Takeda Riken 3788B frequency counter

This frequency counter with nixie tubes uses TTL and RTL IC's, while for the high-speed input stages, discrete circuitry has been used. The display consists of 8 nixie tubes, type CD61 by Hitachi. They have nice large (16 mm) digits. The counter has 4 frequency ranges, using port times of 4, 40, 400 ms and 4 s. There is a potmeter labeled "sample rate" which seems to control the interval between measurement cycles. Although the unit indicator in the display has "μ" and "s" symbols, there don't seem to be ranges for time or period measurements.

When I bought this counter on an Internet auction I was told that it worked perfectly, and that only one knob was missing. Too good to be true. The missing knob turned out to be the entire coaxial input attenuator for the high speed (500 MHz) input. Well, I can do without, if necessary. What was worse, one of the digits was dark and the counter readout was jittery and seemed to be off considerably.

I started to look at the faulty digit first. The display counter consists of 8 identical digit cards, with one nixie tube each and associated counter and driver circuitry.

The photograph on the right shows each card contains three IC's: a Mitsubishi 53290, a TI SN7475 quad latch and a Mitsubishi 53241. The Mitsubishi TTL circuits surprised me, never seen Mitsubishi IC's before. Looking at the circuit more closely revealed that the 53290 is equivalent to the 7490 decimal counter. The 53241 is the decimal decoder and nixie driver. The number "241" suggests it is a successor of the 74141. This must be a 7441 equivalent. The circuit is straightforward so it was easy to discover the anode resistor was interrupted. Indeed, the PCB material round the resistor was slightly charred.

But that was not the only fault. After this, only a small part of the nixie's "0" digit lighted, and very brightly. Some of the numerals in the front lighted very weakly, causing a slight haze in the tube. Switching on the decimal point made the haze disappear. Fiddling with my frequency generator I was able to get a "1" or a "2" lit, but the "2" did not light evenly. Even after taking out the decoder chip, the fault remained This looked like a problem in the tube, and a strange one.

I did not have spare Hitachi CD61 nixies, but in december 2001 I received a few CD66 nixies from NEC, that have the same pinout. Replacing the old nixie by a CD66 went OK. In the process, I noticed how used the old nixie looked. Maybe it is just worn out?

Now it is time to find out why the readings are so bad. It looks like the input circuitry is faulty. Although the input level meter works, I don't get the impression that the counter reacts to the input signal as it should. It might be that the reference oscillator is faulty. The control circuitry may have a problem, although the "test" position of the range switch gives a reading of 1 MHz, indicating that measuring the reference frequency using the reference frequency always tells you that the reference frequency is exactly right.

Checking the 1 MHz reference oscillator proved it worked and was correct within a few Hz.

I will have to check the control circuits. These partly use RTL (resistor-transistor logic) IC's, partly discrete logic. One of the boards is shown on the left. The IC's used are Motorola MC890 and MC824. RTL is an old technology that is quite slow, but this is no problem since most of the control logic runs at clock speeds below 1 MHz. The precision of the gate times may have been a problem.

I started to check the signals on the counter units. In the "check" position of the range switch, everything looks ok. But in the other positions, the signal coming in is not ok.

I checked the output from the input amplifiers. The input signal is amplified and still has the same waveform. This seemed to be ok, but the signal is a bit weak. As I didn't have a manual, I did not know what voltage to expect.

I will have to look into this further. To be continued.