My tube curve tester

[jxjbsd]

Three years ago, I began to conceive of using MCU+DAC+PowerMOS to make a digital instrument for testing the characteristic curve of electronic tubes. I have friends who are electronics enthusiasts who are tube amplifiers. They need such equipment. This is what drives me.
I hope it can test tubes including 12AX7, EF86 and 300B KT66. In this way, four voltages need to be generated: two high voltages for the anode and screen grid, one negative pressure for the grid, and the other for heating the filament. Two currents need to be detected at the same time: anode and screen grid.
The HV range is 0-500V/500mA, and the grid voltage is 0-140V in two stages: 0-25V and 0-140V. The current detection one way is divided into two sections 0-20mA, 0-200mA, and the other way is 0-500mA. Heating filament 1.4-20V/3A.
In December 2018, we started to do design verification, experimented with high-voltage amplification, and initially wrote software for MCU to control DAC and ADC, as well as programs for communication between MCU and PC and PC drawing display. I got the first version of the PCB in March 2019 and started soldering and debugging, which exposed many problems. Of course, this was expected. After the change, the second version was made and it was barely usable. After programming and actual use, it was improved to the current PCB version.
   After two years of continuous programming and improvement, the goal at the time has been basically achieved, and most common electronic tubes can be tested. The user interface is shown below. The problem now is that the software needs to be improved.
   However, I saw Mr. Ronald's work yesterday.（https://www.dos4ever.com/uTracer3/uTracer3_pag0.html） After I watched it on his website for two hours, I decided to abandon my project, In comparison, he is obviously more elegant. I decided to give up my own project, but I will not give up the curve instrument, decided to buy his control part (this part can be sold separately), and process the PCB by myself. Write this post as a memorial to my project.

[jxjbsd]

What it looks like now . Schematic diagram tube_test.pdf (140.22 kB - downloaded 63 times.)

Goals achieved:
1 Va power supply: 0-500V / 0.5A, Vg2 power supply: 0-480V / 0.2A,
Vg1 power supply: 0--140V / 0.01A
2 Ia, Ig2 current detection: 0.5A, 0.2A, 20mA, 3 gears.
3 Heating part 1.4-20V/2A.

[jxjbsd]

This is a picture of some test results

[jxjbsd]

Mr. Ronald's elegant project
https://www.dos4ever.com/uTracer3/uTracer3_pag0.html

I will add other similar links later

[quadtech]

Hello JXJBSD,

Great work - a very nice project and the good thing is that the design uses common parts (may be HCNR201).
You put in the effort to make it work, so you should not abandon it.

About the software enhancements, it would be better to open source the code on github.
You are using a commonly available ATMega32, so others may contribute software enhancements easily.
I also like Ron Dekker's project, but it is too expensive for me.

If your tester board cost is not too high, I am sure many would be interested.
Maybe if you split the design into 2 stackable PCBs of 100mm x 100mm , then it may be cheaper for others
to order from JLCPCB etc and easier to ship also.

[jxjbsd]

hello quadtech,
Thank you for your statement. There are some problems with my design. First of all, I did not make reasonable use of the various resources provided by the MCU. The entire design is like adding a digital process to the analog circuit. The result is that the design is complicated, and the effect is not good. The safety of the high voltage part is not good. If a short circuit occurs, it is difficult to cut off immediately and reliably, at least it cannot be guaranteed to be cut off every time. The current resolution is also very limited, and the current detection method of the screen grid is not linear enough.

[BU508A]

Hello jxjbsd,

nice work. 

Have you seen the Roetest from Helmut Weigl?
This is his website:
http://roehrentest.de/EnglishInfo.html

[jxjbsd]

Hello, BU508A
I went to see his website, his project is very complete, he did take a long time to do it, probably since 2006, such a project can only be done in his spare time, it is great to be able to persist for so long . He used a relay matrix to switch the pin connections. I also bought 72 (9* relays for this. It is necessary to buy some more after opening.
In comparison, my project has just started, or should continue. This will be a long-term job. Thank you

A little improvement：

[TimFox]

Does your design include programmable limitations on screen and anode power dissipation?
Also, it might be useful to include measurement of control-grid current, which (even with negative grid voltage) can have either polarity, and is much smaller.

[BU508A]

Hello, BU508A
I went to see his website, his project is very complete, he did take a long time to do it, probably since 2006, such a project can only be done in his spare time, it is great to be able to persist for so long . He used a relay matrix to switch the pin connections. I also bought 72 (9* relays for this. It is necessary to buy some more after opening.
In comparison, my project has just started, or should continue. This will be a long-term job. Thank you

A little improvement：

The hardware is just one thing. The biggest value in the project of Helmut Weigel's Roetest is the huge database, which covers thousands of vacuum tubes. And it is still growing because of the fanbase of it. Lot of people contribute to the database.

