Contents: The uTracer3 overhaul Plan A

1. The uTracerNXT, a uTracer3 overhaul!

A bit of history

The first ideas for a pulsed tube curve tracer were conceived as far back as Christmas 2010. The first uTracer version used a high-side current sense circuit that did not really work satisfactorily. After a golden tip from somebody following my weblog, I changed the current sense circuit to its present concept, and the uTracer3 was born. Stimulated by my sons “to go commercial,” I designed a PCB for the uTracer3 in 2012 and later that year the first uTracer kits were on their way to the first customers.

How does such an undertaking work in practice? Of course the first experimental versions of the uTracer were constructed with components I had lying around. With the circuit working very satisfactorily, it was not more than logical to copy the circuit one-to-one for the commercial version of the uTracer3. At the time I estimated that perhaps, with a bit of luck, we would sell a few dozens of kits. I absolute could not have imagined that 12 years later there would be over 2600 uTracers in service in 63 countries!

Over the years, I discovered and experienced things that also real, large compagnies have to deal with. One of these things is that once you have a product running, it is not so easy to make changes to it. Over the years there have been two major changes to the design of the original uTracer. The first major change was the introduction of the uTracer3+ which raised the maximum voltage to 400 V, and the second change was a redesign of the PCB. Every change requires adaptation of the manual, changes in the components to be ordered, the production of a test series to make sure the kit is flawless, etc. Additionally, there is the issue of customer support. The uTracer is a rather complicated kit, and unavoidably mistakes happen during its construction. I always try to help everybody as good as I can via email, and if that doesn’t work out I am happy to have a look at the circuit myself, so you really want to limit the number of circuit versions to deal with to the minimum. In short, the choices you make in the beginning in terms of component selection and circuit topology may very well haunt you for years!

One of the things every electronic appliance company that sells products that run over many years must deal with, is that at a certain moment components become obsolete. It happens to big companies, like Philips where I work(ed), and tiny companies like Dos4ever. At a certain moment for example, it was announced that the original processor, the PIC16F874 would be taken out of production. The PIC16F884 was suggested as an almost pin compatible alternative. Nevertheless, it took me days of headache and many software changes to get it working. Also the original MJE350 and KSP99 high voltage pnp transistors in the meantime have been phased out and had to be replaced with suitable alternatives.

The latest addition to the list of “soon to be obsolete” components is the DIL version of the OPA227! Already since COVID times, the availability of the OPA227 has given me a headache, while also since that time its price has skyrocketed! Apparently the SMD/SMT version of the OPA227 will remain in production somewhat longer, so to mitigate the upcoming problem, my first idea was to redesign the PCB to make it suitable for the SMD variant. However, thinking things over I started doubting if this would be the right way to go?

I have always been a strong fan of “old fashioned” through-hole components. Primarily, because most uTracer customers are tube/valve enthusiasts who, very often, do not have much affinity with tiny, fragile, difficult to handle and solder SMD components. Secondly, I always had the idea that servicing a circuit with through-hole components was easier than servicing a PCB with SMD components. I learned that although this certainly holds for IC’s in sockets, where the possibility of rapidly exchanging a part helps to quickly debug the circuit, I discovered that the situation may be completely the opposite for other components. Ever since I started using a hot air soldering station, I realized that especially replacing transistors is much easier if you have an SMD component instead of its through-hole counterpart! On top of that, the harsh reality is that many modern and powerful components are simply not available in through-hole packages anymore!

So, all this made me think and decided me not to go for a straightforward simple modification of the uTracer3 PCB, but instead to take a critical look at the whole circuit and come up with a more modern and future proof overhault: the uTracerNXT!

What’s in a name?

Being a man of exact science (and a bit autistic as well), my first idea was to name the redesign of the uTracer the - you may have guessed it - uTracer3.2. However, after hearing me talk about the uTracer3.2 for weeks, the CEO of our tiny family enterprise (my wife) expressed her discontent with the name. First of all, she thought it looked “ugly,” but she also thought that it was rather nondescriptive, not doing justice to the fact that the new design will also include a substantial number of innovations and improvements. “It’s a bit how kings or popes are numbered” was here objection. Since we already have a uTracer 4, 5, 6, 7 (that not all made it into a kit by the way) I started thinking of an alternative. For a moment I was tempted to call it the “FU2Tracer,” but I am glad I was sensible and didn’t go that route. Playing with words and letters I came up with NXTracer (Next Tracer) to emphasize that the new tracer is not just an upgrade of the old uTracer3+, but really a completely new design. However, then several readers pointed out to me that it would not be a good idea to discard the, in the mean time, well-known brand name uTracer (micro-Tracer), so in the end we settled for uTracerNXT, to be pronounced “Micro-Tracer Next” I tested the new name on my wife and sons, and they all liked it. So, there we have it, the uTracerNXT!

uTracerNXT wish list

Producing more than 2000 uTracer3+ kits, servicing them and getting feedback from users and adding to that new ideas and insights gained from the uTracer6 and the stillborn versions 4 and 5, resulted in a wish-list for the uTracerNXT.

