|
 µTracer/HR in Action
The µTracer board is controlled by a Windows graphical user interface (GUI) which has two modes: Quick Test and Voltage Sweeps. Quick Test is chosen with a button on the main screen while Sweeps are done in the main screen itself, seen by clicking the thumbnail at right. Quick Test lets you set plate, grid and (for pentodes) screen voltages. The settings for these can come from our VTA Tube Settings Chart or directly from standard databooks. Then you run the test to get Gm, Rp, mu and some lesser-known factors for pentodes. In the Quick Test dialog, you can choose either pentode or triode and the fields change accordingly. 
Pentode Quick Test
Click the thumbnail at right to see the pentode screen. (Better to right-click and open in a new window.) On the left side, you set the plate, screen and grid voltages. “Stepping” lets you adjust the percentage change used to calculate Gm, Rp and other derivatives, with the default being 10%.
Clicking the Test button runs it. The example shown is for a 6L6WXT. In the lower left corner you can see the measured plate and screen currents, which are pretty close to the nominal 72 and 5mA. The right side might look confusing but you can just focus on the most important parameters, ignoring the second column. In the first column plate resistance, Ra, is found to be 36.5KΩ. (Plate resistance is more commonly symbolized by Rp.) The next item of interest is gm. Here the value is 6030µ℧ since it’s given in millimhos. (6000 is nominal.) Amplification factor, mu, is given as 220 here. If desired, you can enter nominal values in the fields labeled “Nom” and it will calculate percentage deviations but that doesn’t affect the measurements.
Dual Triode Quick Test
Now for the triode quick test, we plug-in a JJ ECC83 (12AX7) and reconnect the patch panel but things are a little different. The µTracer can test both halves of the dual triode if you connect the screen supply to the second plate and connect the grid supply to both grids. To do this on the µTracer/HR build, I added a short banana cable between the pin-2 and pin-7 test jacks to connect the grids together. I also used the Supp (Suppressor) “source” to connect the second cathode. So the full connection list for the 12AX7 is:
Source Pin Comment
Htr 4
HtrG 5
Cath 3
Supp 8 for the second cathode
Grid 2
Scrn 6 for the second plate
Plate 1
Pin-2 7 banana cable jumper for the second grid
 Now the HtrV switch is set to 12.6V, power is turned on and the tube is allowed to warm up. You can click the thumbnail at right to see the Quick Test screen. Here, we select triode mode, set plate voltages Va1, Va2, and the grid, Vg. In the illustration, I’ve also entered the nominal readings expected. After clicking the Test button we see that the plate currents in the lower left corner are very close to the norm, Gm is excellent and plate resistance, Ra, is close to the nominal value. It seems that JJ has done well with this tube.
Graphing Plate Characteristics
Though Quick Test is more often used for testing tubes, the sweeps and post-processing in the main screen are the most interesting. To see that, let’s get a family of plate curves for the Sovtek 6L6WXT I have plugged into the µTracer/HR at the moment.
To match the GE datasheet plate curves, we’ll select Measurement type as “I(Va, Vg) with Vs, Vh constant”. Settings:
Plate voltage, Va: 2–250V in 50 steps
Screen voltage, Vs: 250V
15 values for Vg: -28 to 0V
[You can get inspiration for these numbers from tube databooks.] Even though I’m using the accurate, “fixed” heater supply addition in the µTracer/HR, I have to click the “Heater on!” button in the GUI so it will change to Measure Curve. I’ve set the HtrV switch to 6.3V and made the pin connections on the front panel.
The high plate currents for Vg=0 are okay because it won’t get hot from the short pulses
Then clicking Measure Curve runs the series and the result is seen above right. That took just under 5 minutes. The graph above is the same but with an overlay of curves from the GE 6L6GC datasheet. The agreement is generally quite good except for the WXT having lower current capability at lower plate voltages. Note that Sovtek is now offering the WXT+, which has increased capabilities.
Lots of Measurement Types
The plate curves shown above are just one of 13 measurement types offered in the drop-down box in the main screen. Dekker has full coverage of those here (section 3.2), so I will only make some general points about the measurements:
- Plate current (Ia) and screen current (Is) are stored for each step of voltage.
- You can plot Ia, Is, both, Gm or Rp. (Gm and Rp are listed as derivatives of current.)
- The X-axis can be Va, Vs, Vg or Vh, determined by the measurement type chosen.
- Measurement types are listed in “shorthand” like this example:
 I(Va, Vg) with Vs, Vh constant
Focus on the I(Va, Vg) part because we know all other voltages are held constant. At right, I used Photoshop to reimagine the full dropdown list with the constant items removed. The first voltage (Va in the example) is the one which is swept in small increments. The second voltage is stepped from a list. So for this measurement type, a chosen current is plotted versus Va, and Vg is stepped between sweeps.
- The most common measurement types are:
- #3 I(Va, Vg) - Plate family, Ia vs Va with Vg stepped
- #1 I(Vg, Va) - Transfer curves, Ia vs Vg with Va stepped
- #4 I(Va, Vs) - Plate curves, Ia vs Va for multiple Vs values
- To shed a little light on the more unusual types:
- +Vg mode refers to using the screen supply to provide positive grid bias for triodes.
