uTracer (Micro Curve Tracer)

[martin manning]

A second bench supply enabled some checking of the variation in measured parameters using the u-Tracer's internal filament supply vs. an external 6.3VDC supply. I tested several tube types and found a significant difference in the anode current (Ia) using the external supply, particularly for small-signal tubes like the 12A_7 series where the error was 15 to 30%! That black-plate RCA 6L6GC I tested earlier is at 91% of spec Ia using the external supply instead of the 81% I got before (an increase of 12%). The other two parameters of interest, Gm and mu, are not much affected in either case.

The internal supply uses the u-Tracer’s power supply voltage (20VDC) to heat the filaments. The power delivered to the filament is regulated by the controller using pulse width modulation (PWM) to adjust it to the appropriate level as specified by a Vf input in the GUI. Some issues with this technique have already been documented for high-current, low-voltage tubes, and the problem is related to the relatively high pulse frequency (19.5 kHz) combined with inductance in the filament circuit. The equivalent filament voltage can't be checked with a meter, not even a true RMS type.

A switch and a pair of banana jacks have been added to my u-Tracer for the external filament supply connection, so it is easy to run a test using the external supply and then quickly switch over to the internal supply to compare the results. An internal supply voltage can then be found such that the measured anode current matches the result using the external supply.

When the switch and jacks were added, I removed two ferrite beads from the filament supply wiring to reduce the filament circuit inductance, and this significantly reduced the required voltage offset. For example, the internal supply voltage originally had to be set to 7.0V in order to match the external supply result for the 12AX7. With the two ferrites removed, the required voltage input was reduced to 6.6V.

The attached plot shows that the voltage offset required to match the external supply result is well correlated with filament current for the tube types tested. This data could be used to find an internal supply voltage setting using the nominal filament current for the device under test, but note that it will be slightly different for each u-Tracer, since they will all have different socket wiring configurations and therefore different filament circuit inductances. The adjusted filament voltage can be saved in the set-up file for each tube type when using the internal supply, so this is a relatively easy thing to deal with, but it does require some extra work to determine what that voltage should be. Alternatively, one could just use an external supply.

Edit: After the first round of tests I discovered that the 1.5AT filament circuit fuse had deteriorated from running the higher current tubes, and the added resistance in the filament circuit increased the voltage offset required in some of the tests. After replacing the 1.5AT with a 2AT, I re-ran everything, and as a result got a slightly flatter correction. This proves once again that it is damn hard to measure anything accurately.

[martin manning]

It's been a while since I've posted anything on the uTracer, but it's been working fine, and a new GUI release came out with some features that might be of interest.

I had been lobbying for the ability to store and display target values for mu, Gm, and Ra for the Quick Test in the setup file to eliminate the need for consulting some outside reference. That is now possible as seen here for a 12AX7-type twin-triode. The improved QT form also has the option to add a title and produce a .txt file report, which should be useful for record keeping.

There is now a distortion analysis form that calculates harmonic distortion components (2nd through 5th) for a given resistive load line placed on previously-traced anode curves. This is interesting and useful for looking at load-line effects for small-signal and single-ended Class-A output tubes at moderate voltages, but of course distortion analysis of a push-pull output stage is not possible. The attached example shows how this looks for a triode-connected 6V6GT, where the load line is in green with the black portion being the specified +/- 5V signal over which the distortion calculation is made. As shown, THD is just over 3% and primarily 2nd order.

Two new measurement types have also been added. UL curves can now be traced with the screen voltage at a specified fraction of the anode voltage, but voltage and current limitations won't allow a full trace of high-power output tubes. Anode curves with "Schade Feedback," where the grid voltage is fed by a resistive divider from the anode, can also be produced.

[premiumplus]

Hi Martin,
What a great thread, and very timely for me. I have a pretty complete lab in my basement, and I've been wanting a tube curve tracer for a long, long time. I have been researching the uTracer and I'm excited to have found an enthusiastic and complete review like this thread turned out to be.
And I compliment your excellent build, beautiful work, sir!
I have a nice old Hickok 536 that has served me well. I added a DMM output to it (no holes drilled!) so I could measure the plate current, but I need more so I am seriously looking at building one of the uTracers for my lab.
Do you think it would be a good idea to use an old tube tester as a chassis? It would supply all the high current switching, and it might eliminate the need for a jumper cable setup like the one you designed for yours. I have an old Eico emission tester that is compact, and might be perfect for the job. Another advantage is that it would prevent accidental shorts from occurring when setting up.
The way you did it sure does make for unlimited configuration options, though.

[martin manning]

Hi, and thanks for the kind words.

I believe I read about someone doing exactly as you are thinking, which saves a lot of wiring. With just six connections from the uTracer to the socket array you might get away with minimal modification to the Eico tester, maybe even use switches, with a DPDT for the filament supply (so you could use either the uTracers filament supply or Eico's), and a 4PDT for the KGSA connections.