I wish you a lot of fun and success with your project.   

[Pawelr98]

A switching approach to high voltages would have not only been more efficient but also cheaper, since no high voltage secondary mains transformer would be required, and those often are expensive.
Off the shelf flyback transformer connected in reverse would be the easiest approach for most applications.

And it would have one more extra point, that is actual overcurrent/short protection.
If there would be, for any reason, a shorted anode or grid (or a gassy tube), then things would end up blowing up.
This thing is limited to 600-700mA, which at 500+V DC is quite the power, that the pass transistor won't be able to handle.

Negative voltage for first grid supply can be obtained from a simple charge pump from the anode supply winding.
One winding less, lower cost. Works for both traditional mains frequency approach as well as switching approach.

As for other improvements. For common modern 6.3V tubes the current setup is fine.
However there is quite a lot of series string tubes rated for 300mA heater, often very usable for audio.
Having a stabilised current source would be nice for testing those.


With modern parts curve tracing is much simpler.
Isolated I2C with specialised chips like INA219 or 226.
I did a basic semiconductor curve tracer using INA226 as my engineer thesis and this particular chip did make things easier for me.
Single chip measuring everything. For tubes it would usable just as a current measurement device.

[jxjbsd]

Does your design include programmable limitations on screen and anode power dissipation?
Also, it might be useful to include measurement of control-grid current, which (even with negative grid voltage) can have either polarity, and is much smaller.

1 There are power and current limits for the anode and screen grid. The power limit is controlled by software, and the current limit software and hardware are both.
2 There is no control grid current detection, because there is only negative pressure now, and now I do not intend to detect negative pressure current, there is no good method. I plan to provide the positive voltage and positive voltage current detection of the control grid in the next step.
thank you very much for your suggestion.

[jxjbsd]

The hardware is just one thing. The biggest value in the project of Helmut Weigel's Roetest is the huge database, which covers thousands of vacuum tubes. And it is still growing because of the fanbase of it. Lot of people contribute to the database.

I wish you a lot of fun and success with your project.   

Yes, this is a big project for individuals. Compared with hardware, the software part is the most cumbersome work. Only if these two do well enough, can they get the support of fans and build a perfect database of electronic tubes together.

[TimFox]

Control-grid current with positive grid bias voltage is important for oscillators, RF amplifiers, and class-B audio amplifiers, and is reasonably large.
Grid current with negative bias is not well-specified in data sheets, but can be important in small-signal amplifier design (typically in the microampere region, and reverses polarity somewhere around -1 to -2 V).

[jxjbsd]

Thank you very much for your suggestion. My design idea in the previous version is very old. There are two reasons: A, do not fully understand the microcontroller; B, do not understand the new device, too little knowledge reserve. Until I saw the design of Mr. Ronald's uTracer3, it was very shocking. Thank you for your suggestion. Controllable high voltage power supply is an important issue. I still need to continue to think and compare. I will look at the INA219 you mentioned. I found it is not expensive, I will try it. Thank you.

[jxjbsd]

I reset my goal for the next step:
1 Va power supply: 0-700V/0.5A, Vg2 power supply: 0-500V/0.2A, Vg1 power supply: +50V/0.1A -(-140)V/0.01A
2 Ia, Ig2 current detection: 0.5A, 0.2A, 0.1A, 50mA, 20mA, 10mA, 5mA, Ig1 current detection 0.1, 50mA, 20mA, 10mA, 5mA. Probably so many gears.
3 Heating part 1-20V/2A。
The goal of this design is to be able to test most of the common electronic tubes (anode consumption is less than 50W Pa<50W), and it is best to maintain reasonable complexity, low cost, and easy production.
Serves most tube enthusiasts.
To deal with various usage errors: short circuit, wrong connection, and possible safety.

[jxjbsd]

Control-grid current with positive grid bias voltage is important for oscillators, RF amplifiers, and class-B audio amplifiers, and is reasonably large.
Grid current with negative bias is not well-specified in data sheets, but can be important in small-signal amplifier design (typically in the microampere region, and reverses polarity somewhere around -1 to -2 V).

I have done some tube amplifiers, mainly in AF. I have no experience with RF. I will consider your suggestions, hoping to have a suitable solution.

[jxjbsd]

Grid current with negative bias is not well-specified in data sheets, but can be important in small-signal amplifier design (typically in the microampere region, and reverses polarity somewhere around -1 to -2 V).

This circuit should be able to solve the small current test under negative bias voltage (-10V to 0V). Of course, a diode with very small reverse leakage current such as BAV99 and an op amp with small bias current such as OPA2277 are required.Of course, it needs to be refined.