The meter would be checking at a low collector current where the leakage would be more significant. In the inverter, the collector current will be much greater and the leakage becomes insignificant.

This inverter is self-oscillating and if the load on it is excessive that can stop it oscillating. I would be checking what the inverter feeds to make sure it is not overloading the inverter.

If you read my posts you will see that I already did that with the 120V winding. It works good on 120V. I'm going to convert it to mosfets. One guy says it won't work but didn't say why.

 

I'm going to convert it to mosfets. One guy says it won't work but didn't say why.

It likely won't work because a MOSFET is a voltage operated device with a typical Vgs on-voltage of several volts, whereas a BJT is current operated with a Vbe on voltage of about 0.7V, so the bias requirements for the two devices are quite different.

 

It likely won't work because a MOSFET is a voltage operated device with a typical Vgs on-voltage of several volts, whereas a BJT is current operated with a Vbe on voltage of about 0.7V, so the bias requirements for the two devices are quite different.

That is true. You are probably right but all I have to loose is a little time. If it works, it should be a little more efficient because mosfets have less on resistance and also the heat of the diode junction is not there. If I have time today or tomorrow I'm going to try it. I like to experiment anyway.

 

Presumably they will need to be P channel?

 

Presumably they will need to be P channel?

Yes, 5305, but they aren't powerful enough. I will have to parallel them. Either that a or switch to N channel. I have lots of N channels with 53 amps which is almost enough.

 

Well, you guys were right. Those transistors were good, only now they're not. There was a broken ground wire but I accidentally hooked to the wrong place and blew both transistors. That will teach me to use a resistor in the power supply when testing! This thing is a little difficult to work on because everything is packed so tight and the wires are difficult to trace.
I found a inverter circuit online but I added some LEDs so I could see the oscillator working. You can see one Mosfet in the heatsink on the left. They have built in Zener diodes. I am surprised that the original circuit didn't have any diodes. I had to add a off and on switch, which is the little red thing in the heatsink. If I had used n channel mosfets, I could have tied into the heater switch. I might change them some day. So I'm glad that section is working again. The transmitter doesn't work and I'll be asking questions about it, but that will have to wait till I have time to work on it.

 

Only the second time and I got it right already!

 

There is a problem, at least with the drawing of the circuit. It wouldn't work properly as drawn. Because of the two inverters in series the MOSFETS will conduct at the same time rather than alternately.

 

There is a problem, at least with the drawing of the circuit. It wouldn't work properly as drawn. Because of the two inverters in series the MOSFETS will conduct at the same time rather than alternately.

You are right! How did I do that? It shouldn't even be working but it does. I have one too many triggers in there! It's a good thing I posted this. I probably would have never noticed it.

 

Ok, I eliminated trigger 4 and used trigger 3, but it barely oscillated and only 200V. Strange! I wonder why? So I used trigger 4 again and used the oscillator as a driver and it worked. With the same setting on the frequency, the voltage is up 80V!

You are right! How did I do that? It shouldn't even be working but it does. I have one too many triggers in there! It's a good thing I posted this. I probably would have never noticed it.

 

Well, I checked it again, and I still didn't get the wires in the right holes. So I got the wire in the oscillator this time and guess what? The voltage jumped up to 1500 volts and the lowest it would go was 1200 volts. So, I will have to add resistance or less capacitance. The original voltage on the output tubes was 800 volts. Before, the transistors were slightly warm; now I can't feel any heat at all. Thanks for the help!

 

You probably cannot get a good reading on the anodes of the tubes like you have the meter, and it is set to ACV.
Measure the DC voltage at the supply output that feeds the tank circuit, NOT on the anodes. And use DC volts range.
You stand a good chance of killing your meter as it is connected. As well the tuning will be upset and the meter lead will radiate.
Make sure you have the radio connected to a suitable high enough power dummy load too!

 

What is the frequency of your oscillator? From the2N4049 datasheet I suspect the transformer is intended to operate at several hundred Hz to a couple of kHz.

If, under load the MOSFETs get hot, you might want to use actual gate drive chips after your oscillator so you can turn the MOSFETs off and on crisply.

