The transformer core is a nice, solid toroid core held in a plastic mount. I can't imagine that it cracked

Do you know the complete history of this particular tacho and can guarantee that at no time has it had a knock? A hairline crack in a ferrite core might open up at high temperature, making the oscillation flakey.

 

Do you know the complete history of this particular tacho and can guarantee that at no time has it had a knock? A hairline crack in a ferrite core might open up at high temperature, making the oscillation flakey.

The photo in post #9 gives enough of s view of the transformer to show that it is in a plastic shell to hold it, and that it is in good shape. But I also see that the 2,2 Mfd capacitor is marked "NP" for non-polarized. Does that matter? Was the original cap also "NP"?? And once again I am thinking that unsoldering one leg of the thermistor to take it out of the circuit should not be a big effort.

 

I will try that. Thanks.

 

One more thought is that the gain of a germanium transistor tends to increase with the temperature, some types more than others. So a look at the data sheets for the new part and the old part is in order, perhaps.

 

I do have a pretty good picture of this history of the tachometer. It came to me years ago after having failed in the car. I determined that the transistor had failed and replaced it with the NTE unit to get it working again. I also replaced the original polarized electrolytic cap out of an abundance of caution. However, that cap seems to be a filter cap. The tachometers will work fine without it under bench condition. I don't like the original design, because the signal does go a little bit negative which seemed to me to be not great design practice, so I replace them with NP units. I have repaired a lot of these tachometers for people and have not had any other ones come back to date.

The transformer core and windings are well insulated against vibration and seem to be in very good shape.

I disconnected the thermistor from the circuit and it still oscillates at higher temperatures, so it seems that it is not the thermistor that is causing the problem.

I have another customer who is asking me what the correct value for a replacement thermistor should be. I have no clue how to calculate what the correct value is. Can anyone teach me how to do that?

I was also wondering if reducing the current through the feedback winding might reduce the gain of the circuit a bit. Do you think increasing the value of the 39 ohm resistor would accomplish that?

 

I have been unable to find a datasheet for the MT59/A original germanium transistor.

 

REducing the gain a bit may be worth considering. IF the tach scale is very linear then an entirely different circuit might be substituted, BUT that is a very big deal change and it might offend some folks horribly. Worse yet, there might not be room in the housing for it.
An interesting experiment would be to lower the supply voltage a bit, maybe to ten volts, and see what that could do for you. I suggest that because ten volt regulators are available.

Another option would be to reduce the gain slightly by shunting the base to emitter with a higher value resistor, try 1 megohm to see if that has an effect. Or maybe a 470K resistor in that location.

But watch the problem wind up being something insane, like mechanical damping of the moving coil portion..
Actually, a capacitor across the meter could have an effect.

 

REducing the gain by shunting either winding with a resistor would work, altering the collector load current will change the calibration.
The challenge at this point is finding which of the original variables has been changed too much.
Can you look at the data sheet for the replacement transistor and see what varies the most with temperature? I think that leakage currents increase.

 

The data sheet is very sparse. There is no temperature related data:



I will try shunting the windings with a resistor to see if that has any effect.

I had a similar oscillation problem with a Gen 3 Smiths tachometer (based on a custom TI IC, similar to an SAK215) that I was never able to solve before the owner wanted his tachometer back. I put a cap in to filter out most of the oscillation, but the tachometer accuracy vs temperature suffered. The owner was ok with that, but I wasn't because I could never find the root cause of the oscillation.

Between surgery recovery and now the oppressive heat with no AC, it is really challenging to get into the lab to work on this problem child...

 

Hello,

I see it handles about a Smiths Tach:
https://accutach.com/smiths-tachs

Bertus

 

Bertus, that's my web site.

 

I was finally able to get back into the lab to work on this tachometer. BTW, don't get a total knee replacement unless you really need it. The recovery is really rough!

I tried shunting the bottom winding with a 500 ohm pot to try to reduce the gain in the feedback loop. I used a hairdryer to force the oscillation and then I reduced the resistance to see if the oscillation would stop. I dropped the resistance down to about 6.5 Ohms. This had no effect on the oscillation. Below 6.5 Ohms, the one-shot failed to work.

Would shunting the winding above that have a different effect than shunting the bottom winding does? I sure wish I understood oscillators better.

