What tube bases to support for a new tube curve tracer

[crj11]

I am in the final stages of designing a tube curve tracer for eventual commercial sale and I need to decide what tube bases to support. The actual connections to the pins are programmable, it is just the physical sockets that are in question.

It will definitely support the standard Octal (KA8) and 9 pin (B9A) bases, as well as the UX4 4 pin base. I was also planning on supporting the 7 pin (B7G) base. However, I plan to use CMC machined pin sockets and I found that they don't make a 7 pin socket. So, I'm wondering whether 7 pin tubes are that popular. Less standard sockets can always be added with a separate adapter, but I am trying to minimize the cost and size of the baseline unit.

So, are there a lot of designs still using 7 pin tubes?

Thanks,

Chris

[Wavebourn]

Yes.

[llwhtt]

How about a five pin for 807s? Plate cap too!

Craig

[kenpeter]

compactron? nuvistor?

Maybe you just need a few standard sockets and a header for future add-on adaptors?

[crj11]

Originally Posted by llwhtt

How about a five pin for 807s? Plate cap too!

Craig

I compiled a list from previous posts of "favorite tubes" and I did notice that there were several 5 pin ones, so that is a possibility. The plate cap is not a problem, since the connections between the test electronics and the sockets are configurable with high voltage sheathed 4mm banana plugs. All you need to do is have an extra cable that terminates in a cap.

[crj11]

I am definitely considering an adapter for other sockets. Worst case you could just make something that plugged into the 8 pins socket.

[Arnulf]

Originally Posted by crj11

I am in the final stages of designing a tube curve tracer for eventual commercial sale and I need to decide what tube bases to support. The actual connections to the pins are programmable, it is just the physical sockets that are in question.

I too am designing a computer-controlled curve tracer (smoking amp's posts in that other thread were a huge inspiration) and intend to make it avaliable as a DIY project with parts not costing more than $50-100 (total, PCB, MCU and transformer included in parts cost estimate). I plan to use computer application for data display (with opto-isolated communication, of course).

Instead of complicationg my life with various socket types I decided to use screw terminals instead so one can wire any sockets in any fashion, be it through rotary switches, relay multiplexers or simply by using alligator clips or some similar method.

Depending on your target customer base (and sale price) you should perhaps consider using fewer socket types by default but keeping your options open instead of blowing costs through the roof by supporting every avaliable socket type out there ?

[porkchop61]

Please let me know when it's available! I was going to build one myself, but just haven't got around to it.
I'm interested

Glenn

[crj11]

Originally Posted by Arnulf

Depending on your target customer base (and sale price) you should perhaps consider using fewer socket types by default but keeping your options open instead of blowing costs through the roof by supporting every avaliable socket type out there ?

Our target price is $599, so the customer base is not as large as that for a $100 kit. But hopefully it will have enough features that people will think it's a good value. A lot of the cost is actually in the physical package and high voltage connectors/cables.

The tentative specs are:

500V@200ma Plate, 500V@200ma Screen, +/-100V@100ma Grid, 2.5-30V@15W Filament, with support for directly heated cathodes. All of the supplies are progammable and voltages are measured with 12-bit A/D converters. It uses high voltage sheathed banana jacks to configure the connections between the supplies and the sockets. As in your case, the interface to the computer will be optically isolated. I'd hate to blow up my laptop when I'm testing it.

We're still in the process of designing the PCB, so it will be a while before the prototype is ready for testing. However, at some point this spring we will be looking for beta testers, if anyone is interested. Also, we'd love to hear any feature requests.

Chris

[smoking-amp]

Those price ranges sure sound a bit optimistic.

For the el-cheapest approach, I would just make a step generator circuit board that one could use with a scope for display. (4 bit counter IC, D/A, HV amp stage and gain control, with a 60 Hz zero crossing comparator/trigger to clock the 4 bit counter) A separate screen supply from the B+ supply allows one to measure "plate" current at the cathode with a small grounded resistor. B+ supplied by rectified, but un-filtered, HV xfmr. Voltage divider off plate for scope input. A collection of plate resistors to jumper in for loads (or see later, a programmable or switch controlled load resistor circuit board). A camera photographs XY displays on scope for capture.

Then for fancy (more expensive) designs with opto-isolated PC control, some nice features (beyond the expected data capture, display and programmable voltages) would be a tube power limit parameter to constrain tube dissipation during scans. A programmable load resistor with near continuous range would save $ (over a bank of power resistors) and be much better utility wise: EDN PDF (bottom article)

The Tek 576 tracer I have is very limited by its small selection of load resistors and voltage ranges. It also suffers from non-contiuously variable display scale factors and step voltages, which I am fixing now, but this makes calibration/readout more difficult. All this is easily done in a PC controlled setup. Tek also included some operator safety features, which generally get over-ridden due to their inconvenience. But a ground fault tripper unit might be a good feature (can be added as an external unit easily anyway).

The usual plate curve display is de-rigueur, but with computational power available, some other displays/readouts would be very useful. Mu curves, gm and Rp plots the most obvious ones (with selection of either grid voltage or plate current or plate voltage as the horizontal axis or the basis for curve steps).

Then a gain plot with a specified load resistor and B+ and bias. This would sweep grid voltage thru a specified signal range (voltage ramp) while displaying gain at each point. Then a set of these curves could be done with discrete steps of load resistance (and a re-bias to some plate current or tube diss. for each). This would indicate the best operating point for a tube quickly (flattest gain curve in the set).

Then tube tester functionality with a gm test at specified current and voltage.

Then gm tests at a group of op. points with some result storage features for tube matching (maybe a super-imposed graph for each tube, for the set of tubes) to make a selection constellation display, where the user can easily pick out the closest tubes.

------------------------------------------------------------------------------------------

The gain plot scheme I mentioned above would also make a nice el-cheapo stand-alone circuit board unit for use with a scope. A small 1 KHz sine wave signal would be summed with a slowish ramp signal for the grid (adjustable ramp limits). Then a frequency (1 KHz) selective (filtered then rectified) voltage readout of the gain from the cathode current sampling resistor. The ramp signal controls the horizontal on the scope, the measured gain signal controls the vertical on the scope.
User adjusts grid bias or plate resistance to find the flattest curve.

A set of such curves could be displayed using either the grid voltage stepper board also or load resistor stepper (below). These would get a count signal from the ramp generator to count up a step each time.

Optionally, a stepped, programmable load resistor could be provided also (maybe a separate PC board for this, so it could be used with the earlier grid voltage stepper board too) which would step thru a range of load resistances with each slow ramp scan (4 bit counter to control the D/A, or a presettable 16 bit counter with modulo 16 count reset). Some adjustable DC grid bias output (second D/A off stepper counter), derived from the Rload stepper, would be needed to adjust the DC grid bias as the load resistor changes.