- The sweeps with “Vs=UL...” are for Ultralinear operation, where k defines the tap position for the screen connection to the output transformer.
- Schade FB: In Schade feedback (may be pronounced “shah-duh”), a small amount of the anode voltage is fed back to the grid.
How Accurate is the µTracer?
I said it earlier on the first page but it bears repeating that, while I will be direct about discussing the limitations of the µTracer, I deeply respect Dekker’s excellent design and I’m very happy to have one, myself. Cost-effectiveness and simplicity were key priorities and those have made it the tube analyzer of choice for over 1400 hobbyists. In contrast, my own Vacuum Tube Analyzer (VTA) prioritized accuracy and became too difficult and expensive for any but the most advanced hobbyists :`(
Please note that for all accuracy testing here, the heater regulator board was used instead of the heater supply on the main µTracer board. As discussed above, additional deviation would be expected if the main board heater supply were used.
Prior to building the µTracer, I compared it to the VTA in the article on that project. At that point, I could only estimate the resolution of µTracer measurements. That was done for four representative tube types at the test conditions in the VTA Settings Chart. Those settings were derived from standard databooks. With the µTracer project built, I can now look at overall accuracy by comparing test results with the VTA. Please understand that I am neither claiming perfection for the VTA, nor trying to brag about it. It’s just the best way I have of evaluating µTracer results. Bear in mind that the VTA has limitations of its own. The comparison uses the same four tube types as the resolution estimates: 6L6, 12AX7/ECC83, EL84/6BQ5 and 6AU6. The particular tubes I chose needed to be reasonably exemplary. Those are:
- Westinghouse 6L6GC - Used but in good condition, borrowed from my Eico HF-20 (#4).
- JJ ECC83 (12AX7) - I wanted to avoid newly-made tubes but µTracer Gm tests must be made at a fixed bias voltage. Normally, I test Gm at the specified current, after checking current at the specified voltage. However, my NOS 12AX7 types tended to have subnormal current with fixed bias, so Gm values would read subnormal. The JJ tube happened to be close to databook norms, so I used it instead.
- Sylvania 6BQ5 (EL84) - Used but in good condition; probably bought as NOS.
- Motorola 6AU6 - Used but in good condition; probably bought as NOS.
Gm and Rp Shootout with the VTA
|
|
|
Gm
|
|
|
Rp
|
|
|
Tube
|
VTA
|
µTracer
|
% Deviation
|
VTA
|
µTracer
|
% Deviation
|
|
6BQ5
|
11540µ℧
|
11020µ℧
|
-4.5%
|
40.68KΩ
|
38.42KΩ
|
-5.6%
|
|
6L6GC
|
5710
|
5640
|
-1.2
|
42.12K
|
39.09K
|
-7.2
|
|
12AX7
|
1270
|
1300
|
2.4
|
78.19K
|
75.35K
|
-3.6
|
|
6AU6
|
4843
|
49701
|
2.61
|
2.439M
|
n/a2
|
n/a2
|
|
|
 The Gm tests showed fairly good agreement between the µTracer and the VTA until it came to the 6AU6. This tube is tested with Vg = -1.06V, where it’s expected to have Ia = 7.6mA. My estimate of Gm resolution was about 2% in this case, with the µTracer operating on its 10mA plate current range. But the accuracy of the Vg dither isn’t known, though Vg resolution is specified at 50mV. The µTracer shifts Vg by a percentage (usually 10%) to measure Gm. The standard 10% dither is only 106mV in this case but that resulted in about 14% deviation relative to the VTA. It worked far better with 20% dither (212mV). More dither increases error from the nonlinear transfer characteristic of the tube, so the lower dither would normally be preferred. The VTA uses a Vg dither of 100mVpp, set by a precision voltage reference. The average magnitude of Gm deviation was 2.7% and for Rp it was 5.5%.
Some limitations of the µTracer were revealed with the Rp tests. The resolution estimate for the 6BQ5 was 7.4% and the deviation from the VTA was indeed -5.6%. The 6L6GC Rp resolution had been estimated at 8.8%, while the measured deviation was -7.2%. The estimate assumes no error in setting the plate voltage dither, whereas we now know that plate voltage resolution is limited by the steps of charge applied to “pump-up” plate voltage to a target value. We don’t know how that impacts the amount of dither.
For small-signal pentodes like the 6AU6, Rp is usually too high for the µTracer to read. That’s because the 10% change in the 250VDC plate voltage would result in only about 10µA change in plate current. The plate current resolution is only about 10µA, so in this case, the method of measuring Rp by differencing measurement pairs doesn’t work well. The VTA has an AC measurement system for Rp and has about 1.5% resolution for the 6AU6.