I also tried a number of different cap values between the collector and ground. A 1uF cap raised the temperature at which the oscillation started up quite a bit, but it did not eliminate the oscillation. Taking it up to 2uF eliminated the oscillation, but affected the waveform to the point where I could not calibrate the tachometer any more.

I can probably get away with adding the cap as a band aid for the customer, but I would really like to get to the bottom of this problem if I can.

I appreciate all of the suggestions and education so far. Any more would be equally appreciated.

 

Shunting the base to emitter with a resistor will divert some of the base current , effectively reducing the gain a bit. The problem is knowing where to start. If you had a one megohm pot you could use the slider to one end and then see if that had any effect. Heat it just enough to start oscillation and then turn the pot to see if that stops it. How hot a temperature does the tach ned to be and sill be accurate?accurate

 

That's a good idea. I am pretty sure I have a 1M om pot. The car is a Volvo P1800. I am guessing that about 110 degrees is about the limit. The dash is under the windshield, so it could get pretty hot in the summer. But if it is 110 under the dash, the driver is going to be pretty uncomfortable. The owner was seeing it oscillate in the car while he was driving it. I figure heating it with a hair dryer shouldn't get too much hotter than a hot car in the summer. I will report the results of the shunting test. Thanks for the suggestion.

 

I found a 200K ohm pot and shunted the transistor with that. I heated the circuit up until it started oscillating and then dialed the pot down until it stopped oscillating. I heated the circuit up to about 110 degrees F and it oscillated again. I dialed the pot down until it stopped oscillating. At this point, the pot is at 93 ohms. Dialing the pot down much further causes the one-shot to stop working.

FYI, the car owner had two identical tachometers and sent me both. I had previously replaced the transistor and cap in both units. The tachometer I have been working on is the owner's problematic one. I was unable to get the other tachometer to oscillate or fail in any way by heating it up as much as possible with the hairdryer.

I am tempted to try replacing the transistor to see if it could somehow be a problem with this particular transistor, since reducing the gain caused the oscillation temperature to go up.

 

How much did that 93ohms shunting affect the calibration?? I am thinking that it would have an effect.

If you have access to a transistor tester then you could compare gain and leakage between the two transistors.

 

Interestingly enough, changing the shunt pot value had no effect on the one-shot pulse width until the one-shot simply stopped working. I don't have a transistor tester at the moment, but I will probably order the transistor tester accessory for my Analog Discovery 2. It's only $50. I am researching how to use it at the moment. Edit: I just ordered the transistor tester adapter. A new toy to play with...

 

OK, now the question is if the resistor raised the temperature of the malfunction enough. If that was a fix then very good. BUT at least it provided a clue, finally.

 

I finally replaced the NTE176 with a different NTE 176 and the one-shot oscillation temperature went up to above 125 Degrees F with the new transistor. I am pretty sure that the owner will be happy with that because he will be pretty uncomfortable driving his car if the interior temperature gets to that level.

A buddy of mine found a mid 1950s masters thesis paper on thermal stabilization of transistorized DC amplifiers. It discusses the fact that germanium transistors characteristics vary a lot more with temperature than silicon transistors do. One suggestion for compensating is to use a thermistor. I had assumed that the thermistor in the circuit was there to compensate for the copper windings in the ammeter. But after reading this paper, it is possible that it also compensates for the transistor too. I have attached the paper in case anyone is a glutton for punishment like me.

My transistor tester accessory for my Analog Discovery 2 has arrived, but I haven't had time to play with it yet. After I get a chance to learn (on a 2N2222), I will try the NTE176 that was in the tachometer and an MT59A that I harvested from a parts tachometer. I may also try a few other units of NTE176 just to see how much they vary from unit to unit.

I really appreciate the help.

 

I do recall the lectures and discussions about the temperature effects on germanium transistors back in 1969, in the "Transistor Circuits 1" class. And that thesis paper is certainly applicable.
What changed between then and now is manufacturing technology, so that the products now are much more consistent in characteristics, and usually exhibit more gain. But germanium is still temperature sensitive.

Finding a simple scheme to replace that transistor with a silicon device would allow a claim to fame, no doubt, but those appreciating it would be few. But certainly you would be a hero to some crowd.