 

You probably cannot get a good reading on the anodes of the tubes like you have the meter, and it is set to ACV.
Measure the DC voltage at the supply output that feeds the tank circuit, NOT on the anodes. And use DC volts range.
You stand a good chance of killing your meter as it is connected. As well the tuning will be upset and the meter lead will radiate.
Make sure you have the radio connected to a suitable high enough power dummy load too!

Oops, I had the meter on AC, not DC! The best I can get is 730V. Voltage is the same at the power supply. I always park the meter at AC voltage, so I don't damage the meter if I forget to set it on the correct range when I start to use it. I've done that before. I put in a resetable fuse for that problem. I guess you didn't see that the transmitter is not working yet, but that problem is for another day.

 

I have never known a transistor failure to cause increased gain - reduced gain, yes.

[Edit] How did you measure the transistor gain?

TRansistor leakage current will give a confusing gain reading if it is not compensated for. So that may be the problem, depending on the testing method that was used.
And looking at the circuit in post #1 I see that the emitters are connected to something not shown, So now questions: Is that transformer with two windings shown the main power transformer, without the rest of the connections shown, or is it an oscillator transformer? What are the emitter voltages when trying to power the set on 12 volts? There appears to be quite a bit that we are not seeing.
Probably with a change in the bias arrangement it could be made to work very well with two suitable PNP silicon transistors. So the very first thing to examine closely is all of the connections, since they do sometimes fail. This will include the connections at the 12 position power connector. I have seen several failures in those connectors, especially in the "AMP" brand ones.

 

(Some text removed for clarity)
TRansistor leakage current will give a confusing gain reading if it is not compensated for. So that may be the problem, depending on the testing method that was used.

Thank you for mentioning that. I had not thought about that, but at low current levels this can be an important distortion of the current gain reading/calculation.

 

TRansistor leakage current will give a confusing gain reading if it is not compensated for. So that may be the problem, depending on the testing method that was used.
And looking at the circuit in post #1 I see that the emitters are connected to something not shown, So now questions: Is that transformer with two windings shown the main power transformer, without the rest of the connections shown, or is it an oscillator transformer? What are the emitter voltages when trying to power the set on 12 volts? There appears to be quite a bit that we are not seeing.
Probably with a change in the bias arrangement it could be made to work very well with two suitable PNP silicon transistors. So the very first thing to examine closely is all of the connections, since they do sometimes fail. This will include the connections at the 12 position power connector. I have seen several failures in those connectors, especially in the "AMP" brand ones.

I have already blown those expensive pnp transistors. I hate PNPs! I did finally get it to work with some old 513 transistors, but they would only put out a low voltage, and they are 20 amp transistors. You are right the problem was a bad ground wire but some how I managed to blow the transistors when testing, and $25 each, is just too much money. The emitters are connected directly to 12V. I posted a pic of the power plug connections. I have abandoned the feedback winding completely with the new oscillator circuit. It works well but for some reason I have 22V where I should have 14V. All the high voltages are there but about 10% low. When I change the power plug to operate the transformer at 120V all the voltages are correct including the 14V winding. Why is the voltage high when operating on the inverter? All voltage readings are on receive mode. I hope the high 22V voltage hasn't already damaged anything yet. Maybe I should slow the frequency of the oscillator till the 14V is correct?

 

This is a guess about the higher voltages when using the inverter. If there is ringing on the waveform of the oscillator the output will rise to that higher voltage. That was a problem with the inverter being discussed, it might be an issue in this one as well. Another possibility is that the transistors that you are using have a lower saturation drop, which would put more voltage into the transformer. And a third guess is that the duty cycle is different with the new circuit.

 

...and if the duty cycles is off the transformer core might "walk up the B/H curve and saturate, blowing a fuse if you are lucky or destroying the transistors if you are not lucky. In the early days of transistorized inverters like this one apears to be it was quickly realized that matching the transistors, especially the saturation voltages was important to minimizing that kind of failure.

 

Thanks for the replies Mr Bill and Dick Cappels2. One of you mentioned something about voltage. You can see in the picture that the voltage is low on the sources durring testing. When I was testing the new circuit I had some resistance in the power supply. I did that to try to not blow more transistors.Maybe that is causing the problem as there is only 8V on the sources? And I might have the frequency too high to compensate? So, when I get a chance I will test at a lower frequency at a full 13.8V.