BJT base too negative?

actually it's just a bit more than -10V, my bad. But the original question still applies.

Thanks

Zenering the BE junction causes progressive degradation in beta. Some types, e.g. the 2SK2704K, tolerate much larger negative V_BE.

Cheers

Phil Hobbs

I don't find that specification on a TIP31B datasheet (ON Semiconductor). To what parameter do you refer?

yes I know that, I asked the question because a) it doesn't seem to be really zenering in this case (how could it go past -10V?) and b) this was used on probably several thousands of this CRT monitor for years and they never issued a "fix" on this. I've used reverse B-E conduction out of BJTs to make nice wideband noise generators in the past. However I think I could add a small series emitter resistor to see if it's going into conduction and if it is indeed zenering, I can add a series diode to the base to stop it.

Thanks Frank

It's V.EBO (emitter-base voltage with collector open). On most transistors it can't be more than 5 or 6V.

Frank

Is that forward or reverse polarity?

Are yopu measuring V_BE or are you measuiring from ground? L900 is going to help by pulling the emitter below ground at the transistor is trying to rurn off.

One-transistor oscillators are notorious for reverse-bias on the base, and (in the era when ICs were expensive) were supported by special transistors that tolerated it well. The CMPT404A tolerates 25V reverse

the original design however uses a TIP31B, nothing special. I have used a different NPN transistor in my reproduction but it has similar V.EBO ratings. This was (and there're quite a few still working nowadays) an arcade monitor, so it saw a lot of hours of use. I don't know, I may make more tests to see if the B-E junction is really going into reverse conduction, but I suspect that would happen around 6V more or less. Thanks Frank

usp=sharing

tter

The trick is basic phasor circuit impedance voltage divider principle. Usin g the 27kHz cited, the base capacitor C906 has a reactance of 87 ohm, the series C907 ground return has a reactance of 60 ohm, emitter inductor (in place to limit current slew rate) has negligible reactance of 0.12 ohm ind uctive, R908 at 4.7R is negligible too compared to square law addition to c apacitive reactance. So the series reactances of C906 and C907 form a volt age divider in the ratio of 87/(87+60)=0.6, resulting in the 10V total ba se voltage dropping to 6V from EB (C906) on Q903. The BE junction won't bre akdown at that voltage. Component values were most likely chosen from avail ability, low cost and "good 'nuf' considerations more than engineering exac titude.

This of course makes sense, I was trying to just explain what I'm seeing on the scope. Keep in mind the scope ground is connected to the oscillator ground and since the emitter is never more than +/-1V from ground (with L900) then I can't make sense of how EB isn't "overdriven". By the way, here're the waveforms with L900 completely removed (so emitter is now at solid ground), always 20V/div on the top trace (collector) and

For completeness, an older version of the service manual also shows the emitter waveform of Q903 and also that corresponds to what I see in my reproduction circuit. I'll be trying adding a series 1N4148 to the base probably later today. Thanks Frank

Ok, that was a quick test to made. Adding a 1N4148 to the base just showed that the total reverse swing of the secondary winding was more or less -20V, so the capacitive divider was limiting this to about -12V as shown on the scope trace. This makes sense since I am using a different turn ratio from the original design, since I only have 12V DC supply, not 25-40V as in the original monitor. The oscillator seems still working fine with the added base diode, at increased drive however since there's no current flowing on that secondary when it swings negative now, but just the recovery spike of the 1N4148. I still can't really tell if the BE was going in reverse conduction, the scope traces don't seem to show a clear "clipping" on the negative peaks. As I said, the scope traces on the original service manual agree with what I have, with the BE going negative to about -10V or slightly more. I wonder if there's a physical reason for the base-emitter junction to tolerate higher reverse voltage when the collector is swinging to a positive voltage. Maybe I should be dusting off the university physics books :) Whatever, if it works, I'll leave it alone.

Frank

You might have mentioned that at the start, and posted the 'new' turns ratio.

I was under the impression that the waveforms were for the original circuit, not for a revision.

Speculations on original design intent are pointless under the conditions presented.

RL

ng on

h L900)

tter

nd

?usp=sharing

ed

It's a rapid charge control turn-on of a couple of uC thru C907. C907 char ges until base is negative enough to initiate similarly rapid turn-off. The main off timing is determined by the reflected magnetizing inductance and current in the base drive winding in combination with the series capacitors C906 and C907 discharging C907 to the point the bias current from R907/R10 initiates another urn-on. If he has a diode blocking the discharge current , it's hard to see how it still works.

funny, it wasn't easy to read all I wrote earlier, I guess. Here're the waveform of the original circuits as shown in the service manual:

forgive them for the V/cm, those were the good old times, you know. Anyway, as you may notice, the collector swings to 40V peak, while on my "new turn ratio", the collector swings to half that. Instead, the base waveform is basically identical, even if I notice now that the waveform of the service manual might be showing the clipping at the bottom swing.

I'm not speculating on the original circuit. Mine works fine, I might even not add a component to prevent BE inverse conduction and live dangerously like the original circuit did. My questions were "simple" enough: is there any condition that prevents BE inverse conduction at V > V.EBO? (If not, how the original design got away with small "zenering" every cycle for years?). The second question was if anyone can see a reason why they used a high voltage BJT for Q900.

Thanks Frank

here you are:

base trace is 2V/div, collector always 20V/div In this case the added diode is between the winding terminal and the junction of base and C906. A better place would be between the winding terminal and C906 junction and the base, in this way, C906 can work as before.

Frank

Those waveforms are no longer phase coherent, for a self-oscillator.

If the service manual gives waveforms similar to those in the revised circuit, then you've probably got something that will work as well as the original.

I notice that you are not monitoring the power switch current. It's combination of current and voltage that will predict switch trouble.

Avalanched EB switching in high voltage bipolar devices, with current-limited drive sources, is/was a normal methode of speeding up turn-off and evacuating stored charge. Normally the reversing dI/dT would be controlled through a small choke. I haven't seen it used at lower voltages (

that's what I noticed too, but it still worked somehow, I won't run the circuit like that anyway, I'll probably stick with the original schematic. By the way, in a previous revision of that circuit (the original monitor's one) there was no capacitor from base to emitter and the reported base waveform on the manual was swinging to -15V, with no evident sign of clipping. I haven't tried to reproduce that anyway, I think they had a reason to revise the circuit.

that's what I think too.

I have to add the DC regulator part still. However I'm going to use a different circuit for that (different from the original one I mean).

I'm actually using a BU406 at the moment, which is surely overkill. I was tempted to use a TIP31B too, since I'm running on a much lower voltage anyway. Do you suggest any particular transistor instead (or any particular characteristic it might be needed in such a circuit?). To me 25 KHz is almost DC anyway :) Required power on this circuit is very low anyway, much less than 10W, but I'll be measuring it once all is connected.

just reproducing part of that circuit to make my own CRT monitor. Since it seemed odd they got away with such a large inverse B-E voltage, I thought I'd ask for hints. I don't think they had many issues with that, given the high number of those monitors that were in use in the early '80s. I've only found two service manual revisions, and the only part of that circuit that changed was the added capacitor from B to E (and they reduced the value of the cap going to ground on the base winding).

Thanks a lot Frank

• Who sez the base the base goes negative like that? The emitter isn't "basically" grounded (only DC no power; that inductor reacts to the transformer drive. Look at the differential E-B voltage over time..

• And all are wrong, exceptions are maybe one out of thousands. Actual value is about 8-9V.

• Maybe that particular transistor (MJE51T) had a lower cost than any other adequate for that.