New features / Modifications:

• Use modern single supply 5V OpAmps
  There are these days many advanced OpAmp that deliver just the right performance in terms of low-offset and slew rate, but (unfortunately ?) only come in an SMD package and with a low (single) supply voltage of 5 V or even less. Switching to a single OpAmp supply is not without issues, but the advantages of using modern OpAmps have become so obvious that the time has come to see if these issues can be overcome.
• Delete the +15 V and -15 V power supplies
  Moving to single supply 5V rail-to-rail output OpAmps automatically eliminates the need for the +15 V and -15 V power supplies, resulting in a considerable reduction of complexity, component count and board space. It also eliminates the need for the over-voltage input protections that are now used to protect the input of the two Programmable Gain Amplifiers (PGAs).
• Extend the control grid bias range 0 to -100 V
  An extended control grid bias range is high on the wish list of many people working on power audio amplifiers. The goal for the uTracerNXT is to extend this range to at least 0 to -100 V and preferably even higher (more negative).
• Redesign of the control grid bias circuit
  With the increase of the control grid bias range, it is a good moment to revise the complete grid bias circuit. The present circuit, which uses one of the PWM outputs of the PIC, is a bit of an odd concept which worked well, but with many affordable and readily available DACs on the market these days, a more straightforward approach is perhaps more logical.
• Reference cathode to ground instead of to the supply voltage
  In the uTracer3 the cathode of the tube is referenced to the plus of the supply voltage rather than to ground. This choice was originally motivated by the fact that normally the output voltage of a boost converter can never be lower than the supply voltage, and to extend the range of the control grid bias circuit to -50 V. However, it appears that with a simple trick the output voltage of a boost converter can be lowered to 0 V, and with the grid bias circuit being redesigned it is a good moment to change to a more conventional configuration whereby the cathode is referenced to ground.
• Use uTracer6 NMOS high voltage switch circuit instead of pnp bipolar circuit
  Because very high voltage pnp transistors are hardly available, I switched to an NMOS based HV-switch for the uTracer6. High voltage NMOS transistors, certainly for voltages at around 600 V are readily available at low cost. Moreover, the uTracer6 high voltage switch has proven to be very reliable and robust. So I plan to use the same circuit for the uTracerNXT.
• Increase maximum anode voltage to (at least) 450 V
  With the tube cathode referenced to ground, and a high voltage switch that can easily handle voltages of more than 500 V, it becomes feasible to use the entire 450 V voltage range of the 100 uF reservoir capacitors. It could even be extended to 500 V if the same capacitors are used as in the uTracer6. Unfortunately, these capacitors have a long lead time, and are sometimes difficult to get. So it would be great if the circuit could be made suitable for both types.
• Easy modification for higher anode/screen currents
  As demonstrated in the uTracer6, the NMOS high voltage switch will allow for higher currents, possibly up to 1A. It would be great if the current range could be increased with minimal changes to the circuit.
• Easy modification for low anode and screen voltages
  Battery tubes usually operate at anode and screen voltages well below 100 V. In the uTracer3+ these tubes could only be accurately traced with some major circuit modification, including a PIC with modified firmware. It would be great if for the uTracerNXT there was a simpler way to test and trace low-voltage battery tubes.
• AC heater synchronization
  In the extension board for the uTracer6, an option was introduced to synchronize the pulsed current measurements of the uTracer with the AC main frequency. This makes it possible to use an AC transformer to power the heater of directly heated tubes, making it possible to trace heavy duty directly heated transmitter tubes. Both the hardware and the software for this option are very simple, and it would be great if it could be added to the uTracerNXT.
• Allow for easy experimentation / modification
  Make it simple to modify the circuit for high voltages, currents etc. e.g. by adding extra resistor positions, jumpers etc.

