uTracer (Micro Curve Tracer)

[martin manning]

Looks great Markus! I've found that while most tube types trace without issues, some will oscillate. Snapping a ferrite on the grid pin patch cable (as shown earlier in this thread) will usually kill it.

[istealth]

I just wanted to make a small update for those who might be interested in the alternative web-based interface for uTracer 3, 3+ and 6, which works on all possible devices (laptops, tablets and phones) and all possible operating systems.

The webpage to try the functionality (with a bit outdated version) can be found here http://utracer.boffin.nl (this is instead of http://utracer.live which is now abandoned).

I just posted the latest release of firmware for 4MB and 16MB flash ESP32 modules, which is more stable and contains some new features (see the changelog for details). For example, there is a new type of Quicktest now, which just with a couple of measurements allows you to find the grid voltage that corresponds to the specified Va, Vs and Ia. One user suggested it (for matching tubes) and several already found it useful. The other features are for example restoring the last quicktest configuration after ESP32 reboot, drawing pinouts for user-defined tubes or easy generation of grid/anode voltage sequences for the "stepping variable".

One more useful feature that I forgot to mention, now during the quicktest, it is possible to choose automatic adjustment of Va and Vs (the option is also visible on the screenshot above, in the light-blue box). So when one sets Va and Vs, for example to 250V, the Ia will be reported for exactly those voltages, and there will be no typical voltage drop due to discharge of caps (which is about 5V for 40mA of Ia in my tests for my uTracer). Now uTracer will take one measurement extra to figure out the voltage drop and then readjust the applied voltages automatically.

[martin manning]

Some recent uTracer tinkering after a long hiatus.

The latest GUI release, 3.12.6, Jan 2019 ( ...I was a bit slow to pick up on that!), fixes a number of small bugs. There remains, however, an error loading saved setups that change the "Vg Low" setting, where the sweep and step data for the trace are set to defaults instead of the saved parameters. If the "Vg Low" setting is pre-initialized, all the saved data will load correctly, or you can load the saved setup a second time and get the missing values (because the VG Low setting has been updated by the first attempt). Annoying, but at least the values are saved, and you can get them loaded without retyping or copy-pasting them.

Also related to the Vg Low mode (which uses the grid loupe accessory described earlier in this thread: https://ampgarage.com/forum/viewtopic.p ... 63#p403463), I finally got around to finding an LM348N quad op-amp that has a reduced Ig offset. The false current is now down at 20uA, where it was at about 80uA. Below are some test runs with a 10k resistor (connected g1 to k) that show good and bad results using different op-amps. Ideally, the trace should show zero Ig all the way up to 0 Vg, and then rise linearly from that point at 100uA/V. Finally, I've attached a full trace of an old 12AX7 made using the loupe, -4 to +1.5V Vg, and Ig measured (shown as Is). Ig stays constant at around 100uA until Vg goes positive, and the grid current lines have a distinct slope with Va. In other cases I have seen grid current start to come up in the last 0.5V below zero, and grid current lines that are more or less flat except for the "ski tip" at very low Va (compare the trace at the linked post).

[martin manning]

I had some fun laying out a Low Vg "Loupe" PCB to replace my prototype. Layout was done with the newly-released KiCad 6, which as you can see has a very nice 3D rendering utility. As for cost, it's really hard to beat free. This board is just a bit bigger than my pad-per-hole version at 46mm x 66mm, but should still fit in the box. A batch of ten is on the way. Extras can go to good homes for cost plus postage.

[dorrisant]

I want one of these boards. Let me know Martin. If you wanted to post the Kicad files (zip, etc), I could take it from there. No worries if not.

[martin manning]

I want one of these boards. Let me know Martin. If you wanted to post the Kicad files (zip, etc), I could take it from there. No worries if not.

Sure, I can send you one. Little boards like this are very cheap, but I don't know why you would want to have another batch made up. I have just confirmed that I have all the components on hand, so as soon as the boards arrive I can stuff one and test it.

[martin manning]

Moving to the other end of the spectrum, I thought I'd try a modification to increase the maximum current capability. This is described in the Lab Notebook pages on R Dekker's web site with a "use at your own risk" caution. The GUI software has inputs for the current sensing resistor values, and resets its current limits accordingly, so it is "approved" to that degree. It's pretty simple, just parallel some resistors and a cap to increase the hardware current limiters, and reduce the value of the current sensing resistors, also by paralleling, to keep the voltage within the limits of the ADC's.

Jacking up the current limit comes at the cost of some measurement resolution, which is the reason I haven't tried this before. Here I wanted to find out if it would be possible to switch the current range from the front panel by adding or removing just the extra current sensing resistors. Unfortunately, t's not. In fact it doesn't work at all, presumably due to lead inductance. It might be possible with a switch or movable jumper on the board, so soldering/unsoldering to change the current range might yet be avoided.

