back!
This project has been terminated as it was found that pulse testing can make weak tubes look better than they actually are; running the heater for a couple minutes then burst testing makes the emission look better than it actually is.
It was also found the grid current on some tubes passed a pulse test and subsequently failed after running 10 minutes at near maximum dissipation. I have started work on a stand alone micropressor tube tester that can run tubes continuously at full power, it uses a receipt printer to print a full report.
LCD 300w