I do like the flexibility of the patch cord set-up. Having access to the board connections gives easy access for calibration and for tracing pentodes in triode connection, and I have traced other things such as varistors as well. The uTracer is short-circuit protected, but I would rather not count on that so I am careful. It takes a bit more time to configure with the patch cords, but two basic configurations cover the most common guitar amp tubes (6V6, 6L6, EL34, KT__, 12A_7) and just one more for EL84's. With pinouts and nominal values now being stored with the setup file, it's not bad at all.

[premiumplus]

Hi, thanks for the reply...you mentioned earlier in the thread that you took out a power transistor from a wiring issue, I think on the grid circuit? Was the short circuit protection added later to the uTracer, or was that event unique somehow? Just curious
So it will cost about $600 to put together a nice example of the uTracer, eh? How long did it take to get the board and components after you ordered, and is there anything else besides a power supply and a laptop that's needed? I assume that I'll supply my own USB interface cable...it looks like I will need to change from RS232 to USB, right? I'm kind of confused about the PC interface.
Sorry for all the questions, but I'm just kind of running a list off in my head, and it's finding it's way into this thread!
Thanks again,
Dave

[martin manning]

The issue with the grid supply is shorting it to the 20V power supply ground. I believe if I were to isolate the enclosure from the power supply and TTL jacks that would be solved. A tube short would take it to +20V.

My cost included a small 30V 3A bench supply, and I subsequently bought another one to use for a filament supply. I think it took about 3 weeks to get the parts kit, but that will depend on whether the current batch of 25 is complete and if they are all spoken for.

The computer interface is USB, but I chose to jump the RS232 interface and go direct from the PIC's TTL connections using a different cable (see below).

[premiumplus]

Very cool, so the uTracer ships with a USB cable that mates with that jack in the photo? There's lots to read, and I see he talks about a serial to usb converter in section 7.3 of the manual. Was that an issue with older computers? I have a 9 month old HP Envy laptop running Windows 8.1; do you know offhand if I will have to cobble something together to make it interface? There's also a youtube from 2012 that talks about testing the GUI. I'm trying to make sense out of the rs232 wiring diagram in in 7.3 of the manual, and figure out how to test the software on my PC. It's already successfully installed, but I have only USB ports on my PC...Do you know what I need to test the software with what I have?
Thanks so much!
Dave

[martin manning]

No cable is supplied with the kit. You can go RS232 to USB, or TTL to USB. The key is getting a cable with the right chipset. I'm running Windows 8 (on a MacBook using Bootcamp) with no problems.

[premiumplus]

[quote="martin manning"]A second bench supply enabled some checking of the variation in measured parameters using the u-Tracer's internal filament supply vs. an external 6.3VDC supply. I tested several tube types and found a significant difference in the anode current (Ia) using the external supply, particularly for small-signal tubes like the 12A_7 series where the error was 15 to 30%! That black-plate RCA 6L6GC I tested earlier is at 91% of spec Ia using the external supply instead of the 81% I got before (an increase of 12%). The other two parameters of interest, Gm and mu, are not much affected in either case.

The internal supply uses the u-Tracer’s power supply voltage (20VDC) to heat the filaments. The power delivered to the filament is regulated by the controller using pulse width modulation (PWM) to adjust it to the appropriate level as specified by a Vf input in the GUI. Some issues with this technique have already been documented for high-current, low-voltage tubes, and the problem is related to the relatively high pulse frequency (19.5 kHz) combined with inductance in the filament circuit. The equivalent filament voltage can't be checked with a meter, not even a true RMS type.

A switch and a pair of banana jacks have been added to my u-Tracer for the external filament supply connection, so it is easy to run a test using the external supply and then quickly switch over to the internal supply to compare the results. An internal supply voltage can then be found such that the measured anode current matches the result using the external supply.

When the switch and jacks were added, I removed two ferrite beads from the filament supply wiring to reduce the filament circuit inductance, and this significantly reduced the required voltage offset. For example, the internal supply voltage originally had to be set to 7.0V in order to match the external supply result for the 12AX7. With the two ferrites removed, the required voltage input was reduced to 6.6V.

The attached plot shows that the voltage offset required to match the external supply result is well correlated with filament current for the tube types tested. This data could be used to find an internal supply voltage setting using the nominal filament current for the device under test, but note that it will be slightly different for each u-Tracer, since they will all have different socket wiring configurations and therefore different filament circuit inductances. The adjusted filament voltage can be saved in the set-up file for each tube type when using the internal supply, so this is a relatively easy thing to deal with, but it does require some extra work to determine what that voltage should be. Alternatively, one could just use an external supply.

Edit: After the first round of tests I discovered that the 1.5AT filament circuit fuse had deteriorated from running the higher current tubes, and the added resistance in the filament circuit increased the voltage offset required in some of the tests. After replacing the 1.5AT with a 2AT, I re-ran everything, and as a result got a slightly flatter correction. This proves once again that it is damn hard to measure anything accurately.[/quote
Those are very interesting observations, and I think that I very well may use a nice outboard heater supply when I build my uTracer. And I couldn't agree more with your observations on how hard it is to accurately measure a given parameter. I have some of the best meters and scopes etc. available and I thought it was always me!