Since I had modified the current limiting circuits, I went ahead and soldered the current sensing resistors on (really under) the board to make a working 600mA-capable configuration. With this it's possible to do full traces (to 0V Vg1) on power pentodes with typical operating screen voltages, including the KT88 shown here at 350V Vg2. I tried a few traces at low current too, and it still seems to be working quite well. I think I'll leave it like this for a while and run some more trials.

[martin manning]

PCB grid loupe has been built up and tested. It looks a lot like the KiCad rendering with the missing bits added. Grid voltages from -10 to +5V match my original version very closely using the same calibration file. This accessory is nice for small tubes like 12AX7 that operate at low grid bias, but it can also be used with power tubes. In the 6L6GC trace below the loupe is used with the high current mod to trace an anode curve at positive Vg1. Unfortunately, the g1 current can't be measured for pentodes since the screen channel is already in use, and this is where the uT3 runs out of options. At some point, the uT6 will be capable of that, once the grid bias extension for it is ready.

[martin manning]

I want one of these boards. Let me know...

Tony, PM me your address. If anyone else is interested in a Low Vg loupe board, I have a stack of them.

[neskor]

I'm interested, just finished my utracer 6
but I'm in Europe, so shipping will be probably expensive
Can you make one populated board or a kit and sell me, that will make more sense to pay extra shipping across the pond

[neskor]

Also one more newbie question.
Is this board compatible with utracer 6 and do i need this board at all with version 6?
I know Ronald made some changes in grid circuit on this version 6.

[martin manning]

I'm interested, just finished my utracer 6

This circuit is for µTracer 3+. Do you have one of those too? If so, there is a guy in the Netherlands who has some boards (different shape, same width as the main board). A similar accessory is being produced for the µT6. As I understand it, there is a waiting list for that which you can get on by e-mailing R Dekker.

[martin manning]

Last word on the grid loupe board: v2 has been created to scrunch the size down a bit, and it is now securely mounted in the enclosure. As with v1, I have extras if anyone is interested.

[martin manning]

A strange thing was discovered when using the high current configuration: The two channels (screen and anode) are not completely independent! The trace below shows what happens when progressively smaller resistors are traced simultaneously and separately using the "I(Va = Vs), Vg constant" test option (as when tracing dual triodes, or pentodes in triode operation). For the separate Ia and Is traces, the opposite channel is left unconnected. The screen trace shifts down at the higher current levels, while the anode trace doesn't. I tried swapping out the high voltage transistors in the screen channel to rule them out, and the effect persisted. This mysterious cross-talk doesn't occur if both channels are in use, however, as they would be when tracing a pentode. The second plot shows the results when an alternate method is used to acquire the data: A trace is made using I(Va = Vs), Vg constant, tracing two resistors simultaneously as before. Then two more traces are made using "I(Va, Vg), Vs and Vg constant," and "I(Vs, Vg), Va and Vg constant." In these traces, both channels have resistors connected, but the "unused" channel is set to a constant low voltage (5V). In this case, the traces repeat nicely, within the calibration difference. I can't say why this happens, but my conclusion is it's a non-issue, and generally speaking it's probably not a good idea to let either channel charge up to high voltage without anything connected to it. At the least, it will be needlessly charging and discharging, wasting time and generating heat.

[dorrisant]

I read the whole thread last night and I had a few questions... Keep in mind, I have a Hickok 605A that is my daily driver. Even if I have a replacement, I doubt I'd ever let it go... No sentiment, just very familiar backup. I have a KS15874-L2 cardmatic that I am currently selling as well as all of the accessories. I have a I-177B that I am on the fence about getting rid of. I could possibly use it for this uTracer project, . There is some great hardware on it that could be recycled for this project, I have to consider it.
That said, will the uTracers be able to replace my Hickok's? I am not so much worried about the gas test or life test, I can see how those are available with some extra thought and setup. But, will it test for shorts? Or do I need to run the tubes through another tester first?
Also, is there a repository of tube files? Like quick test files that I could be dropped into a specified folder for a faster start once the device is functional.
Would you, knowing what you know now, go for the uTracer 6 instead of version 3+? How often do you wish you had that extra plate voltage? How often do you see the extra current being beneficial? I know that you are limited by the maximum current, therefore limiting the plate voltage at times. I have seen tubes make it through two Hickok's only to arc like fireworks when used in the amp (MusicMan). I would think that being able to run big tubes at higher plate voltage would allow one to match the actual working conditions of the amp intended.
I have gotten accustomed to the use of rotary switches for tube pin connections. I think I might want to use that with my own build. I could follow the convention of my 605A making use of the tube charts that already exist. I'd probably install banana jacks for each pin assignment just for the added flexibility. That would be the one thing I would add to what you've done so far. Add a rotary switch for g1... Right in the center of where you mounted the pin jacks, so that you could point the g1 "knob" at the pin jack you want it to feed. No jumper, no possibility of shorting it to anything.
Again, thank you for your contributions on this subject. I'm surprised that they don't just send you a V6... We all know that your posts here have certainly sold several devices for them already. Why not capitalize on that?