• PCB size and (approximate) position of the terminals
  To make the uTracerNXT a drop-in replacement for the uTracer3 with minimal changes.
• RS232 communication, but with connector for serial connection and FDTI cable (uTracer6 style)
  The RS232 communication interface of the uTracers is a point of endless discussions. I still plan to use it for the uTracerNXT because it is a standard, reliable and to keep the uTracerNXT compatible with its predecessor. But, just like for the uTracer6, I intend to make it easier to use with TTL level serial adaptors and FDTI cables.
• PWM controlled heater supply, but at lower frequency
  Again, here I refer to the webpage on the uTracer6.
• The choice of PIC processor, the 16F884
  This is also a point of much debate. Yes, I know, there are more modern processors, but the 16F884 is doing a great job. In the past the migration from the 16F874 to the 16F884 processor has given me a lot of headaches, because although MicroChip claims they are fully compatible, there were many small changes that had to be taken care of. For now, the 16F884 is still listed as “recommended for automotive,” so my guess is, it will be around for some time!

2. Plan A

Figure 2.1 shows a first plan for the analog circuit part of the uTracerNXT. It is by no means a definitive version, but rather a first draft to start working from. In the circuit many of the learnings gained from the uTracer6 but also the work on the uTracer7 have been implemented.

Going from the top to the bottom, the first two “rows” of the schematic as usual depict the anode and screen high-voltage supplies and switches. One of the changes in the uTracerNXT with respect to its two predecessors is that the cathode of the tube is referenced to ground rather than to the + of the supply voltage. The reason for this was that normally the output voltage of a boost converter cannot be lower than its supply voltage. However, by simply adding Zener diode D61 with a breakdown voltage higher than the supply voltage in series with high voltage blocking diode D60, allows the output voltage of the boost converters to go down to 0 V. The potential penalty for this trick is a little but lower efficiency, but we will have to see how serious that is.

Moving to the left, the “normal” diode in parallel to the current sense resistor R62 has been replaced by a Zener diode. The idea is that during the charging of C60 the Zener diode is conducting, just like a normal diode, and that during discharging – that is the measurement pulse – the Zener diode is in reverse bias and in that way protects OpAmp IC61 for high input voltages, e.g. during a short circuit at the output of the anode supply. The circuit around the current sense OpAmp and the PGA IC62 is pretty standard, with the exception of resistor R67, which sinks a small current from the output of the OpAmp to the negative power supply. The idea is that this will help the output of the OpAmp to really go down to its negative supply line, 0 V. The reason is that although most OpAmps these days are advertised as having rail-to-rail outputs, what really is meant is almost rail-to-rail. More in this the next section.

The high voltage switch is identical to the circuit used in the uTracer6. The circuit is extensively described in the uTracer6 weblog. At the writing of this page, there are more than 500 uTracer6 in service in 49 countries. So far, not a single failing high voltage switch has come to my attention. For me reason enough to copy the circuit one-to-one for the uTracerNXT. The only change is that the 1000 V STD2NK100Z NMOS transistors have been replaced by the readily available 700 V IPD70R360P7. As usual the screen circuit is identical to the anode circuit.

Moving down to the bottom part of the circuit, we find the heater, negative grid bias power supply and the +5 V power supply. Missing here, compared to the uTracer3 and 6, are the +15 V and -15 V power supplies, which saves a lot of overhead. The -105 V negative grid bias power supply uses a -400V FQD2P40 PMOS transistor instead of the BD138 pnp transistor in the uTracer3, which I expect to result in a more reliable operation.

Similarly, the grid bias supply also follows the design used in the uTracer6, be it with different components. A LM4040 “Zener” diode, provides a stable 2.5 V reference voltage to the DAC. As DAC the readily available 12 bit MCP4921 from MicroChip is used. The MCP4921 is programmed via a serial SPI bus, and by using the onboard 2x amplifier, the output voltage can be programmed to be between 0 and 5 V. An OPA454 High-Voltage OpAmp from TI is configured as a 20x inverting amplifier which boosts the output voltage from the DAC to a voltage between 0 and -100 V. Just like in the uTracer6 I intend to use the grid bias in pulsed mode by means of T40 to limit the power dissipation in the OpAmp, especially for conditions where the output is for some reason short circuited. The reason I plan to use an OPA454 is because it seems to have a better availability than the LTC6090 from Analog Devices used in the uTracer6. D41 and R45 protect the OpAmp against flashovers in the tube and short circuit conditions.

Figure 2.1 First tentative draft of the (analog part) of the uTracerNXT circuit. For this circuit diagram I used the schematic of the uTracer6 as starting point, so component numbering may be inconsistent.