[martin manning]

R Dekker has devised a modification for the uTracer 3 that increases its capability to 400V on the anode and screen. An Amplitrex can get to 500V, but it is limted to 160 mA, where the uTracer is capable of a bit more than 200.

The upgrade to the 400V "3+" configuration involves swapping out 18 parts, including higher-voltage reservoir capacitors and the high-voltage switching transistors. The other part changes are related to tuning up the speed with which the hardware can shut down in the event of a short, and I believe it is now even better than it was at its original 300V capability. The resulting circuit is a fine piece of work considering it was accomplished without any modification to the PCB, and at a cost of only about $12. A couple of caps are piggybacked on resistors, and a couple of resistors are soldered on the bottom of the board, but it is a neat job. There is interesting reading and scope shots detailing the circuit design on the uTracer web site, V3 web log Section 30: http://dos4ever.com/uTracerlog/tubetester2.html

The board is pretty densely packed, but it wasn't too difficult to first strip the old parts from the top, and then clean it up and install the new ones. The pic below shows the boost converters for the anode and screen supplies, where most of the changes are made. The inductors are at the extreme right, and the HV switching transistors are at the extreme left. All of the blue MF resistors have been replaced, except the ones under the red film caps at the right. Those caps were added previously to improve the accuracy of the Quick Test, and they are the ones discussed earlier in this thread. The 100uF/450V reservoir caps are not installed yet.

In fact not much is happening in that last 100V; the plate curves have straightened out and they are just extended 33% further. This is a nice upgrade, though, and allows measurements at or close to actual operating conditions.

[Phil_S]

Martin,
Most of this is over my head. FWIW, I think this is really impressive and I really like the notion that the item is "low cost" though that might mean different things to different people. As for that last 100V, considering we see lots of amps operating in the mid 300's I am thinking we should not discount the value of getting into that range for tube testing (not that you did.)

Excellent stuff and thank you for sharing with us.

Phil

[martin manning]

I am really not very knowledgable about this sort of thing, but building it has been a great learning experience, and it is a really fantastic tool too. Not a bad return on the investment when you consider both aspects, IMO.

Consider that there is a software program (running on a PC) which tells a digital micro controller how to operate four analog power supplies to step a vacuum tube through its operating range, measure the results, then convert them back to digital for display on the PC... Quite a span of electronics technologies in one little device!

[martin manning]

Here is a 400V trace of a modern (Russian EH) 6CA7. This tube type is said to be a beam tetrode version of the EL34. The curves are not as rounded at the knee as the Tung-Sol 6CA7 data sheet shows (they are almost pentode-like), but they have more slope on the flat portion than the typical American beam tetrodes (6L6 or 6V6).

Here the screen voltage is set at 180 to stay under the 200mA nominal current limit of the uTracer. That "compliance" limit, which is in software, can be switched off, and then the hardware will allow something like 250mA. I would like to see the software limit the current for each grid curve by anticipating the current for the next data point before either running it or moving on to the next grid voltage. As it is, if the compliance limit is encountered a value of zero is stored at that anode voltage and for the remainder of the points for that curve- which makes a messy plot. IMO the values for the last "good" data point should be stored for all of the remaining points on the curve. That would make the current limit less noticeable when making full anode curve traces. The data can be stored and hand-edited, but why not let the program do it for you?

In most cases, data sheet values are provided at a mid-voltage point (often at 250V anode and screen), with Ia, gm, ra, and mu (for triodes or triode-connected pentodes) specified at a particular grid voltage. This is why a 300V capability is really plenty to get a good test that can be compared to the spec.

[JazzGuitarGimp]

Hi Martin,

Here's a question for you: if you were designing a class AB2 stage, and you wanted to know how much grid current would be required of the driver stage at a grid voltage of, say +15V, can the u-Tracer push the grid of a 6L6GC up to +15V and measure and report the required grid current to get there?

From what I can see, it looks like the u-Tracer still isn't quite up to the task of doing R&D in the real world - although bumping the HT supply up from 300V to 400V is a good start. It would be nice to see it be able to go up to 600V plate and screen, or, better yet, 800V plate and 600V screen, with enough current capacity to trace big tubes (KT88, KT120, KT150, 803, etc).

Cheers,
Lou

[martin manning]

Hi Lou,
From the GE 6L6GC data sheet (1959), at Vg2=250, and Vg1=+15, the anode current is about 325 mA, and g1 current is about 35mA at the knee. There are curves for Vg2=400V included too, where Ia is up over 500 mA. You could certainly design an AB2 stage the old fashioned way using these published curves.

Testing at that level of current, and/or at 600-800V is a whole different ballgame compared to anything I've run across, and would require some really serious hardware. Maybe something along the lines of the industrial equipment GE and RCA used back in the day?