The conclusion of the Gm tests is that the µTracer performs well enough for tube testing but you may have to be careful about setting the Vg dither percentage for low-bias tubes. I would try both 10% and 20%. The Rp tests showed a little higher deviation with the 6L6GC. There again, perhaps changing the dither percentage (for Va in this case) could help. Fortunately, Rp isn’t considered a critical parameter for testing tubes, so it’s probably not a big deal.
Additional confidence for either Rp or Gm could be obtained by plotting those versus plate voltage or grid voltage, respectively. For the plate characteristics plot [ I(Va, Vg) ] instead of choosing Ia for the Y1 axis, you can choose dV/dIa, which is Rp. For the transfer characteristics plot [ I(Vg, Va) ] instead of choosing Ia for the Y1 axis, you can choose dIa/dVg, which is Gm. Since these plots are swept through a series of measurements and are smoothed over multiple points, quantization errors are mitigated and large ones tend to stand out as noise or anomalies.
Checking Plate Characteristics Curves
To get an idea of DC measurement accuracy, I ran plate characteristics families, I(Va, Vg) on the four example tubes and then picked 8 points for each tube, to be checked on the VTA. The points were somewhat randomly scattered over the plots but had to be within plate and screen dissipation as well as cathode current ratings. The pulsed measurements of the µTracer allow it to safely ignore such limitations but the VTA measurements are continuous. After taking the DC measurements for the 6BQ5, the results were plotted as on overlay on the µTracer plate family, shown below.
 As you can see, the visual agreement is really impressive! The underlying numbers at right also show good agreement. In the VTA tests for this tube, I noticed that the plate current values started a bit high, dropped slightly over 10-30 seconds and then started rising again. I recorded the lowest value. Viewing such behavior is one of the reasons I wanted the VTA to use continuous-time measurements.
You can see the full list of comparison points here, but I’ll just show the summary below for discussion. As you can see, the 12AX7 had the highest deviation. The µTracer’s 50mV Vg resolution and Vg accuracy in general could explain the greater deviation, coupled with the critical grid voltage of the 12AX7. Keep in mind that at the very low currents of some checkpoints, VTA resolution comes into play. There is also variation of the tube’s performance, which contributes to deviation. You can click the thumbnail below for a plot of the 12AX7 plate curves with VTA checkpoints added.  By the way, I did confirm the somewhat anomalous checkpoint for Vg=0V at the top, near the title. Overall, I would say that the µTracer performs quite well in these brief tests of DC plate characteristics. Roughly, it averages about 2% deviation but can be around 5%.
How Much Does It Cost?
The partial kit gives you all the parts you need to build the main PCB and it sells for €225, including international shipping (about $253 today). As said above, you must provide an enclosure, pin switching, a laptop-type AC adapter and a few other parts. To avoid the limitations of the main board heater supply, this article also includes a custom-designed, precision, switching regulator. It uses remote sensing to hold the heater voltage at the socket precisely to the target value. Offering heater voltages of 5.0 and 6.3V at up to 3A and 12.6V at 2A, it covers most receiving tubes.
The Need for Separate Drive and Sense Lines
You might think that the wiring, patch cords and connectors wouldn’t drop much voltage but when I was testing tubes on the bench for the article here, it was difficult to avoid one or two tenths of a volt loss with tubes like the 5U4, EL34 and 6L6. For the typical 6.3V heater, 0.2V is over 3% error and heater voltage is a critical factor in tube performance. This heater supply keeps it within a couple millivolts at the tube socket. It’s nice to have confidence that it’s rock-solid-accurate, so you don’t feel the need to check it in normal use.
But to support the remote-sensing regulator, the patch-panel for pin connections must carry two separate lines: drive and sense. We need 2-pole connectors for that and they have to be small to fit 12-pins across the panel. The drive line is the main one carrying the current. The sense line carries almost no current and is used to let the regulator accurately observe the voltage at the socket. It’s also handy to let you measure the voltage and monitor signals at the tube. The upshot is that our bill of materials (BOM) and schematic for the µTracer/HR is best used with the heater regulator option. It will work without it (employing the µTracer’s onboard heater supply) but could be simplified if the option isn’t wanted.
The Bottom Line
When built as described in this article, the µTracer BOM totals about $446 and the heater regulator adds about $91, for a total of $537. You can view the BOMs here:
I think the µTracer is an excellent value, even with the heater limitation. It has brought better-quality testing to a huge number of hobbyists at a price most can afford.
What is the minimum cost for a µTracer build? I did an estimate assuming one has a spare, 19V laptop supply, appropriate wire and a power cord. A simple Bud-box was substituted for the enclosure and the heater regulator and associated 2-pole connectors were eliminated. Also removed was the vinyl panel art. The total for the low-cost version is about $341, a $196 savings versus the project as built. You can see the estimate in detail here in PDF format and here in Excel 2002 format. Even if you don’t include the heater supply presented here, I recommend using some kind of accurate heater supply with the µTracer.
Documentation
You can download a zip file of all the documents below by clicking here (21MB). Info on buying a copy of the heater regulator PCB is here.
µTracer Main Board
µTracer/HR Project
Heater Regulator Board
Coming up next: The Construction Gallery with over 30 captioned photos...
|
